CORD-19:ee0a7632371a08727d41e55aad7d621daeceef7d JSONTXT

GENETICS G01. WITHDRAWN G02. ClinVar: A Centralized Database for Interpretations of Both Germline and Somatic Variants Abstract solid organ post-transplant surveillance. We have validated a method based on targeted amplification and next-generation sequencing (NGS) that does not require genotyping of the donor or recipient and is applicable to all single solid organ transplants except those from an identical twin donor (AlloSure). Selection criteria for the 266 targeted SNPs included high minor allele frequency and low fixation index. Quantification of dd-cfDNA for this method is based on the degree of genetic similarity between the donor and recipient, the extent of which differs based on their familial relationship. Other factors, such as HLA matching, may also inform on genetic similarity. Here we evaluated the AlloSure SNP allele distribution in kidney transplant patients with no mismatches at the HLA A, B, and DR loci as determined by low resolution typing. Methods: Kidney transplant recipients with unrelated donors were selected for 0, 5, or 6 mismatches at the HLA A, B, and DR loci. cfDNA purified from plasma was amplified and sequenced to determine the percent dd-cfDNA and also the relative contribution of each of two alleles for each SNP. The number of SNPs determined to be homozygous for the same allele in both donor and recipient was calculated by determining the number of SNPs with only one allele present in the plasma in a set of high %dd-cfDNA samples. The results were compared to expected proportions in unrelated donors. Results: Of SNPs that are homozygous in the recipients, the average number of SNPs expected to be homozygous for the same allele in the donor is 25% based on the SNP selection criteria. A set of 97 samples was identified through which the number of 'homozygous-same' SNPs was not impacted by low numbers of donor molecules in the plasma (above 0.65% dd-cfDNA). For unrelated donor-recipient pairs with 5 or 6 HLA mismatches, the mean percentage of homo-same SNPs is 31%, with a 90% contained within a range of 20% to 43% (n=94 samples, 61 recipients). In the 3 unrelated, 0 mismatch donor-recipient pairs, the homozygous recipient SNPs were also homozygous in the donor at 24%, 25%, and 26%. Conclusions: SNPs used in the AlloSure assay were selected based on data indicating near random population distribution. To optimize successful transplant outcomes, kidney transplants are often matched at the HLA A, B, and DR loci. Only low resolution typing is available due to limited ischemia time required. Patients lacking low resolution mismatches (0 of 6 potential mismatches) had the expected number of donor homozygous alleles matching an unrelated recipient. Therefore, the AlloSure assay is not expected to be impacted by the level of low resolution HLA match. An even smaller number of these drug labels include recommendations for PGx testing, that can be required or recommended prior to drug prescription; or associated therapeutic recommendations. NIH Medical Genetics Summaries (MGS) provide free, comprehensive reviews on the currently available information about genetic impact on drug responses for use in clinical settings. Methods: MGS is available on the NCBI Bookshelf located at www.ncbi.nlm.nih.gov/books/NBK61999. The articles are prepared in a standardized format and include a formal review process. MGS articles combine actionable, evidence-based, and often nonoverlapping information from authoritative sources, including professional societies such as the Clinical Pharmacogenetics Implementation Consortium (CPIC) and FDA drug labels. In 2016, there are 23 MGS that review genetic variants and drug responses. Additionally, NCBI features and links MGS content in related resources, such as the NIH Genetic Testing Registry (GTR; www.ncbi.nlm.nih.gov/gtr) and MedGen (www.ncbi.nlm.nih.gov/medgen). GTR currently lists over 35,000 clinical and research genetic tests, of which over 300 are PGx tests for approximately 161 drug responses. Results: MGS contain recommendations on genetic testing, sometimes not included in drug labels. For example, the warfarin drug label recommends initial anti-coagulation dose ranges in a dosing table based on CYP2C9 and VKORC1 genotypes. However, MGS stress the importance of using a specific online PGx algorithm that takes into account both genetic and non-genetic factors to give a more accurate prediction of dosing. MGS exposes discrepancies in available PGx information. For example, tamoxifen is used in the treatment and prevention of breast cancer. At this time, the FDA drug label does not discuss genetic testing for CYP2D6. The National Comprehensive Cancer Network (NCCN) does not recommend CYP2D6 testing nor do the 2010 guidelines from the American Society of Clinical Oncology (ASCO). In contrast, the Dutch Pharmacogenetics Working Group (DPWG) makes recommendations for tamoxifen therapy based on CYP2D6 genotypes. MGS incorporates their recommendation to consider using aromatase inhibitors for postmenopausal women (for both poor/intermediate metabolizers), and that intermediate metabolizers avoid the concomitant use of CYP2D6 inhibitors. Conclusions: MGS are filling a critical knowledge gap in PGx testing and associated therapeutic recommendations. As PGx testing expands, MGS will also continue to grow and provide the evidence-based information healthcare providers need in this ever-evolving field of medicine. impact of individual mutations on MMR gene function. Methods: We evaluated somatic and germline mutations in 87 tumors with IHC staining patterns suggestive of Lynch syndrome tested using the clinical ColoSeq-Tumor assay. Paired tumornormal samples were used to identify somatic variants and germline variants, and determine normalized variant read allele ratios (VARs) across the tumor. VARs were used to differentiate between putative driver mutations, those likely to be responsible for IHC patterns, and putative passenger mutations in MMR genes. We evaluated distributions of VARs normalized to tumor percentage to determine Bayes factors for pathogenicity of variants identified in tumor. We used intronic mutations in MMR genes to define passenger mutation characteristics and compared these to known pathogenic mutations. Results: Comparing normalized VARs, 35% of normalized tumor reads was an optimal allele fraction cutoff for driver mutations. We found the Bayes factor for pathogenicity for nonsynonymous variants seen above this cutoff was 4.5. Somatic tumor mutations are independent events, so Bayes factors may be combined from events seen in multiple tumors. Conclusions: Evaluation of the mutational spectrum of MMR genes in tumor samples with loss of MMR protein expression by IHC from patients with negative germline results has been shown to alleviate concerns of Lynch syndrome in many patients. Interrogation of the observed mutations in MMR genes in this patient population may aid in the identification and characterization of mutations that lead to absence if MMR expression by IHC and may reduce the need for high throughput functional assays. Z. Fan, C.A. Stolle, J.R. Murrell, A. Santani, M. Luo The Children's Hospital of Philadelphia, Philadelphia, PA. Introduction: Germline inactivating variants in the VHL gene cause dominantly inherited von Hippel-Lindau syndrome characterized by hemangioblastomas, renal cell carcinoma, pancreatic cysts, pheochromocytomas and paragangliomas. Approximately 72% of the pathogenic variants are sequence variants and 28% large deletions. Nearly all individuals with a VHL mutation will develop symptoms by age 65. Due to age dependent penetrance, variable expression, incomplete family segregation information and somatic occurrence, VHL variant classification can be challenging. The purpose of this study is to use newly established guidelines to reclassify variants detected by our clinical laboratory over 14 years. This extensive dataset is used to update the mutation spectrum and aid future VHL variant assessment. Methods: A retrospective review of 3,205 cases submitted to our laboratory for VHL gene analysis since 2002 was conducted. These cases were previously tested by Sanger sequencing for sequence variants and Southern blot analysis, relative quantitative PCR or multiplex ligation-dependent probe amplification for large deletions/duplications. Variants were reassessed using our laboratory's criteria modified from the American College of Medical Genetics and Genomics guideline. Resources used for reclassifications included published literature, databases, and family segregation information focusing on the collection of new evidence that emerged since last assessment. Results: Out of 3,205 cases, 861 had a test finding with 21.6% having partial or whole VHL gene deletions and 78.4% with sequence variants. Reclassification of 186 unique sequence variants yielded 62.3% pathogenic, 21% likely pathogenic, 13% variant of uncertain significance (VOUS), 3.2% likely benign and 0.5% benign calls. Of these, 42 novel variants were identified in our patients that have not been published or reported in public databases and an additional 14 germline variants were previously reported in COSMIC only. Classification of novel variants yielded 55.4% pathogenic, 8.9% likely pathogenic, 26.8% VOUS and 8.9% likely benign calls. Among a total of 155 pathogenic and likely pathogenic variants, 47.1% were missense, 31.6% frameshift, 13.5% nonsense, 5.2% splicing, and 2.6% in-frame deletions/insertions. Conclusions: This study utilized our laboratory's database to reclassify VHL variants using established guidelines and updated information. This knowledge database can be used for future classification of VHL variants, which will be of benefit to clinicians who care for these families. Therefore this information was submitted to ClinVar for public dissemination. The experience gained from this study can also help variant classification in other cancer-predisposition genes. T. Spenlinhauer 1 , C. Zimmerman 2 , J. Stone 2 , R. Winegar 2 , S. Nesbitt 1 , S. Kaul 1 , J. Keierleber 1 , T. Laughlin 3 , J. Starbuck 4 , C. Rundell 1 , J. Gordon 1 1 Maine Molecular Quality Controls Inc, Saco, ME; 2 MRIGlobal, Palm Bay, FL 3 University of Rochester Medical Center, Rochester NY; 4 Cleveland Clinic, Cleveland OH. Introduction: Next-generation sequencing (NGS), provides an opportunity for clinicians to identify multiple mutations in a highly efficient and high-throughput manner. However, this technology requires a complex multistep procedure which presents many challenges for clinical laboratories. Each application and platform is unique and involves challenging, time-consuming, and expensive optimization and validation of 3 complex components -the sequencing platform, the specific assay/test panel and the bioinformatics analysis. Here we demonstrate a practical solution for monitoring the identification of many mutations using a multiplex, synthetic control panel created to monitor the analytical performance of molecular cystic fibrosis testing. The multiple insertions, deletions, and homopolymers of varying lengths and composition in the control make it potentially useful for monitoring the ability of the NGS systems to correctly identify variants found in genes other than CFTR. Methods: Synthetic DNA composed of all 27 CFTR gene exons plus intronic borders containing CF associated variants were designed in silco, ligated into MMQCI vectors and transformed to create stable frozen clone stocks. Multiple plasmids were created with various mutations to represent 186 CFTR variants and mixed to create either heterozygous or homozygous alleles for each variant and diluted to have equivalent copy numbers of the targeted gene as extracted human samples. The plasmid mixes were suspended in buffers and stabilizers, with and without proprietary matrix and tested either as an extractable (with matrix) or non-extractable (without matrix) control panel. Initial studies were performed to determine optimal concentrations for subsequent testing. The extractable samples were extracted by various methods (QiaAmp DNA Blood Mini kit, QiASymphony, and SPRI-TE) and processed the same as a patient sample. The non-extractable panel was tested using 10ul of each sample added to 5uL of Oligo pool and 35uL, of HYB buffer. All samples were tested with the Illumina's MiSeqDX CF 139-Variant Assay. Results: All samples resulted in a valid run with a call rate of 99.26%. The extractable controls, resulted in 99.6% concordance with expected calls across multiple testing sites and various extraction methods. The non-extractable controls resulted in 100% concordance with expected calls. When compared to human gDNA and Coriell samples, relative coverage of the target regions appear similar and generated roughly 1,000X to 1,700X coverage. Conclusion: A highly multiplexed, synthetic and well-characterized molecular CFTR QC reference material can be used as a reliable control to monitor highly complex NGS panels for verification, validation or as a routine control. jmd.amjpathol.org ■ The Journal of Molecular Diagnostics quality nucleic acid with optimal yield, with few steps for molecular applications is an asset and COPAN kit is for nucleic acid preservation for genetic testing. COPAN Kit is CE marked and currently available in the US for research use only. The objective of this study was to validate the performance of the COPAN Kit, an hDNA-free FLOQSwabs and 1 ml tube of eNAT medium, for the collection and storage of buccal specimens for human DNA preservation. Methods: in this study buccal swab samples were collected in duplicate, one using the COPAN Kit and another with Epicentre Catch-All-Sample-Collection-Swab. The entire Epicenter swab and only a portion of 400ul eNAT medium were used to extract genomic DNA using the Thermo Fisher MagMAX DNA Multi-Sample Ultra Kit and the Roche MagNA Pure; DNA yield from both devices was recorded using Thermo Fisher Qubit dsDNA HS Assay. Genomic-DNA from both systems was tested on the current comprehensive panel at Patients Choice Laboratories with Applied Biosystems 7900HT-Fast-Real-Time-PCR System for pharmacogenetics, coagulation factors mutations and genotyping. Data were analyzed with TaqMan Genotyper Software and results for both systems were compared. Results: Optimal quality and concentration of genomic-DNA was obtained from 400ul of eNAT compared to the Epicenter entire swab with the Qubit dsDNA HS Assay. Results generated with eNAT-extracted genomic-DNA showed good amplification and confident calls, all 47 assays of 17 genes showed 100% agreement. Conclusions: COPAN hDNA-free FLOQSwabs and 1ml eNAT medium kit is a convenient device for buccal sample collection, storage, preservation and transportation of nucleic acids. Direct lysis in the eNAT medium inactivates bacterial proliferation, reduces sample processing steps and facilitates automation. The genomic-DNA yielded from a 400ul portion of eNAT medium is pure and adequate for genetic testing. Moreover, the leftover eNAT medium can be stored for additionalrepeat confirmatory testing, avoiding sample recollection. Introduction: For the past three years our laboratory has participated in a study focused on carrier screening in a pre-conception population using whole genome sequencing (WGS), as part of the Clinical Sequencing Exploratory Research (CSER) Consortium. Methods: Approximately 720 genes/condition pairs for autosomal recessive (AR) and X-linked (XL) disorders were selected for an expanded carrier screening panel and were divided into categories of: lifespan limiting, serious, mild, unpredictable outcome, and adult-onset conditions. Participants, recruited from Kaiser Permanente Northwest (KPNW), consented to receive results for lifespanlimiting conditions but could choose whether or not to receive results for the other categories as well as additional medically actionable findings (~120 genes/condition pairs). Our protocol is sequential screening, with females tested first, and if positive, the partner is offered testing. All WGS is performed at Illumina, and sequence alignment and variant annotations were performed at the University of Washington using the SeattleSeq bioinformatics pipeline. Our laboratory filtered, classified, and confirmed carrier variants, and issued a clinical report. We use the ACMG and AMP recommendations for variant classification. Additionally, we perform functional analysis for classification of novel splice variants and have developed an in-house algorithm for classification and confirmation of CNVs. Only pathogenic or likely pathogenic variants were reported. Both pre-and post-test genetic counseling was done at KPNW. Results: Most (~92%) of participants requested all categories of carrier disorders returned, as well as additional medically actionable findings. To date over 150 participants have had results reported with approximately 70% carrying at least one pathogenic or likely pathogenic variant (ranging up to 5 variants) in the selected set of carrier genes and ~1% to 2% having a pathogenic variant in a gene associated with a medically actionable condition. All types of variants have been detected, including copy number variants (CNVs), although the majority (>50%) are missense variants, which present the greatest challenge in variant classification. Conclusions: All variants we reported from WGS would also have been detected by exome or gene panel testing. Our experience suggests such an expanded carrier screening panel for AR and XL conditions, including rare and ultra-rare disorders, using a gene panel approach, or as additional findings in WES, may be an appropriate consideration for clinical laboratories. P.R. Reynolds, J. Abbott, P. Hauk, S. Dirscherl, M. Salfinger, R. Harbeck, E. Gelfand National Jewish Health, Denver, CO. Introduction: In a tertiary referral center with a program in immunodeficiency diagnosis and treatment, it is not uncommon to find patients with overlapping clinical and laboratory phenotypes resulting from distinct genetic mutations. A number of recent studies have associated certain immunodeficiency symptoms with small groups of genes; however, analyzing small numbers of candidate genes via Sanger sequencing has rarely uncovered a causative mutation. Therefore, a panel of 347 genes was created for analysis by next-generation sequencing. All genes have an established relationship to Primary Immune Deficiencies, including complement pathway defects. Methods: Genomic coordinates for all exons (+/-100 nucleotides) of each of the 347 genes were obtained from the UCSC Genome Browser. Coordinates were submitted to Agilent for design and synthesis of a custom QXT library kit (Sure Select Target Enrichment System), which includes Biotin-labeled RNA oligonucleotides for capturing sequences of interest. Libraries were created according to manufacturer's directions. Quality and quantity of each library was established on an Agilent Bioanalyzer, and a MiSeq (Illumina) was loaded at 12 pM total for all libraries. Typical runs yielded ca. 7 Gbase of sequence data, resulting in coverages of 50x-300x for nearly all targets, and 99% overall coverage of the panel. Sequence data was analyzed for variants by NextGENe (softgenetics.com). Variants with "Damaging" predictions via programs such as SIFT and Polyphen and low or zero frequency in 1000 genomes were investigated further by checking on frequency in the EXaC database, reports in the literature and/or OMIM of known or similar mutations, and fit of symptoms/phenotype with mutated gene. Results: A number of disease-causing mutations were identified, both known and novel. The genes containing the mutations include ATM, BTK, CARD11, CTLA4, FOXN1, FOXP3, LIG1, LIG4, NOD2, STAT1, and TAP. Examples: a novel AIRE mutation in a youngster with diabetes and family history of autoimmunity; a CARD11 mutation in a patient with severe eczema and herpes infections; a LIG4 in a newborn with T cell lymphopenia, a NOD2 mutation in a 2-year-old boy with severe arthritis in all joints, FOXP3 mutation in a 12-year-old boy with autoimmune enteropathy, and a FOXN1 mutation in an infant with T cell lymphopenia. Conclusions: With a rapid turnaround time, this panel provides an excellent adjunct to the newborn screening programs developed for statewide testing. Since laboratory testing may be indicative but not diagnostic of a PIDD, genetic testing is now becoming an essential clinical tool since it defines the diagnosis and serves to guide specific therapy. Introduction: Medium-chain acyl-CoA dehydrogenase (MCAD) deficiency is an autosomal recessive disorder that results in defective fatty acid metabolism by preventing the conversion of medium-chain fatty acids to energy. The disease is typically diagnosed during infancy or early childhood. Symptoms may include vomiting, lethargy, hypoglycemia, seizures, breathing difficulties, liver problems, brain damage, coma, and sudden death. The symptoms are commonly triggered by prolonged periods of fasting or by illnesses such as viral infections. Patients are usually identified during newborn screen while final confirmation is made by mutation analysis. The disease manifestations are managed mainly by strict avoidance of fasting, avoidance of medium chain triglycerides and carnitine supplementation in diet. The ACADM gene coding for MCAD deficiency maps to chromosome 1p band 31 and has 12 exons. The most common mutation in the ACADM gene that is found in about 80% of MCAD patients is c.985A>G (p.K329E) and the second most common mutation occurring in approximately 15% of patients is c.199T>C (p.Y67H). In this study we report a 10-month old infant girl who was positive on the newborn screen. Besides the abnormal new born screen, she does not show clinical symptoms of MCAD deficiency. The patient has been metabolically stable at 10 The Journal of Molecular Diagnostics ■ jmd.amjpathol.org population, summarize the phenotypic findings, and determine the frequency of cooccurring copy number variants. Methods: Internal databases of our clinical SNPbased chromosomal microarray (CMA) testing that are used for quality assurance purposes were searched to determine the total number of these tests ordered at our institution between August 2012 and April 2016 and to determine the total number of cases with the 15q11.2 BP1-BP2 deletions. The deletions found in parents or other family members of 15q11.2 BP1-BP2 deletion probands were excluded from the final count of cases with the deletion. Results: During a forty-five month period we detected the 15q11.2 BP1-BP2 deletion in eighteen probands out of approximately 1352 total CMA tests performed, a frequency of 1.3% among patients referred for CMA testing at our institution. The primary clinical indications for study in these individuals included autism (3/18, 16.7%), epilepsy (1/18, 5 .6%), developmental delay (5/18, 27.8%), learning disabilities (2/18, 11.1%), congenital anomaly or anomalies (4/18, 22.2%), dysmorphic features (1/18, 5.6%) , and other (2/18, 11.1%). Under our filtering and reporting criteria, one or more co-occurring copy number variants were reported in 12/18 individuals (66.7%). Three of these cooccurring CNVs fulfilled our criteria for pathogenic (1) or likely pathogenic (2) calls, five were of uncertain clinical significance, and eight were deemed likely benign. One individual was noted to have 4.4% homozygosity across the genome. Conclusions: The 15q11.2 BP1-BP2 deletion was enriched in our patient populations compared to published frequencies in normal control populations. Emerging data continue to support the pathogenicity of the 15q11.2 BP1-BP2 deletion. The incomplete penetrance and phenotypic variability of this deletion may be related to the presence of additional genetic variation in affected patients. S. Lincoln 1 , M. Cline 2 , S. Yang 1 , Y. Kobayashi 1 , S. Topper 1 , D. Haussler 2 , B. Paten 2 , R. Nussbaum 1 1 Invitae, San Francisco CA; 2 University of California Santa Cruz, Santa Cruz, CA. Introduction: As increasing numbers of laboratories offer genetic tests, the potential for inconsistent variant classifications increases. Classification differences between public databases that laboratories may use have been raised as a particular concern. Although real, the clinical impact of these differences is not clear, because experienced and responsible lab directors should never simply copy classifications from public databases. Instead, they critically evaluate underlying evidence and determine classifications following rigorous guidelines. Our recent study (Lincoln et al., JMD 2015) demonstrated high (99.8%) concordance of 975 BRCA1/2 tests classified following ACMG/AMP guidelines and using only public data, compared to tests that also utilized non-public information. Here, we sought to similarly evaluate concordance of BRCA1/2 variant classifications in a larger, inter-laboratory data set. Methods: Over 5000 ClinVar submissions of classified variants from 6 established laboratories (Ambry, Counsyl, Emory, GeneDx, Invitae, and Myriad) were used. Myriad data were submitted by the Sharing Clinical Reports Project. Clinically significant differences were those that would substantially change actionability, i.e., between pathogenic (including likely pathogenic), versus VUS, benign or likely benign; otherwise results were considered concordant. Results: Counting each variant separately, concordance between pairs of labs is high: 98.5% overall. However this calculation greatly underestimates the much higher concordance observed on a per-patient basis. Most discordant classifications are in rare variants that, by definition, are present in very few patients. Moreover, classifications of most rare variants agree (sometimes, concordantly VUS). Based on clinical and population prevalence, we calculate that 99.8% of patients receive net concordant reports, similar to our previous study's results. Even after detailed examination of all evidence underlying the remaining disagreements, the maximally correct classification under current guidelines sometimes remains unclear. Conclusions: Classification concordance needs to be measured carefully to avoid over-counting differences. Whereas substantial differences are seen in few patients, it is important to resolve these collaboratively, not competitively, as is done in other areas of medicine. Thorough peer review of classifications is enabled by public databases like ClinVar and both supports laboratory quality control efforts and helps improve critical care guidelines. R. Kolhe, C. Pundkar, A. Mondal, W. Lee, A. Chaubey Augusta University, Augusta, GA. Introduction: A correct diagnosis is critical in separating spontaneous pregnancy loss and RPL and for determining appropriate therapeutic options. Traditionally, karyotyping has been used in molecular analysis of RPL and IF and been overall very helpful in RPL workup. Unfortunately Karyotype analysis also has multiple limitations such as low diagnostic yield, long turnaround time, missing cryptic changes (<5MB), required trained people to perform analysis and is a subjective method prone to human error. Methods: We performed an internal audit at our institution (2012 to 2015) and out of the 578 samples sent to cytogenetics lab for karyotype analysis only few (< 6%) cases had information which helped patient care. This led us to re-examine our RPL cases (FFPE samples) on SNPM. Archival blocks with slides were retrieved and reviewed. Clinical information was obtained from patient charts under approved IRB protocol. H&E slides were examined and chorionic villi (fetal tissue) was identified and marked for DNA isolation by surgical pathologist. The whole genome SNPM was performed on the DNA isolated from FFPE specimens following manufacturer's protocol (OncoScan FFPE Assay Kit, Affymetrix, Inc.). The raw data was analyzed in Chromosome Analysis Suite 3.0 software and were matched to in silico FFPE reference sets. This platform consists of 274,000 probes including 74 somatic mutations from 9 genes (BRAF, KRAS, EGFR, IDH1, IDH2, PTEN, PIK3CA, NRAS and TP53) . The IF cases were examined on Cytoscan (Affymetrix, Inc.) on blood. Results: All the RPL and IF cases (n=10) resulted in substantial information helping the diagnosis and patient prognosis on SNPM. e.g., Case 1: 38 YO female with history of 7 miscarriages with failed multiple attempts for karyotyping. Her SPMN analysis demonstrated mosaic trisomies of chromosomes 21 (~75% mosaic gain) and 22 (~70% mosaic gain) were observed. Case 2: 32 YO male with infertility/testicular failure with 46XY on karyotype. SNPM analysis showed Y chromosome micro deletion. Conclusion: There is high failure rate in accurately karyotyping products of conceptions (POC) for multiple reasons (Blinded sampling error in OR by D&C, overgrowth of maternal cells resulting in 46, XX, culture failure, microbial overgrowth etc.). SNPM technology has a remarkably higher resolution than conventional cytogenetics (karyotyping and FISH) in identifying cryptic deletions and duplications within the human genome. Having any additional genetic information from POC is very comforting to the patient and extremely beneficial to the reproductive medicine clinic to plan the next pregnancy. We anticipate that this approach of obtaining high-resolution data from an FFPE sample will facilitate studies of RPL which was not previously possible. I. King, M. Zariwala, K. Chao, K. Michael, K. Weck University of North Carolina at Chapel Hill, Chapel Hill, NC. Introduction: Primary Ciliary Dyskinesia (PCD) is an inherited disease causing defects in motile cilia, which manifests primarily with lung and upper airway difficulties. Specialized management of PCD patients has been able to improve outcomes greatly, as lung damage is in PCD is cumulative. However, most PCD patients do not receive a definitive diagnosis because the current gold standard diagnostic methods (electron microscopy and video analysis of cilia) require highly specialized interpretation. We have validated a comprehensive genetic diagnostic panel that will enable higher yield genetic diagnosis for patients suspected of having PCD. Methods: A custom hybridization capture reagent was developed for the Nextera Rapid Capture process (Illumina). This reagent incorporates capture probes targeting all exons of 33 clinically identified PCD genes, as well as one deep intronic site in the CCDC39 gene. All exons of two genes associated with inherited immune disorders that can phenocopy PCD (RAG1 and GAS2L2), were also included. All exons of CFTR were included, as cystic fibrosis is the major differential diagnosis in patients suspected to have PCD. In addition to these clinically reportable genes, all exons of 65 genes hypothesized to contribute to ciliary function were included as capture targets for research purposes. Results: Sequencing under anticipated clinical conditions (multiplexing up to 12x per MiSeq flowcell) yielded coverage of > 0.2x mean coverage for >95% of targets in 5 validation runs. Sporadic target dropout (defined as <20x coverage) was observed, but no one target was consistently under covered. A validation cohort was assembled from 21 PCD patient samples with known mutations from an ongoing PCD research study at UNC. With variants interpreted under a strictly autosomal recessive model of inheritance, no false positive results were obtained, whereas known mutations were correctly identified in all 21 cases. Variants identified by sequencing were filtered for population frequency, alleleic fraction, and to remove known sequencing artifacts and pseudogene interference. With this filtering strategy no more than 5 variants in each sample required characterization. Additionally, one patient, who also suffers from Cri-du-chat syndrome and carries a 5p deletion, was found to have a previously unidentified variant in DNAH5, which provides a potential explanation for the patient's PCD symptoms. Conclusions: This validation has demonstrated the technical and clinical utility of an expanded sequencing panel for PCD, and has identified a potential cause of PCD in one patient with a clinical diagnosis. M. Cleveland 1 , L. Borshuk 1 , J. Zook 1 , M. Salit 2 , P. Vallone 3 1 National Institute of Standards and Technology, Gaithersburg, MD; 2 National Institute of Standards and Technology, Stanford CA. Introduction: The Genome in a Bottle (GIAB) Consortium aims to highly characterize a small number of benchmark genomes. These genomes have been sequenced by several sequencing technologies, each with different biases in detection of indels, SNPs, and structural variants. These highly characterized genomes can thus be used as benchmarks to evaluate new sequencing technologies. As of June 2016, NIST has one highly characterized genome, NA12878, sold as RM 8398, and there are several more genomes in the pipeline to become NIST reference materials. The GIAB consortium would ultimately like to offer more clinically relevant benchmark genomes to the community. Targeted sequencing panels, which cover hundreds of genes, may be utilized to help identify and characterize additional genomes with clinically relevant mutations. Methods: In this project, we use two different targeted sequencing panels (the Illumina Inherited Disease Panel* and the Ion Torrent Ampliseq Inherited Disease Panel*) to examine jmd.amjpathol.org ■ The Journal of Molecular Diagnostics seven different genomes in triplicate (NA12878, GM24385, GM24149, GM24143, GM24694, GM24695, and GM24631). The Illumina Inherited Disease Panel (IDP) covers 550 genes and the Ampliseq IDP covers 328 genes. We examine the concordance of calls between the two inherited disease panels and the agreement of the sequencing with the GIAB base calls. Results: The Illumina IDP and the Ampliseq lDP have an overlap of 96 genes. Of those 96 genes, 29 are 100% covered in the current GIAB High Confidence regions (HCRs). Seven genes (CTNS, F8, HEXA, PEX5, WAS, GBA, GAA) are covered at less than 80% in the GIAB HCRs. There are 8 genes on the Illumina IDP only which are not covered at all in the GIAB HCR (CYP11B1, CYP21A2, HBA1, SBDS, SLC6A8, TUBA1A) and 4 genes on the Ampliseq IDP that are not covered at all in the GIAB HCR (FRG1, HBA2, KRT14, MAPT) . Conclusions: We have demonstrated how the GIAB highly characterized genomes may be used to evaluate different targeted sequencing panels. Targeted sequencing panels may be a cost effective way to characterize a high number of clinically important genomic regions. *Certain commercial products are identified in this paper to foster understanding. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose. To avoid these hypersensitivity reactions, HLA-B*57:01 genotyping prior to ABC administration is considered standard of care. Several HLA-B*57:01 genotyping techniques have been developed; however, most of them have limited specificity and sensitivity for HLA-B*57:01 detection, require special instruments/techniques, and are labor intensive. Here we are reporting a novel HLA-B*57:01 genotyping method using hydrolysis probe-based real-time PCR. Methods: Eleven fully HLA genotyped genomic DNA samples were purchased from the Coriell Institute for Medical Research. Thirty previously-tested patient samples were obtained from a reference laboratory. Primer and probes were designed based on published sequence variations in exon 3 of HLA-B between ABC-sensitive (e.g. HLA-B*57:01) and ABCinsensitive alleles (e.g. HLA-B*57:03 and HLA-B*58:01). There are two polymorphism sites in exon 3. The first one is close to the 5' end of the exon and the other is in the middle of the exon. The first polymorphic site was used for allelespecific PCR amplification. PCR primers were designed to specifically amplify HLA-B*57:01 and its closely related alleles such as HLA-B*57:03 directly from genomic DNA. Most of the unrelated alleles would not produce any product in the PCR reaction. Our hydrolysis probes enable differentiation of HLA-B*57:01 from the other amplified, but ABC-insensitive, ABC alleles based on the second polymorphism. As confirmatory measures, additional hydrolysis probes/primer and PCR-RFLP markers were designed and tested. Results: Our real-time PCR demonstrated 100% accuracy in distinguishing HLA-B*57:01 positive (n=2) and negative (n=9) samples among the genomic DNA samples with known HLA genotypes. Additionally, thirty samples previously tested with a separate HLA-B*57:01 real-time PCR method in a reference laboratory were evaluated with our assay. We found concordant results with all but one apparent false-negative and one apparent false-positive result. Additional real-time PCR with different primer and probe sets and PCR-RFLP showed that the discordant genotypes were correctly called by our assay. Conclusions: The hydrolysis probe-based real-time PCR for HLA-B*57:01 genotyping showed better accuracy compared to previously developed genotyping techniques. Our newly developed test will allow clinical laboratories with real-time PCR capabilities to determine the HLA-B*57:01 genotype in a timely and costeffective manner. A.G. Hadd, S. Morales, A. Anderson, H. Rey, G.J. Latham Asuragen, Inc., Austin, TX. Introduction: Database accessibility to next-generation sequencing (NGS) data from thousands of individuals has substantially increased our understanding of human DNA variation across different populations. However, many SNPs in publically-available databases are invalidated and thus could reflect intrinsic error rates of large-scale NGS. One SNP, rs111485627, reportedly overlaps the FAMlabeled reverse primer of the AmplideX PCR/CE FMR1 Kit, a commonly-used reagent set for FMR1 genotyping. This study addresses testing and secondary analysis of samples reported with this SNP and discusses the implications of the findings. Methods: A total of 20 unique cell-line genomic DNA, 12 female and 8 male, were obtained from the Coriell Cell Repositories. These samples were identified with rs111485627 in either Phase I or Phase III of the 1000Genomes database. DNAs were analyzed using AmplideX FMR1 PCR (Asuragen) using standard primers, and a single-offset base primer designed to avoid the putative SNP, in duplicate PCRs. The primer binding region for each cell-line DNA was independently assessed using Sanger sequencing. Finally, a synthetic construct of the CGG region designed with the SNP was prepared, sequenced verified and analyzed using AmplideX PCR. Results: All 20 samples reported with the SNP (5 from) were successfully PCR-amplified in the FMR1 triplet repeat region and genotyped without loss of signal compared to samples of comparable genotype lacking the SNP. Amplification with a single-base offset primer yielded similar signal intensity and the expected genotyping profiles. Importantly, Sanger sequencing unequivocally showed the wild-type sequence at this location in both males and females. In addition, a synthetic construct containing the G>T substitution was successfully genotyped compared to a wild-type construct and source genomic DNA. Conclusions: A putative SNP in the fragile X gene demonstrated no interference or loss of signal across 20 different samples reported with the SNP. The SNP failed independent verification using Sanger sequencing. Moreover, even if present, constructs with the reported SNP were successfully genotyped for FMR1 CGG repeat sequences. The SNP may be a consequence of error rates in NGS reported for GC-rich DNA regions. These findings provide a cautionary note regarding the accuracy of variants reported in public databases, and their potential impact or lack thereof on nucleic acid assays. D. Vendrone, M. Pieretti, L. Medling, A. Gonzalez BayCare Laboratories, Tampa, FL. Introduction: Clinical mutation assays commonly used to assess predisposition to thrombophilia are: Factor V Leiden (c.1691G>A), Prothrombin c.20210G>A and Methylenetetrahydrofolate Reductase (MTHFR) c.665C>T and c.1286A>C gene mutation. Our laboratory has offered clinical assays for these mutations for several years, but we sought to implement new versions requiring less hands-on, shorter detection, and no extraction to increase cost efficiency. Methods: We chose Focus ASRs designed as Scorpion Primers (fluorescently labeled sequence specific); and enzymes mixes optimized for common thermocycling conditions. Fluorescence thresholds, valid Ct values and other cutoff criteria are not provided and were established during the study. For each assay we used a "training set" of known patient samples to set criteria, and a "validation set" of distinct patient samples to validate them. All patient samples were previously characterized with Hologic Invader-Plus IVD. We established criteria for both whole blood specimens and extracted nucleic acids, intending to use whole blood as the primary method and nucleic acids only as back-up. Fluorescence thresholds for each mutation were chosen so that: 1) The background fluorescence of the negative target would not cross the positivity threshold. 2) For heterozygous samples, the delta Ct value (mutant -wild-type) would be as close as possible to zero. For each mutation, Ct mean and 2 standard deviation values from the "training sets" were used to establish Ct and delta Ct cutoffs for wild-type, homozygous and heterozygous calls. Results: Eighty-nine whole blood samples (20 for Factor V, 27 for Prothrombin and 42 for MTHFR) were used as the "training sets." After all cutoffs were established, an additional 128 samples (32 for Factor V, 33 for Prothrombin and 63 for MTHFR) were analyzed to verify the established parameters. Eighty extracted DNA samples (20 for each assay) were used as "training sets" for the extracted nucleic acid version of the assays. After thresholds and cutoffs were established, an additional 40 samples (10 each assay) were run to verify the established parameters. For both versions and all assays, correlation with the Hologic assays was 100%. Precision, interference studies and DNA concentration/dilution studies, as well as controls' range studies were also performed to further establish the analytical characteristics of these assays. Conclusions: We established Laboratory Developed Tests (LDTs) for Factor V Leiden, Prothrombin and MTHFR gene mutations assays using Diasorin ASR reagents; these tests are frequently used to assess patients risk and management of thrombophilic events. Using Anchored Multiplex PCR and Next-Generation Sequencing B.P. Culver 1 , L.M. Griffin 1 , M.T. Hardison 2 1 ArcherDX, Inc., Boulder, CO; 2 BabyGenes, Inc., Golden, CO. Introduction: Cystic Fibrosis (CF) is an autosomal recessive disease caused by mutations in the CF transmembrane conductance regulator (CFTR) gene. CF is characterized by the build-up of thick mucus resulting in chronic lung infections and airway inflammation. As such, early diagnosis and treatment interventions are crucial to help prevent airway obstruction and lung infections. Therefore, carrier identification and newborn screening have significant implications in the overall prognosis of CF patients. Unfortunately, there is a selection bias in CF diagnosis of white compared to nonwhite populations. This ethnic disparity in CF diagnosis is primarily attributed to differences in underlying CFTR mutations, which were recently shown to vary significantly across ethnic groups by Iris Schrijver et al. Current CFTR genotyping assays detect mutations highly prevalent in white individuals, yet fail to detect mutations that are more prevalent in nonwhite individuals. We present a rapid, cost-effective assay for comprehensive detection of CFTR mutations for pan-ethnic carrier identification and newborn screening. Methods: BabyGenes, Inc, in partnership with ArcherDX, Inc., has developed a targeted next-generation sequencing (NGS) assay based on Anchored Multiplex PCR (AMP) to detect all mutations currently reported in the CFTR2 database. AMP is a library preparation method for NGS that uses unidirectional gene-specific primers (GSPs) and molecular barcoded adaptors ligated to random start sites to enrich for both known and unknown mutations across a panel of target regions. Results: The BabyGenes Introduction: Multiplex PCR and capillary electrophoresis (CE)-based T cell receptor gene rearrangement assay has been widely used as an ancillary test for the diagnosis of T cell lymphomas. Identifying clonal peaks of identical sizes in different specimens from the same patient helps to improve test specificity; however, discrepant results can occasionally be encountered and pose challenges in test interpretation. Next-generation sequencing (NGS)-based clonality analysis reveals the sequences of individual T cell clonotypes in a quantitative way, and can potentially reconcile the differences among the specimens and overcome the challenges associated with fragment size-based testing method. Methods: A total of 10 specimens (4 peripheral blood and 6 tissue) from 4 patients with mycosis fungoides (MF) and discrepant CE-based T cell clonality test results were included in the study. DNA was extracted followed by BIOMED-2 primer-based TCRG and TCRB sequencing library construction. The amplicon libraries with TruSeq adaptors incorporated were subsequently sequenced on the MiSeq platform. After the paired-end reads were joined and the primer sequences were removed, the clonotypes were submitted to IMGT/HighV-Quest for TCR rearrangement analysis. The NGS results were then analyzed to correlate with the CE results. Results: The tissue specimens with clonal patterns identified by the CE method were also demonstrated by the NGS method to harbor clonotypes with predominant frequencies. Clonal peaks of the same sizes identified by CE in different tissue specimens from the same patients were confirmed to be identical clonotypes by NGS. The tumor clonotype present in tissue specimens was also detected by NGS in the blood specimen of a patient with suspected Sézary syndrome. A clonal pattern demonstrated by the CE method in the blood specimen of one patient was not supported by the polyclonal NGS results, suggestive of a false positive result by CE as a result of different clonotypes of the same amplicon size. The tumor clonotype sequences identified in the diagnostic tissue specimens could be used to definitively confirm the presence or absence of tumor clones in blood as well as in other tissue specimens. Conclusions: Compared to the fragment size-based method, NGS-based T cell clonality analysis makes it possible to not only determine the overall clonality pattern, but also sensitively and specifically detect the tumor clonotype in specimens from patients with MF. The discrepancies between the CE-based tissue and blood clonality results can be better interpreted and potentially resolved by the NGS-based approach. The NGS-based method also provides an opportunity to more accurately monitor response to therapy and minimal residual disease in patients with MF. R. Kolhe 1 , W. Lee 2 , C. Pundkar 2 , A. Mondal 2 , J. Chaffin 2 , A. Kornfield 2 , R. Kolhe 2 1 Medical College of Georgia at Augusta Univeristy, Augusta, GA; 2 Augusta University, Augusta GA. Introduction: AML is the most common form of acute leukemia in adults with an incidence of approximately 4/100,000. Upon diagnosis, patients receive induction chemotherapy aiming to achieve complete remission (CR). It known that despite successful induction therapy and completed consolidation therapy, relapse occurs in 60% to 70% of patients within five years. In the first relapse, the likelihood of achieving a renewed CR is lower than at onset of disease, and the duration the second CR is almost invariably shorter than the first. Relapse is a significant reason why the overall prognosis of adult patients with AML remains poor with a 20% to 30% likelihood of 5-year survival after diagnosis. Methods: Archival blocks with slides were retrieved, reviewed and clinical information obtained from patient charts under approved IRB protocol. Several patient/disease characteristics were identified including age, sex, race, body mass index (BMI), and cytogenetics with or without mutational analysis were noted. Theraphy indicators such as treatment received, transplantation status, remission status and relapse status was also noted. Seven al RNA was extracted from ( n=6 with > 5 years of survival) & n=7 with < 6 months of survival) and analyzed with nanoString nCounter using PanCancer Immune Profiling Panel designed to quantitate 770 genes associated with identifying immune cells, assessing immunological function (e.g., immune checkpoint regulation), identifying tumor-specific antigens and housekeeping genes. For mechanistic studies 7 Leukemia cell lines (KU812, HL-60, THP-1, K0562, RS4, MOLT-4, and CCRF-CEM) were obtained from ATCC along with three non-leukemia human controls cells. Results: The gene expression profiles of early relapse vs no-relapse, showed significant (>3 times, p<0.05) upregulation of a set of 8 genes, and down regulation of 8 genes. Interestingly, gene IFI27 was upregulated 13X (p< 0.0012) in early relapse (<6months) patients as compared to >5yrs survival group. To further investigate the mechanism of IFI27 in cellular model RNA was isolated from 7 cell lines and real-time quantitative PCR was performed. THP-1 (acute monocytic leukemia) revealed the most notable upregulation of IFI27 (1,728 fold change). Conclusions: To the best of our knowledge, the current study represents the first gene expression studies investigating immune tolerance exploration of patients with relapse in AML. IFI27 is found to be upregulated in breast cancer, squamous cell carcinoma and hepatocellular carcinoma. Overexpression of IFI27 in ovarian cancer is associated with poor survival. This work is intriguing for the new information it provides about the possible role of IFI27 in AML relapse and treatment failure. M. Steenkamer 1 , A. Stuitje 1 , L. Atanesyan 1 , K. Stouten 2 , M. Werken 2 , R. Castel 2 , S. Savola Amsterdam, Netherlands; 2 Albert Schweitzer Hospital, Netherlands; Amsterdam, Netherlands. Introduction: Myeloproliferative neoplasms (MPNs) are a group of hematopoietic stem cell disorders, characterized by clonal proliferation of blood cells. Relevant diagnostic molecular markers in MPN are recurrent mutations in CALR, JAK2, KIT and MPL genes. Multiplex Ligation-dependent Probe Amplification (MLPA) is a widely used technique for gene copy number detection, while it allows simultaneous identification of known point mutations. However, the sensitivity of standard MLPA is limited to a mutant allele burden of a high sensitivity MLPA assay enabling detection of an allele burden as low as 1% jmd.amjpathol.org ■ The Journal of Molecular Diagnostics for the most frequent mutations in MPN. Our aim was to develop and to demonstrate the feasibility of a modified MLPA assay to artificial positive DNA and cell line samples. Validation of this assay was further performed on diagnostic MPN patient samples. Methods: Novel Salsa MLPA MPN probemix was designed and optimized to detect, in a single allele burden of the eight most frequent mutations in MPNs: CALR, 52-bp deletion and 5-bp insertion in exon 9 (L367fs*46 and K385fs*47), JAK2, deletions in exon 12 (N542_E543del and E543-D544del), JAK2, substitution in exon 14 (V617F), KIT, substitution in exon 17 (D816V) and MPL, substitutions in exon 10 (W515L and W515K). The analytic sensitivity and specificity of this newly developed MLPA assay was determined on artificial positive DNA samples, cell-line titration series (UKE-1) and on healthy human DNA samples (n=143). The assay was further validated using commercial reference DNA with 1% allelic burden of JAK2 V617F and KIT D816V mutations. Validation of the novel probemix was performed in a single-blind setting on diagnostic MPN patient samples (n=167). Results: Results obtained with this high sensitivity MLPA assay were concordant with an allele-specific PCR results in MPN patient samples. Moreover, 2 novel cases with 1.4 % to 5% JAK2 V617F burden, which were not detected by allele-specific PCR, were identified with our novel assay and confirmed by the Ipsogen MutaQuant assay. Furthermore, no false positive calls for mutations were obtained when testing on healthy human DNA samples. Conclusions: Our results demonstrate that our novel high sensitivity MLPA assay is a reliable technique for simultaneous detection of eight frequent mutations in MPNs, even when the mutant allele burden of the patient DNA sample is low (1% to 5%). These results merit further consideration of MLPA as a possible alternative for mutation testing for newly diagnosed MPN patients. Y. Cho, S. Jang, E. Seo, J. Lee, J. Lee, K. Lee, C. Park University of Ulsan, College of Medicine and Asan Medical Center, Seoul, Republic of Korea Introduction: Mutational status and bone marrow histopathology are essential for the diagnosis of myeloproliferative neoplasms. Specifically, bone marrow histopathology is the key to differentiation of essential thrombocythemia (ET) from prefibrotic primary myelofibrosis (PMF) . The aim of this study is to assess the correlation of bone marrow histopathology with mutational profile. Methods: We rereviewed the bone marrow biopsies of 128 patients with either ET (n=69) or PMF (n=59) without knowledge of mutational status or original diagnoses. JAK2 V617F mutations were positive in 68 patients, CALR exon 9 mutations were positive in 31, and MPL W515 mutations were positive in 3. Results: Among the ET group, patients with the CALR exon 9 mutation more frequently showed a higher number of megakaryocytes (P=0.043), dense clusters of megakaryocytes (P=0.021), small megakaryocytes (P=0.003), and bulbous nuclei (P<0.001) than those with JAK2 V617F. A re-review of bone marrow biopsies revealed that 10 of 69 (14.5%) patients originally diagnosed as ET had histopathological features of PMF rather than ET. This finding might indicate that these patients should be re-classified as prefibrotic PMF. However, the histologic diagnoses of patients originally diagnosed as PMF were not changed, even after re-review. Among the PMF patients, including both previous cases and ET cases showing histologic features of PMF, mutational status was not associated with significant differences in overall survival or leukemiafree survival. Instead, patients negative for all three mutations had shorter overall survival (P<0.001) and leukemia-free survival (P=0.009) than patients with one of the three mutations. Conclusions: Patients with JAK2 V617F mutation exhibited a heterogeneous feature in bone marrow histopathology, whereas CALR mutationpositive ET patients showed megakaryocyte abnormalities and exhibited a tendency towards prefibrotic PMF. T. Yang 1 , A. Box 2 , C. Hill 1 1 Emory University, Atlanta, GA; 2 Emory University School of Medicine, Atlanta, GA. Introduction: The diagnosis of myelodysplastic syndrome (MDS) is very challenging in that nonspecific findings, such as mild trilineage dysplastic changes, although important features for the diagnosis of MDS, can be seen in numerous other conditions. Genetic abnormalities, if detected, can provide clonal evidence of myeloid neoplasm and significant prognostic information to guide patient management. With the discovery of MDS-associated mutations, next-generation sequencing (NGS) has become one of the most sensitive methods to provide critical information to assist in the diagnosis of MDS, especially in low grade MDS patients. In this study, we have analyzed the utility of NGS myeloid neoplasm panel (NGS-MNP) to facilitate a definitive diagnosis of MDS. Methods: This study was performed under Emory institutional review board protocol. Clinical data of patients with NGS-MNP testing performed from September 2015 to March 2016 was reviewed. Pre-NGS, post-NGS diagnosis, gene mutations and allele frequencies were tabulated. The data was filtered for patients with a post-NGS diagnosis of MDS, acute myeloid leukemia with myelodysplasia-related changes or if bone marrow examination was performed for a clinical suspicion of MDS. The NGS-MNP was considered to have altered the diagnosis if any MDS-associated mutation was identified and used to confirm the diagnosis, or if no mutation was identified and the result was used to rule out MDS. Results: Between Sept 2015 and March 2016, NGS-MNP was performed on 214 patients. Thirty-one percent (66 cases) of the panels were performed for suspected/presumptive MDS. Of these, 58% (38 cases) of the initial diagnoses were altered based upon the results of the myeloid NGS panel. In cases where the post-NGS diagnosis of MDS was altered, 34% (13 cases) of the cases had no mutations and this result assisted in ruling out MDS. The most common mutations identified in MDS cases with NGS altered diagnoses were TET2 (11; 17%), TP53 (6; 9%), DNMT3A (6; 9%), and SF3B1 (5; 8%). In the 5 cases of low risk MDS, the most commonly identified mutations were SF3B1 (3; 60%) and RUNX1 (2; 40%). A myeloproliferative neoplasm-associated mutation was seen in 8 cases: JAK2 V617F (4; 6%) or CALR in/dels 4; 6%). Conclusion: Myeloid panel NGS testing resulted in a clinically significant change in diagnosis in 58% of the suspected/presumptive MDS cases received at the Emory Molecular diagnostic Laboratory and represent a rapidly expanding and valuable tool for the diagnosis of MDS. Tanzania; 6 University of Utah School of Medicine, Salt Lake City, UT. Introduction: Glucose-6-phosphate dehydrogenase (G6PD) deficiency is a common genetic abnormality known to predispose to acute hemolytic anemia (AHA), which can be triggered by certain drugs or by infection. However, the commonest trigger is the ingestion of fava beans (Vicia faba), causing AHA (favism), which may be lifethreatening especially in children. G6PD deficiency is genetically highly heterogeneous, as nearly 200 different mutations have been observed in the G6PD gene. Methods: We have investigated the hematological and genetic features of acute favism in the Palestinian Gaza community utilizing 131 children hospitalized for G6PD deficiency induced AHA and comparing the findings with indices from the general Gaza population. Results: We discovered the polymorphic coexistence of three different G6PD deficiency genes (G6PD A-, G6PD Cairo, G6PD Med) in the Gaza society. We have found by comparison to the general population (485 adults and 466 newborns) that children with favism, in terms of relative frequency, G6PD A-was under-represented, whereas G6PD Med was overrepresented. We also found that the severity of anemia was significantly greater with G6PD Med and G6PD Cairo than with G6PD A-; and with G6PD Cairo, compared to the other two variants, there was greater hyperbilirubinemia, as well as persistence of mild anemia and reticulocytosis for as long as 4 months after recovery from favism. Conclusion: We conclude that children with G6PD A-deficiency are also susceptible to AHA, but demonstrate in direct comparison within this same population that G6PD Med and G6PD Cairo are more severe forms of deficiency than G6PD A-. Further, we show that the heretofore poorly studied G6PD Cairo may be associated with low-level, chronic hemolysis. This study illustrates favism is a significant public health problem in Gaza due to fava beans as a staple in the diet and the coexistence of polymorphic G6PD deficiency variants in the society. Favism is an easily preventable and manageable genetic disorder with the proper awareness, intervention and education programs. 6 1 Rutgers University, Piscataway, NJ; 2 Rutgers Robert Wood Johnson Medical School, New Brunswick, NJ; 3 Rutgers University, New Brunswick, NJ; 4 Providence Heath and Services, Portland, OR; 5 Moffitt Cancer Center, Tampa, FL; 6 Dartmouth Hitchcock Medical Center and Geisel School of Medicine, Lebanon, NH. Introduction: Next-generation sequencing of cancer tissue is becoming a mainstream technique in clinical laboratories because of its potential to contribute to the selection of patient-specific therapies. Here, we describe the validation of the ThunderBolts (TB) Myeloid Panel from RainDance Technologies for the detection of sequence variants in DNA isolated from bone marrow and peripheral blood samples from patients with myeloid neoplasms. Methods: We received 72 previously tested blood or bone marrow DNA samples from three CLIA-certified and CAP-accredited laboratories: i) Molecular Genomics Laboratory, Providence Heath and Services; ii) Molecular Diagnostics Laboratory, Moffitt Cancer Center; and iii) CGAT, Department of Pathology and Laboratory Medicine, Dartmouth Hitchcock Medical Center. All three labs used the TruSight Myeloid Panel and the MiSeq instrument from Illumina for variant analysis. A minimum of 40 ng DNA was analyzed using the TB panel and PCR libraries from up to 12 samples were pooled and sequenced on the MiSeq. DNA from three well-characterized human cell lines was also used to validate the TB myeloid panel. Nucleotide sequence data were analyzed using the NextGENe software from SoftGenetics. Results: We observed an average of 4,886 reads per analytical sensitivity and specificity of the assay, tested using DNA from three cell lines, were 98.1%. The reproducibility of the assay, determined by analyzing two patient samples in three different runs, was 92.8%. We determined the lower limit of this assay for the detection of sequence variants by mixing cell line DNAs in different proportions. We were able to detect sequence variants at the 2% level in these mixtures. Sequence data were available for 46 samples from the three external labs. These labs did not report any pathogenic variants or variants of unknown significance in 12 of these samples and variants in 5 samples could only be detected by the TruSight panel. The corresponding genes were not part of the TB panel. The external labs reported 50 sequence variants in the remaining 29 samples. The TB panel detected 48 of these variants, giving a concordance of 96%. Our software did not detect 2 variants, but one variant was in the region of a gene (CEBPA) where the coverage depth was low (281X). When we examined this region in more detail, we were able to detect this variant. Including this variant, the concordance was 98%, suggesting excellent agreement between the TB and the TruSight panels. Conclusions: We have validated the TB myeloid panel for the detection of sequence variants in myeloid neoplasms. The assay has high accuracy and it can readily detect minor allele frequencies below the 5%. Myeloproliferative Neoplasia and their Relation with JAKV617F Mutation D. Aguilar León 1 , E. Martínez-Cordero 2 , G. Herrera Maya 1 , C. Lara Torres 3 1 Instituto Nacional de Ciencias Medicas y Nutrición "Salvador Zubirán," Mexico City, Mexico; 2 Instituto Nacional de Enfermedades Respiratorias, Mexico City, Mexico; 3 The American British Cowdray Medical Center, IAP, Mexico City, Mexico. Introduction: Cytokines play an important role controlling cell growth and immune response. Several of them function activating cell membrane associated receptors that rely on JAK2 to transduce signals and regulate transcription of various genes. In mammals JAK family of kinases have four members (JAK1, JAK2, JAK3 and TYK2). JAK2 is a tyrosine kinase that promotes growth of myeloid cell lines through constitutive activation of pathways related with JAK2, STAT5, PI3K, ERK and AKT. JAKV617F mutation may be present in myeloproliferative neoplasia such as; polycythaemia vera (PV), essential thrombocythaemia (ET), and primary myelofibrosis (PMF). This suggest that the immunologic environment may be a factor influencing the evolution of myeloproliferative neoplasia, although little is known. Methods: The study included 9 healthy donors of bone marrow stem cells (controls) and 21 pared samples (bone marrow and peripheral blood) of patients with diagnosis of myeloproliferative neoplasia according to WHO 2014 criteria. Extraction of mRNA was performed with RNA FFPE extraction kit (Promega) and cDNA synthetized with Omniscript Reverse Transcriptase (Qiagen). Primers were design to amplify IL-17, TNF--18, IL-23 e IFNbased on sequences from ensemble.org. Assessment of mRNA expression using Quanti Tect Master Mix Kit SYBR Green (Qiagen) and copy number/ul determined using standard curves. JAK2V617F mutation was determined using PCR with melting curve analysis and conventional PCR with restriction endonuclease enzymes. Melting Mean and standard deviation were determined and the results compared using T-Student test. Results: The distribution of diagnosis was as follows: PV (28%, n=6), ET (58%, n=12), and PMF (14%, n=3). The age of the patients range from 56 to 63 years-old, without differences according to diagnostic group or gender. JAKV617F mutation was detected in 100% of PMF, 50% of ET and 50% of PV patients. The cytokine expression analysis of PV, ET and PMF showed no statistically significant differences between groups or sample types (BM versus PB). However, we identified high levels of IL-4 (both BM and PB) for the three conditions, as well as increased expression of IL-23 and IL-17. Of special interest, dividing the cases by JAK2V617F mutation status we found significant difference in the level of expression of TNF-Conclusion: We identified a possible role of T-reg (Th17/IL-17 expressing) lymphocytes on bcr-abl negative MPN as well as differences in cytokine expression according to JAK2 mutation status. PCR with melting curve analysis had a higher diagnostic yield for JAK2V617K mutation determination compared to PCR followed by digestion with restriction endonucleases. Y. Linnik, F.B. de Abreu, J.D. Peterson, S.A. Turner, P. Kaur, D.L. Ornstein, G.J. Tsongalis, E.Y. Loo Dartmouth Hitchcock Medical Center, Norris Cotton Cancer Center and Geisel School of Medicine, Lebanon, NH. Introduction: The cohesin protein complex forms a ring structure around sister chromatids and helps regulate chromosome separation during meiosis and mitosis. Genes belonging to this complex are recurrently mutated in myeloid neoplasms; but reported disease subtype associations have been varied and sometimes conflicting. We describe the cohesion-complex variants detected in all bone marrow biopsies at a single institution, concerning for a myeloid neoplasm during specimen triage. Methods: All newly diagnosed myeloid neoplasm cases were evaluated using the 54 gene Illumina TruSight Myeloid Sequencing Panel on the MiSeq platform. Basecalling and sequence alignment were performed using the MiSeq Reporter Software and analyzed using VariantStudio v2.1. Results: Seventy-nine cases were evaluated, and variants in cohesion-complex genes were seen in 6/14 (43%) secondary AML (s-AML) and 4/21 (19%) of morphologically non-neoplastic cases (NN); STAG2 (s-AML, n=6; NN, n=3), RAD21 (s-AML, n=1; NN, n=1), and SMC3 were found (s-AML, n=1). No cohesion-complex mutations were seen in our cohort of primary AML (n=16), MDS (n=19), MPN (n=4) , and MDS/MPN (n=5). For STAG2, variants were identified across the transcript; one case had a frame-shift with early termination and loss of the SCD region and remaining transcript, the remaining variants were mostly missense mutations involving non-named/non-conserved regions throughout the transcript. Both RAD21 missense variants occurred in the conserved C-terminalSMC1 binding domain. The sole SMC3 missense variant involved the protein hinge region. Two of six s-AML cases had co-mutation of two different cohesion-complex genes. TET2, ASXL1, and RUNX1 mutations were the three most frequently affected genes in our cohort seen in 27.8%, 20.2%, and 17.7% of cases respectively. RUNX1 was most frequently co-mutated with cohesin-complex genes; in 5/6 s-AML (83%) and 1/4 NN (25%) cases. Co-mutation with TET2 was seen in 3/6 s-AML cases and 1/4 NN cases, and withASXL1 in 3/6 s-AML cases. Conclusions: Our cohort shows cohesin mutations most frequently occurring in s-AML, and often in association with RUNX1 mutations. Prior reports have indicated that cohesion-complex genes were associated with denovo AML as well as with NPM1; but our findings refute this idea, and the discrepancy may be due to the fact that many of these studies included cohorts classified by the old FAB methodology. Our findings support the association of cohesion-complex mutations with s-AML and high-grade MDS, which has also been reported in the literature. Although no variants were detected in our MDS cohort, this may be due to a relatively low sampling of high-grade cases. Y. Kim, H. Kang, Y. Han, S. Kim, H. Yang, H. Yoon, T. Park, H. Lee Kyung Hee University School of Medicine, Seoul, Republic of Korea. Introduction: The t(3;12)(q26;p13) is a recurrent translocation in hematologic malignancy however, rarely observed. The ETV6 gene is located on 12q13 and the 33 translocation partner gene are reported, and the most frequent partner band is 3q26 where the MECOM (complex locus of two genes EVI1 and MDS1) is located. The fusion transcript of ETV6 to MECOM as a result of t(3;12)(q26;p13) is related with development and progression of malignancy by promotion of proliferative capacity of cells. Most cases of t(3;12)(q26;p13) were confirmed with fluorescence in situ hybridization (FISH) and/or RT-PCR however, to the best of our knowledge, only one case proved the fusion site with sequencing and with hemi-nested PCR using two set of primers. In this case, we describe a case of therapy related AML with t(3;12)(q26;p13)ETV6/MECOM, which was confirmed by sequencing using single set primers. A 60-year-old man visited our hospital due to a cough, fever and sputum. He had osteosarcoma, and was treated with four cycles of cisplatin and doxorubicin, and two cycles of ifosfamide and cisplatin. Initial complete blood count showed pancytopenia; hemoglobin level of 6.0 g/dL, white blood cell count of 4.45× 10 9 /L, and platelet count of 68× 10 9 /L. The peripheral blood film smear showed 16% of immature cells among white blood cells. Bone marrow (BM) aspiration showed 26% of leukemic blasts and 36.8% of monocytes, and focal fibrosis were observed in biopsy. Metaphase cytogenetic analysis using BM aspirate revealed 46, XY, t(3;12)(q26;p13),-7 in 16 of 20 metaphase cells. We suspected the ETV6/MECOM fusion. Methods: To further characterize, RT-PCR was performed with primer set; ETV6 F1 (5'-CCTCCAGAGAGCCCAGTGCCGAGT-3') and EVI1R (5'-CTGATCATAACAGCCAGCGA-3'). The second PCR was performed using the same primers. The PCR products were purified and sequenced to further characterize the fusion product using BigDye (R) Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and ABI PRISM 3730XL Analyzer (Applied Biosystems). Results: The RT-PCR showed 212 bp size band, and the sequencing analysis using RT-PCR product showed a fusion between exon 2 of the ETV6 gene and exon 2 of EVI1 gene. The patient was diagnosed with t-AML. A follow up BM study was performed after induction therapy with standard-dose cytarabine and idarubicin however, 38.4% of cells were leukemic blasts and ETV6/MECOM fusion gene was still detected in RT-PCR. Conclusions: Our case of t-AML withETV6/MECOM clearly showed the fusion site with simplified RT-PCR, and the patient showed poor response to induction therapy. Further studies will be necessary for better diagnosis and treatment of ETV6/MECOM fusion related hematologic malignancies. Y. Cho, S. Jang, E. Seo, J. Lee, J. Lee, K. Lee, C. Park University of Ulsan, College of Medicine and Asan Medical Center, Seoul, Republic of Korea. Introduction: FMS-like tyrosine kinase 3-internal tandem duplication (FLT3-ITD) mutations are associated with poor prognosis in patients with acute myeloid leukemia (AML). We performed an analysis of the clinical and prognostic impact of FLT3-ITD mutations according to mutant to wild-type ratio (MR) and duplicated fragment length (DFL). Methods: A total of 122 patients with AML harboring FLT3-ITD were enrolled in the study. GeneScan followed by sequencing was used to analyze the MR and DFL. The MR was determined by dividing the peak area of mutation by that of wild-type. Results: Among 744 consecutive AML patients, FLT3-ITD mutations were observed in 122 (16.4%) patients. FLT3-ITD was found equally in both the FAB classification and World Health Organization subtypes. The MR ranged from 0.01 to 23.31 (median 0.545). The DFL ranged from 15 to 173 (median 45.5). The distribution of these parameters was not different between FAB M3 and non-M3 patients. In the non-M3 group (n=10 significantly higher leukocyte count (P=0.005) and lower platelet count (P=0.009). There were no significant differences with respect to AML subtypes, cytogenetic risk groups, nucleophosmin (NPM1), CCAAT-enhancer binding protein, alpha (CEBPA), jmd.amjpathol.org ■ The Journal of Molecular Diagnostics and mixed lineage leukemia-partial tandem duplication (MLL-PTD) mutations. We observed worse overall survival or relapse-free survival for patients with MR 1.0 or higher (P=0.005 and P<0.0001, respectively). However, the DFL did not provide any additional prognostic significance. In the M3 group (n=17), neither the MR nor the DFS showed clinical difference or prognostic significance. The MR of patients with hematological relapse was significantly higher than that of patients at initial diagnosis (P<0.001). Conclusions: DFL did not confer clinical significance, whereas the MR conferred significant prognostic difference in patients with AML. Therefore, the MR is helpful for performing more detailed risk stratification in patients with FLT3-ITDpositive AML. A higher MR at relapse may indicate selective survival of a FLT3-ITDpositive chemoresistant clone. Introduction: Ph-like ALL is a subtype of B-ALL associated with poor outcomes in children and young adults. These patients have a Ph-like gene expression profile similar to Ph+ ALL, but lack the characteristic BCR-ABL1 fusion. Additional chromosome fusions have recently been identified in Ph-like patients ALL that respond to ABL class tyrosine kinase inhibitors in vitro analogously to BCR-ALB1. These clinically actionable fusions are currently being detected using combined methods of multiplex PCR, kinome capture, or full RNA sequencing, but these assays are laborious, expensive, and time consuming to be useful on a large scale. We report the validation of two additional technologies that are cost effective, accurate, and faster for identifying both known and novel gene fusions. Methods: Two technologies were assessed across three laboratories using retrospective samples from consented patients enrolled on Children's Oncology Group clinical trials. The Nanostring nCounter Leukemia Fusion Gene Expression targeted assay and the Archer FusionPlex comprehensive fusion detection assay were used at both Nationwide Children's Hospital (NCH) and Children's Hospital of Philadelphia (CHOP), whereas Brigham and Women's Hospital (BWH) used the Nanostring assay exclusively. The Nanostring assay uses six fluorescent labeled RNA molecules with barcodes bound to target-specific oligonucleotide probes to potentially detect 972 sequences of interest. The Archer assay uses an anchored multiplex technology with primers flanking genes involved in known leukemia fusions and next-generation sequencing (NGS) to detect known and unknown 5' and 3' partners. Results: Pilot testing of Ph-like ALL cases with and known genomic lesions using the Nanostring assay identified all but one case 46/47 (97.9%) with ABL class, JAK2, NTRK3 or TSLP fusions correctly. Testing was performed in parallel at NCH and Brigham and Women's Hospital with 26/27 cases identified correctly and 27/27 cases correctly, respectively. For comparison, a panel of 29 Ph-like ALL cases with defined genetic lesions was tested in parallel at NCH and CHOP, with 27/29 (93%) fusions identified correctly. Conclusions: The Nanostring nCounter Leukemia Fusion Gene Expression assay performs well and is a fast and efficient method for detecting currently known Ph-like fusion transcripts. The Archer FusionPlex assay is more expensive due to its NGS requirement, but is good for identifying novel fusions involving known kinase genes. Whereas neither assay has the capacity of unbiased RNA sequencing for discovering alterations with novel kinases, both appear to be promising platforms for determining the presence of known, targetable fusions in Phlike ALL. V. Ortega 1 , C. Mendiola 1 , M. Khalil 2 , Y. Qian 2 , G. Velagaleti 1 1 University of Texas Health Science Center San Antonio, San Antonio, TX; 2 University of Texas Medical Branch, Galveston, TX. Introduction: Mantle cell lymphoma (MCL) is a lymphoid tumor derived from naïve CD5+ cells with the cytogenetic hallmark t(11;14) resulting in over-expression of the CCND1 gene. The presence of multiple chromosome abnormalities in addition to t(11;14) is known to be associated with blastoid variants (BMCL) and poor prognosis in MCL. High proportion of these tumors also show deletion of chromosome 11q22-23 or loss of heterozygosity; a region where the ATM gene is located. MCL tumors with bi-allelic ATM inactivation show significantly higher chromosome imbalances compared to MCLs with wild-type ATM alleles suggesting that loss of ATM alleles increases chromosomal instability of the tumor cells. Methods: We report a unique case of BMCL in a 52-year-old male who presented with worsening dyspnea, fever/chills, diffuse lymphadenopathy, splenomegaly and leukocytosis with blasts per differential. CT scan demonstrated extensive retroperitoneal, pelvic and inguinal lymphadenopathy with massive splenomegaly. Peripheral blood and bone marrow morphological, immunohistochemical and flow cytometry studies and bone marrow chromosome, FISH and high resolution microarray studies were performed using standard methods. Results: The blood smear showed leukocytosis due to the presence of "blasts" appearing cells. Bone marrow aspirate showed about 40% abnormal "blast" appearing cells. Bone marrow biopsy revealed remarkable lymphoid infiltrate in interstitial pattern and large lymphoid aggregates, contributing to about 40% of entire cellularity. The abnormal cells were immunoreactive to CD20, PAX-5 and Cyclin D1. Scattered were a few CD3 T-cells. Flow cytometry from the bone marrow aspirate detected a population (about 50% of total events) of lambda monoclonal B-cell population expressing CD19, CD22, CD20, and FMC-7 with coexpression of CD5, but were negative for CD10, CD200, CD43, CD34 and CD38. Based on these results, the patient was diagnosed with BMCL. Conventional cytogenetic analysis showed multiple chromosome abnormalities including t (11;14) and, contrary to published reports, bi-allelic amplification of ATM by FISH and SNP microarray. Apart from several characteristic gains, losses and regions of LOH, SNP microarray also showed high complexity abnormalities suggestive of chromoanagenesis involving chromosomes 7 and 11, especially involving the ATM gene region. Conclusions: Although Bi-allelic ATM mutations are common, gene amplification has never been reported in MCL. Focal chromothripsis as a mechanism of inactivating tumor suppressor genes was postulated in Retinoblastoma and we propose that chromoanagenesis inactivated both ATM alleles in our patient leading to multiple chromosome abnormalities and development of BMCL. Droplet PCR for the Molecular Assessment of Acute Myeloid Leukemia Q. Wei, N. Miltgen, M. Lovell Children's Hospital Colorado, Aurora, CO. Introduction: Library preparation methods are the backbone of any next-generation sequencing (NGS) assay. Specifically with targeted gene panels, the process of generating a library relies on multiplex primer-based enrichment chemistries to generate hundreds of discreet amplicons simultaneously. Currently, a handful of commercially available library preparation kits provide mutational analysis of acute myeloid leukemia (AML)-specific molecular abnormalities. Whereas targeted panel library preparation paired with next-generation sequencing (NGS) technologies have emerged as a powerful duo in the clinical setting, these methods have their limitations. Not surprisingly, it is difficult to optimize or "fine tune" the amplification efficiency for individual amplicons in multiplex PCR assays. Single molecule droplet PCR is a robust methodology for generating NGS libraries with superior amplicon coverage uniformity. Methods: Fifty-one bone marrow samples previously characterized with NGS methods from patients with acute myeloid leukemia/myeloproliferative neoplasm were compared in a double blind fashion to the results obtained using an AML gene panel with 548 amplicons covering clinical relevant regions in 49 genes using single molecule droplet PCR (ThunderBolts Myeloid Panel, RainDance Technologies; Billerica, MA). Sequencing was performed using a MiSeq bench top sequencer (Illumina Inc. San Diego, CA), and bioinformatic analysis was perform using NextGENe 2 nd generation sequence analysis software (SoftGenetics; State College, PA). Results: The panel successfully called all single nucleotide variants (SNV) and small insertions/deletions (indels) in all 51 samples. FLT3-internal tandem duplications were not initially detected with NextGENe software and an indel detection algorithm is needed for their successful detection. Both mean and median amplicon coverage (read depth) of CEBPA gene was greater than 2,500X. Dilution of AML patient DNA showed this panel could detect an allelic frequency down to 5% with a high level of assay reproducibility. Conclusion: Utilizing RainDance's patented droplet technology to ensure single occupancy of DNA templates targeted enrichment allows for the detection of clinically relevant somatic mutations using a simplified workflow with a 7-day turn-around time, requires only 75 ng of patient DNA, and results in superior coverage uniformity, especially in difficult to amplify GC-rich regions such as the CEBPA gene. Bone Marrow Biopsies L. N. Toth, F.B. de Abreu, J.D. Peterson, S.A. Turner, P. Kaur, D.L. Ornstein, G.J. Tsongalis, E.Y. Loo Dartmouth Hitchcock Medical Center, Norris Cotton Cancer Center and Geisel School of Medicine, Lebanon, NH. Introduction: The recently described acquisition of somatic mutations in the absence of abnormal cell counts or dysplasia has been termed Clonal Hematopoiesis of Indeterminate Potential (CHIP), and can reportedly be found in ~10% of persons >65 year of age. Being asymptomatic, these patients would theoretically never be identified as they would not require biopsy. Here, we characterize the cases of clonal hematopoiesis identified in patients who underwent bone marrow biopsy for workup of potential myeloid neoplasm, but were ultimately found to be morphologically non-neoplastic samples. Methods: All cases of suspected myeloid neoplasm during initial bone marrow triage were evaluated using the 54 gene Illumina TruSight Myeloid Sequencing Panel on the MiSeq platform.Base-calling and sequence alignment were performed using the MiSeq Reporter Software and analyzed using VariantStudio v2.1. Results: Of the 79 evaluated cases (36 F, 43 M); 21 biopsies (12 F, 9 M) did not show morphologic criteria for a diagnosis of dysplasia or malignancy. Ten of these 21 cases (~48%) carried at least one mutation in a gene recurrently mutated in myeloid neoplasms (mean 1.8/case, range 1-4), with 16 different variants detected. The average age of those with detected variants was 48 years (range: 17 to 76); all had normal cytogenetic studies. The most frequently reported gene was TET2 (n=4 cases) followed by STAG2 (n=3); other mutations including RUNX1, CUX1, ASXL1, EZH2, PHF6, ZRSR4, MPL, and RAD21 occurred at lower frequencies. All 10 of the ten cases with an identifiable clonal variant had clinical resolution of the issue prompting bone marrow biopsy, either with spontaneous resolution of the issue prompting biopsy (n=2) or identification of a separate exogenous cause for the issue (n=8). Only one of the 11 negative cases continued to have persistent cytoses of undetermined etiology at nine months follow-up. Conclusion: About a quarter of all marrows concerning for a myeloid neoplasm at initial triage were ultimately given a non-neoplastic morphologic diagnosis, and nearly half of these cases showed evidence of clonal hematopoiesis of undetermined significance. Although there is some selection bias, the detection of clonal hematopoiesis was greater than expected. With a mean of about 1 year of clinical follow-up, the detected variants appear to be unrelated to the underlying etiology necessitating bone marrow biopsy in most cases. The long-term consequence of these variants is yet to be determined, but suggest that more extensive studies are necessary before the medical community can reliably determine clinically significant clonal hematopoiesis. Introduction: Whereas the JAK-2V617F allele has been implicated in several of the myeloproliferative disorders, a significant proportion of patients with primary myelofibrosis (MF), or thrombocythemia (ET) are JAK-2V617F negative. One of the factors implicated in these cases is the constitutive activation of JAK-STAT pathway through mutations in the thrombopoietin receptor (MPL) gene. Specifically, mutations at the W515 position are present in approximately 5% of patients with primary myelofibrosis (MF), and in 1% of patients with ET, whereas mutations at the S505 position are common in patients with hereditary thrombocythemia. Mutations of MPL at the S505 and W515 positions have been tentatively linked to severity of myeloproliferative disorders. For this reason, we analyzed MPL sequences of 8700 clinical specimens from JAK-2V617F negative patients suspected to have myeloproliferative disorders. Methods: Clinical specimens were obtained and DNA was sequenced from 8700 patients suspected of having myeloproliferative disorders over an 18 month period. Purified DNA was then amplified by PCR on the ABI 9700 Thermalcycler using primers covering a 667 bp region of MPL that contained both the S505 and W515 mutations. The amplified products were subsequently sequenced using on the ABI 3100 or3730 sequencer in both directions for the covered region. These data were analyzed for mutations at S505 and W515 positions. Results: Of the 8700 patients' DNA analyzed, 104 had identified mutations at S505, W515 or within in the adjacent region. The majority of individuals with a variant sequence carried the mutations in at the W515 position (n=85/104, 81.7%). The majority of these individuals carried the W515L mutation (n=67, 64.4%), with the remainder carrying W515K (n=11/104, 10.6 %), W515A (n=3, 2.9 %), W515S (n=3, 2.9%) and W515R (n=1, <1%). A smaller number of individuals (n=12/104, 11.5%) had mutations at the S505 position, with all of them carrying the S505N mutation. Lastly, a small subset of individuals (n=7/104, 6.7 %) carried previously identified variants of R514S (n=1, ~1%), R514K (1, %), V507E (n=1, ~1%) and 501A (n=2,%). Of greatest interest was identification of two novel indels at position 515 in two different patients: c.1543_1552delinsAAAA, p.W515_PdelinsKT and c.1543_1559delinsAAAACTGCCAC, p.W515FS. Conclusion: This study examines the MPL mutations in JAK-2V617F negative patients suspected to have myeloproliferative disorders. We identified 104 individuals carrying variant sequences in MPL, predominantly at positions W515 (81.7%) and S505 (11.6%) and in adjacent regions (6.7%) in a population of 8700 individuals. Importantly, we have identified two novel indels at position 515 and further studies needed for the significance. A.F. Brown 1,3 , T. Parnell 2 , P. Szankasi 3 , J.A. Schumacher 3 , W. Shen 3 , K. Frizzell 3 , S. Sorrells 3 , J.L. Patel 1 , T.W. Kelley 1,3 1 University of Utah, Salt Lake City, UT; 2 Huntsman Cancer Institute, Salt Lake City, UT; 3 ARUP Laboratories, Salt Lake City, UT. Introduction: FLT3 internal tandem duplication (ITD) mutations are a marker of poor prognosis in cytogenetically normal acute myeloid leukemia (AML) patients. The overall mutation burden has additional prognostic significance. The FLT3-ITD allelic ratio (AR), derived from a PCR and capillary electrophoresis (CE)-based assay, is the current gold standard measurement of FLT3-ITD mutation frequencies. However, much of the testing of AML prognostic markers has moved away from single gene PCR/CE-type assays to next-generation sequencing (NGS) mutation panels. It is unclear if a FLT3-ITD variant allele frequency (VAF) from NGS is comparable to an AR, as currently defined. Here we sought to compare VAFs to ARs in a large set of FLT3-ITD positive cases. Methods: This study was approved by the University of Utah IRB. DNA was extracted from blood or bone marrow from patients with myeloid malignancies (n=1153). A NGS library was prepared and subjected to enrichment for genes recurrently mutated in myeloid malignancies using SureSelect (Agilent) capture probes. Sequencing was performed on Illumina platforms to identify cases that harbored FLT3-ITD mutations. The VAF of FLT3-ITDs and the insert size was evaluated using two informatics algorithms (Pindel and Scalpel) for detecting insertions and deletions (in/dels) in NGS data. FLT3-ITD positive cases were identified (n=63) and 47 were subsequently analyzed by CE and ARs were calculated. Among the 1153 total samples subjected to NGS, PCR/CE was performed on 77 consecutive cases to confirm that NGS did not miss FLT3-ITDs. Results: Initially using Pindel, we identified 63 cases that were positive for a total of 70 FLT3-ITDs by NGS. 21 of these cases were also subjected to analysis by Scalpel. Among the 77 consecutive cases analyzed by PCR/CE, 4 were FLT3-ITD positive (all confirmed by NGS with Pindel) and 73 were FLT3-ITD negative by PCR/CE (all also negative by NGS Pindel). ITD size by NGS and CE were in close agreement. All FLT3-ITD mutations detected by Pindel were confirmed by PCR/CE. However, Scalpel failed to detect 2 of 21 (n=9.5%) large (159 bp and 163 bp) FLT3-ITDs that were detected by Pindel and confirmed by PCR/CE. We observed that the NGS-derived VAFs from both Pindel and Scalpel were consistently lower than the corresponding ARs. This was the case for VAFs from both algorithms. Overall, Pindel showed better correlation with CE compared to Scalpel (r 2 =0.733 vs. 0.351; p=<0.0001). Conclusions: NGS is sensitive for the detection of FLT3-ITDs but the VAF may underestimate the overall FLT3-ITD mutation burden. We hypothesize that this may be due to loss of mutant reads in the informatics analysis, possibly during the alignment step. Introduction: Chromosomal abnormalities in myelodysplastic syndromes (MDS) and acute myeloid leukemias (AML) provide an important diagnostic and prognostic value, assisting in therapeutic choice. Unbalanced chromosomal abnormalities (chromosomes 5, 7, 8, 11, 13, 17, 20 and 21) are often related, but due to karyotype (KT) low resolution it may bypass clinically important smaller rearrangements. As a confirmation technique, FISH is performed for specific regions, however, it is expensive. MLPA, on the other hand, is less costly, evaluates more genomic regions and is performed with genomic DNA, simplifying biological material availability. The aim of this study was to correlate the results of unbalanced chromosomal abnormalities detection in MDS/AML patients by KT and MLPA. Methods: Twentynine MDS/AML patients were tested for KT and MLPA. G-banding KTs were performed in cultured cells, and divided in 3 subclasses: normal (n=5), 1-2 abnormalities (n=12) and complex (n=12). MLPA was performed with genomic bone marrow DNA using MRC-Holland kits P144 and P145. Outcomes were classified in: i) "concordant", if MLPA detected KT abnormalities; ii) "partially concordant", if MLPA detected previous KT abnormalities and extra anomalies; iii) "discordant", if abnormalities detected by KT were not reproduced by MLPA. Results: Overall failure rate for MLPA was 20.68%, which may be explained by DNA heterogeneity, as failures correlate with low DNA quantification. Normal KT was discordant with MLPA in one case (20%), in which MLPA detected a deletion of RUNX1 gene (chromosome 21), confirmed by FISH. Alterations in regions -5; del5q, -7, del7q, +8, del11q, del13q, del17p, del20p, del20q and del21q were most commonly detected. For KTs with 1-2 abnormalities, MLPA was partially concordant in two cases (22.2%): (i) it revealed that the translocation t(6;7)(q25;q22) was unbalanced, as a deletion was detected in region 7q22. 1, and (ii) it showed an amplification in 11q23.3 in addition to the loss of del(5)(q13q31). In complex KTs, MLPA was partially concordant in 3 cases (42.8%), and discordant in one case (14.2%), in which a deletion of 20q was not detected by MLPA. Overall ratios of concordance were: 68.1% concordant, 22.7% partially concordant and 9% discordant. Conclusions: MLPA is a powerful and less expensive technique for the detection of high resolution genomic unbalanced abnormalities in MDS and AML patients, frequently adding more information to KT results, especially in complex cases. Introduction: Acute lymphoblastic leukemia (ALL), the most common cancer in children, is a leading cause of cancer mortality among young adults, and portends a dismal prognosis for adults. Philadelphia-like (Ph-like) ALL is a recently described and high-risk subset constituting 10-15% of ALL whose hallmark features are 1) similarity by gene expression to BCR-ABL1+ ALL and 2) frequent activating ABLand JAK-family kinase fusions that are known or suspected to be responsive to FDAapproved tyrosine kinase inhibitors. However, current diagnostic algorithms for detecting these genetic lesions are multimodal, time-intensive, and not optimally designed to identify the diversity of fusions between kinases and partner genes present in this disease. Therefore, the development of a robust, timely, and efficient method for detecting known and novel actionable lesions in Ph-like ALL would address a major clinical need. Methods: In collaboration with ArcherDX, we designed the Heme Fusion Assay, a single-reaction RNA and DNA next-generation sequencing (NGS) assay targeting kinase fusions and point mutations known to jmd.amjpathol.org ■ The Journal of Molecular Diagnostics occur in Ph-like ALL, as well as genes important for the subclassification of ALL and myeloid neoplasms. This assay, composed of over 800 probes for loci in 58 genes, employs Anchored Multiplex PCR to target relevant kinase domains and identify fusions to known and novel partner genes. Fresh and archived tissue obtained from 25 B-ALL cases at MGH and DFCI/BCH were included in this study, using clinical molecular genetic data or transcriptome sequencing data as a benchmark. Sequencing libraries were constructed according to manufacturer specifications (Archer Universal RNA Reagent Kit v2 for Illumina-8), and 12-15 barcoded sample libraries were sequenced per Illumina MiSeq run. Novel fusion transcripts and gene expression findings were confirmed by RT-PCR. Results: In 25 cases of B-ALL with prior molecular characterization, the Heme Fusion Assay identified all known fusions (n=10, involving ABL1, CRLF2, JAK2, and PDGFRB) and point mutations (n=3, involving CRLF2, JAK2, and KRAS). Unexpected findings within this cohort included the identification and confirmation by RT-PCR of single cases revealing a novel TCF3-FLI1 fusion and markedly elevated PDGFRB mRNA expression without evidence of a fusion event, suggesting an alternate mechanism for kinase activation. Conclusions: Here we present a novel NGS assay designed for detection of kinase fusions and point mutations in Ph-like ALL. In validation cases benchmarked to RNA-seq or clinical molecular genetic data, the Heme Fusion Assay identified all previously detected kinase lesions. Novel findings within this cohort highlight the utility of this assay for targeted discovery in Ph-like ALL. Introduction: NPM1 mutation, typically involving a net insertion of four base pairs in exon 11, is a common molecular lesion in acute myeloid leukemia (AML) patients. Mutated NPM1 confers a favorable prognosis in cytogenetically normal AML patients who lack FLT3 internal tandem duplication. One-third to one-half of AML patients harbor an NPM1 mutation with the most common form being a TCTG insertion designated Type A that is seen in 80% of NPM1-mutated cases. Recent studies show that minimal residual disease (MRD) monitoring of AML patients after chemotherapy provides important prognostic information which is independent of other risk factors and may guide clinical decision making regarding hematopoietic stem cell transplantation. NPM1 mutation is a stable marker of disease in AML and represents a desirable target for MRD assay development. Real-time quantitative PCR is an established method well-suited for detection of low-level variants. Digital droplet PCR (ddPCR) is a relatively new technique that uses allele-specific PCR in single molecule oil-water emulsions to amplify target(s) of interest. Here we describe a comparison study intended to assess the feasibility of ddPCR in the detection of NPM1 Type A transcripts. Methods: This study was approved by the University of Utah Institutional Review Board. OCI-AML3 cells, expressing the NPM1 Type A mutation, were serially diluted 10-fold into Jurkat cells which express wild-type NPM1. RNA was isolated, reverse-transcribed into cDNA, and analyzed in a multiplex assay measuring NPM1 Type A with allele-specific primers and ABL1 transcripts with fluorescent-labeled probes for measuring product formation. For ddPCR, samples were analyzed on the RainDance RainDrop ddPCR system in singlet on two separate runs with varied reagent concentrations. Data was analyzed using RainDrop Analyst II software. For RT-qPCR, samples were analyzed on the Roche LightCycler 480 real-time PCR instrument and data analyzed using LightCycler 480 software. Results: We successfully detected NPM1 Type A transcripts in serial dilutions of OCI-AML3 cells using both RT-qPCR and ddPCR techniques. Preliminary results show that the sensitivity of RT-qPCR (0.01% to 0.001%) exceeds that of ddPCR by 1 log to 2 logs. Conclusions: Both RT-qPCR and digital droplet PCR are viable methods that may be used to detect NPM1 Type A transcripts in RNA samples derived from AML patients. Although ddPCR has been reported to be a highly sensitive technique for detection of low level variants, our preliminary experiments suggest that RT-qPCR remains the gold standard for sensitivity. Further optimization of the ddPCR technique is necessary to provide assay sensitivity at the level required for clinical MRD testing and these studies are ongoing. Introduction: Current comprehensive genetic testing of hematologic malignancies requires multiple testing strategies with high costs. Our previous proof-of-concept study demonstrated the feasibility of using the same NGS data to simultaneously detect both somatic mutations and copy number variants (CNVs) in the targeted genomic regions (Shen et al., 2016) . However, this targeted panel-based approach is limited in its ability to detect CNVs outside of the targeted regions. The limited number of SNPs in the targeted regions also prevented detection of copy number neutral loss of heterozygosity (CN-LOH). Here, we hypothesize that implementation of a genome-wide SNP sequencing backbone will allow for sensitive and comprehensive detection of genome-wide CNVs and CN-LOH by NGS for hematologic malignancies. Methods: This study was conducted in accordance with protocols approved by the University of Utah Institutional Review Board and the Declaration of Helsinki. Genomic DNA was prepared and sequenced as previously described (Shen et al., 2016) . For genome-wide CNV detection, we designed a SNP backbone with 22,762 SNP regions evenly distributed across the entire genome. These SNP regions were selected based on GC content close to 50%, unique mappability and population minor allele frequencies. The read depth of the SNP regions were normalized and converted to a log2 ratio as described in Shen et al., 2016. To detect CN-LOH, SNP genotypes and B allele frequencies were analyzed by UnifiedGenotyper (GATK version 3.4, Broad Institute, Cambridge, MA) . For detection of targeted somatic mutations, sixty-three frequently mutated genes with clinical significance in hematologic malignancies were sequenced simultaneously with the SNP backbone. Results: The normalized read depth of regions in the SNP backbone was highly consistent across different normal reference samples, which allowed for the detection of genome-wide CNVs. The B allele frequencies were accurate for CN-LOH detection. In the limited number of positive cases sequenced thus far, we have detected genome wide CNVs and CN-LOH by NGS in complete concordance with SNP microarray. Conclusions: Our on-going study suggests that the strategy of combining the SNP backbone and targeted mutation NGS panel allows for comprehensive genetic profiling of CNV, CN-LOH and somatic mutations in hematologic malignancies using a single assay. Introduction: Recent advancements of technologies including next-generation sequencing (NGS) have led to discovery of molecular pathogenesis of malignant diseases including hematologic malignancies, and these discoveries are now enabling the beginning of molecular targeted therapy. The MLL-rearranged fusion gene is among the main leukemogenic mutation which is found in both acute myeloid leukemia (AML) and acute lymphoblastic leukemia with a frequency of about 5-10%, and is associated with a poor clinical prognosis. Cooperative mutation study among MLL-rearranged leukemia based on NGS technology has mainly been conducted in Caucasian or western population, and has yet been performed in Asian ethnicity including Korean population. Methods: This study includes total of 24 MLLrearranged AML patients who visited two separate tertiary care hospitals between the period of January 2009 and May 2014. The number of each MLL fusion genes are as follows; MLL/MLLT3 n=12; MLL/MLLT4 n=6; MLL/ELL n=2; other MLLfusion genes n=4. Mutation profile study for 19 candidate genes for cooperative mutation (TET2, DNMT3A, IDH1, IDH2, NPM1, FLT3, CEBPA, ASXL1, BRAF, CBL, KIT, KRAS, NRAS, PTPN11, RUNX1, TP53 , WT1, SETD2, JAK2) was performed using Miseq sequencing equipment (Illumina, San Diego, CA). Results: Among the 24 MLL-rearranged patients, 7 patients (29.2%) had positive results for more than 1 gene mutations which were analyzed for. Positive gene mutations found were in the order of frequency; ASXL1 (n=4), FLT3 (n=2), CEBPA (n=2), KRAS (n=1), NRAS (n=1) andPTPN11 (n=1). Interestingly, 4 out of 6 patients harboring MLL/MLLT4 were found to have additional gene mutations. Two patients had multiple gene mutations whereas one patient had four gene mutations concurrently detected. Conclusions: This is the first cooperative mutation study on MLLrearranged AML patients of Asian ethnicity using targeted NGS technology. Despite small number of cases, a higher incidence of gene mutation among MLL/MLLT4 patient group (4/6) and relatively higher incidence of ASXL1mutation (4/24) being found could be carefully suspected as an ethnical difference in the disease of AML. Further study requires a larger number of MLL-rearranged MLL patients and additional mutation profile study with ethnical comparison. Resistance in CLL J. Lee, S. Sharma, N. Galanina, A. Guo, S. Kadri, C. Van Slambrouck, B. Long, W. Wang, M. Ming, L.V. Furtado, J.P. Segal, W. Stock, G. Venkataraman, W. Tang, P. Lu, Y. Wang University of Chicago, Chicago, IL. Introduction: Ibrutinib (ibr), a first-in-class Bruton tyrosine kinase (BTK) inhibitor, has demonstrated high response rates in both relapsed/refractory and treatment naïve chronic lymphocytic leukemia (CLL). However, about 25% of patients discontinue ibrutinib therapy at a median follow-up of 20 months, usually due to leukemia relapse or progression. Treatment options for these patients are limited and outcomes are dismal with mortality rate exceeding 75% and a median survival of 3 months. Our groups and others have identified the first mutation that drives ibr resistance, BTK C481S. We have also shown that the mechanism of resistance is through disruption of the covalent binding between ibr and BTK at the C481 residual. Recently, additional BTK mutations (C481F/Y/R, T474I, and L528W) associated with ibr resistance have also been reported in ibr refractory cases. All the BTK mutations identified so far are located within the kinase domain of the protein and are predicted to weaken or hinder ibr docking to the BTK ATP-binding pocket. Here we report a CLL patient who relapsed on ibr and evolved a structurally novel BTK mutation that is outside of the kinase domain. Methods: Serial samples were collected from a Richter transformed CLL patient who were treated with ibrutinib, responded and then relapsed. The samples were analyzed using Onco1K, a 1,200-gene next-gen sequencing panel with an average sequencing depth of 420x. The uncovered novel mutation was then validated with Sanger sequencing and characterized with structural modeling. The role of the mutation was further functionally defined with cell transfection followed by assays for cell-proliferation, BrdU-incorporation, and intracellular B-cell receptor signaling. Results: A structurally novel mutation of BTK was identified which was associated with CLL relapse. Functionally, cells carrying the novel mutant BTK show resistance to ibrutinib at both cellular and molecular levels to a similar extent as BTK C481S . Conclusion: Our study lends further insight into the diverse mechanisms of ibrutinib resistance that has important implications for the development of next-generation BTK inhibitors as well as mutation detection in relapsed patients. H25. Optimised, One-Day Hybridization-Based NGS Protocol Yields 1% Sensitivity in MPN Samples, as Quickly and Cost-Effectively as Multiplex PCR G. Speight 1 , E. Chin 2 , L. Georgieva 3 , D. Cook 3 1 Oxford Gene Technology, Begbroke, Oxford, England; 2 Oxford Gene Technology, Tarrytown, NY; 3 Oxford Gene Technology, Oxford, Oxfordshire, England. Introduction: Myeloproliferative neoplasms (MPNs) are a group of diseases that affect normal blood cell production in the bone marrow resulting in the overproduction of one or more cell types. The key MPN driver mutations involve the JAK2, MPL and CALR genes, namely JAK2 V617F, which has an occurrence of 50% to 98% depending on the MPN sub-type; JAK2 exon 12; MPL W515K/L; and CALR exon 9 indels. The SureSeq Core MPN Panel allows NGS-based detection of the above driver mutations and is designed for research into the diagnosis, aetiology and prognosis of MPNs. The aim of this study is to evaluate the SureSeq Core MPN Panel in conjunction with a new streamlined hybridisation-based NGS library preparation protocol. The protocol offers a similar turn-around time to amplicon-based protocols, without the associated disadvantages, such as PCR bias, allelic bias (indels) and drop-outs, as well as poor uniformity of coverage. Methods: The SureSeq Core MPN Panel was validated with the JAK2 V617F Genotyping Sensitivity Panel from the National Institute for Biological Standards and Control (NIBSC) to confirm the lower levels of analytical sensitivity and confidence. The panel was then used to confirm a broader set of variants in 14 clinical samples containing variants for each of the targeted regions, in particular the indels associated with exon 9 of the CALR gene. Sequencing was conducted on a MiSeq (Illumina). Results: The SureSeq Core MPN Panel was shown to accurately detect alleles down to 1% variant allele fraction (VAF) in JAK2 (V617F) at a read depth of >1000x. Analysis in clinical research samples has shown that the panel was able to reliably detect not only single nucleotide variants but also 5 bp insertions in JAK2 (exon 12) and deletions of up to 52 bp in CALR (exon 9). In this region, the uniformity of coverage of the panel allowed key CALR indels to be identified, facilitating MPN stratification. The introduction of an alternative enzymatic fragmentation step and a rapid hybridisation step of just 30 minutes has reduced the overall length of the protocol by almost 6 hours, offering a streamlined, single-day sequencing workflow from DNA sample to sequencer. Conclusions: To detect alleles that contribute only a small percentage of the reads at any locus, a highly uniform and sensitive enrichment is required. We have utilised an optimised 1-day protocol in combination with the SureSeq Core MPN Panel to reliably and routinely detect somatic mutations down to 1% VAF. Researchers can now generate hybridisation-quality NGS data as quickly and as cost-effectively as multiplex PCR. P. Velu, C.C. Hsiung, K. Salafian, S. Luger, A. Bagg, J. Morrissette University of Pennsylvania, Philadelphia, PA. Introduction: Routine use of clinical cancer next-generation sequencing (NGS) in diagnosis and disease monitoring has resulted in sequential mutation profiles of individual patients. Clonal myeloid malignancies such as AML are often characterized and treated based on specific mutation profiles identified at diagnosis. On NGS profiling at interval follow-ups, mutations identified at diagnosis may change, or shift, through disease course. Here we track the AML mutational landscape in the context of therapy, disease progression, clinical course, and pathology. Methods: Patients whose blood or marrow specimens were sequenced using the Penn hematological-NGS panel and found on at least 2 occasions to have mutations were included. An R script was written to track mutations and allele frequencies over time. These data were integrated with corresponding pathology and clinical data to compare mutation timelines to disease progression and therapy timelines. The institutional review board approved this study. Results: Review of clinical NGS data identified 92 AML patients with samples characterized as de novo, recurrent, transformed from MDS, or in remission. NGS data revealed high variability in mutation profiles between patients, but also specific types of mutational shifts between diagnosis, recurrence, and remission of AML. These included no change, gain, loss, and all new mutations. There were several mutations that shifted rarely or never during disease course. Tracking of allele frequencies (>10% change) revealed changes in populations of multiple subclones at different time points. At recurrence, many patients had increased allele frequencies in mutations that were previously present at low levels and/or displayed completely new mutations. Decrease in allele frequency or loss of NPM1 and FLT3 mutations found at diagnosis did not necessarily correlate with remission, as this was associated with the emergence of pathogenic previously low level or undetectable subclones with different mutations. Loss of FLT3 mutations was often associated with FLT3-inhibitor therapy and emergence of an NRAS mutation at relapse. In patients with an initial NPM1 mutation, half lost that mutation but had new NPM1 mutation(s) at recurrence. Conclusions: Monitoring of AML patients by NGS rather than single gene assays is crucial due to the frequent finding of mutational shift and emergence of new disease associated clones at recurrence. Diagnostic mutational profiles can change with divergent allele frequencies, suggesting multi-clonality. The mutations that rarely or never shift are early or initiating mutations. Replacement of FLT3 mutations with NRAS mutations in patients on FLT3 inhibitors is suggestive of AML dependence on mutations in these kinase-signaling pathways. H27. Analytical Evalutation and Applications of the nCounter Vantage RNA:Protein Immune Cell Profiling Panel N. Riccitelli, A. Beams, I. Summit, R. Pollner Genoptix, Carlsbad, CA. Introduction: The NanoString RNA:Protein panel offers the ability to simultaneously examine the expression of 770 RNA targets and 31 proteins in a single reaction. Herein, we examine the analytical validity of the RNA:Protein Immune panel in fresh and frozen human peripheral blood mononuclear cell (PBMC) isolates from healthy donors, characterized cell lines, and chemically-induced samples. Furthermore, we outline a data analysis strategy to maximize interpretable information from low input or poor quality samples. Overall, the RNA:Protein panel exhibits exceptional accuracy and precision, providing an unparalleled level of information simultaneously from precious patient samples. Methods: RNA and protein expressions of both fresh and frozen, as well as phytohaemagglutinin-induced and non-induced human PBMCs, commercial PBMCs, and cell lines were measured using the nCounter Vantage RNA:Protein Immune Cell Profiling panel. Protein fold-change levels observed with the RNA:Protein panel were correlated with flow cytometry measurements for confirmation of results. Correlations between RNA and protein of induced PBMCs were evaluated for consistency with established data. NanoString measurements were compared across multiple runs for precision assessment. A relative count system was established to allow for normalization of RNA and protein expression. Results: Robust correlations were observed between flow cytometry results and NanoString protein fold-change measurements. For induced PBMCs, protein expression by NanoString was increased for expected markers, and correlated well with the respective mRNA expression. Negative and positive control data (n>30) indicate a high degree of reproducibility for the NanoString protein workflow while maintaining exceptional accuracy across independent sample preparations. Conclusions: By combining RNA and protein analysis into one straightforward workflow, the NanoString RNA:Protein panel maximizes the utility of frequently limited sample amounts. Furthermore, robust detection of protein targets was observed for both fresh and frozen PBMC sample types at clinically relevant sample inputs. Whereas the single-cell resolution of flow cytometry will remain critical in many applications, the ability to multiplex over 30 protein targets, as well as hundreds of mRNA analytes, from a single blood draw using the NanoString RNA:Protein panel offers tremendous advantages to researchers and clinicians looking to obtain actionable data and patients seeking to minimize medical visits. Lymphoproliferative Disorders by Massively Parallel Sequencing P. Szankasi 1 , A. Bolia 1 , E.P. Gee 1 , J.A. Schumacher 1 , J.L. Patel 2 , T.W. Kelley 2 1 ARUP Laboratories, Salt Lake City, UT; 2 University of Utah, Salt Lake City, UT. Introduction: Balanced translocations occur in many malignancies and lead to the deregulation of one of the partner genes. Typically, these rearrangements lead to fusion transcripts that can be detected with high sensitivity by reverse transcription-PCR. B-cell lymphoproliferative disorders (LPDs) often harbor translocations between the strong enhancers of the immunoglobulin (Ig) loci and oncogenes, such as BCL2, or MYC. This results in over-expression of the oncogene without the formation of an abnormal fusion transcript. These events are more difficult to detect by classical PCR assays due to the wide distribution of breakpoints over hundreds of kilobases. Here we present a method employing solution capture and massively parallel sequencing that allows comprehensive detection of translocations in DNA isolated from FFPE tissue from B-cell LPDs. Methods: This study was approved by the University of Utah IRB. Genomic DNA was isolated from FFPE specimens and subjected to NGS library preparation and target enrichment using SureSelect capture (Agilent). Capture probes were targeted to all known RSS sites flanking the V, D and J elements of the IGH, IGK and IGL loci, the IGH class switch regions, the IGK deleting elements, and known breakpoint clusters in BCL2, BCL6, MYC, CCND1, MALT1, and AIP2/BIRC3. In addition, 19 genes commonly mutated in B-cell LPDs were captured for mutation analysis. Paired-end sequencing was performed on Illumina instruments. Reads were aligned to the reference genome using BWA. Translocations were identified using DELLY software. Point mutations were identified by the program LOFREQ. Results: We tested 27 patient samples with diagnoses of diffuse large B-cell lymphoma, follicular lymphoma or Burkitt lymphoma, and 10 cell lines. We were able to detect translocations confirmed by other validated methods (FISH and PCR) including t(14;18);IGH-BCL2 (10/10), t(3;14);IGH-BCL6 (3/3), t(11;14);IGH-CCND1 (2/2) and translocation involving MYC (5/6). One discrepant jmd.amjpathol.org ■ The Journal of Molecular Diagnostics sample had very low positivity by FISH. We also detected translocations between BCL6 and 6 non-Ig partner genes. Capture of only one partner gene was sufficient for successful detection of translocations. This allowed detection of translocations with a breakpoint several hundred kb away from MYC and between BCL6 and novel, non-Ig partners. Dilution studies with 4 cell lines indicated that translocations could be detected down to 5% positive cells. Conclusions: We describe a method for comprehensive detection of mutations and translocations using massively parallel sequencing of genomic DNA from FFPE samples. The method allows concurrent detection of multiple translocations in the same sample at 5% positive cells and represents a valuable adjunct to the workup of B-cell LPDs. The ability to detect cancer specific fusion genes is important not only in cancer research, but also increasingly in clinical settings to ensure that the correct diagnosis is made and optimal treatment is chosen. Current methodologies to detect such fusion genes are laborious and time consuming, whereas near patient/point of care testing (POCT) offers a more efficient approach to leukemia treatment and management. Here we report a feasibility study to develop a rapid real-time PCR (RT-qPCR) assay that can detect leukemia fusion genes with high sensitivity and specificity. Methods: The QuanDx ALL Q-Fusion Screening Kit is a novel multiplex RT-qPCR assay for the simultaneous detection of 16 fusion genes with 71 breakpoints. Fusion and control genes are co-amplified and identified with specific Yin-Yang Probes, a novel technology for nucleic acid detection and related applications. The Streck Philisa Thermal Cycler is an innovative high speed PCR instrument designed to simplify and expedite PCR-based applications. An armored RNA positive template control and GUSB internal control were reverse transcribed and used along with a negative template control in both the standard QuanDx RT-qPCR protocol as well as a Streck modified fast protocol to evaluate the feasibility of developing a rapid screen for ALL. These results were then compared with additional data from the same protocol using p190 and p210 DNA plasmids. The total run times and Cq values were evaluated. Results: No significant Cq differences were observed with the Streck modified protocol compared with the QuanDx standard protocol. The standard protocol run time of 140 minutes was reduced to 77 minutes with the modified fast protocol, or 55% of the original run time. The reduction in amplification time is due to the ramp rates of the Philisa instrument, as well as a reduction of touchdown PCR denature and extension hold times. Conclusions: The successful feasibility study for running a rapid multiplex PCR leukemia test with the QuanDx ALL Q-Fusion Screening Kit combined with a modified fast protocol on the Streck real-time Philisa instrument shows the potential for developing near patient/POCT leukemia screening. Rapid diagnosis of ALL and other leukemias would allow for immediate health counseling and treatment options to be put in place for the patient. Future studies are planned for rapid AML screening, and to further reduce the time to results. H30. The ABL and JAK Tyrosine Kinase Pathways Are Co-Activated in BCR/ABL1 Positive Acute Lymphoblastic Leukemia S. Bhagavathi, H. Aviv Rutgers University, New Brunswick, NJ. Introduction: In newly discovered Philadelphia-like ALL or (Ph-like ALL) lacks the BCR/ABL1 fusion, driving event in chronic myeloid leukemia (CML), a subset of acute myeloid leukemia (AML) and acute lymphoid leukemia (ALL). It is characterized by activation of signaling Tyrosine Kinases (TK) and can be responsive to TK inhibitors, similar to the BCR/ABL1 fusion protein. We explored the possibility that ALL with the BCR/ABL1 fusion harbors other driving mutations in the TK signaling pathways. To this end, we studied patients with BCR/ABL1 positive ALL for the co-incidence of other mutations that can be present in the JAK signaling pathways. Methods: From December 2013 to November 2014, all cases positive for BCR/ABL1 by fluorescent in-situ hybridization (FISH) in our database were identified and subjected to FISH analysis using the IGH dual color, break apart probe. Three cases negative for the IgH rearrangement were further analyzed with CRLF2 and JAK2 probes. Results: A total of 18 patients identified; 9 ALL, 7 had CML and 2 AML. Most ALL patients showed complex hyperdiploid karyotypes. The BCR/ABL1 fusion protein was p190 KDa and p210 KDa in all AML and CML patients. Of the 9 ALL patients; 6 showed p190 KDa fusion protein, 2 patients had p210 KDa fusion protein and one patient had no detectable BCR/ABL1 fusion. Whereas no CML or AML BCR/ABL1 positive case showed separation of the IGH gene probe, 6 of 9 ALL patients (4 p190 and 2 p210) demonstrated IGH probe separation. The IGH partner chromosomes were identified in 50% of patients as t(Y;14), t(X;14), and t(14;20), 1 patient each. Hybridization with probe for the CRLF2 gene on the remaining 3 IgH rearrangement negative patients demonstrated the CRLF2/P2RY8 gene fusion in one patient. Two of the 9 patients were negativefor IgH, CRLF2 and JAK2 fusions. Conclusion: We discovered that 6 of 9 Ph positive ALL cases harbor concomitant IGH translocations. The partner gene of these translocations was mostly the CRLF2 gene that resides in the pseudoautosomal region of the sex chromosomes. One additional patient had a CRLF2 and P2RY8 gene fusion. The CRLF2 gene is responsible for activating the JAK pathway of tyrosine kinase signaling. We postulate that ALL with the 9;22 translocation is less responsive to tyrosine kinase inhibitors because in addition to the ABL signaling pathway activation, the JAK signaling pathway is also activated and only a combination of drugs that inhibit both pathways will be effective in stopping progression of ALL with BCR/ABL1 fusion. J.T. Brown, I. Beldorth, W. Laosinchai-Wolf, M. Fahey, J. Hedges, B. Andruss Asuragen, Austin, TX. Introduction: Detection of BCR-ABL1 e13a2/e14a2 fusion transcripts (major breakpoint, M-BCR) of t(9;22) assesses tumor burden in CML. The International Scale (IS) standardized reporting against a common baseline. As newer TKI therapies create deeper responses, analytical sensitivity has become a critical topic in investigations into TKI discontinuation, where researchers require an assay that MR4.5). This has led to various reporting formats, creating non-contiguous monitoring terms: baseline, 10%IS, 1%IS, MMR, MR4, and MR4.5. We describe analytical and clinical validation of a multiplex system reporting continuous MR values via automated analysis, clinical accuracy, analytical sensitivity of MR4.7, and direct traceability to the WHO Primary BCR-ABL1 materials without requiring establishment and revalidation of a conversion factor. Methods: Armored RNA Quant molecules form a blend of nuclease-resistant BCR-ABL1 and ABL1 transcripts to calibrate and control the system. A single 4-point curve using ARQ blends mimics the WHO Primary BCR-ABL1 reference materials and accounts for the relative batch run-specific efficiency of the RT step. cDNA generation and qPCR were optimized, including allowance of high mass of nucleic acid without inhibition. Residual clinical RNAs were tested to estimate LOD at minimum RNA input. Software includes a floating, traceable logic algorithm to ensure that sufficient ABL1 was detected to protect this LOD. A multi-site clinical outcome study included 139 samples. Monitoring was assessed by EFS at 32-40 months against test results at 12-18 months on TKI as estimated by the Kaplan Meier survival function. Results: We generated 1680 measurements across 28 low analyte levels, yielding an LOD estimate of 95% positivity of MR4.7 (0.002%IS). LOQ was similar. Despite deep analytical sensitivity, this system maintains analytical specificity (true negative). Linearity was at least MR0.3 (50%IS) to MR4.7 (0.002%IS). Single-site precision (lot, -site was demonstrated. The EFS difference between Conclusions: The BCR-ABL1 test improves workflow with its streamlined reagent formulation, multiplex assay format, and automated software analysis. It facilitates assessment on the IS without conversion (via integrated ARQ materials traceable to the WHO Primary), reports results on a continuous scale (as both MR and %IS values), and quantifies deep molecular responses. It is clinically validated at MR3 using an EFS model. Introduction: Gene fusions resulting from chromosomal translocations play an important role in driving tumorigenesis in hematologic malignancies, and provide critical diagnostic and prognostic information. Real-time PCR (RT-PCR) or Fluorescent in-situ hybridization (FISH) can be used to detect fusions, but these methods are limited to known fusion gene partners. Advances in next-generation sequencing (NGS) are removing the technical challenges to detecting these critical gene fusions. Anchored Multiplex PCR (AMP) uses molecular barcoded adaptors for universal priming and gene-specific primers, enabling sensitive NGS-based detection of novel fusions from low-input samples. We evaluated the AMP-based Archer FusionPlex system for NGS-based detection of gene fusions in hematological malignancies through identification of fusion breakpoint sequences. Methods: RNA was evaluated from 17 hematologic malignancy blood or bone marrow samples with known or suspected gene fusions previously analyzed by FISH, karyotyping, and/or RT-PCR. All samples were run on the Archer FusionPlex Heme v1panel. Sequencing was done on the Illumina NextSeq 500. All samples were analyzed by ArcherAnalysis v3.0 software. Results: Sequencing detected gene fusions in a total of 14 of the 17 samples. Five of these detected fusions were consistent with RT-PCR results, and 6 were consistent with dual-fusion FISH results. Importantly, sequencing enabled the identification of fusion partners in 3 samples where break-apart FISH probes detected a likely gene rearrangement event, but without information as to the partner gene. Fusions were not detected by NGS in 3 samples. The first of these sequence-negative samples had a very low level BCR-ABL1 e1a2 minor translocation (0.47% by RT-PCR), which is below the detection limit of NGS. Two additional sequence-negative samples had FISH and karyotyping results suggesting a chromosomal translocation, but the gene partners were not defined, and a gene fusion event could not be confirmed by sequencing using the original reagent panel. However, an updated prototype version of the fusion panel enabled detection of a novel KMT2A-MLLT4 fusion in one of these initially false-negative cases and suggested a KAT6A-CREBBP fusion in the other case. These latter two fusion events were not initially detected because these genes were not included in the first version of the panel. However these genes are now included in the updated version of this fusion gene panel, FusionPlex Heme v2. Conclusions: The Archer FusionPlex Heme panel v1 performs well and provides a streamlined workflow for successful NGS-based detection of gene fusions. Improvements in target gene coverage are ongoing based on the ever-changing knowledge base of gene fusions. Introduction: Mastocytosis is a clonal proliferation of mast cells accumulating in one or more organ systems. Systemic mastocytosis, which includes involvement of the bone marrow, can be either indolent or aggressive based on the presence or absence of "C-symptoms." There is currently no cure for systemic mastocytosis. Mastocytosis is a systemic disease, and the microenvironment shaped by mast cells may differ in many aspects from that of anatomically localized, solid malignancies. Whereas those with indolent systemic mastocytosis (ISM) have a normal life expectancy, those with aggressive systemic mastocytosis have historically survival only a few years. We sought to investigate potential biomarkers for both differentiating between ISM and ASM as well as identify potential therapeutic targets. Methods: Archival blocks with slides were retrieved, reviewed and clinical information obtained from patient charts. re used with >50% lesion, n=4 (ASM), and n=3 (ISM). RNA was extracted from pre-treatment FFPE bone marrow slides and analyzed with nanoString nCounter using the nCounter PanCancer Progression Panel designed to quantitate 770 genes from four major biological processes that contribute to increased tumor growth and aggressiveness. Results: The gene expression profiles of ASM versus ISM, show significant (>3 times, p<0.05) upregulation of a set of 4 genes (FN1, SFRP2, PTGS2, and POSTN) and down regulation of 4 genes (TMPRSS6, EPCAM, IFNG, and CLDN3). Interestingly, gene FN1 was upregulated 6.5X (p< 0.0047) in ASM as compared to ISM. Of the four ASM patients, three had an associated hematological clonal non-mast cell disorder (AHNMD) as identified by biopsies performed at our institution. All ASM patients demonstrated significant weight loss and palpable splenomegaly with evidence of hypersplenism. Two patients also showed hepatomegaly with liver dysfunction and osteolytic lesions on skeletal survey. Of the IDM patients, none demonstrated any C-symptoms. Conclusions: To the best of our knowledge, the current study represents the first gene expression studies profiling exploration of aggressive Mastocytosis. Of the upregulated genes, two play a role in wound healing and fibrosis including FN1 (the most significantly upregulated gene) and POSTN. Three of the downregulated genes (TMPRSS6, EPCAM, and CLDN3) encode transmembrane proteins, two of which (EPCAM and CLDN3) participate in cell-cell adhesion. FN1 is upregulated in many cancers, associated with aggressive phenotype in ovarian cancer, and is associated with poor survival in breast cancer. This work is intriguing for the new information it provides about the possible role of FN1 in mast cell migration and aggressiveness in Aggressive Systemic Mastocytosis. H34. Development of a PCR Assay for Detection of JAK2 Exon 14 (V617F), JAK2 Exon 12 and CALR Exon 9 Mutations in Myeloproliferative Disorders D.C. Maxwell, S. Grenier, B. Nwachukwu, M. Jensen, W. Morrison, T.L. Stockley, S. Kamel-Reid, C. Wei Toronto General Hospital, University Health Network, Toronto, Ontario, Canada. Introduction: Myeloproliferative neoplasms (MPNs) are clonal stem cell disorders characterized by proliferation of myeloid lineages. The classic non-CML MPNs include essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (PMF). JAK2 exon 12 and 14 mutations are frequently observed in PV, and mutations in CALR are seen in 67% of ET cases and 88% of PMF cases respectively. We developed a multiplex molecular assay to detect JAK2 exon 14 V617F, JAK2 exon 12 K539L and indels, and CALR exon 9 indels in a single test to expedite the diagnosis, prognosis and management of myeloproliferative disorders. Methods: Genomic DNA extracted from peripheral blood (Qiagen M48) was used in multiplex PCR using custom primers and ABI AmpliTaq Gold PCR mastermix. PCR fragments were analyzed by capillary electrophoresis on an ABI 3500 sequencer. Custom primers were labeled with FAM fluorescent label (blue) for JAK2 exon 14 and CALR mutations, and VIC (green) for the JAK2 exon 12 mutation. Limits of detection were established at 5% for JAK2 exon 12 and CALR and 1% for JAK2 exon 14. Expected normal alleles sizes were 380bp to 381bp for CALR, and 280bp to 281bp for JAK2 exon 12, with mutant indels sizes occurring outside of the normal values. The expected normal allele size of JAK2 exon 12 K539L was 135bp to 136bp and 130bp to 131bp for the mutant allele. The expected normal allele size for JAK2 exon 14 V617F was 98bp and 103bp for the mutant allele. Results: Validation included 25 cases which were previously tested for JAK2 exon 12 mutation by Sanger sequencing and for JAK2 exon 14 V617F mutation and CALR exon 9 indels by multiplex PCR based fragment analysis. The results using the mulitplex assay showed 100% concordance with previous results. The inclusion of JAK2 exon 12 primers to the assay now allows 97% of reported JAK2 exon 12 mutations to be detected. The remaining 3% that are not detected are substitutions, which are either clinically non-significant or synonymous in COSMIC. We identified the co-existence of double mutations in JAK2 exon 12 and exon 14, which would be missed if algorithms are confined to JAK2 exon 14 testing only. Bleed through of FAM and VIC dyes may occur when peak height is exceedingly high, leading to interference with variant calling, so other complementary methodologies such as Sanger sequencing and fragment analysis are employed for further clarification in these scenarios. Conclusions: We present data summarizing development of a comprehensive multiplexed assay for use in MPN's. This type of simple multiplexed assay can be used as an alternative to other methods such as next-generation sequencing, when a shorter turnaround time is required. Multiplex Ligation-dependent Probe Amplification (MLPA) M.S. Hein, K.C. Swanson, P. Lundquist, L. Coon, D. Dawson, J.L. Oliveira, J.D. Hoyer Mayo Clinic, Rochester, MN. Introduction: Beta globin cluster large deletions are heterogeneous and complex. They are classified into groups with important phenotypic differences when present in combination with another beta cluster mutation. These classifications are: 1) epsilon gamma delta beta thalassemia (EGDBT); 2) gamma delta beta thalassemia (GDBT); 3) hereditary persistence of fetal hemoglobin (HPFH); 4) delta beta thalassemia (DBT) and beta thalassemia (BT). High genetic diversity in this genomic region decreases the applicability of routine GAP-PCR methodologies, as novel or less common mutations will be missed. We compared a higher resolution MLPA probe strategy to published deletion breakpoint data to assess improvement in the classification power of our test. Methods: Thirty five probe sets cover the beta globin gene locus and span from 29 kb upstream of the -locus control region (LCR) to 110 kb downstream of the 3' hypersensitivity region (3'HS-1). Probes 1 to 7 cover the LCR and -globin gene (HBE), 8 to 13 the -globin genes (HBG1& HBG2), 14 to 20 upstream and in the -globin gene (HBD), 21 to 35 encompass the -globin gene (HBB) and 3'HS-1. A gene dosage ratio was calculated using the median fluorescent intensity value (Luminex LX200). Case file review identified 301 routine clinical cases with large deletions. Higher resolution MLPA data was applied and phenotypically confirmed. We classified the cases into the 5 groups and projected our probe data to coordinates of 48 defined deletions available in the published record. Results: The 301 cases were able to be classified as follows: EGDBT (n=7), GDBT (n=34), DBT (n=97), HPFH (n=101), BT (n=52). Six cases could not be classified due to ambiguous phenotypes. In addition, our assay showed capacity to further differentiate specific deletion subtypes in 4 /4 EGDBT, 7/8 GDBT, 7/11 HPFH, 7/11 DBT and 8/9 BT cases studied and the ability to assign the Corfu delta gene deletion in 3/3 (100%) cases and identified crossover variants Hb Kenya and Hb Lepore (3 unique subtypes). Conclusions: Accurate classification is critically important in large beta cluster deletions due to the variance in protective phenotype among these groups of mutations. Identifying specific classification subtypes is interesting but less important clinically. By our higher resolution strategic MLPA assay alone we were able to classify deletions in 295/301 (98%) of our clinical cases and further subtype 37/48 (77%) according to published data. It is important to note that although it is possible to classify many deletions using this assay, many disorders have multiple mutations and correlation of phenotypic data remains crucial for accurate assessment of hemoglobin disorders. J. Gregg, M. Chen University of California Davis Medical Center, Sacramento, CA. . Introduction: Hairy cell leukemia (HCL) is a low grade B-cell malignancy with a distinctive immunophenotype. Purine analog therapy is highly effective, with most patients achieving durable complete remissions. The somatically acquired V600E mutation of the BRAF gene has recently been described as a molecular marker of HCL to distinguish it from other B-cell lymphomas with similar clinical and morphologic features, such as the HCL-variant and splenic marginal zone lymphoma. The rare aggressive cases of HCL that are resistant to standard therapy with BRAF oncoprotein inhibitors have prompted the search for additional new HCL associated gene mutations. Case report: Here we report a 67 year-old male with a ten-year history of recurrent HCL. A bone marrow biopsy and flow cytometry analysis revealed prototypical morphological and immunophenotypical features of HCL. The patient was resistant to conventional treatment with cladiribine, pentostatin, rituximab, and splenectomy. Computed tomography showed extensive lymphadenopathy involving cervical, mediastinal, portal hepatic, retroperitoneal, and pelvic lymph nodes. Right cervical lymph node biopsy showed persistent HCL involving the lymph nodes. No evidence of BRAF V600E gene mutation was detected by a genotyping qPCR assay for BRAF codon 600 mutations on formalinfixed, paraffin-embedded (FFPE) tissue. Methods: QClamp is a novel technology to screen for somatic mutations, which utilizes the sequence specific wild-type template xenonucleic acid "Clamp" to suppress amplification of wild-type DNA and selectively enhance mutant template amplification. We applied the DiaCarta QClamp BRAF codon-specific test kit to analyze the FFPE tissue, which resulted in indeterminate call. The BRAF exon 15 amplicons from QClamp qPCR reactions performed on DNA samples extracted from four separate FFPE slides of the cervical lymph nodes were sequenced by the conventional Sanger method. Results: All four jmd.amjpathol.org ■ The Journal of Molecular Diagnostics sequences are identical, which contain no mutation in V600 locus, but have a silent mutation in position A598 with nucleotide change from GCT to GCC (both coding for Alanine). There is also a c.1807C>T (substitution C to T in the R603* codon leading to a nonsense mutation from Arginine (CGA) to a stop codon (TGA). Conclusions: This study reports new BRAF mutation variants distinct from the common V600E in HCL, which is clinically more aggressive and resistant to the conventional therapy for HCL. These findings suggest possible existence of new BRAF mutations causing resistant HCL with prognostic implications. The genetic study of more cases of BRAF V600E negative HCL cases is necessary to further investigate the new BRAFmutants and their potential therapeutic application. S. Loghavi, M.J. Routbort, R. Kanagal-Shamanna, S.A. Wang, J.D. Khoury, C. OK, C.C. Yin, R.R. Singh, Z. Zuo, C.E. Bueso-Ramos, L. Medeiros, R. Luthra, K.P. Patel University of Texas MD Anderson Cancer Center, Houston, TX. Introduction: Guanine Nucleotide Binding Protein Alpha Stimulating Activity Polypeptide (GNAS) is a component of the G signaling pathway. GNAS mutations (codon 201) have been reported in solid tumors. High-throughput sequencing of myelodysplastic syndromes (MDS) have revealed recurrent GNASmutations in <1% of cases, but information about GNAS mutations in other myeloid neoplasms remains largely lacking. Methods: We searched our archives for myeloid neoplasms with GNAS R201 mutations. Cases were classified according to the World Health Organization 2008 criteria. Conventional G-band karyotype analysis was performed. Mutation analysis was performed using DNA extracted from BM aspirates using next-generation sequencing-based mutation analysis of 53-gene hotspot genomic loci (Illumina, San Diego CA). Clinical and laboratory data were obtained from electronic medical records. Results: We identified 16 patients with GNAS p.R201 mutations including 10 men and 6 women with a median age of 67 years at diagnosis (range, 49-79). Cases were subclassified into primary myelofibrosis (PMF), n=4; refractory cytopenia with multilineage dysplasia, n=3; acute myeloid leukemia (AML) with myelodysplasia-related changes, n=3; refractory anemia with excess blasts-2, n=1; AML-not otherwise specified, n=1; therapy-related AML, n=1; blast phase of PMF, n=1; and chronic myelomonocytic leukemia, n=1. One case was a remission bone marrow in a patient with history of acute promyelocytic leukemia. The median bone marrow blast count was 7% (range, 0-79) and the median bone marrow cellularity was 75% (range, 20-100). Fourteen cases showed the p.R201H substitution, 2 cases showed p.R201C and 1 case had both in mutually exclusive alleles. The mutation was present at a median allelic frequency of 0.26 (range, 0.01-0.53). The most common co-mutated genes were JAK2 p.V617F (n=3); TP53 (n=3), IDH1 p.R132 (n=2); NPM1 p.W288fs(n=2); IDH2 p.R140Q (n=1); and KRAS (n=2). GNAS mutation was the sole identified mutation in 7/16 (44%) cases, of note 5 of these cases had non-diploid karyotype. Nine (56%) of cases had a diploid karyotype; 2 cases showed deletion/monosomy of chromosome 20; 2 cases showed trisomy 8 and one case showed deletion 5q. Other non-recurrent chromosomal abnormalities were also observed. Conclusions: GNAS R201 mutations are infrequent but recurrent alterations in myeloid neoplasms. Our data shows that these mutations may be observed in settings other than MDS, including myeloproliferative neoplasms, myelodysplastic/myeloproliferative neoplasms, as well as de novo and therapy-related AML. In our series of cases, the mutation frequently co-occurred with either cytogenetic abnormalities or other somatic mutations. S. Chaudhary, N. Rabade, S. Joshi, R. Mascerhenas, K. Kulkarni, P. Tembhare, P. Subramanian, S. Gujral, N. Patkar Tata Memorial Centre, Mumbai, Maharashtra, India. Introduction: Patients with primary myelofibrosis (MF) carry mutations in the Janus kinase 2 gene (JAK2) or thrombopoietin receptor gene (MPL) in 50% to 60% and 5% to 10% cases respectively. Recently, the presence of somatic mutations in the calreticulin gene (CALR) has been described in JAK2 and MPL negative cases of essential thrombocythaemia (ET) and primary myelofibrosis (MF). Patients with mutated CALR had lower risk of thrombosis and longer overall survival than patients with mutated JAK2. It seems imminent that CALR mutation testing will become essential in the work up of Ph negative myeloproliferative neoplasms (MPN). Here, we describe a fragment length analysis based assay for the rapid diagnosis of CALR positive myeloproliferative neoplasms. This is the first data from India describing CALR mutations in Ph negative MPN. Methods: As somatic mutations in CALR are predominantly insertions and/or deletions we developed a PCR using primers that flanked the exon 9 of the CALR gene. Both blood and bone marrow were used as a sample source. The CALR assay was standardized using genomic DNA as a template and subjected to PCR with fluorescently labelled primers. PCR amplicons were subjected to capillary electrophoresis on an ABI3500 genetic analyser. Results: The assay sensitivity was found to be 5% to 7% and inter and intra run precision was 100%. A total of 21 CALR mutations were detected. Of these 19 patients had MF and 2 had ET. Patients who harboured these mutations had ages ranging between 18 years to 63 years (median 46 years) with a male: female ratio 3.2:1. CALR mutations were exclusively detected in JAK2 and MPL mutations negative cases. The commonest type of mutations were deletions (66.6%) rest were insertions. Type 1 CALR mutation was the commonest followed by type 2 mutation. Other deletions were 2bp, 18bp, 46bp as well as a 164bp and 155bp deletion. Insertions were predominantly of 5bp and 3bp insertions. In addition to these 22 mutations, we also detected a 9 bp CALR deletion that has been described as a polymorphism. Conclusion: To summarize, we describe a rapid, sensitive fragment length analysis based method to detect CALR mutations in Ph negative MPN. G. Zheng, A. Pallavajjalla, L. Haley, M. Lin, J. Eshleman, C.D. Gocke Johns Hopkins University, Baltimore, MD. Introduction: Therapy-related myeloid neoplasm (T-MN) is a complication of cytotoxic chemotherapy and/or radiotherapy with distinct features including higher incidence of TP53 mutation. Subclonal mutations and clonal architecture in T-MN are important but not well-studied: whole exome or genome sequencing lacks sufficient sensitivity and accuracy to identify low-frequency subclonal mutations; a small gene panel achieving sufficient coverage to detect subclonal mutations is too limited to appreciate the mutation landscape. We employed a 643 gene panel with a bioinformatics pipeline designed for somatic mutations including low-frequency variants. With this platform we studied the prevalence of subclonal mutations in T-MN and de novo AML, and compared their mutation landscape and related clinical features. Methods: The study was approved by the IRB at the Johns Hopkins Medical Institutes. From the specimens submitted to the Molecular Diagnostics Laboratory at the Johns Hopkins Hospital between July 2015 and March 2016, we collected 10 consecutive patients with T-MN and 23 consecutive patients with de novo AML. Clinical, hematological, and cytogenetic data were collected from the electronic medical record. The custom panel was designed as a clinical leukemia panel, and covers 643 genes important in oncogenesis. Results: Although the limit of detection depends on variant position, mutations with as low as 1% VAF were detected by our platform. Mutations of a subset of genes appear to be enriched in T-MN cases: TP53 (30% versus 13%), MLL2 (20% versus 4.3%), and BCOR (30 versus 8.7%); whereas mutations of another subset of genes are underrepresented in T-MN cases: TET2 (0% versus 13%), NRAS (10% versus. 21.7%), FLT3 (0% versus. 17.4%), DNMT3A (10% versus. 21.7%), and MSH3 ( 0% versus. 17.4%). Subclonal mutations were found in all cases of T-MN (range of cases per patient: , and in 91% of de novo AML (range: 0 to 10). On average, T-MN has 7.1±7.4 subclonal mutations, whereas de novo AML has 3.3±2.3 mutations (P=0.027). Although mutations of NRAS and TP53 are considered driver mutations, our study identified 4 cases with low frequency clones harboring mutations in those genes: two de novo AML cases with NRAS mutations and two T-MN cases with TP53 mutations. One of the T-MN cases had 25 subclonal mutations, the patient is the only one who received radiation therapy. Conclusion: T-MNs have a different genomic mutation profile from de novo AMLs, and harbor more subclonal mutations than de novo AMLs. The complexity of the clonal architecture may partially contribute to the lack of response to treatment in T-MN patients. The study is limited in case numbers, but the findings will lay the groundwork for a larger study to assess the clinical significance of the subclonal mutations. H40. An Algorithmic Approach to the Diagnosis of Hermansky-Pudlak Syndromes by Using Both Platelet Electron Microscopy and Targeted Next-Generation Sequencing J. A. Majerus, L.M. Coon, E. Turley, W.L. Nichols, R.K. Pruthi, R. He, R. He, D.S. Viswanatha, D. Chen, J. Perez Botero Mayo Clinic, Rochester, MN. Introduction: Hermansky-Pudlak syndrome (HPS) is a heterogeneous group of rare autosomal recessive disorders characterized by oculocutaneous albinism and mild bleeding diathesis due to severe platelet dense granule deficiency. Some patients may present with chronic colitis and pulmonary fibrosis. Currentlythere are 9 HPS subtypes associated with mutations involving 9 genes. Most subtypes are associated with defects in biogenesis of lysosome-related organelle complexes (BLOCs) that participate in endosomal trafficking. Some subtypes (HPS-1, HPS-2, and HPS-4) are commonly associated with chronic colitis and pulmonary fibrosis. HPS-3, HPS5, and HPS-6 form a BLOC-2 complex, and their mutations usually result in milder symptoms. From limited available literature, HPS-5 appears to have a mild clinical course with a lower risk of developing pulmonary fibrosis. Since genotypic variations in HPS may predict clinical outcome, molecular testing is important in this setting. With a combination of platelet electron microscopy (PTEM) and targeted nextgeneration sequencing (NGS), we confirmed HPS-5 in a patient and identified a novel mutation in this rare form of HPS. Methods: A 65 y/o male patient presented with subtle ocular albinism, nystagmus, bleeding diathesis, without evidence of pulmonary fibrosis or chronic colitis. PTEM showed virtual absence of dense granules. A targeted NGS panel encompassing 9 HPS genes (HPS1, AP3B1, HPS3, HPS4, HPS5, HPS6, DTNBP1, BLOC1S3, and BLOC1S6) was designed using SureDesign (Agilent Technologies). DNA library preparation was performed using the SureSelect Target Enrichment System for Illumina Paired-End Multiplexed Sequencing Library (Agilent Technologies). The enriched indexed DNA sample was then sequenced on an Illumina MiSeq platform. All sequence variants were classified following the current American College of Medical Genetics and Genomics (ACMG) Standards and Guidelines. Results: A novel homozygous mutation in the HPS5 The Journal of Molecular Diagnostics ■ jmd.amjpathol.org gene, c.1960A>T (p.Lys654X) , was identified. This mutation results in a premature protein truncation with loss of amino acids downstream of K654 to the c-terminal end of HPS5. The genetic variant was also confirmed by Sanger Sequencing. The truncatedHPS5 protein interrupts the BLOC-2 complex and likely disrupts platelet dense granule genesis. It was therefore interpreted as "likely pathogenic" following the ACMG Standards and Guidelines. The PTEM and NGS findings, along with the clinical history, confirmed the diagnosis of HPS-5. Conclusions: The identification of this novel mutation underscores the importance of the combination of PTEM and targeted NGS testing in accurately characterizing specific variants of HPS, which should lead to more informed anticipatory clinical guidance. A.E. Quesada, Z. Hu, M.J. Routbort, K.P. Patel, R. Luthra, S. Loghavi, Z. Zuo, C. Yin, R. Kanagal-Shamana, S.A. Wang, J.L. Jorgensen, L. Medeiros, C.Y. Ok M.D. Anderson Cancer Center, Houston, TX. Introduction: Knowledge of the mutations that occur in mixed phenotype acute leukemia (MPAL) is limited. MPAL is an uncommon presentation of acute leukemia, and few studies which have assessed MPAL cases for genetic mutations; none have done so utilizing next-generation sequencing (NGS). The aim of this study is to further elucidate the genetic mutations that may occur in MPAL in the hope of finding potential therapeutic targets. Methods: We searched for all MPAL cases tested by a NGS panel over 4 years (2012 to 2016). Only de novo cases that fulfilled the 2008 World Health Organization (WHO) classification criteria for the diagnosis of MPAL were selected. Molecular data were collected. Clinical, laboratory, cytogenetic and bone marrow findings were also reviewed. Statistical analysis was performed using GraphPad Prism with significance set at a p-value < 0.05. Results: We identified 11 cases of MPAL that were examined by a NGS panel. There were 6 (55%) men and 5 (45%) women with a median age of 64 years (range, 28 to 89 years). Two patients had BCR-ABL1 rearrangement and one patient had KMTA2 (MLL) rearrangement. A B-cell/myeloid (B/My) immunophenotype was more common (6/11, 55%) than a Tcell/myeloid (T/My) immunophenotype (4/11, 36%). There was one case with a Bcell/T-cell (B/T) immunophenotype. Mutations in one or more genes were detected in 6/11 (55%) patients. A total of 13 distinct mutations were found in ASXL1, DNMT3A, FLT3, IDH1, IDH2, JAK2, MLH1, NOTCH1, NRAS, RUNX1, TET2, TP53 and WT1. FLT3-ITD was the only recurrent mutation in 2 patients. Mutations were not detected in five patients (45%). Whereas all 4 patients with a T/My immunophenotype harbored at least one mutation, only 2 of six patients with B/My immunophenotype had mutations. Conventional cytogenetics showed a normal karyotype in 3/11 (27%) patients. One (9%) had one chromosomal abnormality, one (9%) had two abnormalities, and 6/11 (55%) had 3 or more abnormalities. Five of 6 patients (83.3%) with a B/My immunophenotype had a complex karyotype, but none of patients with T/My immunophenotype had a complex karyotype. Combining cytogenetics and gene mutational analysis, 10 of 11 (90.9%) patients had either cytogenetic aberration or gene mutation. Only one (9.1%) patient had a normal karyotype without mutations. Conclusions: Our study shows that genomic aberrations in MPAL are complex. Cases of MPAL with a B/My immunophenotype appear to be more cytogenetically complex with less gene mutations, whereas cases with a T/My immunophenotype are the converse, less cytogenetically complex with more gene mutations. Our data suggest that underlying mechanisms of leukemogenesis differ between B/My MPLA and T/My MPAL. R. Faryabi, G.W. Schwartz, Y. Zhou, A.W. Lehman, J.J. Morrissette, K.S. Elenitoba-Johnson, M. Carroll University of Pennsylvania, Philadelphia, PA. Introduction: FLT3 internal tandem duplication (FLT3/ITD) is a common somatic mutation and oncogenic driver in acute myeloid leukemia (AML). Hence targeting mutated FLT3 serves as an important therapeutic strategy in FLT3-mutated AMLs. To this end, several inhibitors of FLT3 (FLT3i) have been developed and are in phase III clinical trials. Despite the effectiveness of FLT3i, the overall response rate is lower than 60%. Although the tandem-duplicated sequences of FLT3/ITD is always in-frame resulting in an elongated protein product, it varies in position, length, number of tandem duplications, and inserted spacer nucleotides in between duplications. We hypothesize that the various forms of tandem duplication and spacer sequences defining the elongated FLT3 protein could modify the response to FLT3i treatment. However, a detailed study of the effect of specific FLT3/ITD class on the clinical outcome is hampered by the complexity of classifying various FLT3/ITD structures. Methods: Here, we present a computational framework, ITDprofiler to identify and classify FLT3/ITD structure. Utilizing efficient implementation of Suffix Tree algorithm, ITDprofiler automatically identifies tandem duplications and the interspersed inserted sequences. ITDprofiler identified two major categories of FLT3/ITDs which we have designated as; typical or atypical. Typical-ITD is defined as tandem duplication without or with insertion of only native exon 14 sequences of the FLT3 genes; whereas an insertion of sequences exogenous to exon 14 of the FLT3 gene is designated as atypical-ITD. Moreover, ITDprofiler automatically defines the duplicated amino acid residues and provides visual representation of the placement of ITD sequences and highlights duplicated and spacer sequences. Results: Using the ITDprofiler, we have analyzed over 200 AML patients sequenced at the Center for Personalized Diagnostics of the University of Pennsylvania and 107 were FLT3/ITD-positive. Our method segregates typicalfrom atypical-ITD patients with more than 99% accuracy compared to manually annotated samples. We observed atypical-ITD in more than 34% of the FLT3/ITDpositive patients. Atypical-ITD patients were more frequently multi-clonal. Our preliminary analysis suggests that the residues of the JM zipper region important for maintenance of the autoinhibitory conformation of FLT3 are frequently duplicated. This observation is consistent with earlier studies in pediatric AML. Conclusions: Our standalone and web-based ITDprofile program enables accurate and efficient characterization of FLT3/ITD structure and provides a clinically validated platform to study the effect of variations in FLT3/ITD structure on the outcome of the treatment with FLT3 inhibitors. H43. Acute Myeloid Leukemia with Concurrent Biallelic CEBPA Mutation and FLT3 Internal Tandem Duplication C. Soderquist, K. Elenitoba-Johnson, A. Bagg, S. Luger, A. Perl, M. Carroll, J. Morrissette University of Pennsylvania, Philadelphia, PA. Introduction: Acute myeloid leukemia (AML) is a neoplasm of hematopoietic precursor cells with altered proliferation and maturation. The current WHO classification of AML is based on various pathologic features including, amongst others, the presence of recurrent chromosomal rearrangements as well as CEBPA and NPM1mutations. Recurrent mutations in a myriad of additional genes, including FLT3, are known to have prognostic and therapeutic implications. Biallelic CEBPA mutations are associated with a good prognosis whereas FLT3-ITD mutations confer a poor prognosis. Concurrent biallelic CEBPA mutations and FLT3-ITD are relatively uncommon. Some studies suggest that AML with biallelic CEBPA mutations are less genetically complex than other AML. Methods: We conducted a retrospective analysis of AML patient samples with concurrent biallelic CEBPA and FLT3-ITD mutations that were identified using a search of electronic databases. Demographic and clinical variables (age, therapy received, disease response), longitudinal molecular cytogenetic data, and bone marrow histomorphology were analyzed. Total numbers of patients concurrently tested for CEBPA and FLT3 alterations were tabulated. Results: One hundred eighty six AML patients from May 1, 2015 to May 15, 2016 were tested with a panel of 68 genes commonly altered in myeloid neoplasms, including CEBPA and FLT3. Of these, 9 (4.8%) showed biallelic CEBPA mutations. Unexpectedly, 6 of the 9 (66%) patients with biallelic CEBPA mutations had FLT3-ITD mutations. In 5 of 6 cases with concurrent mutations, CEBPA mutations were identified at higher allele frequency than FLT3; in 2 of 6 cases, the FLT3-ITD were identified at allele frequencies below 1%. All FLT3-ITD were confirmed by DNA PCR followed by capillary electrophoresis. The nine cases with biallelic CEBPAmutations showed an average of 4.4 additional mutated genes. Of 8 cases with available cytogenetic data, 4 (50%) had a normal karyotype. Of the 6 patients with concurrent CEBPA and FLT3 alterations, AML subclassification varied. Five of 5 patients achieved a remission with induction chemotherapy. Conclusion: Though the sample size is small, and our observations based on a single tertiary institutional cohort may be confounded by a referral bias, our data suggests that co-occurrence of biallelic CEBPA mutation and FLT3-ITD may be more common than previously reported. In our cohort, most CEBPA mutations occurred at higher allele frequency than the FLT3-ITDs, suggesting they may have preceded the acquisition of FLT3-ITD. Additionally, AML with biallelic CEBPA mutations show more genetic complexity than previously reported. A larger cohort with long term follow-up is needed to determine if response to treatment and outcome vary from other AMLs that lack these features. Neoplasms Using a Next-Generation Sequencing (NGS)-Based Assay C. Ho, J.C. Gomez-Gelvez, M.H. Syed, A. Zehir, W. Yu, T. Baldi, M. Ladanyi, A. Dogan, J. Yao, K. Nafa, M.E. Arcila Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: Lymphoid and plasma cell neoplasms are characterized by clonallyrestricted T-cell receptor (TCR) or immunoglobulin (Ig) rearrangements. Across clinical laboratories, this is generally demonstrated with standardized multiplex polymerase chain reaction (PCR) assays, in which V-J or D-J products are separated by fragment sizes on capillary electrophoresis (CE). However, this approach has relatively low sensitivity and does not provide the specific clonal sequence information required for tracking a clone at low level or in minimal residual disease (MRD) setting. In this study, we assessed the performance of a NGS-based assay, LymphoTrack (LT) (Invivoscribe), for detection of low level and MRD in comparison to CE and flow cytometry (FC) assays. Methods: DNA was extracted from bone marrow, blood, and formalin-fixed paraffin-embedded tissue from 30 patients with diagnostic and post-therapy (PT) samples. For clonal Ig rearrangement, PCR primers flanking the IGH conserved framework region 1 (FR1) in VH and conserved JH region were used. For clonal TCR rearrangement, primers flanking the TRG conserved V and J regions were used. The amplified products were sequenced on the Illumina MiSeq platform, and analyzed with the proprietary LymphoTrack analysis software, which provided the quantitation and V-J gene family usages of all unique sequences. The patient-specific diagnostic clonal sequences were used to detect MRD in subsequent samples, and compared to concurrent CE jmd.amjpathol.org ■ The Journal of Molecular Diagnostics and FC results. Results: The study included 30 diagnostic and 37 PT samples, from patients with plasma cell neoplasms (PCN) (n=9), acute lymphoblastic leukemias (n=7), mature B-cell neoplasms (n=8), and mature T-cell neoplasms (n=6). Median number of sequencing reads for PT samples was 440,600, with the expected clonal sequences detected in as low as 0.0045% of reads. In 20/37 (54.1%) PT samples, MRD was detected by CE and/or FC; LT detected MRD in 19/20 of these cases (sensitivity=95%). LT did not detect MRD in a case of PCN, but FC detected 0.0028% plasma cells suspicious for disease. In 8/37 (21.6%) PT samples, only LT detected MRD, whereas FC and CE failed to detect any disease. In 11/37 (29.7%) PT samples, both LT and FC detected MRD, whereas CE did not. In 2/8 cases (25%) with MRD detected by LT only, subsequent patient samples showed overt disease with a median follow-up of 2.5 months. 7/37 (18.9%) PT samples were negative for MRD by all assays, and 6/7 (85.7%) of these patients remained diseasefree with a median follow-up of 3.5 months. Conclusions: Compared to CE and FC, LymphoTrack provides comparable or better MRD detection sensitivity of lymphoid neoplasms, and with increased diagnostic certainty by utilizing patient-specific clonal sequences for MRD detection. Introduction: Congenital neutropenia encompasses a group of disorders broadly defined by a persistent or recurrently low absolute neutrophil count (ANC) caused by a constitutional genetic mutation. A subset of these patients have a form of severe congenital neutropenia (SCN), with ANC levels below 200, and are at increased risk of leukemic transformation. The clinical diagnosis of SCN is challenging given its rarity (10 cases per million people) and overlap with other inherited syndromes that result in low ANCs. Here we describe a novel exome sequencing-based panel for the clinical identification of SCN. Methods: Blood samples were obtained from 10 patients with low ANCs and suspected SCN. DNA was extracted and enriched using hybrid-capture based exome sequencing probes targeting a 54 megabase region. The resulting libraries were sequenced using 2x101bp paired-end reads. Data was aligned and variants were called with a variety of open source tools. Variants were compared to a custom database containing ~1,000 previously reported mutations in 24 genes associated with SCN or related diseases including AP3B1, CSF3R, CXCR2, CXCR4, ELANE, G6PC3, GATA2, GFI1, HAX1, JAGN1, KRAS, LAMTOR2, LYST, NRAS, RAB27A, RUNX1, SBDS, SLC37A4, TAZ, TCIRG1, USB1, VPS13B, VPS45 and WAS. Results: An average of 10 gigabases of sequencing data per case was obtained with an average on target efficiency of 62%, resulting in 97% of the exome covered to a depth of >20x. Of the 10 sequenced cases, 4 (40%) carried a previously described neutropenia-associated mutation. The most frequently mutated gene (2 cases) was SBDS; mutations in this gene have been associated with Shwachman-Bodian-Diamond Syndrome. One case (10%) carried a mutation in ELANE, which encodes the neutrophil elastase gene and is the most commonly mutated gene in SCN. One case carried a mutation in CXCR4 that encodes the C-X-C chemokine receptor type 4 gene that has been associated with WHIM syndrome. Four patients (40%) had variants in genes associated with SCN that have not been previoulsy described in the literature. In 2 patients (20%), no variants with an obvious physiologic relationship to congenital neutropenia were detected. Conclusion: Here we show that clinical exome sequencing can identify canonical SCN-associated mutations in approximately 40% of patients with suspected disease. Given the breadth of gene mutations associated with SCN and the difficulty of establishing a clinical diagnosis of SCN, exome based sequencing is a useful assay for confirming a diagnosis. Further, by cataloging novel mutations, clinical exome sequencing has the potential to discover new genes involved in the pathogenesis of SCN. FusionPlex, a Targeted RNA-Sequencing (RNA-Seq) Assay Using Next-Generation Sequencing (NGS) Technology C. Ho, R. Benayed, P. Sukhadia, K.A. Mullaney, K.M. Rios, L. Wang, M.F. Berger, M. Ladanyi, M.E. Arcila Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: Fluorescence in-situ hybridization (FISH) and quantitative reverse transcription-polymerase chain reaction (qRT-PCR) are common detection methods for gene fusions, but both are laborious, low-throughput, and lack scalability. Here we describe the performance of a clinically validated, targeted RNA-Seq assay for fusion detection in hematologic malignancies. The assay allows the simultaneous assessment of fusions involving 42 genes in a single assay without prior knowledge or need for targeting of the partner genes. Methods: RNA was extracted and reverse-transcribed from 44 blood or bone marrow samples submitted for targeted fusion assessment by qRT-PCR or FISH, based on clinicopathologic history. The cDNA fragments were amplified using Archer's Anchored Multiplex PCR with a custom primer panel targeting 42 genes involved in recurrent fusions in hematologic malignancies. The products were sequenced on the Illumina MiSeq platform and analyzed with the Archer analysis pipeline. Reproducibility and sensitivity studies were performed on replicates and serially diluted samples, respectively. The findings were compared to FISH and qRT-PCR results. Results: A total of 44 samples were analyzed, including B-lymphoblastic leukemia (n=23), chronic myelogenous leukemia (n=8), acute promyelocytic leukemia (n=7), acute myeloid leukemia (n=4), mixed phenotype acute leukemia (n=1), and diffuse large B-cell lymphoma (n=1). The Archer assay detected all common BCR-ABL1 fusions, including e1a2 (p190) (n=10), e13a2/b2a2 (p210) (n=4), and e14a2/b3a2 (p210) (n=6), as well as uncommon fusions not detected by our standard qRT-PCR assays: e13a3/b2a3 (n=1) and e1a3 (n=1). The t(9;22)(q34;q11) was confirmed by FISH in all cases. Other fusions detected included ETV6-RUNX1 (n=10), PML-RARA (n=7), RUNX1-RUNX1T1 (n=1), CBFB-MYH11 (n=1), EIF4A2-BCL6 (n=1), ETV6-ABL1 (n=1), and KMT2A-AFF1 (n=1), which were all fully characterized by this NGS assay in concordance with the expected fusion breakpoints. Reproducibility studies showed excellent interand intra-assay reproducibility. Sensitivity studies showed lower limits of fusion detection at 1% for BCR-ABL1 (quantitated by qRT-PCR and adjusted to the international scale (IS)), and 1% for PML-RARA, ETV6-RUNX1, and RUNX1-RUNX1T1 (quantitated by qRT-PCR and normalized to control cell lines with fusions). Conclusions: Our targeted RNA-Seq assay using Archer FusionPlex technology provides a more comprehensive analysis for gene fusions in hematologic malignancies at the time of initial diagnoses. It provides several advantages, including high multiplexing capability and scalability, and detection of uncommon variant fusions, as well as fusions involving alternative partners, providing further insights into tumor biology. K.R. Bessonen, M. Mai, L.A. Frederick, R. He, D.S. Viswanatha Mayo Clinic, Rochester, MN. Introduction: Somatic hypermutation (SHM) of the immunoglobulin variable heavy chain (IGVH) genes is an important prognostic marker in B-cell chronic lymphocytic leukemia (CLL). SHM is commonly detected by PCR amplification of the clonal IGVH-JH region followed by Sanger sequencing (PCR-SS). Next-generation sequencing (NGS) technology has recently been used to determine SHM in CLL. We present our experience using NGS for the detection of IGVH SHM status in CLL. Methods: RNA was isolated from peripheral or bone marrow samples and reverse-transcribed to cDNA. PCR-SS of cDNA was performed with individual leader (LDR) VH1-VH7 family primers in separate reactions, followed by sequencing (Applied Biosystems Inc., Carlsbad, CA). If a LDR PCR product was absent, a second PCR was performed with individual FR1 family and JH primers. Sequencher version 6.0 (Gene Codes Corporation, Ann Arbor, MI) and the IMGT human IGH reference database (http://www.imgt.org) were used for sequence analysis. NGS was performed using the LymphoTrack IGH somatic mutation kit (Invivoscribe, San Diego, CA). Multiplex LDR and JH region primers were used for initial PCR. In the absence of a clonal band by LDR PCR, the FR1 and JH family primer mix was used and both LDR and FR1 products were pooled and sequenced on the MiSeq platform (Illumina, San Diego, CA). NGS analysis was performed using the manufacturersupplied LymphoTrack software and NCBI IgBLAST reference, as well as additional lab-specific analytic assessment. Results: 45 patient samples were analyzed by PCR-SS and NGS. A minimum 5% clonal B-cells were required for adequacy of the NGS platform. A quality threshold of 10 5 minimum total sequence reads and a dominant clonal population of >10% of the total reads per sample were used. Comparison of PCR-SS with NGS revealed that NGS had an accuracy rate of 95.6%. Two of 45 cases (4.4%) were considered truly discordant likely resulting from differences in single reaction in PCR-SS versus multiplex PCR in NGS. Minor differences in data analysis were also observed in VH gene annotation, likely reflecting variations between IMGT and IgBLAST reference data. We identified 11 distinct data patterns in the NGS platform. These patterns were subsequently employed for improved analysis of the NGS data in the clinical molecular diagnostic scenario. Conclusions: This study demonstrates the NGS technique is acceptable for clinical use in the determination of SHM status in CLL. Results obtained are highly accurate in comparison to the current standard of PCR-SS. Advantages of NGS include direct determination of the IGVH rearrangement and percent SHM. In conclusion, NGS for IGVH SHM analysis in CLL is feasible to implement in the routine clinical molecular diagnostic setting. V. Pattanayak, S. Duraisamy, J.K. Lennerz, D. Dias-Santagata, L.P. Le, A.J. Iafrate, V. Nardi Massachusetts General Hospital, Boston, MA. Introduction: Activating hotspot mutations in RAS family members (primarily NRAS and KRAS) are common in many cancers, including myeloid neoplasms. The majority of RAS hotspot mutations occur at codons Gly12, Gly13, and Gln61. For example, whereas mutations to Asp are most common among NRAS Gly12 hotspot mutations in hematopoietic malignancies, mutations to Ser, Ala, Val, Cys, and Arg are also observed, suggesting functional comparability among different variants. However, concomitant variants are not thought to be common either at the level of the RAS isoform or variant amino acid. Here, we describe the occurrence and dynamics of concomitant RAS mutations in a clinical testing cohort. Methods: Targeted next-generation sequencing data from an assay testing exons and hotspots in 39 genes (SNaPshot NGS), including Gly12, Gly13, Gln61, and Ala146 for both NRAS and KRAS, were evaluated for 391 patients (440 samples) with hematological The Journal of Molecular Diagnostics ■ jmd.amjpathol.org malignancies (predominantly AML, MDS, and multiple myeloma) over a period of clinical testing from October 2014 to March 2016 at Massachusetts General Hospital. Results: NRAS or KRAS mutations were identified in 97/440 (22%) samples (from 76/391 patients). Of the 97 samples with NRAS/KRAS mutations, 74 had a single detectable mutation in NRAS or KRAS and 23 samples had multiple mutations in NRAS/KRAS, often at low variant allele fractions. 16 samples had two, four had three, two had four, and one had seven co-occurring mutation(s). Ten patients who had concomitant mutations had testing performed at multiple time points. Of those patients, only one had no changes in the number and identities of concomitant variants, whereas four had a decreased number of variants, three had an increased number, and two had different variants (but the same number of variants) between subsequent time points. At the variant level, of the patients with concomitant mutations, 16/20 (75%) had at least one NRAS G12 or G13 variant, compared to 16/56 patients (29%) with a single mutation only (p = 0.0013). Of the 20 patients with concomitant mutations, 6/20 (30%) had NRAS Gly12Ser mutations, compared to 3/56 (5%) patients with a single RAS mutation (p = 0.0084). Conclusions: Concomitant NRAS/KRAS mutations are not uncommon in hematopoietic malignancies. NRAS G12/G13 mutations are enriched among samples with concomitant mutations, suggesting that various NRAS hotspot mutations may confer different activation properties. In addition, the variation in NRAS/KRAS mutations among multiple time points in the same patients suggests that the mutations are occurring in different clones, consistent with intratumoral heterogeneity and arguing against a role for NRAS/KRAS mutations as driver mutations in these tumors. Introduction: First recognized as a distinct entity in 1938, Diamond Blackfan anemia (DBA) is a genetically and clinically heterogeneous disorder characterized by pure red blood cell aplasia, variable congenital anomalies and a predisposition to cancer. The genes identified to date that are mutated in DBA all encode ribosomal proteins associated with either the small or large subunit. DBA is typically treated with steroids, red blood cell transfusions, and hematopoietic stem cell transplantation. Herein we report an Asian male presenting with macrocytic anemia, neutropenia, genitourinary malformations and growth retardation due to a novel RPL35A gene mutation. Presentation: This patient was born premature at 33 weeks and was one of the non-identical male twins. At 4 months of age, he was found small for his age (2% WHO growth percentiles) with multiple genitourinary malformations, HGB 7.2 g/dl, MCV 102.1 and HCT 20.6%. He had no significant family history and his twin brother was normal in size. The anemia didn't response to iron supplementation. Bone marrow evaluations revealed relative lymphocytic and megakaryocytic hyperplasia, relative granulocytic hypoplasia, and decreased marrow iron stores. Methods: Genomic DNA was isolated from blood specimen using a standardized kit and quantified. Sequence enrichment of the targeted coding exons and adjacent intronic nucleotides was performed by a bait-capture methodology using long biotinylated oligonucleotide probes, followed by polymerase chain reaction and next-generation sequencing. Ten genes were sequenced, including SBDS, RPL11, RPL35a, RPL5, RPS10, RPS17, RPS19, RPS24, RPS26 and RPS7. Any mutations were confirmed by conventional dideoxy DNA sequence analysis. Results: Sequence analysis revealed a novel p.R76P variant, located in coding exon 3 of the RPL35a gene, resulting from a G to C substitution at nucleotide position 227. The highly conserved arginine residue at codon 76 was replaced by a proline residue. This variant was not reported in population based cohorts in several major databases. In addition, this alteration is predicted to be possibly damaging and deleterious by PolyPhen and SIFT in silico analyses, respectively. A steroid therapy started and this patient's anemia gradually improved; one year later, laboratory tests showed HGB 11.5 g/dl, MCV 103.1 and HCT 33.0%. Conclusions: We report a novel variant of unknown significance in the RPL35a gene (p.R76P) in a patient with DBA characteristic findings. This variant is likely the pathogenic mutation of this patient's DBA, because genetically this variant changed a highly conserved amino acid in available vertebrate species, and clinically this patient responded well to steroid treatment. Leukemia (CLL) I. Hubbard, A. Layton, B. Tandon Molecular Pathology Laboratory Network, Inc., Maryville, TN. Introduction: CLL is the most common leukemia affecting adults in western and European countries. IGHV somatic hypermutation (SHM) status is a baseline prognostic indicator routinely assessed at initial diagnosis and is associated with indolent disease course and favorable overall survival when present. SHM is defined by < 98% sequence homology to most closely matched germline IGHV segments. The current standard for evaluation of SHM in CLL involves RNA extraction from patient peripheral blood or bone marrow samples followed by reverse transcription-PCR and Sanger sequencing. Limitations of the traditional Sanger sequencing-based methodology include the necessity of prompt specimen receipt due to RNA lability and insensitivity in detecting SHM for clonal CLL populations comprising < 50% of total B-cells. We investigated the potential utility of an available next-generation sequencing (NGS)-based assay to determine SHM status using DNA from known CLL samples previously tested at an outside reference laboratory by IGHV Sanger sequencing. Methods: Twenty-three CLL patient specimens were studied. NGS testing consisted of target enrichment PCR using LymphoTrack primers (Invivoscribe) with adapters for sequencing on the Ion Torrent PGM. The LymphoTrack PGM v2.0 software was used to process and analyze sample results. Mutation rates for the most abundant sequence reads were evaluated based on comparison to partial V genes by the re considered positive for SHM. Results: Of the 23 specimens analyzed by IGHV Sanger sequencing, SHM status could not be determined in 5 cases (22%) due to inability of the assay to detect a dominant clonal population. In contrast, the NGS assay successfully yielded informative results for all 23 specimens. For the 18 samples with reported Sanger sequencing results, 6 (33%) were hypermutated and 12 (67%) were non-hypermutated, with 100% concordance noted between the Sanger Sequencing and NGS assays. The NGS assay also demonstrated complete concordance with expected results based on correlation with other CLL metrics including flow cytometry demonstration of clonal B-cell subsets and ZAP70 expression patterns. Conclusions: NGS may provide an effective replacement for IGHV Sanger sequencing. The utilization of DNA instead of RNA as a starting material, and greater sensitivity of detection of clonal populations, may also represent improvements over the traditional Sanger Sequencing based methodology. The technical feasibility of utilizing patient DNA, as opposed to RNA, also raises the possibility of obtaining clinically valid results from FFPE tissue biopsy specimens. Introduction: Chronic myelomonocytic leukemia (CMML) is characterized by persistent absolute monocytosis (>1 x 10 9 /L) in the peripheral blood (PB) and dysplasia in >1 lineages. In the absence of dysplasia, an acquired clonal cytogenetic/molecular abnormality is required or causes for reactive monocytosis have to be excluded. Oligomonocytic CMML (OM-CMML) showing increased monocytes in BM (>10% of the cellularity) but no absolute monocytosis in the PB occurs occasionally. These cases often show relative monocytosis and are likely classified as myelodysplastic syndrome (MDS) or myelodysplastic/myeloproliferative neoplasm, unclassifiable (MDS/MPN, U). A subset of these cases eventually develop overt CMML. Better characterization of OM-CMML is essential since the distinction between CMML and MDS is clinically relevant. Methods: Thirty OM-CMMLs (>10% monocytes in BM and >10% PB monocytes with absolute monocyte count of 0.5-1x10 9 /L) and 20 overt CMMLs (11 CMML0, 6 CMML1, 3 CMML2/AML) were compared, of which 21/30 and 20/20 cases were subjected to 21-or 45-gene mutation analysis. Results: OM-CMML included 21 men and 9 women (med age: 69 y/o, range 19-87). OM-CMML had lower WBC, lower absolute PB monocyte count and lower percentage of BM monocytes than overt CMML (p<0.05). The demographic distribution, the remaining CBC values and the degree of dysplasia were not significantly different. 14/28 OM-CMML and 1/20 overt CMML had abnormal karyotypes (p<0.01). 7/30 OM-CMML had prior chemotherapy (CT) whereas 1/20 overt CMML had prior CT. 6/7 OM-CMMLs with prior CT showed abnormal karyotypes. 9/31 (29%) OM-CMMLs progressed to overt CMML (med: 3 months, range 1 months to 58 months). Mutations in ASXL1, TET2 and SRSF2 were frequently seen in OM-CMMLs and overt CMMLs (13/21 versus 17/20; p=0.16). Overt CMML contained more cases with >2 concurrent mutations in the 3 genes when compared with OM-CMML (11/20 versus 5/21, p=0.058). 2/5 OM-CMMLs with >2 concurrent mutations in the 3 genes developed overt CMML. The remaining 3 cases showed persistent relative but no absolute monocytosis at follow-up. Mutations in CBL and RUNX1 were found more frequently in overt CMML. Conclusions: Mutations in ASXL1, TET2 and SRSF2 were frequently found in OM-CMML, indicating the genetic similarities between overt CMML and OM-CMML. The findings suggest at least a subset of OM-CMMLs likely represent early phase CMMLs, which was further substantiated by the development of overt CMML in some OM-CMML cases. Rearrangements in Hematopoietic Malignancies. D.P. Simmons, F.C. Kuo, E.P. Garcia, N.I. Lindeman, L.E. Macconaill, M.D. Stachler Brigham and Women's Hospital, Boston, MA. Introduction: Next-generation sequencing provides a breadth of knowledge about the alterations that drive cancer, including single nucleotide variants, copy number changes and chromosomal rearrangements. The latter is accomplished by selective inclusion of intragenic regions known to be involved in rearrangements to allow detection of structural alterations, including novel translocation partners not detectable by FISH. Our panel targets exons of genes (N=300) known to be important in many types of cancer and intragenic regions known to be involved in selected rearrangements (e.g., IGH, ABL, N=36 genes). We report the findings from jmd.amjpathol.org ■ The Journal of Molecular Diagnostics this targeted panel to identify translocations in 243 patients with various hematopoietic cancers. Methods: Samples were selected from patients with hematopoietic malignancies consented under the Dana Farber institutional review board approved protocol 11-104. Genomic regions of interest were targeted by hybrid capture for next-generation sequencing on the Illumina HiSeq platform. Single nucleotide variants, copy number variants, and structural variants were processed through a bioinformatics pipeline and interpreted by a pathologist or geneticist. The results from these interpretations were further analyzed. Results: Two hundred forty three (243) patients with hematopoietic malignancies were evaluated, including 75% with lymphoid neoplasms and 25% with myeloid disease. Structural variants were reported in 33% of patients, and potentially actionable SNV or CNV findings were reported in 61% of individuals. Structural variants were reported in 36% of individuals with other actionable SNV or CNV findings and 27% of individuals without other findings. BCR-ABL translocations were identified in chronic myelogenous leukemia (N=2, 100%) and acute lymphoblastic leukemia (N=2, 50%). In acute myeloid leukemia, MLL rearrangements (N=2, 8%) and FLT3 internal tandem duplications (N=4, 17%) were reported. A variety of findings were reported in B cell non-Hodgkin lymphomas including IGH-BCL2 (N=25, 17%), IGH-CCND1 (N=3, 2%) and IGH-MYC (N=2, 2%) rearrangements as well as SOCS1 deletions (N=3, 2%). Novel translocations identified included a RAF1-MTAP rearrangement in a histiocytic dendritic cell lymphoma, as well as three different IGH rearrangements in a single case of diffuse large B cell lymphoma. Conclusions: Structural variants were identified in patients with a broad spectrum of hematopoietic malignancies using a targeted next-generation DNA sequencing panel. Targeted capture of introns that are frequently involved in rearrangements in human cancer allows for identification of previously reported and novel structural variants within a sequencing panel and can provide additional valuable diagnostic information. Introduction: A commercial (HemaVision) multiplex RT-PCR assay is used widely for detecting 28 fusion transcripts in acute leukemia. However, it is difficult to find other recurrent fusion transcripts, especially with novel or cryptic chromosomal rearrangements. The purpose of this study was to identify significant leukemia fusion genes using targeted or untargeted next-generation RNA sequencing. Methods: We selected 10 acute leukemia patients with novel translocations by G-banding who were investigated by untargeted RNA sequencing and 10 acute leukemia patients with normal karyotype and negative HemaVision result who were investigated by targeted RNA sequencing and one ALL patient with BCR-ABL1/t(9;22) and one AML patient with CBFB-MYH11/inv(16) as positive controls. For untargeted RNA sequencing, total RNA was extracted from leukemia cells and cDNA libraries were constructed with TruSeq RNA kit. Paired-end sequencing was performed on HiSeq2500. For targeted RNA sequencing, total RNA from each bone marrow sample was converted to cDNA using Ovation cDNA Module (NuGEN). Multiplexed enriched libraries were produced by Ovation Fusion Panel Target Enrichment System V2 (NuGEN) which includes all exons of 502 known cancer fusion genes, and NGS was performed on HiSeq2500. Reads were aligned with TopHat/BowTie, and two algorithms such as deFuse and FusionCatcher were used to detect fusion transcripts. The candidate fusion transcripts were confirmed with RT-PCR followed by Sanger sequencing. Results: From the 10 acute leukemia patients with novel translocations by G-banding, we found 5 in-frame fusion genes exactly matched on translocation breakpoints from 3 AML patients and 1 B-ALL patient: USP34-ASAP3/t(1;2)(p36.1;p11.2), MAZ-MKL1/t(16;22)(p11.2;q13), MLL-SEPT6 and SEPT6-CDCA5/t(X;11)(q24;q13), and RCSD1-ABL1/t(1;9)(q24;q34), respectively. In the targeted RNA sequencing, as positive controls, we were able to identify the predicted BCR-ABL1 and CBFB-MYH11 fusions with both deFuse and FusionCatcher algorithms. The patient with BCR-ABL1 had also PAX5-ZCCHC7 fusion. Six of 10 patients with normal karyotype were found to have leukemiaassociated fusion genes; NUP98-NSD1 in two AML, PAX5-ZCCHC7 in B-ALL, MLLT10-C5 in AML, TAL1-PDZK1IP1 in AML, and MTA1-EPS15 in AML. Conclusions: Using targeted or untargeted next-generation RNA sequencing, we have discovered 5 candidate fusion genes from 4 patients of 10 acute leukemia patients with novel translocations (40%) and 2 known fusion genes and 3 novel fusion genes from 6 patients of 10 acute leukemia patients with normal karyotype (60%). Targeted cancer gene panel should be useful for the detection of recurrent fusion genes in acute leukemia patients rather than multiplex RT-PCR. values and long-term outcome was determined. Methods: The QuantideX qPCR BCR-ABL IS Kit from Asuragen uses standard TaqMan chemistry to quantitate BCR-ABL1 and the ABL1 reference gene RNA's. Associated software reports an international scale BCR-ABL1 value and a log-transformed MR value, with a 3 logreduction from pre-treatment baseline represented as 0.1% IS and MR3.0. Three laboratories performed BCR-ABL1 testing on 96 chronic phase CML patient's banked RNA specimens collected from two hospitals and drawn 12 to 18 months after starting TKI therapy. Clinical events (TKI therapy change, loss of complete hematologic or cytogenetic response, progression to accelerated phase/blast crisis, kinase domain mutation, or death) were recorded through 36±4 months after starting TKI. Two operators per site also tested serially-diluted reproducibility samples (range MR1.0 to MR4.0) in multiple replicates over 5 days. The 95% limit of detection (LOD) for the assay was defined as the median measured %IS value of four analogous serially-diluted specimens. Results: Fifty one patients had MR<3.0 at 12 to 18 months post-TKI. Of these 51 patients who did not achieve a major molecular response (MMR), 20 had a subsequent clinical event, 17 had no event, and 14 were lost to follow-up (LFU). Forty-five 18 months post-TKI. Of these 45 patients who did achieve MMR, 8 had an event, 28 had no event, and 9 were LFU. Kaplan-Meier survival curves demonstrated a 22% prolongation of eventfree survival (95% CI 2% to 42%) at 3 years in the MMR (versus non-MMR) group [p=0.028; 58% (95% CI 44% to 75%) for MR<3 versus 80% (95% CI 68% to 93%) reproducibility %CV (log-transformed) between 1.9 and 4.2%. The 95% LOD for both breakpoint transcripts (e13a2 & e14a2) was MR4.7 (0.002%IS), allowing sensitive detection of the MR4.5 cutoff that defines "complete molecular response" in ongoing treatment-free remission clinical trials. Conclusions: The QuantideX qPCR BCR-ABL IS Kit has excellent reproducibility and analytical sensitivity, and the achievement of MR>3 (major molecular response) by this assay predicts prolonged event-free survival in TKI-treated CML patients. Sequencing Method in Acute B-Lymphoblastic Leukemia S. Cheng, G. Inghirami, W. Tam Weill Cornell Medical College, New York, NY. Introduction: Minimal residual disease (MRD) is a powerful predictor for relapse of acute B-lymphoblastic leukemia (B-ALL), and its detection has been integrated as part of modern B-ALL therapy protocols. Traditional MRD assays rely on flow cytometry (FACS) or PCR, and recently NGS has emerged as a novel and more sensitive method. However, the current NGS assay is proprietary and its protocol is difficult to interpret and follow by a routine clinical laboratory. A simpler but highly sensitive and reliable NGS assay is needed. Methods: Here we applied LymphoTrack platform (Invivoscrible) with minor modification to capture all immunoglobulin heavy chain VDJ rearrangements from evaluable pre-treatment and post-treatment B-ALL samples, and developed a novel computer algorithm to identify tumor-associated clonotypes. Briefly, B-ALL genomic DNA was amplified with barcoded primer sets in a single PCR reaction. Resulting libraries were purified and quantified, followed by Miseq sequencing. LymphoTrack software (Invivoscribe) and our custom algorithm were used to analyze NGS data to generate MRD values. Assay performance characteristic, including analytical sensitivity, intra-run and interrun reproducibility, and linearity, were assessed by using B-ALL DNA serially diluted with normal PBMC genomic DNA. For concordance with FACS, we tested >50 B-ALL clinical samples, and compared the NGS results with those of FACS. Results: The assay can detect as few as 5 tumor cells among one million normal leukocytes. At a dilution of 1 in 250,000, intra-run analysis showed a mean (± SE) patientspecific tumor clone frequency of 0.00074% (± 0.00017%); at a dilution of 1 in 100,000, inter-run analysis showed a mean (± SE) tumor clone frequency of 0.00093% (± 0.00027%). Linearity studies demonstrated a high correlation (R 2 =0.999, a slope of 0.92) between reference and observed MRD levels. FASC versus NGS results showed concordance in 25 (78.1%) of 32 MRD + or MRD -B-ALL samples. In 6 of 26 (23.1%) samples that were MRDor inconclusive by FACS, MRD was positive by the sequencing assay. The median MRD level of these Miseq-MRD + samples was 0.0424% with a range of 0.006-0.35%. Our results indicate ultrasensitivity and excellent quantitative accuracy of the assay with high intra-run and inter-run reproducibility. Conclusions: We described here a simple, reliable, highly sensitive and accurate approach for MRD detection in B-ALL. It is advantageous over FACS because of its increased sensitivity as well as ease and consistency of interpretation. Above all, this method can be adopted by any clinical laboratories equipped with the basic physical and human resources for performing NGS. This method will be a useful adjunct in clinical management of patients with B-ALL. samples from patients with known/suspected hematologic malignancies. NGS was performed using a 35 gene panel targeting commonly mutated genes. 200ng sheared genomic DNA was prepared using a SureSelectXT capture reagent (Agilent, Santa Clara, CA) and the library sequenced on the MiSeq platform (Illumina, San Diego, CA). NGS data was processed through a proprietary bioinformatics pipeline (Mayo NGS Workbench) for alignment, base calling, and insertion/deletion (indel) detection. The processed data was annotated for final classification and clinical significance reporting. Results: One acute myeloid leukemia (AML) patient had a CEBPA c.318del involving 100% of the sequence reads (SR), with no coverage depth loss at this locus. Chromosomal microarray demonstrated a 26 megabase (Mb) copy-neutral loss of heterozygosity (cnLOH) event encompassing CEBPA from 19q13.11 to 19q13.43 . A second patient with B-lymphoblastic leukemia (B-ALL) harbored an IGH/BCL2 fusion and trisomy 8 (92%) and a low level BCR/ABL1 fusion (7%) by FISH. Manual review of the NGS data revealed a concurrent FLT3-ITD in c.1817_1818ins87 in 30% of the SR, providing additional information for targeted therapy. The last patient with AML had decreased coverage in a region of ASXL1 associated with clipped (unmapped) SR. Analysis of the clipped SR showed 100% homology to the TSHZ2 gene 20.5 Mb telomeric to ASXL1 on chromosome 20q. These findings were consistent with FISH data showing del(20q) in 46% of nuclei. The genomic break disrupts exon 14 of the ASXL1 gene and joins to the 3' UTR of the TSHZ2 gene. Conclusions: Standard NGS pipeline analysis identifies single base and small indel variants in a straightforward manner. Although wider genomic analysis (e.g. whole exome) is desirable for comprehensive detection, it imposes practical limitations of sample throughput, sensitivity and large datasets. Our experience with a targeted NGS platform suggests that even with small-scale targeted panels, recognition of novel pathogenic associations or analytic anomalies can enhance interpretation of data and may help guide clinical intervention. N. Patkar, R. Kodgule, G. Raval, C. Kakirde, S. Joshi, S. Chaudhary, R. Mascerhenas, K. Kulkarni, P. Tembhare, P. Subramanian Tata Memorial Centre, Mumbai, Maharashtra, India. Introduction: Recent whole genome and exome sequencing studies have identified key driver mutations that are harboured by myeloid malignancies. Amongst these, acute myeloid leukemia (AML) is perhaps the most biologically heterogeneous entity comprising of chromosomal abnormalities and somatic mutations at a DNA sequence level. It is important to recognize these mutations as some of them are of prognostic and therapeutic value. Here, we identify somatic mutations in AML using a 54 gene panel using targeted NGS on an Illumina MiSeq platform. Methods: Ten patients of AML (diagnosed as per the WHO 2008 criteria) were sequenced. FISH and conventional karyotyping was done to risk stratify the patients. Approximately 50 ng of genomic DNA was used as a template. Oligonucleotides were hybridized to target regions of the genomic DNA. This was followed by extension and ligation. Barcode indices and sequencing adapters were added by PCR based amplification. Once the libraries were prepared they were purified, normalized and pooled. Sequencing was done on a MiSeq system using the V3 chemistry. Base calling was done by the MiSeq Reporter software. Sequence alignment and generation of VCF files was done using Basespace and analyzed using VariantStudio v2.2.1. Results: Four patients could be classified as intermediate cytogenetic risk whereas the rest were favourable cytogenetic risk, (mean age: 31.4 years; M:F=3:2). Analysis of data quality indicated that all of the samples met the predefined acceptance criteria (mean Q30 of >91.2). The average amplicon mean coverage depth was 7463x. Mutations were detected in 9/10 samples (90%), which included base substitutions as well as indels at a sensitivity of 6%. An average of 2.2 mutations were detected per sample. Mutations affected diverse biological pathways such as signalling pathways, genes regulating epigenetic pathways, transcription pathways and genes encoding the RNA spliseosome. Mutations were detected in KIT (4 samples) followed by GATA2 (3 samples), ASXL1 (2 samples), DNMT3A (2 samples), RUNX1 (2 samples), TET2 (2 samples), FLT3 (1 sample), IDH2 (1 sample), NPM1 (1 sample), NRAS (1 sample), SMC3 (1 sample), U2AF1 (1 sample) and WT1 (1 sample). Patients with core binding factor leukemia predominantly harbored KIT and ASXL1 mutations. A single intermediate risk patient harboured a coexisting NPM1 and DNMT3A mutation. Conclusions: The TruSight myeloid sequencing panel is a comprehensive test that identifies mutations in myeloid malignancies. The assay has a relatively simple and straight forward workflow and is very suitable for application in a clinical setting. Our results indicate importance of comprehensive molecular profiling of AML in the era of personalised medicine. A. Hill, J. Peters, G. Hosler, K. Murphy ProPath Services, Dallas, TX. Introduction: There are 3 main subtypes of myeloproliferative neoplasms: polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF). Somatic mutations in the calreticulin (CALR) gene have recently been identified in the majority of ET and PMF patients that lack a JAK2mutation. There have been over 50 different CALR mutations described to date. These genetic alterations consist of insertions, deletions, or a complex combination of insertions/deletions (indels) within exon 9 of the CALR gene. The vast majority of the currently reported mutations leads to a +1 base pair (bp) shift in the open reading frame. This frameshift results in a common novel amino acid sequence of the Cterminal domain and elimination of the KDEL amino acid sequence required for endoplasmic reticulum retention. Given that most known mutations result in the same downstream reading frame, it is likely that the sequence plays a role in the oncogenicity of the mutant protein. Methods: CALR mutational analysis was performed by PCR amplification of exon 9 followed by fragment analysis via capillary electrophoresis (CE). The same region of exon 9 was then sequenced by the Sanger method. Results: CE results were consistent with a 1 bp insertion mutation. Sanger sequencing of the same region revealed 2 separate mutations within exon 9, a 2 bp insertion and a 1 bp deletion occurring 76 bases downstream of the insertion. The sequencing data was in agreement with the CE result of an overall 1 bp gain. Taken together, it can be concluded that this patient has a 2 base insertion and 1 base deletion on the same allele of CALR exon 9. The 2 bp insertion found in the patient's sample produces a +1 reading frame shift, resulting in the common mutation sequence for the first 13 amino acids. However, the downstream deletion changes the reading frame to +2 compared to wildtype, generating a novel amino acid sequence. This novel mutant sequence extends 10 amino acids longer and results in a greater addition of positively charged amino acids compared to the common mutant sequence. Conclusions: To our knowledge, a CALR mutation with an initial +1 and subsequent +2 open reading frame has not been previously described. The unique nature of this mutation may provide some insight into critical sequence for oncogenesis. This mutant does contain the first 13 amino acids of the common mutant sequence; therefore, sequence downstream of this core may be less important for oncogenicity. Alternatively, it is possible that the additional positively charged amino acids may enhance the oncogenic effect of this unique mutation. Hypermutation Status in Splenic Marginal Zone Lymphomas Using Next-Generation Sequencing J.C. Gomez-Gelvez, C. Ho, W. Yu, M.H. Syed, A. Zehir, T. Baldi, A. Dogan, M. Ladanyi, J. Yao, K. Nafa, M. Arcila Memorial Sloan-Kettering Cancer Center, New York, NY. Introduction: Splenic marginal zone lymphoma (SMZL) is a rare indolent B-cell neoplasm involving spleen, bone marrow (BM) and, frequently, blood. Its distinction from similar indolent B-cell malignancies may be often challenging, particularly when diagnosis must be based on the BM findings alone without the support of spleen histology. Prior studies have shown that SMZL exhibit specific immunoglobulin heavy variable gene (IGHV) gene biases which are distinct from other entities and thus ancillary testing could be potentially utilized to aid in the diagnosis or further stratifying this disease. This assessment is, however, often not feasible in the clinical setting as current methods are laborious and not performed in most laboratories. In this study, we explore the utility of next-generation sequencing (NGS) for the clinical characterization of IGHV in a cohort of SMZL and compare it to other subtypes of marginal zone lymphomas (MZL) reported in the literature. Methods: BM samples from patients with an established diagnosis of SMZL and submitted for routine clonality assessment were selected for the study. After establishing the presence of IGH clonality by capillary electrophoresis, the samples were analyzed using an NGS assay targeting IGH-FR1 (Lymphotrack, Invivoscribe) and sequenced by Illumina MiSeq. Data was analyzed using the LymphoTrack IGH-FR1 and Somatic Hypermutation (SHM) software. Clinical and other ancillary laboratory data were collected from the electronic medical records. Results: A total of 20 BM samples were analyzed. Patients included 8 women and 12 men with a median age at diagnosis of 68 years (range: 47 to 87). IGHV families most frequently rearranged were IGHV3 (11/20, 55%) and IGHV4 (6/20, 30%). The IGHV genes most frequently rearranged were IGHV4-34 (4/20, 20%) followed by IGHV3-23 (2/20, 10%), IGHV3-30 (2/20, 10%), IGHV3-33 (2/20, 10%), IGHV3-73 (2/20, 10%) and IGHV1-3 (2/20, 10%). Using a 98% identity cut-off value, 12/20 cases (60%) showed SHM. Review of the literature showed similar pattern of IGHV usage to other subtypes of MZL. Conclusions: We confirm that SMZL have a biased IGHV gene usage, which is in keeping with prior literature. This usage, however, has significant overlap with other subtypes of MZL considered in the differential diagnosis and therefore does not provide a means of discrimination for diagnostic purposes. This finding however, suggests that the pathogenesis of SMZL may involve epitopes or an antigenic trigger common to other indolent lymphomas. Whether particular molecular characteristics of the IG receptors might be associated with clinical outcome, genetic or phenotypic features is an area that deserves further study. Introduction: Incidence of non-Hodgkin lymphoma (NHL) is increasing in sub-Saharan Africa (SSA), primarily due to human immunodeficiency virus (HIV) and population aging. Many aggressive NHL subtypes are causally associated with Epstein-Barr virus (EBV). We hypothesized that plasma EBV DNA is a potential tumor marker associated with clinical outcome and risk of recurrence in patients with jmd.amjpathol.org ■ The Journal of Molecular Diagnostics aggressive NHL in this setting. Methods: The Kamuzu Central Hospital Lymphoma Study is a prospective cohort of patients with pathologically confirmed lymphoproliferative disorders receiving treatment under local conditions at a national teaching hospital in Malawi. Plasma EBV DNA was measured at enrollment, midtreatment, and treatment completion, using a real-time qPCR. Overall survival (OS) and progression-free survival (PFS) were evaluated by Kaplan-Meier methods, and survival differences by baseline plasma EBV DNA were assessed using the log-rank test. Results: EBV viral loads were available from 70/81 (86%) adults with aggressive NHL enrolled between May 2013 to October 2015. Diagnoses included diffuse large B-cell lymphoma (DLBCL, N=43), Burkitt lymphoma (N=6), NK and Tcell lymphomas (N=6), plasmablastic lymphoma (N=4), and other high grade B-cell lymphomas (N=11). The majority of patients (59%, N=41) were HIV-positive. Baseline plasma EBV DNA was detected in 37/70 (53%) at a median level of 3.9 log10copies/mL. Baseline plasma EBV DNA >3.0 log10copies/mL was associated with inferior OS (P = 0.007) and PFS (P = 0.009), and these associations were similar when analyses were restricted to patients with aggressive B-cell NHL (N=64, P = 0.039 and 0.048 for PFS) or diffuse large B-cell lymphoma (P = 0.007 for OS and 0.023 for PFS). In addition, serial assessments during treatment demonstrated that clinically relapsed or progressed NHL was typically associated with rising plasma EBV DNA levels. Conclusions: EBV viral load monitoring is an implementable tool which may have utility as a prognostic and response assessment tool for aggressive NHL in SSA. Introduction: Acute myeloid leukemia (AML) carries a high mortality rate and economic burden. Elucidating the heterogeneity of AML will aid in understanding the hematopoietic stem cell (HSC) self-renewal and differentiation. Though AML is classified as a myeloid neoplasm, we were interested in determining the prevalence of clonal rearrangements within the immunoglobulin heavy (IGH) and light chain (IGK), T-cell receptor gamma (TRG) loci in AML patient samples. Methods: DNA was extracted from a random sampling of 200 AML anonymized patient residual peripheral blood (PB) or bone marrow (BM) specimens. Each DNA sample was tested for 6 different PCR master mixes: IdentiClone IGH Tubes A, B, C, which target the framework (FR) 1, 2, and 3 regions, respectively; IGK Tube A, IGK Tube B, and TRG 2.0. Amplicon products were analyzed using the ABI 3500 XL instrument. Based on the florescent signals, clonal (positive) or polyclonal (negative) were assessed. Results: The IdentiClone IGH assay identified 23 (12%), 14 (7%) and 16 (8%) clonal positive samples for FR 1, 2 and 3, respectively. Combining all 3 IGH tubes increased the clonal detection rate to 28 (14%) with 172 (86%) samples determined to be negative. The IdentiClone IGK assays identified 17 (9%) and 11 (6%) clonal positive samples for Tube A and Tube B, respectively. Combining the two IGK tubes increased clonal detection to 23 (12%), with 175 (88%) determined to be negative. Combining all 5 IGH + IGK tubes, 35 (18%) clonal positive samples and 165 (83%) clonal negative samples were identified. The TRG 2.0 assay detected 85 (43%) clonal positive samples and 114 (57%) clonal negative samples. Overall, using 6 tubes of PCR MM across IGH, IGK and TRG assays, 99 (50%) samples were identified as clonal positive and 101 (50%) samples were identified as clonal negative. Conclusions: Approximate 50% of AML samples demonstrated at least one clonal IG or TCR gene rearrangement. Although it is unclear if it is the malignant myeloid cells or companion lymphoid cells that harbor these somatic gene rearrangements, the relatively high percentage in AML makes this an area worthy of further investigation. Introduction: Next-generation sequencing (NGS) of hematopoietic and lymphoid neoplasm genomes promises to revolutionize oncology, with the ability to design and use targeted drugs, to predict outcome and response, and to classify patients' responses to treatment more thoroughly using more predictive combinations of mutations. It is of critical importance that the NGS platform chosen, the chemistry utilized and the well vetted bioinformatics are then applied consistently for future adoption in clinical decisions. To illustrate the increased capacity and resolution of NGS for the comprehensive characterization of patients with hematologic cancers, we sequenced both clinical patient samples and contrived cell lines using a novel specific targeted strategy involving DNA and RNA. Using contrived cell lines, we utilize maximum gene coverage, long read lengths, and higher sequencing depth to accurately detect variants, indels and breakpoints critical in both hematologic development and for tracking minimal residual disease (MRD) and clonal architecture. We demonstrate limit of detection that allows for complete characterization of the molecular changes, with high sensitivity and specificity, thus allowing clonal architecture discovery. Methods: To examine these cancers, we targeted coding exons (571 genes) and potential genomic breakpoint regions within known somatic gene fusions (360 genes) comprising the MyHeme gene panel. We sequenced target loci on the Illumina MiSeq platform to an average depth of coverage 1000x for cell lines and patient samples. Using a custom bioinformatics pipeline, we performed thorough mutation detection analyses to identify single nucleotide variants (SNVs), indels, inversions and translocations. In addition, we calculated allelic frequencies to investigate potential aneuploidy, LOH and clonality. Using contrived samples, we were able to determine limit of detection rates, sensitivity and specificity. Results: Our analyses of targeted sequencing results from cell lines identified the published genomic variants within MyHeme targeted genes. Critically, our assay enabled detection of variants as low as 5%. In many cases, these variants were more fully characterized for their precise genomic breakpoints and inserted sequence content. Conclusions: We demonstrate that by specifically targeting driver genes using the MyHeme gene panel, we can comprehensively characterize mutations for AML, ALL, Non-Hodgkin's, Multiple Myeloma cell lines and patients with these hematologic conditions. Our results show this assay can comprehensively characterize the cancer genome of patients, identifying not only primary clones, but secondary clones that are present in at least 5% of the patient's sample. Introduction: By implementing novel technologies on the Ion S5 sequencing platform, read lengths of 600 bp were achieved, equivalent to Sanger capillary sequencing platforms. Methods: Leveraging innovations in isothermal amplification, Ion Sphere particles templating, and sequencing chemistry, we demonstrate efficient, high quality sequencing with a 600 to 700 base mode. These advancements were used to sequence human leukocyte antigen (HLA) libraries, a haplotyping application requiring long reads for phasing information. Results: Using TypeStream auto-analysis, concordance rates of 99.7% were achieved for HLA-A, B, C, DRB1, DQB1, DPB1, and DRB345 loci. The higher output of the Ion S5 system allowed us to type 96 samples on one Ion 530 chip with substantially improved typing resolution. Conclusions: We believe that these longer reads have the ability to span the 500bp intron, linking exon 2 and 3 of the Class 1 genes, leading to less ambiguity and proper gametic phasing. With long contiguous, accurate reads, we demonstrate the potential to dramatically reduce sample cost and time by multiplexing up to 192 samples on one chip and delivering results in a single day. Introduction: MYD88 L265P, a diagnostic marker for Waldenström macroglobulinemia (WM) /lymphoplasmacytic lymphoma (LPL), has also been described in IgM-monoclonal gammopathy of undetermined significance and in other types of lymphomas, therefore abating initially predicted specificity. We recently demonstrated a novel approach to assess MYD88 L265P mutation status (Burnworth et al. AMP 2015; platform presentation) . Combining flow cytometric cell sorting (FACS) with molecular analysis increased the specificity for WM/ LPL. Mutation analysis supports a diagnosis of WM/LPL with high confidence only if the MYD88 L265P is detected in both the plasma-cell and B-lymphocyte fractions. Cases with MYD88 L265P present only in either the plasma-cell or B-lymphocyte clones may represent two independent neoplasms. The current study compares analysis for MYD88 L265P in plasma-cell and B-lymphocytes populations sorted by FACS to plasma cells enriched with magnetic beads. Methods: Four specimens with known WM/LPL and 4 specimens with suspected LPL were analyzed. By flow cytometry, all specimens had restriction for the same immunoglobulin light chain in both the plasma-cell clone and the B-lymphocyte clone. FACS isolated clonal plasma and B lymphoid cell fractions, whole bone marrows as well as magnetic bead enriched CD138+ collections were tested for the presence of MYD88 mutations by Sanger sequencing. Results: MYD88 L265P was detected in the FACS isolated plasma and B lymphoid cell fractions as well as in the CD138+ magnetic bead collection of all specimens with confirmed WM/LPL. Magnetic CD138+ collections tested positive for MYD88 L265P in all specimens revealing positive MYD88 mutation status in both the FACS plasma and FACS B-cell fraction. In contrast, enriched CD138+ cell fractions collected by magnetic beads tested positive for bone marrow aspirate specimens harboring MYD88 L265P in only the FACS isolated CD19+ B lymphoid cells but not in FACS isolated plasma cells. Conclusions: Confirming the presence of MYD88 L265P in both the plasma-cell and B-lymphocyte populations is an important prerequisite to distinguish LPL/WM from related disorders. In this study we demonstrate that flow cytometric cell sorting (FACS) is required for specific MYD88 mutation status of the plasma cell component. Magnetic CD138+ bead enrichment will collect cross-contaminations of other cell types, such as B lymphocytes. Therefore, lymphoid malignancies carrying MYD88 L265P will result in false-positive test results in plasma cell collections enriched by magnetic beads. FACS rather than enrichment with magnetic beads is required for separate analysis of the plasma-cell and B-lymphocyte populations for MYD88 L265P, which is necessary for a specific diagnosis of WM/LPL. C. Moung 1 , Y. Liu 2 , J. Intrieri 1 , L. Borsu 1 , K. Nafa 1 , M.E. Arcila 1 1 Memorial Sloan Kettering Cancer Center, New York, NY; 2 Weill Cornell Medical College, New York, NY. Introduction: The identification of the BRAF V600E mutation in hairy cell leukemia (HCL) has both diagnostic and therapeutic utility. The presence of increased reticulin fibrosis in all bone marrow (BM) samples of patients with HCL constitutes a distinct challenge for mutation assessment. In the vast majority of cases, BM aspirates are scant or result in a "dry tap." In a proportion of patients, the BM may also be hypocellular. Testing methods should therefore be highly sensitive to allow detection of the mutation. In this study, we compare two techniques which incorporate locked nucleic acid (LNA) probes to increase the sensitivity of the V600E mutation detection in HCLs. Methods: HCL cases with archived DNA were selected for the study. All concurrent biopsy sections, flow cytometry and IHCs were reviewed to ensure correlation. Molecular testing was performed by Sanger sequencing with and without an LNA probe to suppress the amplification of wild-type DNA. Digital PCR (dPCR) was performed using LNA probes for both wild-type and mutant DNA on a RainDrop system. Results were compared to standard Sanger sequencing and IHC (when available) using the BRAF V600E mutation specific antibody (VE1). Results: A total of 21 HCL cases were tested by dPCR and the LNA-PCR sequencing assays (18 BMs and 3 peripheral bloods (PB)). Seventeen cases (81%) were positive for the BRAF V600E mutation and 4 were negative. There was 100% concordance between the two high sensitivity assays. Both methods proved superior to both standard Sanger sequencing and IHC. Five of 17 positive cases were not detected using standard Sanger sequencing whereas only 1 of 9 positive cases tested by IHC was negative. The sensitivity of both high sensitivity assays is 0.1%. Conclusions: The detection of the BRAF V600E in BM and PB samples is often challenging due to intrinsic characteristics of the leukemia which result in low tumor yield in the specimens that are tested. LNA probes used in LNA-PCR Sanger sequencing and digital PCR are extremely valuable tools for increasing the sensitivity of a standard assay. Both high sensitivity assays showed equal performance. dPCR has multiple advantages compared to LNA-PCR Sanger sequencing as it requires significantly less DNA and allows for a much rapid turnaround time of 1 day post extraction versus 3 days with LNA-PCR. dPCR is also quantitative, providing valuable information for monitoring of patients. gene are the most common mutations found in acute myeloid leukemia (AML) and are characterized by an aggressive phenotype with a high prevalence of relapse. Internal tandem duplication (ITD) mutations within the juxtamembrane domain are the most common mutations in the FLT3 gene. The development of a sensitive and specific assay for FLT3/ITD mutations represents a significant advancement in guiding treatment decisions. Methods: The next-generation sequencing (NGS) MRD assay was designed to target exons 14 and 15 of the FLT3 gene with a single PCR amplification. Amplicons from up to 24 samples were purified, pooled and sequenced before being analyzed using proprietary software developed by Invivoscribe. Validation was performed by spiking in fixed amounts of mutant DNA into wild-type DNA to establish a sensitivity equivalent to detection of at least one ITD-containing cell out of 10,000. The DNA input of the assay was 700 ng (>100,000 cell equivalent). The assay was applied to bone marrow DNA from patients with FLT3/ITD AML. Results: The FLT3/ITD MRD assay can detect mutations with a mutant cell sensitivity of 10 -4 (1 mutant cell in a background of ten thousand normal cells) which is equivalent to an allelic sensitivity of 5x10 -5 when a single mutant allele is present. The linearity of the assay is excellent in the mutation/total reads range of 10 -1 to 10 -5 . Excellent precision and reproducibility of the assay was demonstrated in the range of 10 -3 to 10 -5 by testing DNA from two cell lines and a clinical sample. Fifteen clinical follow-up samples determined to be negative for the FLT3/ITD mutation by the standard CLIA-certified assay were tested by the NGS MRD assay. There was no detectable FLT3/ITD mutation in six of these samples by the MRD assay, which are concordant with clinical outcomes that these patients are disease free. TheFLT3/ITD mutations were detected with read frequencies in the range of 1.38x10 -6 to 3.67x10 -3 in the rest of the 9 samples. The ITD lengths detected by NGS were the same as these detected by the CLIA-certified assay in the original diagnosis samples. These nine patients either relapsed or are still under treatment. The NGS FLT3/ITD MRD assay is highly specific and is at least two orders of magnitude more sensitive than current commercially available assays. The results of clinical sample tested by the NGS MRD assay showed 100% concordance with clinical outcomes. This assay provides a reliable tool to assess MRD in leukemia patients. A.R. Carson, Z. Xie, V. McClain, J.E. Miller, T. Stenzel Invivoscribe Technologies, Inc, San Diego, CA. Introduction: The nucleophosmin (NPM1) gene is an important marker for acute myeloid leukemia (AML) stratification. NPM1 is one of the most commonly mutated genes in AML, with mutations seen in roughly 35% of patients at diagnosis, and in approximately 60% of adult cytogenetically normal AML patients. Importantly, NPM1 mutations generally confer a slightly more favorable outcome in AML patients and can mitigate poor prognoses when they occur concurrently with activating mutations in FLT3. As a commonly mutated gene with prognostic value in AML, NPM1 is an appropriate biomarker for minimal residual disease (MRD) monitoring. MRD detection has proven to be valuable in the clinical management of patients with AML. NPM1 is of particular importance, as it was recently shown that the presence of NPM1 mutants after chemotherapy was associated with a greater risk of relapse. Thus, the development of a sensitive and reliable assay to detect NPM1 mutations at low frequencies represents a significant advancement in guiding treatment for AML patients. Methods: The assay targets exon 12 of the NPM1 gene using a single optimized PCR amplification that was developed to overcome inherent challenges caused by repetitive sequence across this locus. Using DNA input of 700 ng (>100,000 cell equivalents), up to 24 samples are amplified per run. The PCR libraries are then purified, pooled and sequenced using a next-generation sequencing (NGS) platform to a depth of at least 100,000x. Sequenced reads are then analyzed using proprietary software developed by Invivoscribe. The specificity and sensitivity of the NPM1 mutation detection were established and validated by spiking fixed amounts of mutant DNA into wild-type DNA. Results: We have developed an amplicon-based NGS assay to detect NPM1 mutations with a detection sensitivity that can be used to track MRD in AML patients. This assay can detect mutations with a mutant cell sensitivity of at least 10 -4 (1 mutant cell in a background of ten thousand normal cells), which is equivalent to an allelic sensitivity of 5x10 -5 when a single mutant allele is present. We performed linearity and show that the assay's performance is excellent in the in the allele frequency range of 10 -1 to 10 -5 . In addition, we assayed cell lines and clinical samples multiple times (>20 replicates each) to establish high precision and reproducibility for the detection of different NPM1 mutations, including the most commonly seen 4bp insertion (COSM17559). Conclusions: The NPM1 MRD NGS assay is a highly specific test that can detect NPM1 mutations with a sensitivity at least two orders of magnitude greater than current commercially available assays without sacrificing specificity. This assay provides a reliable tool to assess MRD in AML patients. Introduction: Burkitt lymphoma is one of the most common non-Hodgkin lymphomas in children and adolescents. The genetic hallmark of Burkitt lymphoma is chromosomal translocations involving the MYC oncogene. Up to 90% of the cases have a translocation between MYC and one of the three immunoglobulin loci: IGH, IGK, and IGL. A small group of MYC-negative lymphoma cases sharing similar pathomorphology, immunophenotype, and clinical presentation of Burkitt Lymphoma have been described. However, the genomic alteration of these Burkitt-like lymphomas have not been well illustrated. We present the comprehensive genomic analyses of a MYC-negative Burkitt lymphoma case using conventional cytogenetics fluorescence in situ hybridization (FISH) and high resolution SNP array technologies. Methods: A 13 year-old boy presented with right side neck swelling and an enlarged right tonsil. Tonsillectomy specimen showed malignancy resembling Burkitt lymphoma. Chromosomal analysis, FISH with CEP8/MYC/IGH tricolor dual fusion probe and KMT2A dual color break apart probe (Abbott Molecular), and high resolution SNP array were performed on the specimen. A literature review of MYCnegative Burkitt lymphoma was also conducted. Results: FISH analysis found no evidence of IGH/MYC gene rearrangement but three copies of KMT2A signals. Chromosomal analysis revealed two related abnormal clones: the stemline showed a derivative chromosome 11 consisting of an inverted duplication of 11q12.3-11q24.1 and a translocation between the derivative chromosome 11 at band 11q24.1 and the chromosome 12 at band 12p11.2; the subclone displayed an extra copy of chromosome 19. High resolution SNP array studies demonstrated a 59.5Mb mosaic duplication at chromosome 11q12.3-q24.1 followed by a 12.7Mb terminal deletion at 11q24.1; and two discontinuous duplications at 12p13.33 and 12p13.32p11.22. Array analysis also identified a copy number neutral loss of heterozygosity (cnLOH) involving almost entire long arm of chromosome 18. Conclusions: We present a case of MYC-negative Burkitt lymphoma with partial duplication and partial deletion of chromosome 11q, unbalanced translocation between chromosomes 11 and 12, and cnLOH 18q. Literature review found that duplication of proximal 11q region with or without telomeric deletion have been described in patients with Burkitt-like lymphoma without MYC rearrangement. Our case adds additional evidence supporting the association of chromosome 11q aberration and MYC-negative Burkitt lymphoma. The clinical significance of additional genomic findings in this patient remains to be explained. B. Van Deusen, M. Bessette, L. Johnson, A. Berlin, M. Banos, L.M. Griffin, E. Reckase, J.A. Stahl, A. Licon, B. Kudlow ArcherDX, Inc., Boulder, CO. Introduction: Acute myeloid leukemia (AML) oncogenesis is thought to require multiple somatic mutations in a "two-hit" process to 1) increase proliferation and 2) prevent maturation of myeloid cells. Whereas FLT3 and KIT mutations are associated with increased proliferation, NPM1, CEBPA and several other mutations can be associated with maturation inhibition. The most common mutations in AML are internal tandem duplications (ITDs) in FLT3, which are detected in more than 20% of pediatric and adult AML cases and are associated with an aggressive phenotype. As FLT3-ITD expressed kinases are sensitive to tyrosine kinase inhibitors, they are of considerable interest for the development of novel AML treatments. Capillary gel electrophoresis can detect ITDs but cannot be easily coupled with assays to detect other mutation types common in AML. Next-generation sequencing (NGS)-based methods enable comprehensive detection of multiple mutation types. However, detection of ITDs by NGS is particularly challenging, in part because of their highly variable nature and the difficulties of mapping repeated sequences to a wild-type reference. Anchored Multiplex PCR (AMP) is a target enrichment strategy for NGS that uses molecular barcoded adaptors and genespecific primers, permitting open-ended capture of DNA fragments from a single end. We present an approach based on AMP technology and a novel bioinformatics algorithm to detect ITDs in FLT3 as well as other mutation types common in AML. Methods: We developed the Archer VariantPlex Core AML library preparation assay for NGS to detect FLT3-ITDs from genomic DNA extracted from clinical samples. We designed AMP probes to cover the commonly mutated juxtamembrane domain and tyrosine kinase domain 1. We further developed a novel de novo sequence assembly algorithm based on over 2000 in silico datasets representing a large range of known ITDs. Results: In silico datasets enabled optimization of the VariantPlex Core AML analysis algorithm, resulting in the detection of over 98% of in silico ITDs with no false positives. The VariantPlex Core AML library preparation assay in conjunction with the optimized analysis algorithm enabled sensitive NGSbased detection of ITDs in 16 AML-positive blood samples. These results were consistent with results obtained from standard capillary gel electrophoresis. In addition, point mutations in the TKD of FLT3 and insertions in NPM1 exon 11 were detected in 2/7 and 5/7 FLT3-ITD positive blood samples. Conclusions: Our data show that AMP enables accurate NGS-based detection of FLT3-ITDs from clinical DNA samples. As this approach can detect multiple mutation types from a single sample, our VariantPlex Core AML kit enables simultaneous detection of multiple mutations relevant in AML. J.T. Kim, K.J. Newsom, H.Y. Wang, P. Starostik University of Florida Health, Gainesville, FL. Introduction: Enteropathy-associated T-cell lymphoma (EATL) is a rare type of malignant peripheral T-cell lymphoma (PTCL) typically affecting the small intestine. Clinically, EATL has generally shown a poor response to chemotherapy and no personalized targeted-therapies exist. EATL is known to be highly associated with celiac disease, harboring the same genetic background as celiac disease. About 90% of PTCLs carry cytogenetic abnormalities although none is specific for PTCL. In one study, EATL was characterized by frequent complex gains of 9q31.3-qter (70% of cases), or by an almost mutually exclusive 2.5-megabase loss of 16q12.1 (23% of cases). Amplification of NOTCH1 and ABL1 gene loci has been observed in EATL, but the gene mutational profile is largely unknown. The aim of this study was to analyze EATL for SNVs and short indels with a next-generation sequencing (NGS) assay. Methods: Genomic DNA from FFPE tissue was extracted using phenol/chloroform and sequenced using a LDT, GatorSeq. The assay is a hybridization capture-based next-generation targeted resequencing assay for deep sequencing of 1207 exons of 76 key cancer genes encompassing 242,205 bp of genomic DNA. 24 EATL tumors were analyzed on the Illumina NextSeq instrument. Sequence data were then processed using a customized analysis pipeline designed to accurately detect base substitutions and short insertions/deletions. Results: The patient population consisted of 24 cases with even split between males and females with average age 59.3 years. The most frequently detected alterations were seen in the following genes: BCOR, BRAF, EZH2, FAM5C, IDH1, MLL, MYH11, NOTCH1, NUP214, RUNX1, and TET2. Conclusions: EATLs display multiple genomic alterations known to be present in other hematologic malignancies. GatorSeq allows for a rapid and cost-effective mutational profiling of hematologic malignancies such as EATL. The information generated by the NGS technologies enables clinicians to make improved diagnostic and treatment decisions. (CLL) have been shown to bear mutations which contribute to oncogenic signaling and therapy resistance, or associated with biologically relevant subsets and prognosis. Therefore, we developed and analytically validated a next-generation sequencing (NGS) panel capturing 220 genes relevant to these diseases for use in a clinical diagnostic lab. Methods: Target enrichment for the NGS assay (FOCUS::Lymphoid) was developed using a hybrid-capture approach. Probes (N=4086) were designed (Nimbledesign, Roche, Inc.) to capture all coding exons of the 220 genes. Sequencing was performed on either a Miseq or Nextseq500 sequencing system. Library preparation and sequencing methods were established using a set of 210 formalin-fixed, paraffin-embedded (FFPE) or freshfrozen (FF) DNAs from DLBCL, FL, MCL, or CLL patients. Target performance metrics were set using 95 samples from within this set. Analytical validation of the assay used an additional 60 FF or FFPE DNAs and normal controls. A diluted Jurkat cell line was used as a positive control in all runs. After sequencing, alignment and variant calling were performed usingCLC Genomics Workbench v2.1 (CLCbio) according to build hg19/GRCh37 reference genome with variant annotation using Cartegenia Bench Lab NGS v4.2.2 (Agilent). Results: The average read depth was found to be 590X (FF) and 540X (FFPE). Target performance metrics were with >10% targets not achieving this metric were excluded from analysis. Two genes (CTBP2, PDPK1) and 7 targets failed to consistently achieve this and are not analyzed. The sensitivity of the assay is 5% allele variant frequency (AVF) at coverage >200X, and 10% between 60X to 200X. Comparison of DNAs (N=39) run on the NGS panel versus other CLIA-approved sequencing tests showed 100% concordance. Precision of the assay was tested using 7 FF and 4 FFPE DNAs where 99% concordance was achieved between 104 variants SNVs, 14 indels, 1 MNV) from samples run on different days. Intra-run assessment (N=3) showed 29 variants (24 SNVs and 5 indels) were consistently detected, yielding 100% concordance. Hapmap and normal DNAs (5 FF and 3 FFPE) were found to harbor no pathogenic mutations. Variants are reproducible with starting material as low as 50 ng (FFPE) and 25 ng (FF). Conclusions: The Focus::Lymphoid NGS panel enables detection of somatic variants relevant to mature B-cell neoplasms. Analytical validation demonstrates the robustness of the assay and can be used for the management of these diseases. Introduction: Bordetella pertussis (B. pertussis), the etiologic agent for whooping cough is still a public health issue despite widespread vaccination of most children in the US. This is due to the failure of some parents to vaccinate their children, and waning immunity, typically between the ages of 11 to 18 years. The very contagious nature of B. pertussis is responsible for localized community outbreaks of whooping cough requiring the need for a rapid, highly specific and sensitive test. Luminex has developed the ARIES, a sample to answer automated instrument, which is designed for their proprietary MultiCode PCR technology. The ARIES uses test cassettes into which the sample is added. The MultiCode PCR primers are manually added to a tube that clips on to the cassettes. Once placed into the ARIES instrument, nucleic acid extraction and PCR analysis are fully automated. The purpose of this study was to determine the suitability of the ARIES platform for detection of B. pertussis and B. parapertussis DNA directly from patient samples compared to our current Light Cycler-based test. Methods: B. pertussis and B. parapertussis MultiCode PCR primers and ARIES test cassettes were obtained from Luminex (Austin, TX). Testing was performed, according to standard instrument settings using Luminex SYNCT software, on 200 μL of patient sample (nasopharyngeal swabs in M4 transport media). Our current method uses the same MultiCode primers and reagents with analysis performed on the Roche Light Cycler (LC). For the latter DNA was extracted from 200 μL of patient sample using the bioMerieux EasyMag, and eluted into 50 μL of EasyMag Buffer. Five (5) μL of DNA was added to the master-mix for a final volume of 25 μL. PCR amplification and melting point analysis were performed according to the manufacturer's recommendations. Results: Forty patient samples collected as above were tested for B. pertussis and B. parapertussis DNA on the ARIES and LC instruments. Agreement between the two PCR methods was very good. One sample negative on the LC was positive by the ARIES. Three samples previously positive by the LC were negative by ARIES. These samples had a very high Ct value on the LC (>40) and had been stored in the freezer long The Journal of Molecular Diagnostics ■ jmd.amjpathol.org term. Otherwise the ARIES Ct values for nearly all the samples tested were lower than for the LC by an average of 3.2 cycles. This is expected since the ARIES utilizes a higher input of the sample DNA than the LC method. Conclusions: The ARIES instrument is a robust, simple to use, sample to result platform. The performance of the ARIES B. pertussis assay compared very well to results obtained using the Light Cycler. The major advantage of the ARIES method is the elimination of all sample processing steps. A. Rezaei 1 , N. Guaring-Angulo 1 , J. Walker 2 , J. Graham 1 , N. Dhiman 1 1 med fusion, Lewisville, TX; 2 Roche Diagnostics, Indianapolis, IN. Introduction: As one of the largest regional laboratories and an integrated Roche Molecular Center of Excellence for molecular diagnostics in Texas, our objective was to consolidate and fully automate our routine viral load testing. We performed a comparative workflow analysis of the Cobas p630 Cobas AmpliPrep/Cobas TaqMan (CAP/CTM) with the Cobas 6800 System as a high throughput molecular testing solution. Our aim was to integrate best practices with lean workflow principles and automation to maximize efficiency, improve turnaround time (TAT) and reduce waste. Methods: Current workflow was evaluated in terms of total volumes, batch sizes, TAT, waste generated and hands-on and walkaway time during pre-analytical, analytical and post-analytical testing phases. These data were analyzed in a performance matrix to identify opportunities for error, overproduction, wait times, waste and spaghetti maps. The outcomes were used to drive recommendations for improved workflow using the Cobas 6800 System. Results: As a leading reference laboratory for transplant centers and AIDS clinics, we consolidated Cytomegalovirus (CMV), Human Immunodeficiency Virus (HIV), Hepatitis B and C (HBV and HCV) viral load assays on CAP/CTM platform in 2013 on International Scale. Current CAP/CTM usage for one shift operation is at a median of 235 (range 171 to 491) samples and 5 (range 2 to 8) runs per day, posing a challenge to TAT. The current workflow had 56 process steps and the performance matrix showed inefficiencies at batched rack preparation, work list review and manual pipetting workaround for short volumes. There were a few efficiency advantages, including complete chain of custody and directionally LIS interface. We observed a ~63% reduction in process steps (35 out of 56) in transitioning to the COBAS 6800, including elimination of 70% of the potential error points. This had a projected time savings of ~1.7 hours per day, improved total TAT by ~4 hours and reduced solid wastes by 80%. Conclusions: The Cobas 6800 System dramatically streamlined and standardized the process for viral load testing by reducing the number of process steps, eliminating potential error points and reducing TAT, hands-on time and waste. Other efficiencies gained were the consolidation of seven instruments into a single instrument, which lead to space and maintenance savings. In conclusion, the Cobas 6800 System was identified as a lean option for molecular viral load testing for a large reference laboratory. T. Sundin 1 , D. Seidman 1 , K. Slaughter 1 , M. Aunchman 1 , L. Grissom 1 , J. Inman 1 , R. Post 1 , A. Shean 1 , K. Gonzalez 2 , N. Suhail 1 1 Sentara Healthcare, Norfolk VA 2 Focus Diagnostics, Cypress CA Introduction: Herpes simplex virus 1 and 2 (HSV-1 and HSV-2) are relatively large double-stranded DNA viruses associated with a wide range of clinical infections including genital, dermal, ocular, and central nervous system. Prompt detection and diagnosis of HSV illnesses are critical for infection control and treatment. Due to numerous biological areas known to harbor HSV, a vast array of clinical specimen sources may be obtained to test for accurate diagnosis of HSV infection. In this study, performance of the Focus Diagnostics HSV real-time PCR assay for quantitative detection of HSV-1 and -2 in whole blood and CSF and qualitative detection of HSV-1 and -2 in APTIMA GEN-PROBE and ThinPrep samples was evaluated. Methods: A total of 441 samples (blood/plasma [n=180], CSF [n=72], ThinPrep [n=92], APTIMA swabs [n=97]) were tested utilizing a combination of retrospective samples and control-spiked samples. Samples were extracted using Roche MagNA Pure 96 system with the DNA Isolation Kit DNA/Viral NA SV 2.0 and subsequently tested by the Focus 3M TM Integrated Cycler using Focus HSV-1 or HSV-1 GEN II and HSV-2 Primer Pairs and 2.5X Universal Master Mix. Linear ranges, accuracy, analytical sensitivity, analytical specificity, precision, and limit of detection (LOD) were assessed. Results: Real-time PCR testing of 441 specimens, revealed the Focus HSV assay has excellent analytical sensitivity and analytical specificity [HSV-1 100% positive predictive value (PPV)/ 100% negative predictive value (NPV); HSV-2 99.32% PPV/ 99.03% NPV]. Out of 441 samples, only 2 discrepant results were found. Referee testing on discrepant samples with a third party also using the Focus HSV assay were concordant with our results. Inter-assay variability was measured using log values from AcroMetrix HSV-1 and HSV-2 low and high positive CSF controls, with average coefficients of variation (CV) of 1.8% and 1.1%, respectively. Quantified AcroMetrix HSV-1 and HSV-2 Plasma Panels were analyzed with 5 replicates of each dilution to determine intra-assay variability [HSV-1 2.9% (CV); HSV-2 4.4% (CV)] and linearity (250-1,000,000 copies/mL). Additional serial dilutions of the panels were evaluated to determine an LOD of 250 copies/mL. Conclusions: These results demonstrate the benefit of the Focus Diagnostics HSV assay for quantitative results on whole blood and CSF samples, and qualitative results for ThinPrep and APTIMA GEN-PROBE samples from patients with suspected HSV infection. Furthermore, qualitative and quantitative assays can be run simultaneously in a clinical laboratory improving workflow and decreasing control costs. Use of this molecular assay will ensure efficient, reliable and reproducible analysis of HSV infections in a single real-time PCR assay. Kidneys from hepatitis C virus-positive (HCV+) organ donors are almost exclusively transplanted into HCV+ recipients. Because the number of HCV+ kidneys far exceeds the number of patients willing and able to accept such organs, nearly two-thirds of these kidneys are discarded every year. The development of new therapies for HCV that cure more than 95% of patients, coupled with the vast shortage of available kidneys, have led some transplant professionals to consider transplanting HCV+ kidneys into HCV-negative recipients. One potential barrier to doing so is the ability to rapidly genotype organ donors for HCV, since viral genotype guides therapy and no pan-genotypic antiviral regimens exist that can safely be used in patients with impaired kidney function. Also, it is not known whether donor stabilization procedures, such as fluid or drug infusions, affect genotyping ability (e.g., by reducing viremia or introducing PCR inhibitors into circulation). Our aim was to determine whether archived plasma samples from past organ donors could be accurately genotyped for HCV. Methods: A local organ procurement organization performed a retrospective review of their database to identify previous organ donors that were HCV+ on a qualitative nucleic acid screen. Archived frozen plasma samples were obtained for 17 such donors, along with donor demographics. RNA was extracted from the thawed samples using the QIAamp DSP Viral RNA kit on the QIAcube (Qiagen, Valencia, CA). HCV genotypes were detected using the eSensor HCVg Direct Test (RUO) and XT-8 System (GenMark Diagnostics, Carlsbad, CA). All samples were also genotyped by Sanger sequencing (Retrogen, San Diego, CA) and subjected to quantitative HCV viral load testing (COBAS AmpliPrep/TaqMan system, Roche Diagnostics, Indianapolis, IN). Results: The samples obtained were collected over a 10 month period primarily from Caucasian donors, both male and female, between 21 and 57 years old. All samples were successfully extracted and genotyped. The majority of samples (n=14) were HCV Genotype 1a with the remainder being Genotype 2b (n=1) or Genotype 3 (n=2). All genotyping results were concordant with those obtained via Sanger sequencing. The average HCV viral load across the samples was ~ 1.6 million IU/mL (range: ~16,000 IU/mL to 7 million IU/mL). Conclusions: Albeit a limited sample size, we found that viral RNA from plasma of organ donors can be successfully extracted and genotyped for HCV. Treatment provided to stabilize donors prior to organ procurement does not appear to interfere with the ability to extract or amplify viral RNA. The capacity to provide an accurate HCV genotype suggests that clinical trials designed to transplant HCV+ kidneys into HCV-negative recipients could be feasible. Introduction: Mortality rates for women with cervical cancer in low-and middleincome countries (LMICs) are approximately three times higher than in high income countries. Limited access to high-risk HPV (hrHPV) screening and adequate followup are major contributors to these statistics. Successful use of molecular analysis for hrHPV testing has been limited to established laboratories experienced with highly complex testing techniques, effectively creating a significant barrier to increased hrHPV screening of at-risk populations in LMICs. Here we report the development of a simplified point-of-care (POC) field protocol and results from field-testing of an hrHPV molecular assay in the LMIC of Honduras. Methods: A total of 662 women from outreach clinics in a textile factory in San Pedro Sula (402), and a rural village clinic in Locomapa (260), were screened for 14 hrHPV types in portable rudimentary POC field laboratories. Three cervical specimens from each women were collected using standard sampling techniques. One sample was immediately preserved for follow-up Pap smear testing at La Liga Contra el Cancer in San Pedro Sula and one sample was inactivated and shipped to the Laboratory of Clinical Genomics and Advanced Technologies (CGAT) at Dartmouth-Hitchcock Medical Center for confirmation studies. DNA from the remaining sample was extracted into a crude jmd.amjpathol.org ■ The Journal of Molecular Diagnostics lysate and hrHPV testing was performed onsite using the QuanDx High-Risk HPV Genotyping Assay on the ZeeSan SLAN-96S RT-PCR System following a modified protocol, including applying crude extract directly to lyophilized reagents. The presence of any of the 14 hrHPV types was determined through melt curve analysis. Results: Of the 662 samples tested, the total validity rate was 76.2% (505/662), with 73.1% (294/402) and 81.4% (210/260) from the factory and village respectively. Swabs tested by a similar method in the CGAT laboratory had a validity rate of 97.1% (136/140) in a controlled clinical laboratory environment. 13.0% (86/662) of samples tested in Honduras and 14.3% (20/140) of samples tested at CGAT were positive for hrHPV. Conclusions: The successful establishment of two portable field laboratories and implementation of high throughput molecular testing has shown the potential for same-day diagnosis of hrHPV in LMICs. Rapid diagnosis would allow for immediate health counseling and services for patients who can be challenging to track due to poor communications infrastructure and poverty. Increasing the access for large numbers of women to molecular testing for hrHPV could dramatically impact mortality rates of cervical cancer in LMICs by correctly identifying women at highest risk for cervical cancer and immediately directing clinical follow-up as appropriate. Introduction: Persistent infection with high risk (HR) human papillomavirus (HPV) is a principal cause of cervical cancer and is associated with greater than 99% of cervical cancer cases worldwide. Numerous FDA approved or cleared laboratory testing platforms exist for the detection of HR HPV in gynecological specimens. The goal of this study was to verify and implement the Cobas HPV Test (Roche Diagnostics, Inc., Indianapolis, IN) in our laboratory and to evaluate the effects on various aspects of specimen workflow. Methods: Verification studies, using 421 patient specimens, were performed to switch clinical HR HPV testing from the Cervista HR HPV and the HPV 16/18 genotyping assays (Hologic, Inc., Marlborough, MA) to the Cobas HPV Test. The Cobas assay was always performed in the HPV 16/18 genotyping mode. Discrepant results were resolved by repeat testing as well as by using the Linear Array HPV Genotyping Test (Roche Diagnostics, Inc.). Data were compared for various workflow metrics between the two testing platforms. Results: The overall, positive, and negative agreements between the Cervista and Cobas HR HPV assays were 92.9%, 87.9%, and 94.3% respectively. Notably, our high positive agreement was due in part to our laboratory's data analysis algorithm of Cervista HR HPV results. The majority of discrepant results were HPV positive on the Cobas and HPV negative on the Cervista assay, suggestive of increased analytical sensitivity of the Cobas assay. The percent positivity by cytologic diagnosis was evaluated after Cobas implementation and ranged from approximately 3% positive for negative for intraepithelial lesion or malignancy (NILM) to approximately 85% positive for high-grade squamous intraepithelial lesion (HSIL). Upon switching testing platforms various workflow improvements were observed. For a subset of samples, the Cobas assay eliminated the need for a reflex HPV 16/18 genotyping assay which reduced TAT by approximately three days and the physical distance traveled across our Cytopathology and Molecular Pathology laboratories by 1.5 miles. The Cobas assay also eliminated secondary review of results and reduced the number of indeterminate/inconclusive results as well as the repeat testing rate. All of these improvements resulted in decreased laboratory costs. Conclusions: Improvements following this laboratory testing methodology change were observed in specimen workflow, analytical testing metrics, and overall laboratory operations. Taken together, these improvements translate into an earlier HR HPV laboratory result for patients and clinicians within our health system with a potential for overall improved patient care. Recently, different classes of promising new anti-TB agents, as the Benzothiazinones (BTZs), that results active against M. tuberculosis drugsusceptible, MDR and XDR clinical isolates, were discovered. Many of these drugs hit the same target, DprE1, an enzyme involved in arabinogalactan biosynthetic pathway. Based on the analysis of dprE1 gene, two ready-to-use assays were developed, one for TB diagnosis and identification of MTBC or MAC infections and one to identify single point mutations associated with resistance in M. tuberculosis. Methods: The 2 assays were based on Real-Time PCR analysis. Each test tube contains all the required PCR components in a freeze-dried form, exploiting STAT-NAT technology that allows the room-temperature storage of the mix for 2 years. The MTBC/MAC screening kit include specific sets of primers and probes, one to amplify a fragment of M. tuberculosis IS6110 direct repeat region and one to amplify a fragment of M. avium dprE1 gene. For drug resistant TB assay, the STAT-NAT technology was associated with a novel quenching probe (Qprobe). The assay allows the discrimination between the DprE1 inhibitors and Rifampicin (RIF) sensitive/resistant strains. Specific sets of primers and Qprobes were designed to amplify fragments of M. tuberculosis dprE1 gene and rpoB gene. In each assay, another set of primers and probe was included as an internal amplification control. Results: The LOD of MTBC/MAC screening assay was 1 and 10 copies/reaction for MTBC and MAC species, respectively. The assays did not cross-react with any of the other mycobacterial species tested. Regarding the drug resistant TB assay, the genomic DNA obtained from M. tuberculosis DprE1 inhibitors-resistant mutants, M. tuberculosis MDR and wild-type strains, was tested using this assay. The results showed unambiguous discrimination between mycobacterial strains that are resistant or sensitive to RIF or DprE1 inhibitors. Conclusions: The high-sensitivity and specificity of these two assays, associated with the ready-to-use and room temperature storage of the STAT-NAT technology, would have a direct impact on the early and correct management of the affected patients. Overall, these technologies would be a valuable tool in the rapid diagnosis of other tuberculosis drug-resistance. Introduction: Escherichia coli is a common cause of septicemia and urinary tract infection, and the increasing rate of quinolone resistance has become problematic. The global spread of quinolone-resistant E. coli sequence type (ST) over the past decade has been assessed using DNA sequencing technique. However, they have limitations in discriminating genetically similar microbes and requires time and cost. Therefore, we investigated the applicability of Raman spectroscopy which measures the spectral phase of molecular vibration as a typing tool to discriminating ciprofloxacin (CIP)-nonsusceptible E.coliSTs. Methods: CIP-nonsusceptible E. coli isolates, collected from blood cultures, were pre-characterized regarding the minimum inhibitory concentration (MIC) of CIP and extended-spectrum betalactamase (ESBL) production. E. coli ATCC 25922 was used as susceptible control, and the resistance E. coliSTs between ST131 and ST1193 isolates were analyzed and compared with Raman spectroscopy technique. All Raman spectra were obtained through gold-coated surface-enhanced Raman scattering (SERS) technique using a SENTERRA confocal Raman system (Bruker Optics). The enhancement factor of this gold-deposited substrate was roughly calculated 2.4×10 7 . The principal component analysis (PCA) algorithm and Multi-support vector machine (SVM) classifiers with a one-against-one approach onto this two-dimensional plot were used to discriminate between ST131 and ST1193 isolates. Results: The optimal separating lines could be obtained from the SVM classification algorithm using the training data (n=20) from each group. A linear kernel on the score plot of PC1-PC3 clearly showed a difference between the control and two CIPnonsusceptible E. coli STs. The score plot of PC1-PC2 showed individual characteristics of three groups, which yielded a sensitivity of 100% and a specificity of 100% for discriminating between strains of ST131, ST1193 and controls. Conclusions: This study suggested the high sensitivity of Raman spectroscopybased identification method for discrimination of different CIP-nonsusceptible E. coli STs without additional labeling. H. Lee, Y. Kim, J. Choi, H. Kanga, S. Cho, T. Park Kyung Hee University, School of Medicine, Seoul, Republic of Korea. Introduction: The microbiome study using NGS-metagenomics on the human body was widely performed on the gut, skin, vagina, oral cavity. However, respiratory tract has not been actively studied, and the microbiome of healthy lungs is very poorly understood due to difficulties in obtaining proper respiratory samples. Few studies on lung microbiome showed diverse results owing to various specimen types, different sampling methods and study designs. We investigated the microbiome on healthy trachea with a controlled sampling method. Methods: The subjects were adult patients who needed general anesthesia for operation no longer than 2 hours. We excluded the patients with any conditions related with respiratory tract (infection, smoker, allergy, cancer and decreased pulmonary function). Patients with infections on other sites, metabolic or nutritional abnormalities were also excluded. We collected tips of intubation tubes used for general anesthesia. The depth of intubation were 21cm for female and 23cm for male. The collected tips were washed with DEPC water and the spin-down pellets were used for NGS-metagenomics with Mi-Seq system (Illumina Inc.). Paired-ends fastq data were merged using FLASH (ver. 1.2.11). Pre-processing and clustering of reads were done with CD-HIT-OTU software. Taxonomic assign and diversity analysis were done with QIIME-UCLUST. Cultures were performed in both aerobic and anaerobic conditions with media routinely used in clinical microbiologic laboratory, and obtained colonies were identified with MALDI-TOF MS (Bruker) to the species level. Results: The included patients were eight females with breast cancer stage under T1 and one male with hydrocele. All of the samples showed plateau in rare fraction curve and sufficient reads were obtained. The median (range) number of identified species and genus by MALDI-TOF MS were 11 (8 to 17) and 5 (3 to 8) whereas the operational taxonomic unit (OTU) by NGS-metagenomics were 64 (53 to 71) and 53 (45 to 52), respectively. The top 5 frequent OTU in genus level (% total, range) were Prevotella The Journal of Molecular Diagnostics ■ jmd.amjpathol.org (23.1, 31.9 to 9.6), Streptococcus (12.2, 25.2 to 4.6), Veillonella (8.9, 10.8 to 5.7), Neisseria (8.3, 27.7 to 1) and Alloprevotella (8, 15 to 2.4), respectively. The samples showed relatively consistent patterns of OTU composition among enrolled subjects. Conclusions: Our controlled sampling method using the endotracheal tube was useful for collecting healthy respiratory tract specimens without ethical issue. The OTU and their compositions shown in our study were consistent with previously reported studies on healthy lung microbiota. To establish the diagnosis of acute HIV infection is extremely important from a public health point of view. This study aims to detect acute HIV infection in patients seen in emergency cares with signs and symptoms of mononucleosis or aseptic meningitis syndrome, but who have negative results to serological tests for mononucleosis, CMV and toxoplasmosis; as well as for the detection of enterovirus and HSV I / II by nucleic acids amplification techniques (NAT). Methods: This study was conducted from February 2014 to February 2015 in 5 different hospital settings of the Hospital Israelita Albert Einstein, four exclusively private patient intake sites and one site for admitting patients from the public health system. It was included patients over 18 years with signs and symptoms suggestive of mononucleosis syndrome (fever, sore throat, cervical nodes) in which the doctor asked for mononucleosis, cytomegalovirus and toxoplasmosis serologies whose results were negative; as well as patients with aseptic meningitis (fever, headache, nausea or vomiting and neck stiffness) who underwent lumbar puncture with Enterovirus and Herpes virus I and II detection request by NAT with negative results. The patients were analyzed for the presence of anti-HIV antibodies as well as the presence of viral RNA (with a sensitivity of 20 copies / ml and linearity ranging from 47 to 10,000,000 copies/ml). Results: In a collected sample of 30 patients during the years 2013-2015, two of them were positive for HIV serology and viral load. The sample population consisted of 16 females and 12 males aged between 18-49 years and the two positive cases were males with 32 and 45 years old, respectively. Acute HIV infection incidence was 6.67% in the target population of the study.In the study, the incorporation of the following laboratory tests were necessary: HIV viral load ($ 73.00) and anti-HIV serology ($ 17.26). Conclusions: Our results demonstrate that it is important to detect acute HIV infection to prevent HIV spreading to other patients and to help acute HIV patients to receive an appropriate antiretroviral therapy Introduction: Increasing prevalence of antibiotic resistance (AbX) bacteria in the last decade is of great concern in healthcare settings. The most common type of blaCTX-M. The CTX-M genes encode for a complex and non-homogenous group of enzymes that are seen in Enterobacteriaceae. Klebsiella pneumoniae carbapenemase (KPC) is the most common carbapenemase gene type reported in the US. KPC is produced by the blaKPC gene in Enterobacteriaceae, Acinetobacter and Pseudomonas species and confers resistance to all (ß) lactam agents. Of greater clinical significance worldwide are the carbapenemase metallo-(ß)-lactamases (MBL) variants: blaNDM, blaVIM and blaIMP seen in Enterobacteriaceae and P. aeruginosa. We have designed a multiplex PCR assay that detects most variants for these five genes that confer AbX in various gram negative bacteria. Methods: Organisms were ordered from American Type Culture Collection. Samples were extracted on the MagNA Pure LC or MagNA Pure Compact. Nested PCR and detection was done with either dual labeled hydrolysis probe or LC Green signal and melting curve analysis on the Bio-Rad CFX96 Touch System. Nested PCR primers for outer PCR1 and inner PCR2 were designed for; IMP, KPC, NDM, VIM and CTX-M groups; 1, 2, 8, 9 and 25. All 5 assays were multiplexed using the outer PCR primers for 25 cycles comprising PCR1. The amplicon was diluted 1:100 and added to 6 subsequent PCR reactions with inner primers and either a hydrolysis probe or LC Green. The CTX-M types 1 and 9 were multiplexed together for PCR2 and CTX-M groups 2, 8 and 25 were multiplexed together. Eighty two bacterial and viral isolates were run for all 5 AbX markers. We designed 9 positive control, 500 base pair gBlocks that were synthesized by Integrated DNA Technologies. All positive detections were confirmed by Sanger sequencing. Results: The assay was performed for 82 bacterial and viral isolates. Eighteen isolates had previous ESBL, carbapenemase or MBL detection by other molecular methods. Some samples had multiple resistance genes. All eighteen expected positive samples (12 CTX-M, 2 IMP, 1 KPC, 1 NDM, and 4 VIM) were detected by the multiplex assay and confirmed by Sanger sequencing. No unexpected negatives or positives were encountered. All assay controls passed. Limit of detection for each gene was determined to be at least 1000 copies per reaction. Conclusions: We have designed a real-time PCR assay that is sensitive and specific for 5 antibiotic resistance gene families. The assay can be run in 3 hours. It is sensitive and specific for the target genes with no observable interference from human genomic DNA. This assay is not FDA approved. J.R. Rebello Pinho 1,3 , A.R. Marra 2,3 , D.T. Malheiro 2 , S.C. Oller 2 , R.A. Santana 2 , R. Sitnik 2 , R.C. Petroni 2 , G.F. Dastoli 2 , E.A. Rossetto 2 , T.Z. Camargo 2 , P.D. Gonçalves 2 , M.D. Martino 2 , J. Pasternak 2 , C.L. Using the Cepheid SmartCycler S.J. Deharvengt 1 , S.A. Turner 1 , L.S. Kennedy 2 , G.J. Tsongalis 1 1 Dartmouth Hitchcock Medical Center, Norris Cotton Cancer Center and Geisel School of Medicine, Lebanon, NH; 2 Norris Cotton Cancer Center and Geisel School of Medicine, Lebanon, NH. Introduction: Cancer of the uterine cervix (CC) is the most common cancer in women in Honduras. The main risk factor is infection with one or more high-risk human papillomavirus (hrHPV) types. The aim of this study was to develop an affordable assay to detect the genotype-specific prevalence of the 14 hrHPV types for CC. Methods: Primer3 Input and pave Papilloma virus genome database websites were used to design primer sets. The uMelt SM Software was used to predict the melt curve shape and melting temperature (Tm). DNA samples included human genomic DNA (gDNA) (Promega), gBlocks gene fragments of the 14 hrHPV types (IDT) and DNA extracted from previously tested patient samples. The enzyme used for PCR amplification was SsoFast EvaGreen (Biorad) on the Cepheid SmartCycler. Results: We selected one primer pair for each of the 14 hrHPV types producing a specific amplicon and giving a single peak melt curve. Primer sets were combined and tested with only gDNA to confirm that no cross-reaction occurred between the primers. This reaction was included in every run as a negative control. The primer combinations without a high background were used with up to 5 gBlocks as template in one reaction. Melt curves were obtained with Tm differences that could differentiate hrHPV types. This reaction was included in every run as a positive control. The multiplex PCR reactions were then run using DNA extracted from cervical swab specimens collected in Honduras, previously tested with the QuanDx HPV assay. The results were highly consistent between the two methods. Conclusions: The SmartCycler can be used for multiplex Real-Time PCR melting curve analysis for identifying up to five hrHPV types in one reaction. The main advantages of this technique are its affordability and flexibility. 1 1 Louisiana State University Health Sciences Center, New Orleans, LA; 2 University Medical Center, New Orleans, LA. Introduction: Chlamydia trachomatis and Neisseria gonorrhoeae (CT/NG) are the most common sexually transmitted bacterial infections in the United States. Infection with either organism can be asymptomatic and is associated with significant urogenital and reproductive tract sequelae in men and women. CT/NG infections among men who have sex with men (MSM) often involve in the rectum and pharynx, and commonly occur without concomitant urethral infection. Currently, CT/NG infections are most commonly detected using nucleic acid amplification tests (NAATs) due to enhanced analytical sensitivity, specificity and ease of specimen transport. The Roche Cobas CT/NG v2.0 test is an automated, in vitro NAAT for the qualitative detection of CT/NG DNA in several urogenital specimen types. Using this test on the Cobas 4800 system, the objective of this study is to validate CT/NG detection from rectal and pharyngeal swab specimens. Methods: Using the CT/NG v2.0 and the Aptima Combo 2 CT/NG assays, CT/NG detection was performed on 20 dual rectal swab samples and 18 dual pharyngeal swab samples collected from the HIV Outpatient Clinic (HOP) in New Orleans, LA. An additional comparative assessment using the CT/NG v2.0 was performed using a panel of known positive/negative rectal (n=15) and pharyngeal (n=19) specimens from a secondary lab. Lastly, CT spiking experiments were performed to further assess assay performance in pharyngeal swab specimens. Results were analyzed using Cohen's kappa statistics. Results: Sensitivity and specificity values for both rectal and pharyngeal CT/NG were 100% between the CT/NG v2.0 and the Aptima Combo 2 CT/NG assays. In the secondary comparison of known positive/negative specimens on the Cobas CT/NG v2.0 test, sensitivity and specificity values for pharyngeal CT/NG and rectal CT were both 100%. One rectal specimen deemed NG-positive at the reference lab was not detected in our testing, yielding a sensitivity and specificity of 89% and 100%, respectively. CT spiking experiments showed excellent interassay reproducibility with coefficients of variation that were less than 5%. Kappa coefficients we're classified as very good or perfect for all comparisons. J. Li 1 , S. Favaloro 2 , C. McGowin 1 , C. Kletecka The results confirm that CT/NG detection from rectal and pharyngeal swab samples can be performed successfully on the Cobas 4800 system. Introduction: Cryptococcus is a human pathogenic fungus and is implicated in pneumonia and meningoencephalitis. Cryptococcus spp. have been reported to cause 15% of AIDS-related mortality. C. neoformans is found worldwide and C. gattii has been associated with tropical as well as temperate locations such as British Columbia and Northwest US. Molecular methodologies including PCR in its many forms are essential for fast, sensitive detection jmd.amjpathol.org ■ The Journal of Molecular Diagnostics of C. gattii and C. neoformans infections. Methods: Two independent nested realtime PCR assays were each designed to detect both C. gattii and C. neoformans. The cox2 and atp6 genes were chosen because they are located in the mitochondrial genome of Cryptococcus and are present in multiple copies, increasing sensitivity of detection. Each PCR assay was designed with an outer primer pair, an inner primer pair, and a hydrolysis probe for fluorescent signal generation. The inner primers were tailed with M13 tails to facilitate sequencing. The outer primer PCR product was diluted 1:10 and added to the inner PCR reaction for an effective 1:100 dilution. Both assays were evaluated for C. gattii and C. neoformans identification using a panel of reference strains representing all 4 genogroups of C. gattii and various strains of C. neoformans. Further testing of 25 culture-confirmed clinical cerebrospinal fluid (CSF) specimens was performed. CSF quantitative culture, cross-sectionally collected during antifungal therapy, ranged in burden from 10 CFU/mL to >100,000 CFU/mL CSF. Nucleic acid was extracted from 200 μL CSF and 10 μL were used for PCR. Results: The COX2 and the ATP6 realtime nested PCR assays detected all C. gattii and C. neoformans reference strains including the 4 genogroups of C. gattii. The limit of detection of both assays was between 100 copies/reaction to 1,000 copies/reaction when compared to synthetic template. The COX2 and ATP6 assays did not detect any of the non-cryptococcal organisms tested. Compared with culture, the COX2 and ATP6 assays each showed 80% (20 of 25) concordance. Of 18 culture-positive CSF specimens from patients with confirmed Cryptococcus infection, both assays resulted in 13 (72%) positive detections. Bi-directional sequencing revealed that the COX2 assay did not amplify a C. gattii strain at 660 CFU/mL and the ATP6 assay did not amplify a C. neoformans strain at 120 CFU/mL. The COX2 and ATP6 assays tested negative in all culture negative CSF. Conclusions: The two PCR assays are able to detect C. gattii and C. neoformans species faster than culture, suggesting reduction in mortality and morbidity with prompt treatment. A comparison with quantitative culture shows the described COX2 and ATP6 assays each perform well for testing CSF specimens. These assays are not FDA-cleared. Introduction: Fever is a common medical problem that accounts for a large proportion of pediatric hospital visits. Symptoms displayed are often non-specific and it is difficult to distinguish between viral syndromes, serious bacterial infection (SBI), or non-infectious causes of fever. Even though the majority of cases are thought to be viral in origin, many patients receive extensive evaluation including blood, urine, and cerebrospinal fluid (CSF) cultures and then require hospitalization until an SBI can be ruled out. A rapid, easy-to-use, comprehensive diagnostic test could benefit patient care by potentially reducing antibiotic use or influencing hospital admission/discharge decisions. To aid in viral identification associated with pediatric fevers, BioFire Diagnostics is developing the FilmArray (FA) Febrile Infant (FI) Panel for use on the FA System. The FA FI Panel simultaneously tests for ten viruses using 200 μL of plasma. Two minutes of hands-on time are required and comprehensive results are returned in about an hour. Methods: One hundred ninety six (196) fever (reported temperature > 38.0°C) or a blood/CSF culture where an SBI was suspected, were evaluated in this study. This study was approved by participating institution's review boards. Residual standard of care (SOC) specimens as well as prospectively collected samples were evaluated by the FA FI Panel and results were compared to SOC testing. Results: Eighty viruses were observed in 196 plasma specimens (41% positivity). The most frequently detected viruses were human herpesvirus 6 (19), adenovirus (14), human herpes virus 7 (13), and Epstein-Barr virus (13). The FA FI Panel provided a positive virus identification in 34% (15/44) of cases where all SOC test results were negative. Many viruses included in the FA FI Panel are not part of SOC testing, however the FA FI Panel was concordant with positive SOC results for adenovirus and enterovirus in 60% (4/7) of cases. Discordance may be due to different sample matrices (nasal swab v. plasma). Conclusions: Because febrile pediatric patients often exhibit non-specific symptoms, discriminating between viral, bacterial, or non-infectious causes of high fever is challenging. Testing with the FA FI Panel resulted in a virus detection in 41% (80/196) of samples. Detection of some viruses, such as enterovirus, is known to be associated with a reduced probability of SBI. These results suggest that the FA FI Panel could be a useful system to rapidly aid in identifying pathogens causing fever in infants. ID17. Evaluation of the ARIES HSV-1/2 Assay for the Detection and Differentiation of Herpes Simplex Virus 1 and 2 from Blood, Respiratory, and Ocular Specimens M. Espy, C. Irish, M. Binnicker Mayo Clinic, Rochester, MN. Introduction: Herpes simplex virus types 1 and 2 (HSV-1/HSV-2) can be found in a wide range of clinical specimens, including genital, dermal, cerebrospinal fluid, respiratory, ocular, and blood. The use of real-time PCR is a sensitive and specific method for the detection and differentiation of HSV-1/2 in these specimens. Historically, specimens submitted for real-time PCR are processed to extract nucleic acids, followed by amplification and analysis by real-time PCR. Recently, Luminex Corporation (Austin, TX) developed an automated platform (ARIES) that can perform all of these functions on a single instrument. The ARIES is FDA-cleared for testing genital/dermal specimens. In this study, we sought to evaluate the performance of the ARIES HSV-1/2 system using a variety of non-FDA cleared sample types, including blood, respiratory, and ocular specimens. Methods: One hundredseventeen retrospective clinical specimens (respiratory [n=66], ocular [n=28], and whole blood [n=23] that were determined to be positive for HSV-1 (n=72), HSV-2 (n=9), HSV non-typeable (n=6) or HSV negative (n=30) by routine testing were included in this study. Routine testing consisted of extraction of 200 μl of sample on the MagNA Pure (Roche Diagnostics), followed by analysis of 5 μl of extract using the Roche HSV-1/2 analyte specific reagents (ASR) on the LightCycler 2.0 (Roche). An aliquot of each sample (200μl) was also tested by the ARIES assay per manufacturer's instructions. Results were analyzed by comparing the ARIES results to the results obtained by routine testing, which was considered the reference standard for this study. Results: Following testing of the one hundred-seventeen specimens, the ARIES HSV-1/2 system demonstrated a sensitivity of 100% for HSV-1 (72/72) and HSV-2 (9/9) when compared to the Roche HSV-1/2 ASR. The six samples that were resulted as "HSV non-typeable" by routine testing were determined to be positive for HSV-1 by the ARIES system. The specificity of the ARIES HSV-1/2 assay was found to be 100% (30/30). Conclusions: Testing of blood, respiratory and ocular specimens by the ARIES HSV-1/2 assay showed comparable performance to routine specimen processing and testing. The results of this study suggest that the ARIES system may serve as an option for clinical laboratories that are seeking an automated platform that performs all current PCR functions on a single instrument. Kubasek 2 , R. Widen 2 1 Federal University of São Paulo, Sao Paulo, Brazil; 2 Tampa General Hospital, Tampa, FL; 3 Children's Healthcare of Atlanta, Atlanta, GA. Introduction: Achromobacter xylosoxidans (AX), Burkholderia cepacia (BC), Pseudomonas aeruginosa (PSA) and Stenotrophomonas maltophilia (SM) are non-fermentative Gram negative rods frequently isolated from the respiratory tract of Cystic Fibrosis (CF) patients. The aim of this study was to develop a multiplex PCR test to detect AX, BC, PSA and SM directly from CF patient's respiratory samples using the open mode system of the BD MAX platform (BDM). Methods: A total of 402 CF respiratory samples, with concurrent culture, were tested by the new CF BDM PCR test. A specific set of forward and reverse primers, as well as a probe for each one of the four targets tested, were designed inhouse. In addition, a set of primers and a probe to detect the beta globin (BG) gene was used as an internal control. Before testing, samples were transferred into a 500μL tube of SL solution (Copan Diagnostics, California) and held for 15 minutes at room temperature for liquefaction. Liquefied samples were treated with Proteinase K at 60°C for 30 minutes and at 95°C for 5 minutes. After treatment, 250μL of each sample was inoculated into the BD MAX Sample Preparation Reagent Tubes. Extraction and multiplex PCR were performed by the BD MAX system, using the BD MAX ExK TNA-2 extraction kits, the BD MAX TNA MMK master mix and the specific primers and probes. Cycling conditions were as follows: 80°C for 10min and 40 cycles of 95°C for 15s and 60°C for 60s. The Limit of Detection (LoD) of each target was evaluated by testing seven 10-fold serial dilutions. PCR performance, interference and inhibition were evaluated by spiking eight negative clinical samples with more than one ATCC strain, each one representing a different target. Specificity was performed by testing different species of mycobacteria (n=26), aerobic bacteria (n=15) and candida (n=7) commonly identified in respiratory infections. Results: Out of 402 samples tested, 227 were identified as negative and 175 as positive by culture, for at least one of the targets tested by PCR. Among culture positive samples, the new PCR was able to identify 21 out of 27 (78%) AX, 4 out of 5 (80%) BC, 138 out of 140 (99%) PSA and 29 out of 34 (85%) SM. In addition, the new CF BDM test was able to identify 35 samples as positive that were initially negative by culture (8 AX, 2 BC, 12 PSA and 13 SM). The LoD was 1.5 x 10 2 CFU/mL for AX, BC and PSA and 1.5 x 10 3 CFU/mL for SM. The new multiplex test demonstrated no cross reactivity to related organisms and correctly identified the species correspondent to each PCR target. Conclusion: The new multiplex test performed on the BD MAX System proved to be a specific and sensitive test to detect AX, BC, PSA and SM by real-time PCR on an automated sample-in results-out platform. Introduction: Diagnosis of lower respiratory bacterial infections largely rely on timeconsuming culture-based identification and quantification of bacteria from heterogeneous lower respiratory specimens. Background levels of colonizing flora vary between specimens. PCR assays that provide quantitative information of bacterial loads can potentially be useful to distinguish lower respiratory tract infections from colonization. We designed, evaluated, and compared two multiplex quantitative PCR (qPCR) assays. Each multiplex qPCR assay was designed to detect 16 pathogens. qPCR assay results from 82 clinical specimens (23 tracheal aspirate, 23 BAL, and 36 sputum) were compared to clinical laboratory results. Methods: Organisms evaluated include Acinetobacter baumanniicalcoaceticus complex (ABC complex), Escherichia coli, Enterobacter cloacae, E. aerogenes, Klebsiella pneumoniae, K. oxytoca,Haemophilus influenzae, Proteus sp., Moraxella catarrhalis, Pseudomonas aeruginosa, Stenotrophomonas maltophilia, Serratia marcescens, Streptococcus pyogenes, S. agalactiae, S. pneumoniae, and Staphylococcus aureus. Two individual multiplex qPCR assays were designed to detect these using 2 separate gene targets for each analyte. Extracted nucleic acid from 82 clinical specimens were tested and quantified by both qPCR assays. Data from each assay were compared for number of positive and negative detections, quantitative titer, and agreement with culture results. Results: The limit of detection for the majority of analytes was estimated to genomic equivalents/reaction. Testing the same clinical specimens by both qPCR assays showed good correlation between each quantified analyte. Compared to culture, percent positive agreement (PPA) was 100% for all analytes except K. pneumoniae (70%, both qPCR assays), S. maltophilia (75%, one qPCR assay), and P. aeruginosa (90%, both qPCR assays). Concordant analytes included 19 P. aeruginosa, 7 K. pneumoniae, 7 S. marcescens, 6 Proteus sp., 5 E. coli, 5. E. aerogenes, 4 S. agalactiae, 4 S. maltophilia, 4 ABC complex, 3 K. oxytoca, 3 E. cloacae, 2 H. influenzae, 2 S. pneumoniae, and 1 M. catarrhalis. Percent negative agreement was greater than 80% for all analytes except ABC complex, S. maltophilia, and E. coli. Conclusions: Both multiplex qPCR assays have near perfect PPA with culture when testing lower respiratory specimens for bacterial detection. A higher number of detections was observed using multiplex qPCR than culture, however PCR quantification suggests sub-clinical levels in some specimens. Quantification can be incorporated to study clinically relevant limits for organisms that commonly colonize the lower respiratory tract. These assays are not FDAapproved. M. Alikhan 1 , V. Tesic 2 , K. Kaul 3 , S. Das 3 1 NorthShore University Health System/University of Chicago, Chicago, IL; 2 University of Chicago, Chicago, IL; 3 NorthShore University Health System, Evanston, IL. Introduction: Invasive fungal infections (IFIs) can be life-threatening, especially in immunocompromised patients, making rapid and reliable identification of causative species essential for early and adequate therapy. Microscopic identification of fungi from formalin-fixed, paraffin-embedded (FFPE) specimens may provide presumptive identification, and routine culture can take days to weeks. Rapid identification of fungi directly from FFPE tissues could be of immense benefit in patient care. Adequate and efficient extraction of DNA from FFPE tissue is paramount to ensuring purity for subsequent molecular genetic studies which can provide more robust speciation. Herein, we describe the use of the PinPoint Slide DNA Isolation System (Zymo Research) as a convenient and effective method of DNA isolation and extraction from FFPE tissues for definitive identification of fungal microorganisms by subsequent Sanger sequencing. Methods: IFIs from 23 patients were selected from 2 institutions, of which 21 showed fungal organisms by Gomori methanamine silver (GMS) stain on histology. The sources of infections included sinus contents, liver, bone, lung, and skin. Areas with the highest organism burden were selected from each slide, and DNA was isolated and extracted using the PinPoint Slide DNA Isolation System according to the manufacturer's instructions. The recovered DNA was subjected to polymerase chain reaction (PCR) analysis using specific primers against the internal transcribed spacer-2 region of the ribosomal RNA gene. The resultant sequencing data was compared against a homology search using BLASTn. Concurrent fungal culture results were compared to sequence data. Results: Fungal DNA was successfully amplified in 18 of 23 specimens with histologically-proven IFIs (78%). Sequencing showed a variety of pathogens including Aspergillis spp., Pseudoallescheria boydii, Fusarium, Curvularia, Alternaria, and Candida spp. Of the 23 cases, 9 had no growth on fungal culture; 6 of these had positive identification of a fungal pathogen on PCR analysis, and, in an additional 2 cases, identification was made by a reference laboratory. Two culture-positive cases were negative by PCR (9%). Cultures and PCR were concordant in 11 of the 12 remaining cases (92%). Conclusions: Direct sequencing from FFPE specimens provides a rapid method of fungal species identification in IFIs and is superior to conventional culture methods. The extraction system evaluated in this study provides adequate and good quality DNA for sequencing. The method can provide rapid identification of the pathogen at the species level and help guide therapy. Additionally, FFPE tissues with IFI can be stored and DNA retrieved for future studies years later, providing a valuable source for research. Introduction: Respiratory viral infections are the most common worldwide cause of infectious disease. We investigated the incidences and age-related/seasonal variations of respiratory virus infections in Korea. Methods: A total of 3,467 respiratory specimens from patients with acute respiratory infection symptoms in Hwaseong, Korea were tested for respiratory viruses using a multiplex real-time PCR kit during 2013 to 2015. Results: At least one virus was detected in 2,561 of the 3,467 specimens (73.9%), and 706 specimens (20.4%) were sitive for two or more viruses. Two viruses were simultaneously detected in 604 specimens (17.4%), 96 specimens (2.8%) had three viruses, and 6 (0.2%) had four viruses. The most frequently detected viruses were rhinovirus (23.9%), respiratory syncytial virus B (15.5%), and adenovirus (12.5%). Most of the patients (with and without a detected virus) were children, and young children (<5 years old) were significantly more likely to have two or more viruses, compared to older individuals (P < 0.0001). Most viruses exhibited seasonal variations. Coronavirus and respiratory syncytial virus were prevalent during the late autumn and winter, metapneumovirus was common during the spring, bocavirus was common during late spring and early summer, and parainfluenza was common during late spring and summer. Conclusions: This study revealed that the incidence of respiratory virus infections and co-infection rates were high, especially among children, and most viruses exhibited seasonal variations. These findings can enhance our understanding of the distributions of respiratory viruses according to patient age and season. A. Ricketts, C. Netti, S. Kazi Qnostics, Glasgow, Scotland, United Kingdom. Introduction: Accurate and Reproducible viral load determination plays a critical role in clinical diagnostics particularly in monitoring patients' response to treatment and disease progression. In this study, 4 transplant associated viral targets were analysed over multiple sites and molecular platforms to assess whether the presence of a recognised standard helped improve results. Assay reproducibility and accuracy were examined by comparing how different platforms quantified identical samples. Two of the targets; Human Cytomegalovirus (CMV) and Epstein Barr Virus s), the other targets, John Cunningham Virus (JCV) and BK Virus (BKV) had no international standards available at the time of this study. Methods: Control materials were prepared in plasma at a titre that falls within the linear range for most clinical assays. The controls were characterised internally using in house PCR and digital PCR (BioRad). Labs were asked to treat the materials as a clinical sample and to return quantitative data along with information on the assay used. The data were separated by assay, anonymised, collated and non-compliant results removed. Datasets were included if more than five laboratories reported compliant data. Results: Five hundred and seventy-eight CMV results were collected across 7 assay groups, 67% were commercial kits, 37% were expressed in IU/ml. 260 EBV results were collected across 5 groups, 55% commercial, 30% in IU. The consensus values reported in c/ml and IU/ml were closely aligned (CMV 3.86 c/3.92 IU, EBV 4.09 c/3.97 IU). 190 JCV results were collected across 3 groups, 33% were commercial kits. 445 BKV results were collected across 5 groups, 47% were commercial kits. Standard deviation of the consensus mean was comparable for all 4 targets, (CMV 0.36, EBV 0.40, BKV 0.47, JCV 0.46). JCV and BKV were higher, although this may be due to fact that there are currently more commercial kits used in CMV and EBV detection. Values for dPCR were in line with the observed consensus. CMV, JCV and BKV t 10 of the observed consensus. For EBV dPCR was 0.4 Log10 c/ml lower. All dPCR values were within 1 standard deviation of the consensus mean. Conclusions: The use of recognised international standards is essential in supporting and improving the traceability and comparison of results across laboratories which in turn can help improve the accuracy and reproducibility of results. However in the absence of International standards well characterised control materials and good QC practice are essential in helping the laboratory to monitor and improve the accuracy and reproducibility of their assays on a daily basis which also helps define future International Standard requirements. A. Hindupur 1 , J. Evans 2 , C. Maity 3 , S. Raines 4 , B. Loeffler 4 , S. Elagin 4 , V. Slepnev 1 Meridian Bioscience, Cincinnati, OH. Introduction: Current molecular methods for testing Zika virus face several challenges such as 1) simplification of sample preparation from blood, 2) reagent stability under ambient conditions, 3) ease-of-use for the end-user and 4) affordable pricing. Reverse transcription coupled Loop mediated isothermal amplification (RT-LAMP) is a highly sensitive, rapid molecular method which can be used to detect Zika virus RNA. We report on the performance of a simplified Zika assay in an easy to use LAMP platform. Methods: The Meridian illumigene Zika (Research Use Only, RUO) reverse transcription coupled DNA Amplification Assay, uses one step RT-LAMP to detect the RNA of Zika virus. During LAMP amplification, magnesiumpyrophosphate accumulates, changing the absorbance of the reaction mixture. The absorbance change is measured by the Meridian illumipro-10 instrument. A one step purification based on chemical lysis and gel filtration was designed to extract Zika virus RNA from whole blood. Purification procedure uses gravity-driven gel filtration column, Meridian M-prep that produces amplifiable RNA within 5 to 7 minutes. Simulated blood samples were prepared by mixing particles of Zika virus with donor blood samples. Viral DNA was purified using M-prep and directly amplified in one step RT-LAMP assay. Amplification of target was detected using illumipro-10 readers. Results: The Limit of Detection (LoD) of illumigene Zika assay is 840 Copies/mL. The assay also detected the synthetic templates representing highly polymorphic seven strains of Zika virus. In addition, the assay also detected the RNA of commercially available strains of Zika virus. No cross-reactivity was observed with synthetic templates of closely related Dengue type 1 and 2. Conclusions: The RT-LAMP based illumigene Zika assay was demonstrated to detect Zika virus from whole blood samples with high analytical sensitivity using an extremely simple procedure, in less than one hour. No blood fractionation or additional lab equipment (centrifuge or heat blocks) was necessary. The described RT-LAMP may be particularly suitable for deployment in resources limited settings. This provides a much needed alternative to the more complex molecular test for Zika diagnosis. NOTE: The illumipro-10 and M-prep are not cleared for use with the illumigene Zika RUO assay. M-prep is CE Marked, not cleared for use in the United States of America. (68) samples were evaluated. All samples were nasopharyngeal swabs (NPSW) collected in BD viral transport media (VTM), stored at 4°C. Primary patient testing was performed on either the BioFire (BF) FilmArray Respiratory Panel or the Cepheid Xpert FLU/AB/RSV assay. The maximum time from collection to evaluation was 72h. Testing on all three platforms was performed on the same day. Alere is CLIA waived for direct nasal swabs, CLIA moderate for swabs in VTM. Sofia is CLIA waived for nasal swabs and NPSW tested directly, and NPSW in VTM. Xpress is CLIA waived for NPSW in VTM. All three assays were run according to manufacturers' instructions. Evaluation Criteria: Samples had to be positive with at least two systems (BF, Alere, Xpert, Sofia) to be included in the study. Several operators performed testing and evaluated hands-on time for each assay. Results: Ten samples were negative for Flu A,B,RSV; 4 samples were Flu A-, B-, RSV+; and 6 samples were Flu A,B,RSV-and positive for other targets. Twenty-seven samples were Flu A+; 21 samples were Flu B+. Of the 68 samples, 22 were discrepant on at least one of the 3 systems. Four of the discrepant samples positive for Flu A in BF and negative with all three comparison assays were excluded from further evaluation. Forty-six samples agreed across all systems: 20 Flu A/B-, 16 Flu A+/ B-, and 10 Flu B+/ A-. Sensitivity: Alere 82.6% A+, 81% B+; Xpress 100% Flu A,B; Sofia 69.6% A+, 47.6% B+. Specificity: 100% for all platforms. Hands-on/testing times for each platform are: Sofia 2.5 min set up time, 15 min incubation and read totaling18 min. Alere 15 minutes directly from swab and 18 min if testing swab in VTM; about 2 min hands-on time. Xpress 1 to 2 min set-up time, 60 min incubation. Conclusions: Timely testing at the POC with a rapid accurate influenza test will directly impact appropriate treatment. The Xpert Xpress platform had the highest sensitivity (100%) with results in 1 hr. The sensitivity of Alere was 82.6% and 81% for Flu A and B respectively, results in 18 min. Sofia had the lowest sensitivity with 69.6% and 47.6% for Flu A and B respectively; results in about 18 min. When considering rapid systems, accuracy, hands-on time, incubation time, and interface capability should all be considered for the final system selection. Y. Lu, S. Adhikari, L. Liu, K. Norman Thermo Fisher Scientific, Fremont, CA. Introduction: EBV is the pathogenic agent for infectious mononucleosis, as well as a variety of lymphoid and epithelial malignancies in immunocompetent and immunosuppressed individuals. EBV DNA load measurement by quantitative PCR has been shown to be a potential diagnostic and monitoring tool. Despite the availability of the WHO standard for EBV since 2011, variability of assay amplification target sites, particularly between different laboratories, still leads to variability in test results. In this study, we evaluated the potential contribution of reference material diversity to EBV test variability. Methods: The performance of EBV strain B95-8 obtained from different commercial sources was compared across different platforms (including ABI, Roche, and EliTech), different laboratories, and different EBV target genes (including IR1, LMP2, BNRF1, EBNA1, and BKRF1). Results: Results were normalized to the WHO EBV standard to evaluate which source materials exhibited similar performance to the WHO. Surprisingly, up to a 6fold difference in titers was observed across assays for the same material. Differences in titers were also observed from different EBV sources run on the same assay. Conclusions: Our data indicate that gene copy number varies across B95-8 source materials, which affects traceability to the WHO depending on the assay used. This was especially apparent when value assignment was based on a qPCR assay targeting variable regions of the EBV genome, which caused either under-or over-estimation of titer relative to the WHO standard by 2 fold. One EBV B95-8 source performed consistently across assay platforms when compared to the WHO standard (CV at 24.4% compared to the 44.9% from another source). Conclusion: Significant test variability can be artificially introduced through the use of secondary standards that are not genetically similar to the WHO B95-8 EBV strain. EBV reference material testing against multiple targets and laboratories has allowed for the determination of the most consistent EBV B95-8 source for use as a secondary reference material. Introduction: Hepatitis C virus (HCV) is a common chronic viral infection worldwide and a major cause of liver disease. Direct-acting antiviral therapies used to treat HCV infection are typically based on HCV genotype. Here, we describe our validation of the research use only (RUO) eSensor HCVg DirectTest (HCVD) (GenMark Diagnostics, Carlsbad, CA) to change from our laboratory's previous method, the Versant HCV Genotype 2.0 Assay (LiPA) (Siemens Healthcare, Malvern, PA). The main advantages of HCVD (objective resulting, shorter turnaround time), as well as its drawbacks (difficulty with mixed infections), have previously been described. During our validation, we uncovered an additional limitation of the assay not previously reported, as well as a potential way to improve genotyping of mixed infections. Methods: Archived plasma samples (n=81) and commercially available HCV genotyping panels (n=2) extracted using the QIAamp DSP Viral RNA kit and QIAcube (Qiagen, Valencia, CA) were used to evaluate the performance characteristics of HCVD and the XT-8 System (GenMark). HCVD was assessed for analytical sensitivity and specificity, concordance, and precision. Discrepant samples were resolved through repeat testing, altering the extraction method (GenMark), and Sanger sequencing (Retrogen, San Diego, CA). Results: HCVD was able to accurately genotype all commercial control samples (n=15, genotypes 1-6). The approximate analytical sensitivity was an HCV viral load of 1,000 IU/mL. Precision studies showed concordance of results for multiple samples over several replicates. HCVD demonstrated 100% analytical specificity (n=20) and an overall concordance rate of 90% (55 of 61 clinical samples previously genotyped by LiPA or Sanger sequencing). The 6 discrepant results were from samples with mixed infections (n=3), a low viral load (218 IU/mL) (n=1), a genotype 4f infection (n=1), and a prior indeterminate result by LiPA (accurately genotyped by HCVD). We found that altering the extraction method allowed HCVD to detect 2 of the 3 mixed infections previously not called. Additionally, we discovered that viral nucleic acid from genotype 4f cross-reacts with reagent signal probes for genotype 1b. This led a genotype 4f sample to be genotyped as a Type 1 plus Type 4 co-infection and then as a Type 1b in subsequent reactions. Conclusions: Overall, the RUO eSensor HCVg Direct Test performed favorably relative to LiPA; after discrepant resolution, the overall concordance rate between the two methods was 93% (n=57 of 61). Whereas HCVD had difficulty with type 4f and mixed infections in our hands, modifying the extraction method allowed for genotyping of previously missed coinfected samples, suggesting a potential solution for laboratories encountering similar issues. Introduction: HPV is the most common sexually transmitted infection. Persistent infection with oncogenic high risk HPV (hrHPV) types causes virtually all cases of cervical cancer. HPV detection and prevention algorithms are moving targets, with evolving guidelines and diagnostic techniques. Recent studies have demonstrated regional and population differences in the distribution of hrHPV genotypes. Bronx, NY is ethnically diverse with 52% Hispanic, 32% African-American, 12% White, 2% Asian, and 2% others. Almost one-third (31.8%) of its residents are foreign-born and a majority (50.5%) of its births are to foreign-born mothers. Methods: Utilizing a retrospective query of LIS database we studied SurePath cervicovaginal cytology and Cobas HPV results reported between 10-5-15 and 5-6-16. Patients aged 16-95 (average age 43), with racial distribution including: African-American 32.4%, Other mixed, Hispanic35%, Caucasian 14.4%, Asian 0.7%, Declined/Unknown 17.5%. A total of 18,333 or 73.2% underwent testing for hrHPV, which separately reports HPV 16, HPV 18 and a pool of 12 other hrHPV types. Results: Among all samples tested, 3901 or 21.4% were hrHPV positive. All 27 of COBAS-tested HSIL cases were hrHPV positive. In keeping with the increased risk of progression reported for HPV16, the percentage that were positive for HPV 16 alone or in combination was 37% in HSIL versus 14.8% in patients overall. However, only one of the HSIL cases was HPV 18 positive; majority (66.7%) was OHR positive. Further evaluation is The Journal of Molecular Diagnostics ■ jmd.amjpathol.org needed to determine if this OHR pool includes individual genotypes that in our population carry a higher risk of persistence and progression to cancer. Conclusions: In this diverse population, the 21.4% hrHPV is higher than Athena study. Percentage of hrHPV positive results, both overall and for individual genotypes, increases with increasing level of cytologic abnormality. HPV 16 only, is present in 1.6% of cases diagnosed as NSIL and increases to 22.2% in cases diagnosed as HSIL. OHR types only, are present in 13.6% of NSIL and 59.3% of HSIL. Moreover OHR was the only type identified in one case that proved to show invasive cancer on biopsy. This seems contrary to prior reports that HPV 16 and 18 account for a majority of invasive cancers. Given the ethnic diversity of population, and in light of prior studies indicative of regional differences in prevalent genotypes, further genotyping with the most abnormal cytology may reveal additional genotypes. This study suggests that there is a need for more studies in ethnically diverse populations to identify other high risk genotypes, and plan risk and ethnic specific prevention as well as tests for earlier diagnosis. Secondly to compare the cost, TAT and diagnostic yield of different algorithms for the detection of respiratory pathogens. Finally, to test the Simplexa assay on BAL samples, which has only been validated on nasopharyngeal swabs (NTS) so far. Methods: We collected 125 NTS and 25 BAL samples from symptomatic immunocompromised patients. Samples for which Simplexa and TAC results were discordant underwent verification testing using the multiplex real-time FTD Flu/HRSV assay (Fast-track Diagnostics). The TAC assay is based on singleplex RT-PCR, targeting 24 viruses, 8 bacteria and 2 fungi simultaneously. Results: As expected, the overall sensitivity was significantly lower for DFA testing than for the two molecular methods (p<0.05). However, when considering results for each pathogen separately, the difference in performance was only statistically significant for Flu A. The Simplexa direct test missed one RSV, one Flu A and two Flu B positive samples in comparison to the TAC assay and verification PCR. One sample was found strongly positive for Flu A by Simplexa, but was negative by DFA, TAC and verification testing. Nevertheless, the differences in individual and overall sensitivity and specificity of Simplexa testing were not significant compared to TAC testing (p>0.1). For BAL samples only (n=25), the sensitivity and specificity of the Simplexa assay was 100%. In total, DFA identified 14 samples (9.3%) and Simplexa testing found 24 (16%) samples positive with one pathogen each. The TAC assay identified 93 samples with one or more respiratory pathogens (62%). More than half (54%) of Simplexa negative samples were positive by TAC for other pathogens than RSV, Flu A and B. A co-infection rate of 15.3% was found by TAC. The estimated costs and TAT were 8.2€ and 2 hours for DFA, 31.8€ and 1.5 hours for Simplexa and 56€ and 6 hours for TAC testing. Conclusions: Based on these results, performing a first line molecular method such as the Simplexa test instead of DFA would be necessary to obtain an acceptable overall sensitivity, albeit at a higher cost generated in the laboratory. Performing the TAC as a second line test for patients with a negative Simplexa result would increase the diagnostic yield significantly, albeit at an even higher cost. Introduction: Human monocytic ehrlichiosis and human granulocytic anaplasmosis are tick borne bacterial diseases, found in several states in the U.S. We developed and evaluated a multiplexed, real-time PCR assay to detect A. phagocytophillum (ANA) and several Ehrlichia species (E. chaffeensis, E. ewingii and E. muris) (EHR) in whole blood specimens using the Luminex ARIES instrument. Selective primers, labeled with several different fluorescent dyes, facilitate the detection of ANA, EHR and an internal control in whole blood specimens in a single-step, multiplexed, qPCR assay using the ARIES instrument. This will allow for rapid diagnosis with increased sensitivity in less time. Methods: Specimens: Multiplexed, real-time PCR methods, based on the unique MultiCode base pair (isoC:isoG), were developed for diagnosis of tick borne diseases. Analytical specificity studies were assessed by testing whole blood, collected from separate volunteers, spiked with 10 fold dilutions of ANA and EHR. Primers were added to the ready mix tubes, attached them to the cassettes and then loaded the samples on to the cassette. Reagents: Primer pairs are designed to include a fluorescent reporter labeled primer with an isoC on the 5'end and an unlabeled primer and obtained from IDT. The primer sequences for ANA are: forward 5'-CAG TCG TGA ATG TAG AGG GAA AAA C-3'; reverse 5'-GGA ATC CCC CTT CAG GAA CTT G-3' and for EHR are: forward 5'-AAT GCT TCT ACT GCT ACT GT-3'; reverse 5'-GCT CCA CCA TGA GCT GG-3'. Ready mix and cassettes were purchased from Luminex. The cassette contains all reagents needed to run PCR and all the steps including extraction, purification, amplification, detection reagent and sample processing control are contained in the cassette. Instrument: ARIES is an in vitro diagnostic medical device for detection of nucleic acids by fluorescence based PCR. Results: The limit of detection (LOD) was determined for both targets and was defined as the lowest concentration of each organism in the 10 spiked samples that produced a CT value <40. Thus, the overall LOD for each ANA and EHR in whole blood was shown to be 36.7±1.5 and 36.06±0.7, respectively (n=5), at a concentration of 100 copies per ml. In a limited study, 17 specimens (2 EHR positives and 15 negatives) correlated with an in-house PCR assay for the same targets. Conclusions: The availability of an ARIES detection system involves minimal technologist time and increases the ability to rapidly diagnose with improved sensitivity of tick borne infections. A. Sanchez 1,3 , B. Rodic-Polic 2 , K. Culbreath 3 1 University of New Mexico, Albuquerque, NM; 2 DiaSorin Molecular LLC, Cypress, CA; 3 Tricore Reference Laboratories, Albuquerque, NM. Introduction: Norovirus, the most common cause of acute gastroenteritis in the United States, causes epidemic outbreaks among children and adults, leading to 19 million to 21 million new cases every year. Though 90% of adults are seropositive for Norovirus antibodies, immunity is not long-lasting and reinfection can occur. The illness tends to be self-limited, comprising of nausea, vomiting, non-bloody diarrhea and abdominal cramps. The symptoms, caused by transient malabsorption, can last up to 6 weeks following a viral challenge, but usually resolve within 72 hours. Currently, there are 6 recognized Norovirus genogroups, all belonging to the Caliciviridae family-the GI, GII, and GIV being the genogroups most commonly affecting humans. Reliable detection of Norovirus has the potential to improve patient management decisions and facilitate prompt infection control measures. This study evaluates the performance characteristics of the FOCUS Diagnostics RT-PCR molecular assay using analyte-specific reagents (ASR) for the detection of Norovirus GI/GII. Methods: To evaluate the accuracy of the assay, we compared our results with those obtained with the reference method. One hundred and twelve formed and unformed stool samples were analyzed during this study: 30 previously tested positive samples and 25 previously tested negative samples as well as 57 samples which were collected for the detection of other gastrointestinal pathogens, and comprised the random sample group. Nucleic acid extraction was performed using the NucliSENS easyMAG system (bioMerieux, Durham, NC). Extracted samples were subjected to RT-PCR qualitative analysis on Integrated Cycler. Analytical specificity of the assay was evaluated by testing a panel of most common gastrointestinal pathogens. Analytical sensitivity was determined by performing serial dilutions of Norovirus GI/GII quantitated controls. Results: Among 30 known positive samples previously tested, 26 were confirmed positive by the FOCUS Diagnostics assay. The remaining 4 discrepant samples were subsequently repeated by our method and then confirmed negative by the reference method, resulting in 100% positive agreement. Of the 57 randomly selected samples, 4 were positive for Norovirus GII. No cross reactivity with members of the specificity panel was observed. Conclusions: The preliminary results of our evaluation study indicate that Focus Diagnostics RT-PCR assay for qualitative detection of Norovirus I/II is a sensitive and reproducible molecular assay. This assay proved to be robust and easy to perform which makes it suitable for the reliable detection of Norovirus I/II in the clinical laboratory. C. Cheng, Y. Parocua, B. Torres, M. Tabb DiaSorin Molecular LLC, Cypress, CA. Introduction: Enteroviruses (EV) and human parechoviruses (HPeV) are single stranded RNA viruses from the Picornaviridae family. EV are the most common pathogens associated with aseptic (viral) meningitis. However, HPeV have been reported to cause up to 5% of these cases in the pediatric population. The Simplexa EV & HPeV Direct assay (Simplexa assay) is a sample-to-answer assay that detects EV and HPeV from 50 μl of CSF without any additional specimen preparation or extraction steps. The test result is available in about 90 minutes. The goal of this study was to evaluate Simplexa assay performance for analytical limit of detection (LoD), microbial inhibition, method comparison and competitive interference. Methods: A LoD study was performed on 5 EV and 2 HPeV strains. For the microbial inhibition study, contrived samples containing EV and HPeV were spiked with 24 interfering pathogens. A panel of 101 CSF specimens was used for the method comparison study. Due to lack of HPeV positive specimens, 29 contrived and 1 HPeV clinical sample were used. Each specimen was tested in singlicate using the Simplexa assay and results were compared to previously reported PCR results (reference method). The competitive inhibition panel consisted of near LoD EV and HPeV samples contrived using synthetic spinal fluid (SF), challenged with jmd.amjpathol.org ■ The Journal of Molecular Diagnostics high concentrations of competing virus. Results: LoD concentratons for 5 EV strains (Coxsackievirus A3, A17, Echovirus 21, Enterovirus 68 and Poliovirus 2) were confirmed at 0.0325, 0.0125, 0.0325, 6.5 and 26 TCID50/mL respectively. LoD for HPeV strains (HPeV-1 and HPeV-3) were confirmed at 1.25 and 4.7 TCID50/mL respectively. No inhibition was observed with 24 potentially interfering organisms in the microbial inhibition study. Positive and negative agreements in method comparison between Simplexa and the reference method were 98% (50/51) and 100% (50/50) respectively for EV. All 30 HPeV samples were detected as positive (30/30) with the Simplexa assay. In competitive inhibition, all replicates containing low concentrations of one target virus (EV or HPeV) were detected in the presence of high concentrations of the second virus (EV or HPeV) . No inhibition to detection of low virus concentrations was observed. Conclusions: The Simplexa Direct assay was able to detect EV strains at LoDs ranging from 0.0125 to 26 TCID50/mL and two HPeV strains at 1.25 and 4.7 TCID50/mL. The assay demonstrated 98% positive and 100% negative agreements for EV detection and 100% positive detection for HPeV. No cross reactivity or competitive inhibition was observed using the bacteria or viruses tested. The Simplexa assay is currently in development and is not currently for sale and not FDA cleared. Introduction: PCR primers used to determine fungal species identification cited from phylogenetic studies often need to undergo optimization and validation for use in the clinical laboratories. Except for the "universal primers" which target the internal transcribed spacer (ITS) as well as domains 1 and 2 of the large ribosomal subunit -TUB) and calmodulin (CAL) genes are also frequently used in phylogenetic studies and have been applied in clinical diagnostics. We retrospectively reviewed 593 clinical and -TUB and CAL primers for identification of Aspergillus species by Sanger sequencing. Methods: DNA was extracted from 593 isolates of various Aspergillus spp. representing clinically significant species by bead-beating -TUB and CAL regions were then PCR amplified using M13 tagged BT2a/BT2b, and cmd5/cmd6 primers respectively. PCR products were sequenced bi-directionally with M13F/M13R primers using the BigDye sequencing reagents. Raw sequences were aligned and edited using the Lasergene software (DNASTAR, Inc). Finished sequences were queried through the GenBank nr database. The top 50 scoring hits were combined with phenotype data to identify each isolate to the species level. Results: -TUB (Bt2a/Bt2b) and CAL (cmd5/cmd6) primers were selected to test 94% and 84% of 593 Aspergillus isolates. Sequences were successfully obtained in 94% (527/560) and 72% (355/496) of isolates -TUB and CAL primer set, respectively. Among these 593 isolates, 33 species and 4 species complexes were identified. The major pathogenic Aspergillus species found were A. fumigatus (44%; 260/593), species in the Aspergillus section Nigri, (25%; 146/593), A. flavus (8%; 49/593), as well as Aspergillus terreus species complex (6%; 35/593). Failure of PCR amplification -TUB) and 28% (CAL) isolates, which was mostly caused by biological issues (lack of primer binding sequence in a given species). However, species and species complex level identification using both -TUB and CAL sequence data in combination with phenotype data was successful in all 593 cases. Conclusions: In the molecular identification of Aspergillus spp. -TUB (Bt2a/Bt2b) primer had the highest success rate in producing sequence data, but did not produce a species level identification for some aspergilli (e.g., Section Nigri) due to lack of phylogenetic differences and/or sufficient coverage in the GenBank database. The CAL (cmd5/cmd6) primer alone was able to identify isolates to the species level in 72% of the cases, thus making this primer useful when used in conjunction with -TUB primer. Simplexa Bordetella Direct Assay Y. Xie, H. Mai, J. Chen, M. Tabb Diasorin Molecular LLC, Cypress, CA. Introduction: Bordetella pertussis is the main cause of whooping cough, however other Bordetella species, such as Bordetella parapertussis and Bordetella holmesii, can cause similar symptoms. The Simplexa Bordetella Direct assay is in development as a sample-to-answer assay performed on the Integrated Cycler instrument. Nasopharyngeal swab specimens collected in transport media are loaded directly onto a Direct Amplification Disc without extraction or other specimen preparation. The Simplexa Bordetella Direct assay was developed to detect and differentiate B. pertussis, B. parapertussis and B. holmesii. The goal of this verification study was to evaluate the performance of the Simplexa Bordetella Direct assay. Methods: Limit of detection (LoD), analytical reactivity, reproducibility and substance interference studies were performed to evaluate Simplexa Bordetella Direct performance. Limit of detection (LoD) studies were performed to determine the analytical sensitivity of the assay. Ten additional B. pertussis strains were evaluated for analytical reactivity. A reproducibility study was performed with medium and low positive panels. A panel of potentially interfering substances was tested to determine whether any inhibition was observed. Results: LoD studies showed that the Simplexa Bordetella Direct assay detected B. pertussis B. parapertussis B. holmesii at 60 CFU/ml. All 10 B. pertussis Inter-and intra-assay reproducibility assays yielded <4.0% coefficient of variation. No inhibition or interference was observed from any of the substances tested. Conclusion: The Simplexa Bordetella Direct assay was capable of directly detecting and differentiating B. pertussis, B. parapertussis and B. holmesii without up-front nucleic acid extraction from nasopharyngeal swab specimens. The assay and instrumentation provide a compact system for rapid (~80 minutes) detection directly from nasal swab samples. This assay can be performed simultaneously with Simplexa Flu A/B & RSV Direct and Simplexa Respiratory Virus Direct to detect a panel of 9 respiratory pathogens (Flu A, Flu B, RSV, adenovirus, parainfluenza virus (1, 2, 3 & 4) , hMPV, B. pertussis, B. parapertussis and B. holmesii) from one patient sample. Simplexa Bordetella Direct and Simplexa Respiratory Virus Direct assays are in development; they are not currently available for sale and are not FDA cleared. There are estimated to be at least 150,000 cases a year in the USA, with Arizona accounting for at least 90,000 (60%) of these. The PathoGene Coccidioides Assay is a qualitative real-time PCR-based assay that detects Coccidioides spp. target DNA from bronchial alveolar lavage (BAL) or bronchial wash (BW) samples. Extracted DNA is added to an assay cartridge that contains all the reagents for amplification and detection of the Coccidioides target DNA and an internal control. The cartridge is placed in the GeneSTAT instrument and the assay is initiated by the user; all subsequent steps of the assay are performed by the GeneSTAT without user intervention. Assay turn-around time from cartridge loading to results generation is 1.5 hours. Methods: The Coccidioidesspecific PCR assay targets a 106-bp sequence that is present in multiple copies within the C. posadasii and C. immitis genomes (assay patent licensed from TGen, Phoenix, AZ). The human RNase P gene acts as an internal control for the assay. Amplified PCR products are detected with FAM-(Coccidioides) and AP593-labeled (RNase P) hydrolysis probes. C. posadasiispherules and extracted DNA from both Coccidioides strains were obtained from TGen North (Flagstaff, AZ). Results: Limit of Detection (LoD) for C. posadasii spherules was 50 spherules/mL. The LoD for extracted DNA from both Coccidioides strains was 10 genome copies/mL. The assay was 100% specific when screened against 47 different bacterial, viral, and fungal species. Clinical performance was determined vs. current gold-standard culture testing at three external sites, and the overall clinical sensitivity and specificity of the assay was 100% (55/55) and 98.9% (275/278), respectively. Conclusions: The PathoGene Coccidioides Assay is a sensitive and specific method for direct detection of Coccidioides DNA from BAL/BW samples. The assay may offer a significant improvement over current serology and culture-based testing methods for VF: Serological testing is rapid but is problematic in that the immune response is delayed after infection and this can lead to false negative results. Fungal culture is specific but it can take 3 weeks or longer to confirm a negative culture. The PathoGene assay is able to generate accurate test results the same day as sample collection, and this should result in improved patient outcomes and reduced medical costs. For research use only. Not for use in diagnostics procedures. Introduction: Accurate and rapid identification of Aspergillus species is often complicated by overlapping phenotypic features. Species of the Aspergillus section Nigri are significant pathogens causing allergic reactions and invasive infections. Molecular methods are often used in combination with phenotype data to finalize a species level identification in a clinical setting. We retrospectively reviewed 146 Aspergillus section Nigri -TUB) and 2 sets of commonly used calmodulin (CAL) primers for molecular identification by Sanger sequencing. Methods: DNA was extracted from 146 Aspergillus section Nigri isolates by bead-beating and the EZ1 DNA Tissue Kit (Qiagen). Regions of the --TUB (BT2a/BT2b), as well as CAL (CL1F/2AR & cmd5/cmd6), primers respectively. PCR products were sequenced using the BigDye sequencing reagents. Raw sequences were aligned and edited using the Lasergene software (DNASTAR, Inc). Finished sequences were queried through the GenBank nr database. The top 50 scoring hits were combined with phenotype data to identify each isolate to the species level. Results: -TUB (Bt2a/Bt2b) and CAL (CL1F/2AR) primers were used when Section Nigri was suspected, whereas the CAL (cmd5/cmd6) primer set was used when this was not the case. Sequences were successfully obtained for all isolates using the Bt2a/Bt2b primers (118/118) and CL1F/2AR primers (84/84), whereas sequences were successful in only 77% (51/66) cases when cmd5/cmd6 The Journal of Molecular Diagnostics ■ jmd.amjpathol.org primers were used. Sequence alignment of the calmodulin amplicons indicate that differences in the consensus sites used to generate CAL primers and the actual calmodulin sequence of these species account for the disparate success rates for -TUB, CL1F/2AR primer sets and phenotype data, species in Aspergillus section Nigri were able to be identified in 86/146 cases as A. brasiliensis (2), A. japonicas (2), A. luchuensis (1), A. neoniger (1), A. niger (24), A. tubingensis (37), and A. welwitschiae (19) . Sixty isolates could not be differentiated to species. Conclusions: Whereas -TUB had a high success rate in producing sequence data, -TUB alone lacks sufficient phylogenetic information to fully resolve section Nigri, highlighting the importance of using multiple genetic regions to make species level identification for these fungi. The CAL cmd5/6 primer was shown to be a useful marker to identify many Aspergillus spp. in other sections but with a relative higher failure rate. In contrast, the CAL CL1F/2AR was found to be the superior marker for generating sequence data used to resolve the isolates to the species level in Aspergillus section Nigri. , also known as yeast infection, accounts for about a quarter of reported cases of vaginitis. Whereas CV is typically treated with standard antifungal azole treatments, some C. glabrata infections are resistant to azoles and are instead treated with polyene drugs. Accurate diagnosis of CV infections therefore improves treatment outcomes. This study evaluated the clinical and analytical performance of a research use real-time transcription-mediated amplification (TMA) test for the qualitative detection of 5 common Candida species on the automated Panther System. Methods: Vaginal swab specimens were collected from 362 women symptomatic for vaginitis or vaginosis at 11 clinical sites in the United States. The samples were tested with a research use real-time TMA assay for 5 Candida species (Candida group: C. albicans, C. parapsilosis, C. tropicalis, C. dubliniensis; and C. glabrata). Test results were compared to blood agar and Chromagar results. Subjects from whom specimens yielded a blood agar score of 3+ or 4+ were considered positive for Candida vaginitis. Speciation was determined by Chromagar results. The results were also compared to BD Affirm assay results when available. Results: Compared to the culture reference standard, the CV TMA assay had 97.5% sensitivity and 91.9% specificity, whereas the BD Affirm assay had 71.4% sensitivity and 97.5% specificity. The CV assay had 77% positive predicted value and 99.2% negative predicted value, whereas the BD Affirm assay had 88.7% positive predicted value and 92.4% negative predicted value. Analytical sensitivity of the CV TMA assay (in CFU/mL) was C.alb: 1521, C.dub: 2085, C.par: 1786, C.tro: 387, C.gla: 56. Testing with 44 non-candida microorganisms showed no cross-reactivity or inhibition. Conclusions: The Candida real-time RUO TMA assay showed good agreement with traditional culture methods in clinical sample testing and has been demonstrated to be both sensitive and specific for detection of Candida vaginitis. To address this, reagents have been developed and evaluated. DiaSorin Molecular's ZIKV Primer Pairs are designed for real-time PCR amplification and detection of conserved NS-1 and ENV genes. This study was done to determine the crossreactivity, analytical reactivity and ZIKV primer pair's ability to multiplex with Chikungunya Primer Pair. Methods: Each ZIKV Primer Pair contains forward primer, reverse primer, and a CFR610 labeled probe. The CHIKV Primer Pair contains forward primer, reverse primer, and a FAM labeled probe. A panel of 46 bacteriae/ viruses was tested to evaluate cross reactivity. Five ZIKV strains (Asian/ African lineages) were evaluated for analytical reactivity. A set of 40 ZIKV contrived and blinded plasma and urine samples were tested. Samples were extracted on the Roche MagNA Pure LC, using the Total Nucleic Acid Isolation Kit and tested in a real-time PCR assay. Results: No cross-reactivity was detected when closely related flaviviruses or other pathogens were tested. Analytical reactivity studies demonstrated that the ZIKV primer pairs detect both Asian and African lineages efficiently. The primer pairs correctly identified 40 /40 contrived samples including Martinique and MR766 strains. The ZIKV/CHIKV multiplex demonstrated 100% correlation compared to CHIKV primers run as singleplex. Conclusions: DiaSorin Molecular has developed two primer pairs for real-time PCR amplification and detection of the conserved ZIKV NS1 and ENV genes. Specificity studies demonstrate that this primer pair does not cross-react with other pathogens found in plasma and urine. Analytical reactivity studies demonstrate detection of known lineages of ZIKV. Both ZIKV Primer Pairs are compatible for multiplexing with CHIKV primer pairs. These analyte specific reagents should be compatible with all modern thermocyclers. Introduction: Treatment and management of HIV-1 infection has improved dramatically in the recent decade. With more people utilizing antiretroviral treatment options, and with markedly longer life spans, the importance of infection monitoring has risen dramatically. Options for the quantitative monitoring of HIV-1 infection must be sensitive, inclusive of many genotypes, and easy to introduce in the work flow of the diagnostic laboratory. The Hologic Aptima HIV-1 Quant Dx assay (Aptima), was compared to the Abbott RealTime HIV-1 assay (RealTime) for the ability to sensitively detect HIV-1 in human clinical specimens. Methods: Clinical specimens from HIV-1 infected individuals were collected in EDTA-plasma collection tubes and processed. Plasma specimens representing a broad range of HIV-1 RNA copy numbers (10e1-10e6), both frozen and non-frozen, were tested by RealTime and Aptima according to the manufacturer's procedures. Commercial panels, including the Acrometrix HIV-1 copies/ml, the Seracare Worldwide HIV and the WHO 3 rd HIV-1 IS were also analyzed on both platforms to assess accuracy and inclusivity, respectively. Acute HIV-1 specimens, found to contain HIV-1 RNA in the absence of HIV-1 antibody were also assessed quantitatively on both platforms. Results: One hundred and fifty-five (155) samples were tested by both methods, showing an overall agreement of 80.0% (124/155) with a Kappa statistic of 0.628 (SE 0.052; 95% CI 0.527 to 0.730). With regard to discordant samples: 12 Samples <30 copies/mL by Aptima were not detected by RealTime. Eight (8) samples that were detected <40 copies/mL by RealTime were not detected by Aptima. Ten (10) samples quantified by RealTime that were Detected <30 by Aptima. In an analysis of 92 clinical samples with quantitative results in both assays, the slope 95% CI was equal to 0.9921 to 1.076 with an intercept 95% CI of -0.1700 to 0.2023. All subtypes tested (A, B, C, D, F, G, H, CRF01-AE and CRF01-AG) from a panel of the most prevalent subtypes worldwide were accurately quantified by Aptima. Workflow evaluations comparing the Aptima assay on Panther to the RealTime assay on m2000 revealed that Aptima was measurably easier to use, and required substantially less hands-on time than the compared system. Conclusions: The Aptima assay is capable of accurately detecting and quantifying HIV-1 RNA in clinical samples. The test generates results highly comparable to an existing FDA approved system. Aptima is capable of accurately detecting and quantifying HIV-1 RNA from a wide variety of subtypes, and is easier to use than existing cleared methods. The system appears to be highly effective for use in the assessment of viral loads in HIV-1 patients, as it establishes more facile workflows for laboratorians. K.B. Pierce 1 , A. Hopper 2 , S. Holt 2 , A. Blaschke 1 , K. Ampofo 1 , K. Korgenski 2 , A. Phillips 2 , M. Dickey 2 , R. GrandPre 2 , J. Daly 2 1 University of Utah, Salt Lake City, UT; 2 Primary Children's Hospital, Salt Lake City, UT. Introduction: Pharyngitis is one of the most common reasons for visits to health care providers. Rapid, accurate identification of the presence or absence of Group A Streptococcus (GAS) as a cause of pharyngitis is important for both appropriate treatment when it is present, and to reduce inappropriate use of antibiotics in cases of viral pharyngitis. Groups G (GGS) and C Strep (GCS), though part of normal human flora, are also know to cause pharyngitis and invasive disease, particularly in the immune compromised, and the presence of either entity warrants antibiotic treatment in the presence of symptoms. Microbiologic culture remains the gold standard for diagnosis of strep pharyngitis, but takes 24 hours to 48 hours to yield definitive identification. New molecular tests are challenging culture's efficacy in this role. The Solana Strep Complete Assay (Quidel Corporation, San Diego, CA) is a rapid molecular test designed to identify the presence or absence of GAS and GCS/GGS in throat swabs from symptomatic patients via qualitative isothermal helicase-dependent amplification (HDA) endpoint detection by fluorescent probe within one hour of swab receipt: Methods: Throat swabs obtained for GAS and GGS/GCS testing at Primary Children's Hospital (PCH) in Salt Lake City, Utah per standard of care and at physician discretion over the time period of April through June of 2016 were included in the study. Data collection is still underway. Each throat swab was first plated using PCH standard procedures and then tested with the Solana Strep Complete Assay. Pledgets from culture swab transport tubes were forwarded to Quidel for culture-based testing for GAS and GGS/GCS and for testing with an alternative nucleic acid test (PCR). PCH and Quidel culture outcomes determined the consensus culture result. PCR testing was used for discrepancy resolution between consensus culture and Solana. Results: To date, a total of 176 swabs were tested by both Solana and consensus culture. Forty-eight (27%) were culture positive for GAS or GGS/GCS. Of these, 42 (24%) were positive on Solana testing. Of 128 (73%) culture-negative swabs, 119 (68%) were negative by Solana. In 4 cases where culture was negative and Solana positive, PCR showed the presence of GAS or GGS/GCS DNA within the sample. In 6 culture positive cases where Solana was negative, PCR identified no GAS or GGS/GCS DNA in the sample. After discrepancy resolution, sensitivity, specificity, accuracy, PPV and NPV of the Solana Strep Complete Assay were, respectively, 100%, 96%, 97%, 90% and 100%. Conclusions: Use of this rapid method for identification of GAS and GGS/GCS in clinical samples has the potential to refine current methods of diagnosis for strep pharyngitis. S. Das, K.A. Mangold, J. Behles, K.L. Kaul, R.B. Thomson NorthShore University HealthSystem, Evanston, IL. Introduction: Identification of mycobacteria by traditional biochemical and phenotypic characteristics is slow and laborious. Additionally, with over 150 species of atypical mycobacteria, accurate phenotypic identification is also difficult to achieve. Molecular methods are often sought as they are rapid, can impact patient care, and provide reliable identification to the species level. We instituted a rapid molecular method for identification of mycobacteria directly from Mycobacterial Growth Indicator Tube (MGIT) specimens. We analyzed the performance of this assay in comparison to other techniques available for the identification of mycobacterial species. Methods: A PCR-melt curve analysis (melt) method has been validated in our laboratory for rapid identification of mycobacteria. This realtime assay uses FRET probes targeting the hsp65 sequence region and can be performed from both MGIT tube and solid media. The assay is currently used as an initial identification method as soon as growth becomes evident in a MGIT broth. Confirmation of melt results if warranted is usually achieved by additional tests either alone or in combination. These include phenotypic characteristics such as growth rate, DNA probes, 16S rRNA or rpoB gene sequencing, and matrix assisted laser desorption and ionization-time of flight (MALDI). Herein, we review identification results of positive cultures for mycobacteria over a 2-year period to determine the performance of the melt assay. In a subset of isolates, we also compare results of species identification obtained using the different methods. Results: A total of 395 positive cultures were identified. In 89.6% of the isolates (including M. gordonae, Mycobacterium avium complex, M. chelonae, M. fortuitum and M. abscessus), melt curve results provided an initial identification that matched the phenotypic characteristics or results from DNA probes or MALDI. We compared identification in a subset of 30 positive cultures where an alternate identification method was performed. The melt results agreed with the alternate method in 15 cultures, but failed (n=1), or was indeterminate (n=1) in two isolates, Indeterminate melt curves obtained in 11 isolates were identified as new species not included in the analysis scheme (e.g., M. neoaurum, M. parafinicum, M. elephantis, M. yongonense, M. celatum). Conclusions: The real-time PCR-melt curve analysis can be performed from MGIT broth immediately after growth becomes evident as opposed to the MALDI or sequencing which requires growth on solid media. Although our study includes limited number of isolates with alternate identification, the melt assay appears to be rapid and reliable for the initial identification of commonly isolated mycobacterial species. Y. Chen, D. Hsia Danner Laboratory, San Diego, CA. Introduction: M genitalium was first identified in the early 1980s. According to 2010 study by University of Washington, the prevalence of M genitalium for combined men and women was 1.0% compared with 0.4%, 4.2%, and 2.3% for gonococcal, chlamydial, and trichomonal infections, respectively. For men, epidemiology studies indicated M genitalium responsible for up to 45% of nongonococcal and nonchlamydial urethritis, and approximately 30% of cases were persistent or recurrent. For women, the organism can be found in the vagina, cervix, and endometrium. It could be attributable for 10%-30% cases of women with clinical cervicitis and 2%-22% cases of PID (median: 10%) depending on the setting. M genitalium diagnostic specimens can be obtained from both male and female urine as well as cervical and vaginal swabs. Genetic testing can speed up diagnosis and improve treatment decisions since it is a slow-growing organism which can take up to 6 months to grow in the culture. Danner Laboratory has validated a Transcription-Mediated Amplification (TMA) based assay using the automated Panther System enabling a fast turnaround for M genitalium diagnosis. Methods: A batch of 92 urine and vaginal swab samples, 62 positive and 30 negative of M genitalium, was kindly provided by Springfield-Greene County Health Department and used for the validation. The samples were collected in tubes with a solution which released RNA and protected it from degradation during storage. The assay was performed with the Panther System in three major steps which all take place in one single tube: capturing the targeted rRNA, target amplification by TMA, and RNA amplicon detection by hybridization protection assay. In addition, analytical performance of the assay including sensitivity, specificity, and reproducibility were evaluated. Results: It has been determined that the assay has 94% sensitivity and 100% specificity. The inter-assay and intra-assay variability were determined as 1% and 1.2 % CV, respectively. The assay was specific and did not cross react with any of the common genital tract pathogens. Limit of detection (LOD) of the assay was at 278 copies/mL. Conclusions: The assay was validated using a panel of characterized urine and vaginal swab specimens. It shows excellent clinical and analytical sensitivity, specificity, and reproducibility. In addition, the automated Panther System was capable of accomplishing the tests in 3 hours with minimal hands on time. Therefore, it is an ideal assay to replace the traditional M genitalium detection by culturing techniques. A. Greninger, C. Johnston, D. Koelle, M. Huang, K. Jerome University of Washington, Seattle, WA. Introduction: Herpes simplex virus 2 (HSV2) is the most common cause of genital herpes, causing 417 million infections worldwide among adults aged 15 years to 49 years old. Phylogenomic analysis of viral infections is a powerful technique for understanding transmission of the virus and detecting variants that may be associated with clinical phenotypes, i.e., antiviral resistance. HSV2 genomics is challenging due to its incredibly high GC% (70%), presence of multiple repeats, and presence at low concentration in a high concentration of human DNA background. Methods: We developed a tiling IDT xGen capture panel to HSV2. DNA libraries were prepared using NEB fragmentase, end-repair/dA tailing, barcoded ligation protocol and pooled 8-ways prior to capture based on a relative HSV2 to human beta-globin DNA qPCR. Libraries were sequenced on an Illumina MiSeq to achieve ~100X coverage of the HSV2 genome. Reads were aligned using Geneious v9.1 software. Results: Approximately 130 genomes from HSV2 strains from clinical swabs and cultured isolates have been recovered. Capture sequencing allowed for the recovery of whole genomes from clinical swabs with concentrations below 10^3 copies per mL, an increased level of detection of 4 to 5 logs beyond shotgun metagenomics. Comparison of sequences between genomes recovered from clinical swabs versus 1X cultured passage virus often revealed no mutational changes, consistent with the slow molecular clock of herpesvirus. The HSV2 panel recovered for approximately 30% of the genome from HSV1 isolates. Strikingly, multiple clinical swabs demonstrated recombination between HSV1 and HSV2 sequence in drug targets such as the ribonucleotide reductase and DNA polymerase. Conclusions: We demonstrate rapid and cost-effective recovery of HSV2 whole genomes direct from clinical samples. These sequences demonstrate a complex genomic relationship between HSV1 and HSV2 sequences that require further biochemical characterization. E. Rakhmanaliev 1 , P. Ariyaratne 1 , A. Yeo 1 , C. Lee 1 , P. Nimitsuntiwong 2 , C. Wathtphan 2 , Z. Rui 1 , E. Passomsub 2 , W. Chantratita 2 , G. Michel 1 1 Vela Research Singapore Pte. Ltd., Singapore; 2 Ramathibodi Hospital Mahidol University, Bangkok, Thailand. Introduction: Resistance of HIV to antiretroviral drugs is the most common cause for therapeutic failure in people infected with Human Immunodeficiency Virus (HIV). Objective of this study was to compare 2 sequencing-based HIV-1 drug resistance monitoring systems: a CLIP-based system (TruGene HIV-1 Genotyping Kit) and a novel next-generation sequencing (NGS)-based test (Sentosa SQ HIV-1 Genotyping Assay). Methods: We used an automated NGS-based integrated workflow, comprised of 1) a robotic liquid handling system for nucleic acid extraction and NGS library preparation (Sentosa SX101); 2) instruments for deep sequencing; 3) kits for RNA extraction, HIV NGS library preparation and deep sequencing, and 4) data analysis and reporting software. Reporting includes 86 Drug Resistance Mutations (DRMs) across the Reverse Transcriptase (RT), Protease (PR) and Integrase genes. 111 prospective plasma samples from patents infected with HIV-1 were tested for this study. Results: All samples were tested on both systems. 97.3% (108/111) samples were subtyped as CRF01_AE. In total, 647 DRMs were detected (435 in the RT, 199 in the PR and 13 in the Integrase genes). The Sentosa HIV Assay detected 100% (199/199) and L10I 20.7%(23/111). Although DRMs are reported by the system the software does not provide interpretation of the data with regard to usage in patient management. Conclusions: Detection and reporting of DRMs is critical for drug regiment and can minimize the development of resistance to antiviral drugs. High sensitivity (up to 5% mutation frequency) and the comparatively The Journal of Molecular Diagnostics ■ jmd.amjpathol.org short turnaround time of 2.5 days make this NGS-based workflow a promising new tool for detecting relevant mutations in HIV-1 treatment targets. Introduction: Metagenomic next-generation sequencing (mNGS) for pan-pathogen detection has been successfully tested in proof of concept studies to aid in difficult diagnostic cases, but is not available in a clinical setting to date. The UCSF Clinical Microbiology Lab has developed and validated a clinical mNGS assay for diagnosis of infectious causes of meningitis and encephalitis from cerebrospinal fluid (CSF). We evaluated performance characteristics of the assay using contrived and excess patient samples. Methods: Two libraries per sample were constructed from DNA and RNA portions of extract and prepared using Nextera XT, with 5-20 million sequences generated per library. Internal controls consisting of DNA and RNA phage particles were spiked in prior to extraction. An external positive control made of 7 representative organisms (DNA virus, RNA virus, Gram-negative rod, Gram-positive coccus, yeast, mold, and parasite) was developed. Positive control dilutions in synthetic CSF were used to determine limits of detection. To document the effects of inhibition, hemolyzed RBC and human DNA and RNA were added prior to extraction to the positive control and sequenced. The accuracy study included 105 CSF samples (known positives: 28 DNA viral, 11 RNA viral, 5 bacterial, 14 fungal, and 1 parasitic). For bioinformatics analysis, the SURPI pipeline for pan-pathogen detection from mNGS data was optimized to report positives within established cutoffs after normalization (SURPIclin). An algorithm for automatic reporting of detected pathogens was developed with final interpretation by laboratory physicians. Discrepancy testing was conducted when possible using additional orthogonal molecular tests when mNGS detected organisms not routinely tested for by reference methods, and to establish the presence or absence of microorganism nucleic acid. Results: We established quality metrics, threshold values, and limits of detection of between 0.01-130 genomic copies or colony forming units per mL. Gross hemolysis and high white blood cell counts reduced assay sensitivity; however, the phage-spiked internal control was a reliable indicator of sensitivity loss. The assay exhibited 76.4% sensitivity and 98.3% specificity compared to reference lab results. 27/87 (31%) of organisms undetected by mNGS were rare isolates by culture or Ct >37 by qPCR; when removed from the accuracy study, assay sensitivity improved to 90.9%. Conclusions: A mNGS assay for pan-pathogen detection in the context of samples from patients with encephalitis or meningitis was developed and clinically validated. Performance characteristics are highly promising. Clinical utility is being evaluated in a 300-patient prospective study supported by the California Initiative to Advance Precision Medicine. Introduction: Viral respiratory infections are a significant cause of morbidity and mortality during the flu season. Rapid virus identification provides critical information for the treatment, management and isolation of infected patients, and can offer substantial economic savings to health care systems. Previously we modeled various strategies of implementing rapid viral detection in our health care system and calculated the predicted effects on laboratory workflow, turn-around time (TAT) and cost. Here we review the observed impact on these laboratory metrics following the implementation of a rapid molecular viral detection assay as the initial testing modality for selected patient populations. Methods: During the 2015 to 2016 flu season, rapid virus identification for influenza A/B and respiratory syncytial virus (RSV) was performed using the Cepheid Xpert Flu/RSV XC assay for patient samples originating from the emergency department (ED) and the perinatal evaluation center (PEC). Rapid test negative/inconclusive samples were reflexed to our laboratory developed, real-time PCR based nine analyte respiratory virus panel (RVP) for confirmatory testing. For RVP reflexed specimens, concordance rates between the two platforms were calculated. The effects of rapid viral identification on laboratory workflow, TAT and cost were evaluated. Results: Over 700 respiratory specimens were analyzed using the rapid molecular assay, accounting for one fifth of all specimens requiring viral testing during this past flu season. One fourth (26%) of samples tested by the rapid molecular assay were positive for either influenza A/B or RSV. Compared with the RVP assay, the Xpert had a sensitivity of 99% for influenza A, 100% for influenza B, and 90% for RSV. Discordant samples displayed a high Ct value on the RVP assay, suggesting a low viral load in these samples. Laboratory costs increased by 1.5 or 2.5 fold for rapid test positive and negative samples respectively. A median reduction in TAT of 20 hours was observed for rapid test positive samples. For rapid test negative samples that underwent additional RVP testing, the final TAT was not statistically different compared to samples that did not undergo rapid testing (unequal variances t-test, P>0.05). Conclusions: The Xpert assay results were highly concordant with our RVP results for influenza A/B and RSV. Furthermore, initial rapid molecular testing resulted in a drastic reduction in TAT for influenza or RSV positive samples, without significantly altering the TAT for rapid testing negative samples that required additional RVP testing. These findings suggest that the implementation of rapid testing improved the management of ED and PEC patients during flu season and may significantly reduce overall health care costs. Introduction: Viral respiratory tract infections are some of the most common illnesses and cause significant morbidity and mortality worldwide. Among noninfluenza respiratory viruses, Human Bocavirus (BVS), Human Metapneumovirus (hMPV), Parainfluenza (PIV), and Respiratory Syncytial Virus (RSV) are the most frequent. Molecular methods have significantly improved the diagnosis of acute respiratory tract infections as they offer high sensitivity and specificity. VIASURE Real-Time Detection kits (CerTest) are novel assays which contain all the necessary components for the PCR in a stabilized format, which allows its shipment and jmd.amjpathol.org ■ The Journal of Molecular Diagnostics storage at room temperature. The aim of this study is to compare Real-time PCR assays and multiplex PCR followed by visualization in low-density arrays. Methods: A prospective comparative study was performed in 320 throat swab samples from patients with clinical suspicion of viral respiratory infections collected at Hospital Clínico Universitario Lozano Blesa (Spain) during three winter seasons: 2014, 2015 and 2016. These samples were tested with VIASURE monoplex assays for the detection of BVS, hMPV, PIV 1, 2, 3 and 4, and RSV A and B. The results were compared with the data obtained by a microarray method, CLART PneumoVir (Genomica), and confirmed by liquid-format commercial Real-Time PCR kits: "RealStar RSV RT-PCR Kit" and "RealStar hMPV RT-PCR Kit" (Altona Diagnostics), "RIDA GENE Parainfluenza" (R-biopharm) and "FTD HAdV/HMPV/HBoV" (Fast Track Diagnostics). Results: Two hundred and eighty samples were positive for some of these viruses by Genomica: 36 for BVS, 50 for hMPV (A and B), 8 for PIV1, 3 for PIV2, 41 for PIV3, 19 for PIV4, 61 for RSVA and 62 for RSVB; and 40 samples were negative for all of these pathogens. These samples were tested using the pertinent VIASURE monoplex assay obtaining: 28 positive samples for BVS by VIASURE out of 35 positive samples for BVS by Genomica (28/36), 45/50 for hMPV, 3/8 for PIV1, 3/3 for PIV2, 34/41 for PIV3, 13/19 for PIV4, 59/61 for RSVA and 59/62 for RSVB. Therefore, the sensitivity reached for VIASURE may be estimated 70%-100% for BVS, hMPV, PIV3, RSVA, and RSVB and <70% for PIV1 and PIV4 (probably due to low sample size). However, the commercial qPCR kits used leading to the similar results and then, showing a 98.2% concordance rate with VIASURE. Conclusions: "VIASURE Real-Time PCR Detection Kits" represent an efficient diagnostic tool for the detection of noninfluenza viruses as they show a high sensitivity and specificity. Moreover, the results obtained by VIASURE were supported by commercial Real-time PCR kits. These data suggest that visualization of PCR in arrays platforms may result in a higher sensibility than fluorescent signal or leads to putative false positives results. Introduction: Genotypic and phenotypic resistance assays are used to assess viral strains and inform selection of treatment strategies. Standard genotypic drugresistance testing in ARV-naive persons involves testing for mutations in the reverse transcriptase (RT) and protease (PR) genes. If transmitted Integrase strand transfer inhibitor (INSTI) resistance is a concern, providers should supplement standard genotypic resistance testing with an INSTI genotype test the same approach is used in persons failing INSTI-based regimens. In this study we evaluate performance of ACL LDT HIV Integrase genotype assay for drug-resistance detection. Methods: Forty HIV positive plasma specimens were tested using ACL LDT HIV Integrase genotype (HIVING) assay and compared to results provided by ARUP Laboratory. Method 1: ACL LDT HIVGT assay: Abbott ViroSeq HIV-1 Integrase Sample Preparation Kit extraction, amplification on ABI 9700, detection by (36 cm) CE Sequencing on ABI 3130xl instrument, data assemble and analysis by ViroSeq analysis software provided by Abbott (last updated in 2011). Method 2: ARUP Laboratories/comparator. HIV Integrase sequencing data of forty (40) clinical samples belonging to a cohort of treatment-experienced patients was generated using three software; RUO ViroSeq HIV Integrase assay (Abbott Molecular, US 2011), RUO ViroScore-HIV system from Advanced Biological Laboratories 2013 and Stanford HIVdb v7.0.1 February 2014. Results: The ACL HIV Integrase assay performance was evaluated by assessing analytical specificity, sensitivity, interfering substances, and accuracy (clinical sensitivity and specificty) based on HIV positive specimens (in plasma, with or without drug resistance mutation(s)). ACL LDT HIVING produced 90% clinical sensitivity (18/19) and specificity (19/21) all samples were genotype M, subtype B (40) with viral loads ranging from (1000 to 150,000 cp/mL), analytical sensitivity 100% at VL=1000 cp/mL for HIV subtypes (B, G, C, CRF01_AE) and per in silico analysis HIV genotypes (D, F, A, CRF02_AG), analytical specificity 100% (11/11) in comparison to potentially cross-reactive microorganisms/viruses. Reproducibility was assessed by testing 20 times positive and negative control (expected result for positive QC were B genotype, amplicon length 864bp and (G123S, A124T, R127K, N232D, A265V) variants detected) all replicates matched 100%. Precision of the assay is 100% (in range 1000 cp/mL of viral load). Conclusions: This study demonstrates that the ACL LDT HIV Integrase sequencing assay is a sensitive and reproducible method for detecting HIV Integrase genotype and determination of drug-resistance status in clinical specimens. Introduction: HPV genotyping is relevant for studying the natural history of HPV infections and the diseases associated therewith, for monitoring vaccine efficacy, and for identifying novel HPV candidates for vaccine development. There is currently a need for a sensitive and specific HPV genotyping assay that is cost effective for testing large sample sets. The HPV prevalence at oral, penile, and anal sites is not well documented in Greek men. We developed a next-generation sequencing (NGS) assay for HPV genotyping, and applied it to test a cohort of Greek male samples. Methods: The assay uses PGMY09/11 primers modified by lllumina adapter and patient-specific index sequences to amplify the HPV L1 consensus region. Input DNA requirement is 10 ng. Up to 200 samples can be sequenced per flow cell on the Illumina MiSeq. HPV genotyping is obtained via a customized bioinformatics pipeline, which removes ambiguous and off-target sequences, aligns the filtered data to 179 HPV genomes from the Papillomavirus Episteme (PaVe) database, and scores each read based on its alignment to a specific HPV genome. A test cohort of 29 samples with previously characterized HPV status by a CLIA-certified multiplex qPCR and linear array, including HPV 16 (n=2), HPV 18 (n=2), HPV 33 (n=1), HPV 39 (n=1), HPV 45 (n=2), and 22 negative high-risk HPV specimens was used to assess the assay performance and to establish HPV positive and negative cutoffs. We then applied this methodology to test 882 samples of oral mucosa, glans penis, and anal mucosa from 294 Greek men with uncharacterized HPV status and genotype that were collected during their routine STD clinic visit at A. Syggros Hospital in Athens, Greece. Forty-three of these samples were genotyped by qPCR for comparison. Results: Concordance between NGS and qPCR was 96.5% (28/29) in the test cohort. One sample was positive only by NGS for HPV 59, which is not targeted by the qPCR method. The discordant sample was p16 positive by IHC. In 49% of Greek subjects, at least one HPV subtype was detected in one or more of the tested sites. HPV 6 was the most commonly found HPV type in this population. Genital warts were reportedly present in 42/53 (79%) of men with an HPV 6 or 11 infection. Correlation of NGS and qPCR for the Greek samples was 91% (40/44). Phylogenetic analysis showed that all HPV types found were found in the Alphapapillomavirus branch of the HPV tree. Conclusions: This NGS assay may be a viable platform for population studies and for routine diagnosis of HPV-associated cancers in a clinical setting. Results of this study could be used to create various intervention programs for Greek men, and to provide justification for expanding the current HPV vaccination schedule to men. Introduction: PCR tests for influenza and respiratory syncytial virus (RSV) must be sensitive and accurate to effectively guide patient care, ideally with the capability to be performed in multiple healthcare settings. We sought to 1) compare the overall sensitivity and specificity of the Cepheid Xpert Flu+RSV Xpress Assay to that of the Prodesse ProFlu+ test, and 2) evaluate concordance of results with those of ProFlu+ when performed by individuals with no formal laboratory training in point of care (CLIA-waived) environments. Methods: Nasopharyngeal (NP) swabs consisting of fresh-prospectively collected samples and pre-selected frozen specimens were obtained and tested from patients with flu-like symptoms presenting to emergency departments (EDs), outpatient (OP) and urgent care clinics (UC), and from hospital inpatients (IP) at 12 clinical sites (10 categorized as CLIA waived) during the 2014 -2015 respiratory flu season. Failed assays were retested one time. Performance of the Xpert Flu+RSV Xpress test was compared to the PCR-based ProFlu+ Assay and percentages of agreement were determined. Results: 2435 NP swabs (2176 fresh, 259 frozen) were obtained from a patient population consisting of 422 (17.3%) < 5 years, 421 (17.3%) 6-21 years, 1302 (53.5%) 22years of age. Samples were collected from 1120 (46%) ED; 565 (23.2%) OP; 688 (28.3%) UC patients and 62 (2.5%) IPs. Xpert Flu+RSV Xpress Assay tests for 94.9% (2329/2453) of specimens were successfully completed on the first attempt. Following a single retesting of 124 failed specimens, an overall 99.3% (2435/2453) assay success rate was achieved. Compared to the ProFlu+ Assay, the Xpert Flu+RSV Xpress Assay demonstrated a high positive, negative, and overall percent agreement, for detection of flu A, flu B and RSV for all samples tested in moderately complex and CLIA-waived settings (Tables 1 and 2). Introduction: K. kingae is a fastidious Gram-negative bacillus that causes bone and joint infections in young children. It can also cause bacteremia and endocarditis in adults and children. The primary method for diagnosing K. kingae infections is culture, which is, unfortunately, insensitive. We designed a LightCycler real-time PCR assay to detect K. kingae in blood and synovial fluid which can potentially be used to improve the diagnosis of infections caused by this organism. Methods: The K. kingae PCR assay amplifies a 156 bp portion of rtxB, a single-copy gene within the RTX toxin locus. The assay consists of a single forward primer, two internal FRET hybridization probes, one labeled with fluorescein and the other with LC Red 610, and a single reverse primer and is performed on the LightCycler 2.0 instrument. A plasmid containing the amplified sequence was created and used to determine the assay's limit of detection. Genomic material from a library of bacterial organisms, including non-kingae, Kingella species, was used to assess for cross reactivity. Thirty EDTA blood specimens and 30 synovial fluid specimens were spiked with a reference strain of K. kingae (DSM 7536) to a final concentration of approximately 130 CFU/μL. The spiked specimens were extracted and assayed along with 10 negative blood and 10 negative synovial fluid specimens. Results: Gel analysis of the DNA amplified from genomic DNA of K. kingae produced a single band at the appropriate size, indicating the desired target had been amplified; sequencing confirmed amplification of the rtxB gene. The limit of detection was 1 target copy/μl, as confirmed in 6 out of 6 replicates tested at this concentration. Cross reactivity studies were negative, which included Moraxella bovis which contains an RTX toxin locus with homology to the RTX locus present in K. kingae. The molecular assay detected K. kingae in all spiked blood and synovial fluids and yielded no false positive results in the negative blood and synovial fluids tested. We developed a highly sensitive and specific assay with the potential to improve the detection of K. kingae in blood and synovial fluid specimens. Future studies are needed on clinical specimens containing K. kingae. -lactams are the antibiotics most widely used all over the world, have also given rise to a continuous increase of resistance and driven diversification of the resistance --lactams. The extended--lactamases, such as CTX-M and SHV, hydrolyze cephalosporins and monobactams; whereas K. pneumoniae carbapenemases -lactamase genes reside in plasmids, transferring between different stains or species. Methods: Routine clinical MicroScan and Etest methods were used to profile the susceptibility of antimicrobial resistance in K. pneumoniae clinical isolates recovered from patient specimens during 2005-2014. Additional PCR amplification followed by Sanger sequencing was used for gene detection of blaCTX-M and blaKPC. A combination of short-and long-read whole genome sequencing was used for three isolates positive to both blaCTX-M and blaKPC. Comparative genomic analyses were further conducted to characterize detailed resistance gene structures and genomic features. Results: We identified three K. pneumoniae isolates, CN1, CR14 and NY9, carrying both blaCTX-M and blaKPC. Through de novo assembly, we obtained three complete genomes. In CR14 and NY9, blaCTX-M and blaKPC were carried in distinctive plasmids with distinctive incompatibility. In contrast, CN1 harbors in chromosome one blaKPC-2 and three blaCTX-M-15 genes, in addition to one blaSHV-11. Whereas blaCTX-M-15 genes were transferred via the same insertion sequence ISEcp1, blaKPC-2 gene was mobilized in the context of insertion sequence Tn4401a, which was found conjugative with a PsP3-like prophage in CN1 chromosome. Intriguingly, downstream of the newly generated Tn4401a-prophage genomic island, CN1 carried a CRISPR-Cas array with four spacers targeting a variety of K. pneumoniae plasmids carrying antimicrobial resistance genes. It suggests the evolving CRISPR-Cas, a bacterial immune system preventing phage infection and plasmid transfer, might have induced the mobilization of resistance genes from plasmids into chromosome under the selective force. Conclusion: We observed the emergence of K. pneumoniae isolates -lactamase genes. Our comparative genomic analyses revealed their shared genomic features and dynamic evolutionary events occurring in chr -lactam resistance from plasmids to chromosome depicts the currently complex pandemic scenario of multidrug resistant K. pneumoniae isolates. Introduction: Acute respiratory infections are a significant public health issue; therefore rapid diagnosis is crucial for appropriate patient care. The accuracy of these rapid multiple pathogen detection panels is critical to start and guide treatments. Best practice and regulations require laboratories to establish quality control programs for every assay they perform. The routine use of quality controls that are consistent lot to lot assists the laboratory in identifying shifts, trends, and increased frequency of random errors caused by variations in the test system, such as failing reagents. Maine Molecular Quality Controls Inc (MMQCI) has developed a synthetic, multiplexed control to simultaneously monitor all 12 common pathogens detected by BioFire's FilmArray RP EZ Assay. Methods: Synthetic, multiplex molecular constructs containing indidual genome segments of all respiratory pathogens were designed in silico to create a singular piece of synthetic DNA, ligated into MMQCI vectors and transformed to create stable frozen clone stocks. RNA transcripts were generated and quantified by 260/280 UV spec and formulated in MMQCI's proprietary matrix. The control was titrated to an optimal level for positive detection on the FilmArray RP EZ Assay to ensure reliable simultaneous detection of all targets included in the panel, at a concentration close enough to LOD to monitor the test system. Results: Three different manufactured lots of RP EZ control were tested across three FilmArray RP EZ assay pouch lots (n=35), with 100% detection of all targets with a standard deviation of 2 Cps. RP EZ Negative control panel gave accurate results across three BioFire pouch lots. Clinical data supplied by BioFire, consisted of 10 days of testing at three sites of three positive and three negative controls run per day (total of 60 control runs per site). Multiple operators participated in testing at each of the three locations. Three lots of external control material and 3 lots of RP EZ pouches were tested across all sites. 100% concordance was seen for all three positive controls and 97.8% concordance for all negative controls from 182 RP EZ test results. Conclusions: MMQCI's proprietary matrix and stabilization buffers allow for the synthetic RNA to be maintained as a stable and reliable control that can be carried through the entire molecular diagnostic assay including the extraction process. The RP EZ Control Panel offers a ready-to use, non-infectious control material designed to monitor multiplex assays with just two controls.This qualtitative, synthetic molecular control material provides a wellcharacterized reference to be used for verification, validation or as routine controls. Panel with High-Throughput BioCode MDx3000 Instrument J. Kirchner, C. Knoth, M. Henrie, B. Shah, T. Ul-Hasan, D. Mantzke, A. Pham, M. Aye Applied BioCode, Inc., Santa Fe Springs, CA. Introduction: Gastroenteritis is the second most common illness after the common cold. Globally, diarrhea accounts for approximately 2 million annual deaths in children under 5 years old, or about 2,000 child deaths per day (Boschi-Pinto C, Velebit L, Shibuya K. Bull. World Health Organ. 2008; 86(9):710-7). High-throughput multiplex assays allow for both rapid identification of diarrhea-causing pathogens and enhanced infection control in healthcare settings. Using proprietary barcoded magnetic bead (BMB) technology, Applied BioCode has developed a molecular diagnostic assay for detection of gastrointestinal pathogens. Concurrently, we have developed the BioCode MDx3000, an automated high-throughput instrument with a 96-well format. Methods: The BioCode MDx3000 instrument integrates and automates PCR, post-PCR sample handling and detection steps in a 96-well format. Unpreserved stool or Cary-Blair samples were subjected to bead lysis and extracted with an automated instrument. Extracted samples were combined with RT-PCR chemistry and loaded onto the BioCode MDx3000 instrument where they were amplified by PCR, transferred to a 96-well plate and captured by target-specific probes coupled to BMBs. The presence of captured target sequence(s) was detected by a fluorescence conjugate and qualitative results were determined by median fluorescence intensity (MFI) values relative to assay cutoff. Results: The BioCode GI Pathogen Panel is a high multiplex assay with preliminary limit of detection (LoD) comparable to currently available molecular assays (STEC 2.4E+03 CFU/mL, Rotavirus 6E+02 TCID50/mL, Cryptosporidium (hominis and parvum) 2.5E+04 oocytes/mL, C. difficile A+B+ 7.7E+02 CFU/mL). The panel is highly specific and gave robust MFI for 60 targeted GI pathogens, including synthetic jmd.amjpathol.org ■ The Journal of Molecular Diagnostics mimics of norovirus GI 1-8 and GII 1-14, while not cross-reacting with the off-target organisms tested in this study. No well-to-well contamination was observed. Equivalent detection was observed between unpreserved stool and samples in Cary-Blair transport medium. Evaluation of performance with a set of clinical specimens showed 91% positive agreement and 94% negative agreement compared to a FDA cleared test. Conclusions: The BioCode MDx3000, automated for integration of RT-PCR, post-PCR sample handling and detection steps, enables multiplex molecular detection in 96-well format whereas BioCode GI Pathogen Panel specifically detects and discriminates 18 bacteria/toxins, viruses and parasites. In combination, this platform and assay will allow users to perform multiplex molecular detection in a high throughput format, thereby simplifying the workflow, reducing hands-on time and minimizing contamination risk. (>80%) between the three methods. The ART assay detected 12 %and 14% additional patient samples that CAP/CTM and altona missed, respectively; whereas CAP/CTM detected 3% and 10% patient samples that ART and altona missed, respectively. Method bias ranged from -1.2 log IU/mL between Roche and altona, -0.2 between Abbott and altona and finally 0.8 log IU/mL between Abbott and Roche (Table 1 ). High degree of agreement was observed between methods at the CAP/CTM and altona LOD (>79%) whereas relatively low level of agreement was observed at the ART LOD (>7%). Conclusion: Method comparison demonstrated the importance of further understanding differences which contribute to assay variability in light of the 1 st CMV WHO IS. Overall method bias ranged from 0.8 log IU/mL to -1.2 log IU/mL. Additionally, this study demonstrated that the ART assay detected CMV residual viremia more readily than the other two commercial assays and in turn would be able to identify early therapy failures or disease progression. Dilutions of viral material were tested and probit analysis was subsequently performed to determine the limit of detection (LOD) for each sample type. Patient samples positive for other viruses commonly identified in plasma and urine specimens were tested to determine analytical specificity. Inter-assay and intra-assay precision studies involved a panel of 6 specimens tested independently by 3 technologists and tested in triplicate within the run. Additionally, method comparison analyses were done to determine the degree of agreement between the assays. Results: Compared to the ViraCor and Scott & White assays, the clinical sensitivity using the Focus ASR assay was 98.3% (59/60); 100% (40/40) for plasma and 95% (19/20) for urine. Clinical specificity was 97.1% (68/70); 96% (48/50) for plasma and 100% (20/20) for urine. Regression analysis of plasma and urine results revealed a high correlation (R 2 = 0.9804 and 0.9467, respectively). A strong correlation was also seen using Pearson correlation; r = 0.9901 for plasma and 0.9730 for urine. Bland-Altman plots for plasma and urine determined the mean bias to be -0.227 log10 and -0.616 log10, respectively. The LOD with a 95% confidence interval was determined to be 54 cp/mL for plasma and 119 cp/mL for urine. Furthermore, no cross-reactivity was observed for any analytical specificity samples tested and all precision studies produced equivalent results, yielding coefficients of variation less than 6%. Conclusions: The Focus ASR assay performed comparably to the assays performed by ViraCor and Scott & White and provides accurate and reproducible results for the detection and quantification of BKV viral load in plasma and urine specimens. (BKV) replication is recommended to identify post-transplant patients at risk of BKV-associated diseases and to guide therapeutic intervention. BKV quantification in urine by real-time PCR is noninvasive and can be informative, as a high level of BK viruria usually precedes viremia and potential nephropathy by 4 to 12 weeks. However, the heterogeneity of qPCR methods among different laboratories limits interlaboratory comparison and hampers the establishment of thresholds to distinguish asymptomatic viruria with clinical disease. This study aimed to evaluate the performance of the QIAsymphony RGQ system for BKV DNA quantification (BK-Q) using artus BKV QS-RGQ reagents. Results were compared to our Laboratory-Developed Test (LDT) for correlation and agreement. Methods: Total nucleic acid extraction and qPCR reaction set-up were performed using an integrated mode on the QIAsymphony SP/AS instruments. The QIAsymphony DSP Virus/Pathogen Mini kit in combination with the Complex200 protocol (input volume: 200 μL, elution volume: 60 μL) were used for extraction. The artus BKV QS-RGQ reagents (Qiagen) were used to run the qPCR on the Rotor Gene Q instrument. Calibration was determined using serial dilutions of a commercially available BKV (ExactDx, Fort Worth, TX). Linearity, precision and lower limit of quantification (LoQ) were evaluated using a serial dilution panel from a patient sample nominally quantified in-house and then diluted in urine negative for BKV. Correlation and bias were assessed using 31 urine specimens (21 positive, 10 undetectable) from transplant recipients previously quantified by the LDT. Results: Linearity was observed from 3.18 to 7.0 Log10 BKV copies/mL (R 2 =0.99). The LoQ was 1,510 copies/mL which is well below the clinical significant values. The assay showed high precision with standard deviations from 0.05 to 0.20 Log10 (CV: 1% to 8%). Quantitative data obtained with BK-Q correlated (Pearson's r = 0.95) but was The Journal of Molecular Diagnostics ■ jmd.amjpathol.org not identical to LDT, with BK-Q values trending lower. All undetectable samples by LDT were below the BK-Q LoQ. Conclusions: The BK-Q demonstrated reliable performance. Monitoring of BKV replication with standardized system may enable clinical laboratories to assess quantification with greater reproducibility and precision. The liquid Amies appears to be a useful culture medium to improve stability and viability of certain organisms. Our data shows that ES 480C could be used as a universal collection medium for molecular testing in a clinical microbiology lab. Step Algorithm for the Diagnosis of C. . In addition to the FDA cleared HSV1&2 assay, the ARIES system also has the ability to run ASRs and LDTs by using either ASR primers or primers purchased from a third-party vendor, ARIES extraction cassettes and ARIES Ready Mix. Objective of this study was to verify the performance of HSV1&2 assay, and validate ASR primers for Candida albicans (CA), Candida glabrata (CG), Gardnerella vaginalis (GV) and Trichomonas vaginalis (TV) on ARIES platform. Methods: The vaginosis panel (VP) comprised of CA/CG run in one ARIES extraction cassette and GV/TV run in a second cassette. Primers were loaded into ARIES Ready Mix tubes prior to snapping them onto the ARIES extraction cassette. ARIES HSV IVD cassettes were used per the manufacturer protocol. All 3 assays used 200μL input volume. Results were reviewed using cycle threshold (Ct) and melt curve (TM) analysis on ARIES for the HSV IVD tests per manufacturer protocol and in the SYNCT software using the UDP (User Defined Protocol) for VP. The limit of detection, clinical accuracy, precision, specificity and stability were evaluated. For clinical accuracy, the VP panel on ARIES was compared to either BD Affirm VPIII (for GV, CA and TV) or Aptima TV assay on Hologic Panther platform (for TV) whereas HSV1&2 assays were compared to ROCHE LightCycler HSV 1/2 Qualitative assay. Results: ARIES uses Multi-Code real-time PCR technology to identify and discriminate targets using melt curve analysis. Compared to Affirm, ARIES was not only able to accurately identify Candida species but also sub-type all 3 patients as CA. BD Affirm VPIII detected 7 of the 8 GV positives detected on the ARIES but missed a TV sample which was positive on the ARIES . ARIES detected 90.9% (10/11) of the TV positives detected by Aptima TV assay. ROCHE LightCycler HSV 1/2 Qualitative assay missed one of the 8 HSV1 samples that were positive on ARIES. In summary, ARIES was more sensitive than Affirm in accurately identifying TV and equally sensitive compared to Aptima TV assay. Sensitivity for HSV1&2 was similar to ROCHE assay. Conclusions: ARIES offers a sample to answer solution with versatility, flexibility and scalability for a growing molecular lab. It also appears to have good sensitivity as compared to some of the other real-time PCR platforms currently available. The VP assay on ARIES is easy to perform and has better sensitivity compared to currently widely used BD Affirm with ability of also discriminating between Candida species. typically receive empirical treatment with broad-spectrum agents, which are inadequate in up to 30% of cases. Inadequate therapy is associated with increased mortality, morbidity, and length of hospital stay. Rapid identification of BSI pathogens and resistance markers directly from clinical specimens, bypassing the requirement for lengthy culture upstream, would facilitate appropriate therapy to improve survival rates, reduce hospital costs, and diminish the development of antimicrobial resistant organisms. We report the sensitivity of a novel method to detect BSI pathogens and resistance genes directly from whole blood and some validation results of the method on clinical specimens. Methods: The sensitivity of component steps in the Genalysis Blood Stream Infection Identification (G-BSI) assay was established for a large range of BSI pathogens. Sample prep sensitivity was measured by spiking 10 mL samples of healthy donor blood (n=20, per pathogen) at nominal concentrations of 1 CFU/mL. Recovered cells were plated on solid medium and quantified after overnight incubation at 37°C. Sensitivity of nested PCR to species-specific ID was measured by spiking genomic copies of C. albicans, S. aureus, E. faecalis, E. coli, and mecA, in the presence of 500 ng human DNA in a titration series into a preamp reaction, with PCR detection by measurement of pH changes by ISFET sensors on a CMOS sensor chip using specific primers. End-to-end functionality was confirmed by testing clinical specimens obtained from consented patients at the University of New Mexico Hospital. Results: Sample prep sensitivity to 1 CFU/mL was demonstrated for nine Gram-positive bacteria, eight Gram-negative bacteria, and three yeasts. Nested PCR showed sensitivity down to a single genomic copy for the four microorganisms and mecA resistance marker. End-to-end functionality was demonstrated on a set of clinical specimens; 7 specimens were correctly identified as negative (concordant with blood culture) and 2 specimens correctly identified as positive (K. pneumoniae, S. aureus). Conclusions: Rapid detection of pathogens in spiked and clinical samples was demonstrated with Genalysis. The current manual process allows pathogen ID by molecular methods, at the 1 CFU/ml level, in <5 hours. Ongoing automation of the process is expected to reduce the total process time to <3 hours, providing therapeutically actionable information in a clinically relevant time frame. Screening for GBS colonization in antepartum women between 35 and 37 weeks' gestation, followed by intrapartum antibiotic treatment for women with positive colonization status has proven to be an effective mechanism for prevention of perinatal GBS disease. Here we describe the implementation of a polymerase chain reaction (PCR) based GBS Assay (Great Basin Scientific), a qualitative in vitro diagnostic test (IVD) for the detection of GBS DNA from vaginal/rectal swabs from antepartum women. Methods: Samples included 20 known GBS, and 20 known E. faecalis QC organisms performed over 20 separate days, and 26 patient specimens run concurrently with the Illumigene Pro GBS Assay (Meridian Bioscience) which was the procedure in place. All samples were initially cultured in Hardy Carrot broth (Hardy Diagnostics #Z140) which uses the Granada medium reaction and contains the necessary components for pigment detection of betahemolytic GBS and produces positive results in as little as 6 hours. Following incubation in enrichment broth (utilizing Hardy Carrot broth as the enrichment broth rather than LIM broth) for 18 -29 hours, samples underwent automated sample preparation and PCR on the PA500 Portrait Analyzer System to amplify a cfb gene sequence specific to the GBS genome which is detected by hybridization probes immobilized on a silica chip surface according to manufacturer's instructions. Results: Of the 20 known positive and 20 negative control specimens tested, all yielded the expected result. Of the 26 patient specimens, 19 were negative by both Illumigene and Great Basin GBS. Of the 7 positive specimens, 3 were positive by pigment change in Carrot broth, 2 were positive by culture, CAMP and/or Vitek GP ID, and 2 were positive by Illumigene. Conclusions: The Great Basin GBS assay demonstrated 100% concordance with expected results over 66 specimens and three methodologies. The Carrot broth has proven an effective enrichment media for the assay. In our current workflow, negative Carrot broth specimens are tested using the Great Basin GBS Assay to detect non-hemolytic strains of GBS. (2008) was conducted on 90 clinical plasma or serum specimens. Results: All 35 of the published ZIKV strain sequences showed 100% homology with at least one of the two amplification reactions. No amplification was observed in ZIKV-negative clinical specimens. No cross-reactivity was observed with any of the pathogens tested and no significant sequence homology was found for any of the 51 organisms evaluated in silico. An assay LoD of at least 0.05 U/mL (TCID50) was established in both plasma and serum on each of the three thermal cyclers. In the method comparison, both assays detected ZIKV in each of 34 specimens and did not detect ZIKV in each of 44 specimens. The VERSANT Zika RNA 1.0 Assay (kPCR) detected ZIKV in an additional 10 specimens that the comparator CDC assay did not, whereas the comparator CDC assay detected ZIKV in 2 specimens that VERSANT Zika RNA 1.0 Assay (kPCR) did not. The VERSANT Zika RNA 1.0 Assay (kPCR) qualitatively detects ZIKV RNA. The assay recognizes a broad spectrum of published ZIKV RNAs in silico, has high analytical sensitivity, is specific to ZIKV among Flaviviridae, and has excellent performance with clinical specimens. (CDI) is a leading cause of hospitalassociated gastrointestinal illness. Many clinical labs use molecular diagnostic assays for C. difficile to screen unformed stool samples from patients with signs and symptoms of CDI. However, a future potential application of such assays is for use in screening asymptomatic patients (either on admission or during a hospital stay) for presence of the C. difficile organism using unformed stool or peri-rectal swabs. This will assist hospitals in isolating asymptomatic carriers, especially from immunocompromised in-patients, to prevent the spread of infection. The goal of this study was to compare the performance of unformed and formed stool samples in the Simplexa C. difficile Direct assay, and compare the Simplexa assay to Cepheid Xpert C. difficile. The reproducibility of the Simplexa Direct assay with formed stool as well as the ability of the Simplexa Direct assay to detect a wide range of C. difficile isolates was also tested. Methods: Limit of detection (LoD) panels were created by serially diluted C. difficile strains ATCC 43255, NAP1A and BAA-1805 and spiking into negative formed stool to mimic clinical samples. The LoD panel was screened with both molecular assays (Simplexa and Xpert) in triplicate to determine LoD. A total of 33 C. difficile isolates ranging in different toxinotypes and ribotypes were evaluated for analytical reactivity in the Simplexa assay. A reproducibility study was performed for the Simplexa assay using medium and low positive contrived samples in formed stool matrix. Results: LoD studies showed that the Simplexa C. difficile Direct assay had similar performance with the three strains tested in formed stool. The approximate LoDs for ATCC 43255, NAP1A and BAA-1805 for Simplexa were 3.1, 400 and 119 CFU/mL and were 1.6, >410 and >121 CFU/mL for Xpert. The Simplexa assay detected all 33 characterized isolates. Conclusions: The Simplexa C. difficile Direct assay was capable of detecting C. difficile strains spanning a wide range of toxinotypes and ribotypes. Comparison between the Simplexa and Xpert assays showed that the assays demonstrated similar sensitivity when tested with formed stool. The Simplexa assay also demonstrated reproducible performance for formed stool samples and may be useful for screening asymptomatic carriers. Simplexa C. difficile Direct is in development; the assay is not currently available for sale and is not FDA cleared. Introduction: The first report of HIV anti-retroviral therapy (ART) resistance occurred in 1989 after the release AZT. Since this initial report, HIV resistance testing, also referred to as HIV genotyping, has become an integral part of HIV care. HIV-ART resistance has significant clinical implications for choosing effective ART regimens. Genotypic resistance assays used by most clinical laboratories detect the presence of specific drug resistance mutations in regions of the HIV genome encoding protease (PRO), reverse transcriptase (RT), and integrase (INT). Results are reported as individual mutations (eg, M184V, the signature mutation for lamivudine resistance) with comments such as "susceptible," "possibly resistant /intermediate," or "resistant" for each antiretroviral agent based on known impact of each mutation reported. Currently, a single FDA cleared assay (Abbott Viroseq) is available for HIV genotyping. We report the validation/comparison of a laboratory developed procedure (LDP) HIV genotype protocol, incorporating the use of SmartGene HIV module, for determination of ART resistance. Methods: A total of n=94 unique HIV cases representing n=15 HIV-1 subtypes were amplified using the Viroseq assay and compared to the HCMC-HIVgt assay spanning PRO, RT, and INT regions. Amplification efficiency was partially determined by concentration of the amplified product(s) prior to Sanger sequencing using the ABI 3500 genetic sequencer. A single extract, using automated EasyMag extraction was used for all amplification reactions. Comparison/correlation of i) amplification efficiencies, ii) failed sequence runs, iii) resistance mutations detected iv) interpretative resistance/susceptible reporting using Viroseq and SmartGene software and v) limit of detection were compared. Results: Overall, 83% correlation in interpretive reporting, including all forms of associated ART resistance was observed. For cases involving "Major HIV Associated Drug Resistance Mutations" correlation improved to 95%. Eight cases (8.5%) failed to produce reliable genotype results using the ViroSeq assay compared to five cases (5.32%) using the HCMC-HIVgt assay. Conclusions: A 95% correlation in detection of clinically significant HIV resistance mutations between the two genotype protocols was observed. Use of the SmartGene HIV-1 module allowed streamlined reporting of all HIV ART drug classes within a single, consolidated patient report and allowed for longitudinal resistance analysis and reports for 100% of cases. The HCMC-HIVgt assay used with the SmartGene HIV-1 module represents an alternative approach for reliable HIV genotyping using LDP protocols for determination and reporting of HIV-1 resistance and a highly cost-effective approach for our institution. Introduction: To comply with CLIA/CAP regulations for NGS data analysis, we developed a bioinformatics pipeline to process raw sequencing data from ampliconbased target enrichment NGS panel. The pipeline has two parts: a Galaxy workflow to generate a processed variant call file (VCF), and variant filtering and annotation using third party software. A modified GRCh37 reference that contained artificially integrated variants was used to generate synthetic sequencing reads in FASTQ format. These FASTQ fileswere processed to assess accuracy, sensitivity, specificity, and precision. These files can be used to re-validate updated pipeline versions. Methods: Synthetic FASTQ files: Wild-type sequences were extracted from hg19/GRCh37 referencing the NGS panel BED file. Spike-in variants were sourced from known mutations in Coriell samples and also contained a number of random variants from dbSNP (v144). Zygosity of Coriell mutations were based on observation in samples and all selected dbSNP variants were randomly assigned zygosity based on a 1:3 homozgous to heterozygous ratio. To mimic the diploid human genome, homozygous variants were integrated into one set of wild-type sequence, heterozygous and homozygous variants were integrated into another set of wild-type sequence. Primers were subsequently added to create synthetic amplicons. In silico sequencing reads in FASTQ format were sampled from these synthetic amplicons by ART using Illumina sequencing error and quality data profile. Ten synthetic samples were generated to assess sensitivity, whereas three wild samples and four duplicate samples simulated by different random seeds were also included for specificity and inter-assay repeatability assessment. Informatics Pipeline: FASTQ files were processed by FASTQ Parallel Groomer tool and Trimmomatic to remove low quality reads. The remaining reads were mapped tohg19/GRCh37 reference genome using BWA-MEM. The alignment files were further intersected by the NGS panel BED file and once the amplicon region was retained, primers were removed. Variant calling for SNVs and short indels was performed with Freebayes. The resulting VCF was uploaded to Bench (Cartagenia, MA) and variants are filtered based on call quality, population frequency, panel artifacts, and known pathogenicity. Results: The following assay metrics were calculated: Accuracy, 99.99%; sensitivity, 98.58%; specificity, 100%; precision, 100%. Conclusions: Use of synthetic FASTQ files enables evaluation of the accuracy of an NGS bioinformatic pipeline independent of sequencing run quality. This approach provides a highly efficient, reproducible, cost-effective approach for QC/QA in a clinical laboratory. High-Resolution Melting Z.L. Dwight 1 , C.T. Wittwer 2 1 University of Utah Health Sciences, Salt Lake City, UT; 2 University of Utah, Salt Lake City, UT. Introduction: High-resolution melting is a rapid, low-cost diagnostic tool for genotyping. This technique visualizes the thermodynamic differences in PCR products with fluorescent melting curves. Melting curves can differentiate and identify types and subtypes of viruses. The same virus can exhibit sequence variation between patients and even within a patient due to high rates of replication and mutation. Designing assays that offer specificity and sensitivity can be challenging, forcing inefficient 'shotgun' approaches to primer design and experimental testing. The software presented uses systematic frequency analysis to identify primer pairs that will amplify the intended targets with the highest specificity possible while optionally excluding unwanted types or subtypes. Methods: Multiple sequences for the intended target are input in FASTA format. All sequences are divided into an exhaustive set of possible primers that are within a Tm range. Any redundant primers are removed. Each unique primer is then compared to each target sequence. If a perfect match is found, the consensus score is set to 100%. If a mismatch occurs, both the mismatch Tm and a weighting factor based on mismatch/bulge position and type lowers the consensus score. The average consensus score across all target sequences are ranked and primer sets that match a product size and/or Tm specified by the user are then listed. This consensus portion of the algorithm provides primer pairs that will amplify the intended target (descending order of average consensus rank) that are within selected primer Tms, products sizes, and product Tms. Optionally, users may define unwanted targets by importing a second set of sequences. That is, the desired primers need to avoid amplifying this second set of sequences. Similarly, this segregation score will identify primers with high consensus to the target set and high differentiation of the unwanted set. Results: The software was tested with the goal of amplifying all types of Hepatitis C with sequences obtained from the Los Alamos HCV Database. The software was also tested against Zika sequences obtained from NCBI's Zika Virus Resource. In the case of Zika, NCBI's PrimerBlast was also used to discriminate against other viruses. For HCV and Zika, the top four assays were selected from the software results, and all successfully amplified the intended target. Conclusions: This software provides a straightforward approach to designing assays to amplify and differentiate difficult viral targets. Focusing on assay design rather than primer characteristics in the context of the intended targets greatly reduces the parameters often believed to be required in primer design. Introduction: Whereas laboratory information systems (LISs) have grown increasingly complex to meet the demands of expanding anatomic and clinical pathology laboratories, they continue to neglect the nuances of the molecular laboratory. As a result, basic functionalities necessary for molecular workflow are glaringly absent. This study determined functionality gaps of laboratory information systems (LISs) in molecular laboratories and the associated impact to workflow, efficiency, and security by collecting anonymous survey data from clinical laboratory professionals. Methods: A 34 question survey (30 required + 4 optional questions) was compiled using an online survey tool (Survey Monkey, Palo Alto, California, USA). Participants were recruited through professional molecular society listservs (AMP, ASHI, API, AMIA) and given four weeks to complete the survey. Data collected included participant demographics, scope of testing, software capabilities for the LIS and molecular instruments, and comments. Results: Eighty respondents completed the entire survey. LISs included basic anatomic pathology (26), custom (20), basic clinical pathology (19) and molecularspecific LIS modules (14). 55% of respondents could not record nucleic acid extraction data into the LIS. In contrast, 87.5% listed this ability as important. For sign-out, 65.0% manually type the result into the LIS because no electronic transfer was available. Only 12.5% reported full electronic transmission of results from the instrument to the LIS. Conversely, 96.3% listed this function as important. HIPAA requires that any software with identifiable patient information have audit trails and unique user logons. Only 68.8% of respondents reported ability to access audit trails in the LIS, with the remainder stating that they had to use other mechanisms to get this data. Strikingly, only 16.3% of participants stated that unique user logins were required in all instruments, whereas 41.3% and 42.5% reported unique user logins in some and no instruments, respectively. Conversely, 100% of respondents listed HIPAA compliance as important. Conclusions: The respondents indicated that many basic functionalities are lacking in the majority of LISs. Further, because of the lack of reported access to audit trails and enforcement of unique user logins, these results raise concerns that both molecular laboratory instruments and LISs are not compliant with federal HIPAA security law. Collaboration between molecular professionals and software vendors is necessary to correct these major deficits and is critically important to ensure the continued high quality and safety of molecular practice. Introduction: Current primer trimming approaches in amplicon next-generation sequencing (NGS) data is a double-edged sword, since to minimize variant allele frequency (VAF) dilution effect to mutations at primer binding sites through trimming may compromise the detection of mutations near primer binding sites due to nonalignment at the edge of amplicon. Since the current approaches work in prealignment level, both pre-trimming and post-trimming NGS reads of a given sample need to be analyzed in parallel. The effort in analysis and interpretation is doubled for optimal detection of mutations both at and near primer binding sites. Methods: We developed a post-alignment primer trimming bioinformatics algorithm BAMClipper, which trims primer sequence by soft-clipping BAM alignments of original NGS reads. VAF dilution effect and alignment edge effect are minimized simultaneously to cater for mutations both at and near primer binding sites, respectively. The Perl-based algorithm reads and writes in standard BAM alignment format and needs only primer genomic positions that are commonly provided by commercial panel vendors. It is independent of gene panel or organism and can be readily incorporated into any bioinformatics pipeline without any compilation process. We evaluated BAMClipper with Illumina MiSeq sequencing reads of QIAGEN GeneRead DNAseq Human Breast Cancer Panel. Frozen tissue sample of 6 breast cancer patients with a known germline BRCA1/BRCA2 mutation was analyzed by the 44-gene panel. Original and BAMClipper-processed BAM alignments were processed by the same bioinformatics workflow on a Cray XC30 supercomputer. VAF of the known mutation were compared. Results: Among the 6 known BRCA1/BRCA2 mutations, 5 mutations were overlapped with primer binding sites of nearby amplicons and thus susceptible to VAF dilution effect. VAF was underestimated by 16% to 82% before primer trimming. In an extreme case, VAF was 9% before trimming (51% after trimming) to the extent that common germline mutation callers could miss the heterozygous mutation due to low VAF (false negative). The remaining mutation was not near or at any primer site. Trimming did not affect its VAF (51%). Conclusions: Primer trimming at post-alignment level restored otherwise underestimated VAF of mutations at primer binding sites, without compromising detection of any mutations near the sites due to alignment edge effects. Unnecessarily duplicated computational analysis and human interpretation efforts can be avoided while mutation detection performance is maintained. The algorithm is particularly relevant to clinical molecular pathology laboratories utilizing amplicon-based NGS panel for germline and somatic mutation detection. Introduction: Screening of polymorphisms in PCR primer binding sites is essential during evaluation of any PCR-based diagnostic assay. Otherwise, sequence mismatches between primer and template may result in sub-optimal PCR reaction and potentially false positive or negative results. Although public catalogs of human variation data (e.g. 1000 Genomes Project Phase 3 (1KG) and Exome Aggregation Consortium (ExAC)) and possibly in-house catalogs are rapidly growing, there is a lack of high-throughput tool for laboratories to evaluate hundreds of primer pairs at once. The popular online tool SNPCheck does not support the latest public catalogs and it is also recently announced that the service will wind down. Methods: We developed a user-friendly bioinformatics tool PanelQC for polymorphism screening of multiple primer pairs at once. In addition to the latest 1KG and ExAC catalogs, it also supports any additional catalogs in the standard VCF format. Laboratories can keep the tool on premises to utilize their in-house private catalogs. Allele frequency threshold (e.g. 5%) can be configured to flag any primers potentially affected by polymorphisms within primer binding sites. Various in-house and commercial PCR primer sequences were evaluated by PanelQC with 10% 1KG global minor allele frequency (GMAF) as flagging threshold for demonstration. Results: We tested PanelQC with 296 pairs of laboratory developed assay primers in our ISO15189 accredited molecular pathology laboratory as part of internal regular checking exercise. The automated screening was completed in 15 seconds. We also performed a preliminary screening of a commercial next-generation sequencing (NGS) gene panel (QIAGEN GeneRead DNAseq Human Breast Cancer Panel) and PanelQC completed in 11 minutes. The 44-gene panel comprised 2915 amplicons, of which 121 amplicons (4%) was potentially affected by at least one polymorphism with GMAF >10%. However, whether or how these polymorphisms affect the PCR targeting efficiency remained to be clarified because presence of any countermeasure in the actual reagent was unknown (e.g. degenerate base, additional primers). Conclusions: A new tool PanelQC was developed for efficient screening of polymorphism in multiple PCR primer pairs. As highly multiplexed PCR primer panels (e.g. commercially available NGS gene panels) become increasingly popular, the tool should be particularly useful to facilitate comprehensive polymorphism screening in thousands of primers for high-quality molecular diagnostic service. Introduction: Performance evaluation of next-generation sequencing (NGS)-based clinical molecular diagnostics needs reference materials that are sustainable, comprehensive and realistic. In silico simulation of NGS data with the target mutation profile is a complement (but not replacement) to fully experimental approaches, particularly for the bioinformatics part of evaluation. Full simulation of NGS data from scratch is sustainable and comprehensive but requires panel-specific data modeling to become realistic. Although semi-simulation approach (in silico adding mutations to real NGS reads) looks more flexible and realistic, there are few available and userfriendly tools to perform such semi-simulation. Methods: We developed an easy-touse bioinformatics tool MutationEngineer for semi-simulation. Given a mutation list (in standard VCF format) and BAM alignments of any sample, the tool will return modified BAM alignments with mutations added. Resulting BAM alignments per se or FASTQ reads after back-conversion can be fed into any bioinformatics pipelines of interest. Optionally locus sequencing depth and mutation allele frequency (AF) can be chosen. The tool written in Perl was tested to be compatible with standard Linux system or a Cray XC30 supercomputer without compilation. Semi-simulation of mutations at various depths and AF was performed for a hereditary breast and/or ovarian cancer (HBOC) gene panel (germline mutation) and a myeloid neoplasm gene panel (somatic mutation). Results: We applied semi-simulation on the 4-gene HBOC panel data (Illumina MiSeq) of a breast cancer patient with known mutation status by Sanger sequencing of BRCA1 and BRCA2 full genes. For a panel of 4 mutations (BRCA1, BRCA2, TP53 and PTEN), 24 sets of FASTQ reads are semisimulated corresponding to 4 different depths (20X, 50X, 100X and 200X) and 6 different mutation AF (20%, 25%, 30%, 40%, 50% and 100%). In-house bioinformatics pipeline for clinical reporting detected all 4 mutations in all tested depth-AF combinations. Likewise, JAK2 V617F semi-simulation was performed on the 54-gene myeloid panel data (Illumina MiSeq) of a healthy adult with normal complete blood profile. In-house bioinformatics pipeline detected the mutation in all 18 depth-AF combinations (depth: 100X, 200X, 500X; AF: 10%, 20%, 30%, 40%, 50% and 100%). Conclusions: Mutation semi-simulation was demonstrated to be a feasible approach in evaluating bioinformatics aspect of NGS-based molecular diagnostics. MutationEngineer facilitates molecular pathology laboratories to scale up evaluation to a wider mutation spectrum and depth-AF combinations. Here we aim to assess the prognostic utility of mutational signatures in ovarian high grade serous carcinoma. Methods: Open access whole exome sequencing data of 15,439 somatic single nucleotide variants of 310 ovarian high grade serous carcinomas from The Cancer Genome Atlas (TCGA) are used to construct a Bayesian model to classify each cancer as either having or lacking a BRCA1/2 mutational signature. We evaluate the association of the BRCA1/2 signature with overall survival on the TCGA dataset and on an independent cohort of 92 ovarian high grade serous carcinomas from the Australian Ovarian Cancer Study (AOCS). Results: BRCA1/2 mutated ovarian carcinomas exhibited a distinctive mutational signature, characterized by a decreased enrichment of C>T transitions at CpG sites compared to BRCA1/2 wildtype carcinomas (11.5% versus 17.8%, p = 0.0004) as well as a relative increase of T>A mutations at GpTpC sites (0.81% versus 0.33%, p = 0.0001). Patients from TCGA with tumors harboring a BRCA1/2 mutational signature have improved survival (57.8 months versus 38.0 months), which is independent of BRCA1/2 gene mutation status, age, stage and grade (HR = 0.52, p = 0.0003). In the AOCS dataset, the BRCA1/2mutational signature is also associated with improved overall survival (46.7 months versus 22.7 months) independent of stage (HR = 0.50, p = 0.004). Conclusions: A BRCA1/2 mutational signature is a prognostic marker in ovarian high grade serous carcinoma. Mutational signature analysis of ovarian cancer genomes may be useful in addition to testing for BRCA1/2 mutations. These analytical methods may expand the clinical indication for the broad sequencing of cancer genomes. Introduction: Germline and somatic mutations in the BRCA1 and BRCA2 genes are highly involved in hereditary and non-hereditary breast and ovarian cancers. A test that detects these mutations from clinically relevant FFPE samples is tremendously valuable for both research and future clinical diagnostics purposes. Large rearrangements (LRs) represent an important portion of BRCA1/2 mutations, in addition to single nucleotide mutations and small insertion/deletions. The sizes of LRs make them difficult to detect using traditional sequencing approaches thereby requiring additional tests such as multiplex ligation dependent probe amplification (MLPA). Recent reports showing feasibility using amplicon-based massively parallel sequencing methods to detect LRs either were not designed for use with FFPE samples, or lack data analysis methods optimized for such an application. We have developed an amplicon-based NGS approach* for FFPE samples that can detect SNVs, small mutations and LRs simultaneously. We have implemented a comprehensive bioinformatics algorithm that detects LRs at high sensitivity, even in the absence of control sample(s). This significantly reduces the cost and labor for BRCA1/2 genetic analyses. Methods: Ion AmpliSeq, a targeted, multiplexed amplification technology was used in combination with the Ion PGM and Ion S5 sequencing platform. An assay was designed to cover 100% of all exons of BRCA1 and BRCA2 with 263 amplicons (targeted regions). Torrent Suite Server and Torrent Variant Caller were used for analysis of small mutations. For LRs, a standalone algorithm was developed that includes custom normalization, detection of whole gene deletions, sample and run quality control metrics as well as exon-level copy number detections. NGS data from libraries made from cell line genomic DNA and FFPE derived DNA was used to train and verify the assay. To test this approach, NGS data was generated from an early access study comprising more than 20 independent research organizations. At least 4 sites included samples with LRs that were previously characterized by MPLA. Results: We demonstrate high sensitivity and specificity for both LR and small mutation detection from cell line, blood and research FFPE samples. We detected a range of LRs, including heterozygous whole gene deletions, single and multiple exon heterozygous deletions, single and multiple exon duplications, and homozygous multiple exon deletion. Conclusions: We have developed an NGS assay and comprehensive data analysis approach capable of detecting both small mutations and LRs simultaneously from FFPE samples with high sensitivity and specificity. For research use only. Not for use in diagnostic procedures. A. Momeni Boroujeni, E. Yousefi, R. Maglantay, R. Gupta, N. Chen State University of New York Downstate Medical Center, Brooklyn, NY. Introduction: FSHD Region Gene 1 is located on chromosome 4q35. This gene is implicated in facioscapulohumeral muscular dystrophy (FSHD). The protein product of the gene is believed to play a role in spliceosome formation and pre-mRNA processing. FRG1BP is the pseudogene counterpart of FRG1 which is located on chromosome 20q11. We have evaluated mutations in FRG1BP in different cancers. Methods: Mutational and copy number alteration data from 814 cancer patients were extracted from the TCGA database. The frequency of mutation and copy number alterations of FRG1BP was evaluated in these cases. We used subnetwork analysis, network power analysis to determine the main subnetworks within the mutational profiles of different cancers among our sample. We further performed mutual exclusion analysis between genes implicated in spliceosome. Results: Our results show that FRG1BP is mutated or has copy number alterations in a significant percentage of chromophobe renal cell carcinoma, uterine carcinosarcoma, papillary renal cell carcinoma, adrenocortical carcinoma, pheochromocytoma and melanoma. Subcluster analysis shows FRG1BP is mutated in a significant sublcluster of patients which have unique mutational profiles when compared to other patients with the same cancer. Power centrality analysis showed that mutations in this gene have a power centrality (2.97) comparable to PTEN and PIK3CA (2.38 and 2.31 respectively). We also demonstrated that alterations of FRG1BP are mutually exclusive from FRG1, PRPF19 and CDC40 (p<0.001). Conclusion: To date a comprehensive review of the mutations of FRG1BP has not been performed. Our study shows that FRG1BP is mutated in a considerable subcluster of cancer patients and it appears to be an important gene in some cancers. Our mutual exclusiveness test shows that FRG1BP is likely to play a role in spliceosome formation pathways. Further investigation of FRG1BP's function and role in cancer pathogenesis is needed. has the potential to benefit public health by facilitating the early detection of pathogen outbreaks. Efficient tracking of ID requires 1) broadly-distributed, comprehensive diagnostic testing and 2) rapid electronic collection, analysis and distribution of data. Several FDA-cleared diagnostic platforms are in use in clinical laboratories, which have assays for many of the large groups of infectious agents known to cause diarrhea. BioFire's FilmArray (FA) is one such system. The FA GI panel detects 22 pathogens. Whereas the first condition for tracking GI disease has been met, the second condition has not; there is no general, automated electronic The Journal of Molecular Diagnostics ■ jmd.amjpathol.org mechanism for aggregating GI test results from across the United States in realtime. Methods: We have implemented a cloud-based epidemiology network, FA-Trend. The system connects FA Instruments directly to the cloud, automatically exporting electronic de-identified test results to a secure, HIPAA-compliant database. Web-based views of the aggregated data are accessible to various user groups: clinical users can track institutional and local trends, and the public can monitor bacteria, viruses and parasites causing infectious GI disease. This automated approach does not require labor intensive manual processing or data extraction from information systems that vary by hospital. Results: Four US sites are participants in the initial GI FA Trend pilot study, Medical University of South Carolina (SC), Primary Children's Medical Center (UT), UC San Diego Medical Center (CA), and NYU Langone Medical Center (NY), with more than 15 FA Instruments in the cohort. Automated export of the electronic results and archival data began in the fall of 2015 and collectively the group will contribute over 4,000 test results to the project this year. Data presented will include: 1) GI pathogen prevalence, including fluctuations in the prevalence of diarrheagenic E. coli/Shigella among other organisms; 2) Polymicrobial detections, which indicate, at the population level, interactions between pathogens that occur in a patient; 3) Rotavirus prevalence trends compared to CDC NREVSS Rotavirus surveillance. Conclusions: We demonstrate that, when appropriate care is taken to remove protected health information from ID test results, it is possible to address hospital data security concerns and patient privacy issues involving real-time, automated data export. With this infrastructure in place it is straightforward to connect FA Instruments directly to the internet and export deidentified GI pathogen results. The resulting data stream gives us unprecedented visibility of the prevalence and spread of GI infectious disease. The application of DNA sequencing in clinical cancer genomics has proven to be an invaluable tool to characterize acquired tumor mutational load, inform therapeutic routes and to predict treatment response in oncology patients. Despite these powerful benefits, this approach lacks the ability to quantitatively characterize downstream gene expression levels or capture non-genomic transcript variation-two important avenues currently under characterized in next-generation sequencing based clinical diagnostics. RNA sequencing (RNAseq) addresses these issues by quantitatively capturing the entirely of the expressed transcripts in a tissue, while additionally detecting all of the expressed genomic alterations as seen in DNA sequencing. One of the barriers to its implementation in a clinical setting is the availability of computational pipelines and bioinformatics tools to comprehensively characterize the broad range of variation present in transcriptome data. Methods: We have developed a cloud-based RNAseq computational workflow to detect expressed variation including SNVs, INDELs, structural variants, gene fusions, differential gene expression, allele specific expression, and novel alternative allele isoforms in a single pipeline. Results: We apply our method to characterize the transcriptomes of pediatric sarcoma and lung adenocarcinoma to identify novel variation. Further, we demonstrate that our approach compares favorably to DNA based sequencing methods when performing variant calling on somatic patient samples.Conclusions: Our cloud-based transcriptome profiling pipeline serves as a proof-of-principle for RNA-guided comprehensive tumor profiling, and may serve as a parallel bioinformatic avenue to analyze, interpret, and apply genomic information for personalized oncology patient care. J. van Riet, N.M. Krol, P.N. Atmodimedjo, J.W. Martens, L.H. Looijenga, G. Jenster, H.J. Dubbink, W. Dinjens, H.W. van de Werken Erasmus MC Cancer Institute, Rotterdam, Netherlands. Introduction: Exploration and visualization of next-generation sequencing (NGS) data originating from targeted multi-gene panels is crucial for analysis of genetic aberrations in both research and clinical settings. However, software for simple, robust and dynamic web-based visualization of single nucleotide polymorphisms (SNPs) in the regions of these targeted multi-gene panels is lacking. Methods: we developed a lightweight Shiny-based web-application, called SNPitty, for interactive visualization and interrogation of single-and multi-sample Variant Call Format (VCF) files. SNPitty is best applicable with data from NGS targeted multi-gene panels to display loss of heterozygosity (LOH) and copy number variation of genetic heterozygous markers. Moreover, SNPitty is capable of generating predefined reports which summarize and highlight the targets-of-interest based on LaTeXtemplates. Results: Here, we apply SNPitty on patient-derived tumor-only glioblastomas (n=3) to assess LOH and amplification. DNA was sequenced on an Ion-Torrent platform using a diagnostic multi-gene panel targeting known genetic aberrations associated with tumor formation and progression. VCFs were subsequently generated using Torrent Suite. Amplification and LOH in regions for known genetic aberrations were clearly indicated for the glioblastoma samples and reported by SNPitty. Conclusion: SNPitty can be used freely at https://ccbc.erasmusmc.nl/SNPitty. Source code and documentation for SNPitty can be obtained under the terms of the GPL-3 open-source license through BitBucket at https://bitbucket.org/ccbc/snpitty. Introduction: Karyotyping, the practice of visually analyzing and recording chromosomal abnormalities, is commonly used to diagnose and determine treatment strategies for patients with malignant neoplasms. Karyotypes are written in the International System for Human Cytogenetic Nomenclature (ISCN), which has a linguistic structure. Thus, assessing karyotypes to identify recurrent cytogenetic abnormalities is performed in a non-computational manner, by visually inspecting long strings of often complex nomenclature. Because ISCN is currently computationally intractable, much of the genomic data stored in tens of thousands of karyotypes remain inaccessible, and potential applications unrealized. To mine these data, we have developed a cytogenetic platform (CytoGPS) that transforms humanreadable ISCN karyotypes into a machine-readable model (Loss-Gain-Fusion [LGF]). As proof of principle, we analyzed karyotype data for chronic lymphocytic leukemia (CLL) in the Mitelman Database of Chromosome Aberrations and Gene Fusions in Cancer (http://cgap.nci.nih.gov/Chromosomes/Mitelman). We chose CLL because it contains recurrent, clinically relevant cytogenetic aberrations. This provides us with known aberrations, as well as the opportunity to discover novel cytogenetic aberrations. Methods: Parsing and mapping of ISCN karyotypes was performed in CytoGPS using grammatical rules for writing karyotypes in the ISCN. Applying these rules, we extracted two key pieces of information: the location of the band where an event occurred and the biological result of that event. We classified all events as having three possible outcomes: loss, gain, or fusion. Thus, each ISCN text-based karyotype is represented as a binary vector, to which cluster analysis can be applied. Results: We applied CytoGPS to parse 2051 CLL karyotypes from the Mitelman database into the LGF binary vector model. We then performed cluster analysis on the binary vector karyotypes obtained from these data. We recovered the known recurrent cytogenetic abnormalities in CLL: trisomy 12, deletion 13q, deletion 11q, deletion 17p, and deletion 6q. We also discovered many unusual and novel aberrations, including the co-occurrence of trisomies 18 and 19 and the cooccurrence of monosomies 21 and 22. Conclusions: Karyotype data, which contain clinically important genomic information, are difficult to mine because they are not computationally readable. We describe a method to translate text-based karyotypes into a computationally usable binary vector that retains the biological meaning of the karyotype while facilitating modern informatics analyses. This model will enable researchers to use karyotype data in computational studies in a new and powerful manner. Introduction: Polymerase chain reaction (PCR) followed by high-resolution melting (HRM) of DNA samples is a simple, cost-effective, fast, and powerful, method for genotyping applications. With increased knowledge about DNA sequence variation in genetic targets and the ability to perform high throughput DNA testing, the amount of data is becoming overwhelming for manual interpretation. Automated HRM analysis software has thus emerged as an effective solution toward either assisting a user make the genotyping call or replacing user's interpretation. The goal of this investigation was to develop a novel, scalable software solution for automated genotyping and to assess the genotyping accuracy of the software. Methods: We prototyped a novel software solution to perform automated genotyping based on HRM data. The software is based on nucleic acid thermodynamics theory and apriori information, and was designed so that no user input or manual data manipulation is required. It can operate without any algorithm modification for either small-amplicon or probe-based assays. The software requires a-priori information that is specific to the target being tested as well as melting curve data of a control wild-type sample. A companion software application was developed to allow a user to easily generate the necessary a-priori information for any new HRM assay using a training set of melting data. In this investigation, we performed accuracy testing of the prototype software on two HRM assays: a) small-amplicon MTHFR A1298C and b) probe-based CYP2C9*2 assays. For each assay, a set of melting curves representative of all three genotypes, wild-type (WT), heterozygous (HET), and homogeneous (HOM), as determined by an expert in HRM analysis was collected and used for testing the software. The software genotyping accuracy was derived by comparing the software results with the genotyping calls made by the expert. Results: We collected a) 102 melting curves for MTHFR A1298C consisting of 43 WT, 34 HET, and 25 HOM samples, and b) 50 melting curves for CYP2C9*2 consisting of 25 WT, 15 HET, and 10 HOM samples. All sample melting curve data were acquired on Canon's prototype genotyping platform. For both assays, our software resulted in 100% genotyping accuracy. Conclusions: A novel, scalable software solution to perform automated genotyping based on HRM curves has been prototyped. Results revealed excellent genotyping accuracy of our software and further demonstrated that the underlying methods and mechanics of our software are sound and reliable. J.D. Peterson, C.I. Amos, F.B. de Abreu, W.A. Wells, G.J. Tsongalis Dartmouth Hitchcock Medical Center, Geisel School of Medicine and Norris Cotton Cancer Center, Lebanon, NH. Introduction: In recent years the NGS sample preparation and sequencing processes have become increasingly streamlined, whereas converting raw sequencing data into reportable results remains highly labor intensive. For this reason, automating and integrating the data analysis and variant interpretation processes can be extremely effective in reducing total analysis time. CLC Biomedical Genomics Workbench (BMGW) is a comprehensive NGS data analysis platform designed to create automated bioinformatic workflows for a wide spectrum of sequencing applications. Here we describe a custom BMGW pipeline for analyzing data generated using the Ion Torrent Cancer Hotspot Panel v2, and compare the results with our clinically validated in-house bioinformatic pipeline. Methods: FASTQ files from 30 FFPE samples previously analyzed using our inhouse bioinformatic pipeline were re-processed using a custom BMGW analysis pipeline. Our in-house pipeline utilizes Torrent Suite (v4.0) for alignment and variant calling, and Golden Helix SVS (v.8.3.4) for variant annotation. The BMGW analysis pipeline includes read mapping (hg19), local-realignment and variant calling, and variant annotation from a number of public databases. Identified variants were subjected to a multi-tier filtering process of varying stringency designed to isolate high-quality clinically significant variants passing the minimum validated thresholds (>500x coverage, >5.0% allelic frequency). The BMGW analysis pipeline also incorporates a module to automatically upload filtered VCF files to QCI-interpret for variant curation and reporting. Results: For the 30 FFPE samples, we observed high concordance in total variants called between the custom BMGW and in-house bioinformatics pipeline (>98.0%, 50/51 variants). One point mutation was identified using our in-house analysis pipeline that was not called using the BMGW. In addition, one 3bp deletion was detected using the BMGW pipeline that was not called using our in-house pipeline. Overall, the custom BMGW pipeline showed superior accuracy in annotating complex variants (INDELs and compound substitutions). Conclusions: As laboratories are faced with increasing pressure to both maintain testing turn-around-times, and increase the amount of clinically relevant data that is reported to clinicians, creating highly efficient analytical processes is more critical than ever. The BMGW is a powerful analysis platform for creating automated NGS bioinformatic pipelines for a variety of sequencing applications. The ability to consolidate the data analysis and interpretation processes, while maintaining sensitivity and accuracy in variant calling, suggest it is ideally suited for integration into a clinical testing environment. Introduction: Pathology reports in narrative text format require a primarily manual abstraction of key data elements for entry into cancer registries. Due to delays in the timely receipt of reports and the need for manual abstraction, cancer surveillance reports are often released years after scientific and care standards have changed. This motivated the College of American Pathologists (CAP) and the California Department of Public Health (CDPH) to undertake a pilot project to dramatically reduce the receipt and abstraction time for surgical pathology and biomarker (BMK) data, which currently takes up to 24 months per case, while increasing the quality of the data for analysis. The project uses 13 CAP electronic Cancer Checklist (eCC) BMK templates, which follow national standards to provide required data elements (RDEs) required for patient care and accreditation bodies. These templates are used by vendors to standardize the data entry forms used in the pathologist workflow and to standardize the collection and reporting of cancer RDEs. As reporting via cancer BMK templates may soon become standard practice, we describe the experience of the California Cancer Registry (CCR) in using two CAP BMK templates (breast and colon) for structured data reporting. Methods: mTuitive xPert software, using eCC BMK templates, was installed for pathologists at the 10 hospitals within St. Joseph Health (SJH), to create and transmit standardized BMK reports to CCR. mTuitive and Meditech software generated messages containing the eCC BMK RDEs per the NAACCR volume 5 (version 4) guide. Real-time eCC data feeds into the CCR were captured and parsed by the Eureka system (a CCR registry software suite). Data presented here were collected from live transmissions to the CCR from February 2015 to May 2016. Results: SJH is now automatically transmitting eCC BMK data daily into Eureka via HL7 messaging. A total of 570 Breast BMK reports from 248 unique patients (range 1 to 4 reports/patient) were received by the CCR during the study period. Analysis of Breast BMK data revealed hormone receptor and HER2 status. 432 of 518 cases tested positive for ER, 376 of 522 cases tested positive for PgR and 38 of 384 cases were positive for HER2 by IHC. A total of 113 colon cancer BMK reports were received. The time course for collection was also analyzed. Conclusions: Automated use of CAP eCC templates for BMK reporting allowed for operational improvement and user satisfaction at SJH and CCR to report and collect standardized cancer case information. The eCC BMK templates facilitate the recording and rapid automated transmission of standardized BMK RDEs, so that inferences from registry data can be drawn in a timely fashion and based upon the most current CAP RDEs. (Lincoln et al., JMD 2015) showing that technically challenging variants for NGS were a substantial fraction of the pathogenic findings in a representative population of 1062 patients receiving a 29-gene hereditary cancer test. We sought to explore this observation in a larger, more diverse data set and to address the challenges these variants pose to test development and validation. Methods: We examined panel test results for over 30,000 patients across 1000 genes associated with cancer, cardiovascular, neurological and pediatric disorders. We examined the Genome in a Bottle (GIAB) dataset (Zook et al. Nature Biotech 2014) for the targeted bases of these same genes, and we similarly examined NGS coverage in exome sequences. As samples with "hard" variants are difficult to obtain, we developed large plasmid inserts bearing specific variants engineered into GIAB reference sequences. These were spiked into GIAB genomic DNAs at concentrations that mimic heterozygous changes and sequenced using standard NGS protocols. Multiple assays including linked reads and long-read single molecule sequencing are also being used to improve the GIAB reference data in difficult regions. Results: Consistent with our prior data, challenging variants were prevalent: Of roughly 5000 patients with a pathogenic variant, 2.9% had single-exon CNVs,1.8% a large indel or complex sequence change, and 5.8% a variant in a region of high-GC, low complexity or poor mappability. Even at >250x average depth, 3% to 4% of target bases had low (<20x) coverage in the exome sequences, compared to 0.0% to 0.1% for optimized panels at similar depth. About 20% of targeted bases lie outside the published GIAB highconfidence regions, and, unsurprisingly, very few examples of "hard" GIAB variants are present. However, new draft GIAB data increase the high-confidence regions from 78% to 89% of the genome, improving its utility. Moreover the GIAB samples with engineered fragments generated BAM files resembling those of patients carrying the same events, suggesting these may be a useful tool for test development, optimization, and validation. Conclusions: Most currently published validation studies include few, if any, of the most challenging types of variants, which are indeed diagnostically important and are prevalent across a range of disorders. As standard bioinformatics methods are not designed to detect such variants and standard targeting chemistries provide limited coverage of important genes, laboratories need to do additional work to develop and validate sensitive tests. To help with this, new GIAB data and GIAB samples with engineered variants will be broadly available to the AMP community by the 2016 Annual Meeting. Introduction: Minimal residual disease (MRD) monitoring using ultra high sensitive PCR assays for BCR/ABL1 and PML/RARA is critical for the management of patients with chronic myelogenous leukemia, acute promyelocytic leukemia and other hematologic disorders.The laboratory workflows for generation, interpretation, reporting and management of these high-volume assay results are often cumbersome, redundant, and error-prone due to manual transfer of alphanumeric data. Retrospective review of test results is critical for appropriate test selection and clinical interpretation, which adds to the complexity. Using principles of laboratory automation and web technology, we aimed to optimize the assay workflow. Methods: The ABI 7500 real-time PCR module was used for generating the raw quantitative PCR data. A web-based application was developed using ColdFusion and SQL Server 2012 database to automate several components of the existing workflow. Briefly, tab-delimited files, containing the raw data, were exported from the ABI 7500 real-time PCR module and uploaded to the web application. The raw data was processed and presented for review and interpretation via a web user interface (UI). Algorithms for clinical interpretation of test results were incorporated. The application was hosted in our institution's datacenter using appropriate security protocols. Results: This web-application significantly reduced manual transfer of patient and sample information and test results (at least 4 manual steps). Interoperability with our Laboratory Information System (LIS) enabled automated integration of molecular test results with patient and sample information. The interactive UI enabled single view of all historic test results for a given patient in a tabular and graphical format eliminating the need for redundant and time-consuming manual lookup in LIS. Clinical reports in HTML format was generated after review by pathologist and transferred to the LIS using clipboard function for electronic signout. Calculation of clinically reportable information (logarithmic change in transcript levels and International Scale) was also automated. The application was clinically validated using a total of 56 cases and at least 10 data points per case. The validation cohort constituted an appropriate mix of cases with negative, positive and indeterminate The Journal of Molecular Diagnostics ■ jmd.amjpathol.org test results. Conclusions: This web-based application is a user friendly, secure, reliable and efficient tool for management of molecular test results and clinical information. Although significant optimization was achieved, few manual steps could not be eliminated, such as review of printed copy of real-time PCR amplification curves and manual transfer of clinical reports. Further development is underway to address these issues. A. Momeni Boroujeni, R. Maglantay, R. Gupta State University of New York Downstate Medical Center, Brooklyn, NY. Introduction: Endometrial carcinomas have two main subtypes: serous and endometrioid. It has been known that serous adenocarcinomas have a worse prognosis than endometrioid carcinomas. They are also believed to be genetically different from endometrioid carcinomas. In this study we aim to cluster the endometrial carcinomas based on their mutational profile and compare the survival statistics of the mutational versus morphologic clustering. Methods: Five hundred and forty-eight cases of endometrial cancers (serous and endometrioid) from The Cancer Genome Atlas were studied. Their survival and mutational data for the most common mutations in endometrial carcinomas was extracted. A network analysis method was used to cluster the cases based on their mutational profiles. The survival of mutational clusters and morphologic clusters were compared using survival statistical tools including Kaplan-Meyer, Log rank and Cox regression tools in R programming language. Results: After adjusting for clinical stage, morphologic clustering of endometrial carcinomas was found to have a better correlation with survival compared to mutational clustering (p value: 0.007) with endometrioid carcinoma having the best prognosis (average survival: 110.55 months) followed by mixed endometrioid and serous carcinomas (average survival: 36.1 months) and serous adenocarcinoma (average survival: 30.09 months). Of note, a subcluster of endometrioid endometrial carcinoma cases with a unique mutational profile (with mutations in PTEN/FGFR2/NRAS/CCND1/CYLC1 and no mutation in PIK3CA and none of type I endometrial carcinoma mutations) have a significantly better prognosis (p value: 0.022) compared to the other endometrioid carcinomas irrespective of clinical stage. Copy number alterations is another predictor of survival (p value: 0.032) whereas overall number of mutations has no meaningful effect on survival (p value: 0.102). Conclusion: Our results show that although mutational clustering of endometrial carcinomas is possible, these clusters do not conform to histologic grouping of endometrial carcinomas because of shared mutations between endometrioid and serous adenocarcinomas. Morphologic grouping is a better predictor of outcome compared to genetic clustering. Genetic profiling, however, identified a unique subcluster of endometrioid carcinomas which confer an excellent prognosis, irrespective of stage. Introduction: Proficiency testing (PT) is essential to assess the accuracy and reproducibility of clinical reporting. In 2015, a consortium of 6 academic clinical laboratories was formed to test their respective bioinformatics workflows. Each lab, which performs clinical laboratory developed tests for solid tumors using the same commercial library preparation kit, exchanged 24 FASTQ datasets. As described in a publication earlier this year, 100% concordance in reporting clinically significant single nucleotide variants (SNV) was achieved at all sites who performed the analysis. However, the discordance rate for calling insertion/deletion (indel) mutations was as high as 60% (3/5) for a clinically significant mutation in the EGFR gene and was not uniform among all other submitted challenges. To address this, we chose to re-challenge bioinformatics PT focusing specifically on the detection of indel mutations. Methods: A preliminary series of 10 (8 new, 2 previous) FASTQ datasets produced with the Illumina TruSight Tumor 26 kit and MiSeq sequencers was distributed to 4 sites. The two previous datasets had clinically significant indel mutations that were not detected by 2/4 sites in our published study. Data analysis was performed independently by 4 sites and clinically significant mutations were reported. The results were reported back to each site, and an assessment of the performance of the bioinformatics workflows were discussed. An additional 10 FASTQ datasets with complex indel and SNV mutations are in the process of being analyzed by 5 sites. Results: Analysis of our preliminary series of 10 FASTQ datasets demonstrated a dramatic improvement in the identification of clinically significant mutations. We achieved 100% concordance in 9/10 datasets. Only one mutation, a 12 bp insertion in ERBB2, was missed by one site that had previously missed the same mutation. Although the detection rate of indel mutations dramatically improved, nomenclature discrepancies were noted. An additional 10 FASTQ datasets are in the process of being analyzed and the results of all 20 FASTQ datasets will be presented. Conclusions: We observed that despite disparate clinical bioinformatics workflows, consistency in detecting indel mutations is achievable. One site that previously performed poorly in detecting indels upgraded their entire workflow, dramatically improving results. A second site that correctly reported 9/10 datasets, but missed a clinically significant mutation in ERBB2, was able to identify the cause of their software problem. PT via FASTQ exchange provides an essential function in assisting laboratories to achieve robust and reproducible clinical NGS assays. T.M. Pearce, M.N. Nikiforova, S. Roy University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: Determining the clinical significance of any given sequence variant requires an integrated evaluation of information from multiple genomic data sources, such as variant location mapping, predicted cDNA and amino acid sequence change, minor allele (population) frequency, insilico prediction scores, somatic and germline mutation databases, and comparison with previously identified variants (inhouse knowledgebase). Software tools to visualize the mapping of sequence variant to peptide structure have been developed and used predominantly in research projects. We investigated the scope for applicability of data visualization techniques to assist in the interpretation of sequence variants from NGS-based somatic mutation testing. Methods: We developed a modular, browser-based visualization widget using JavaScript using widely-used, freely-available libraries such as jQuery. The widget can consume publicly-available or custom web services via AJAX (Asynchronous JavaScript and XML) to obtain data for visualization for a variety of annotations. The interactive graphical user interface utilizes HTML, SVG and CSS, and is encapsulated in an HTML element for easy integration with any web application. An application programming interface (API) provides developers with a variety of configuration options for programmatic control over the widget. Intuitive user interface and easy integration into existing clinical workflows were important design considerations. Results: The data visualization widget was successfully incorporated into a clinical variant management and reporting application used at our institution with minimal change in codebase. It rendered an integrated visualization of the variant location, functional domain mapping, and details and prevalence of a variant of interest when compared to in-house knowledgebase or external databases. For example, during review of uncommon or rare somatic variants, the widget provided visual overlay of the existing variant with variants from the COSMIC database while preserving the positional mapping to the functional protein domains. During routine clinical case workup, it minimized the requirement for separate review of multiple data sources such as pFam, uniprot, COSMIC and other data sources. Conclusions: Data visualization is powerful approach for presenting and interpreting complex genomic data, in clinical as well as research settings. We developed a modular, browser-based widget, which was easily configurable and incorporated into our existing clinical application. Integrated data visualization provided a novel and valuable user experience for variant interpretation, especially with high test volumes. Introduction: Somatic copy number alterations (SCNAs) have significant diagnostic and prognostic value in many types of cancer, including glioblastoma (GBM). SCNAs are often identified by fluorescence in situ hybridization (FISH), but massively parallel sequencing (MPS) can also be used to infer copy number based on read depth. It is, however, challenging to predict SCNAs from targeted re-sequencing of limited gene panels due to uneven coverage of the genome and variation in the efficiency of molecular baits, particularly when no matched normal control is available. Methods: Fifty-four GBM specimens (tumor cellularity ranging from 60 to 97%, average 84%) were subjected to MPS following enrichment of a 516kb target by hybrid capture at Genomics and Pathology Services at Washington University. Target space encompassed the coding regions of 131 genes, selected introns, and a tiled 35kb genomic backbone spaced at 10Mb across all chromosomes. Alignment files generated within the clinically validated pipeline were fed to CNVkit for copy number prediction using both on-and off-target reads. Fifteen cases with no discernable SCNAs were included as an averaged control for normalization. Results were compared with FISH profiles to determine accuracy. Results: Using standard parameters, CNVkit performed robustly on the assay, providing fine resolution of ontarget regions and as well as a broad view of the entire genome. Polysomy of chromosome 7 was detected from MPS data in 45 of 54 GBMs, including 19 of 20 cases with focal EGFR amplification (7p11.2). All cases with EGFR amplification also showed monosomy of chromosome 10 and focal loss of CDKN2A (9p21.3), whereas these aberrations were found in 68% and 56% of cases with unamplified EGFR, respectively. TERTpromoter mutations were common in both EGFR-amplified and unamplified GBM (69% of all cases), whereas TP53 mutations were more common in cases with unamplified EGFR (65% versus 5%). MPS-based detection of focal EGFR amplifications was concordant with FISH in the 39 cases tested with jmd.amjpathol.org ■ The Journal of Molecular Diagnostics both methods. Conclusions: Overall, massive amplifications (e.g., EGFR) were detected with greater accuracy from MPS data compared to lower copy number polysomies and single copy loss. Factors influencing the reliability of SCNA predictions include the quality/quality of input DNA, tumor cellularity, and tumor heterogeneity. I24. An Integrative Multi-Software Approach for FLT3 Internal Tandem Duplication Detection in Hybrid Capture Next-Generation Sequencing Data S. Kadri, I. Mujacic, B.C. Long, C. Zhen, M.N. Wurst, B. Ameti, N. Niu, S. Benhamed, L.V. Furtado, J.P. Segal University of Chicago, Chicago, IL. Introduction: Detection of large insertion mutations with high sensitivity remains a substantial challenge in clinical next-generation sequencing (NGS) diagnostics. At large sizes, it may not be possible for NGS reads to contain the entire insert, leading to frequent failed mapping. Of all the common clinically relevant insertions, FLT3 internal tandem duplications (ITDs) remain potentially the most difficult due to their large size, clinical importance, and sequence characteristics. We have previously presented Amplicon Indel Hunter (AIH), an alignment-independent tool for highly sensitive indel (>5bp) detection in amplicon assays. However, hybrid capture data requires different methodologies. Here, we present an integrated approach combining results from multiple complementary informatics systems, including novel software (ITDHunter) designed to improve sensitivity of ITD detection. Methods: DNA from blood/bone marrow specimens previously tested either by FLT3 fragment length analysis or amplicon-based NGS with AIH were selected. Briefly, DNA was fragmented (Covaris), subjected to library preparation (Kapa Biosystems), pooled and captured (Roche SeqCap EZ) as part of the protocol for UCM-OncoPlus, a 1212 gene hybrid capture sequencing panel which includes 119 clinically reported genes. Captured libraries were sequenced on the HiSeq 2500 system (Illumina). Data processing for FLT3 ITDs included three software systems: (i) indel-realigned software (Abra), (ii) published pattern growth software (Pindel) (iii) ITDHunter, a custom software to discover mis-aligned reads in the FLT3 ITD hotspot region. Results: FLT3 ITDs were detected in all 15 specimens previously observed to contain an ITD via amplicon NGS/AIH or via fragment length analysis. No ITDs were detected in 85 previously tested samples documented to be negative for the presence of an ITD. High concordance between UCM-OncoPlus and previous assay mutant allelic fractions (MAF) were observed. Both perfect and imperfect duplications were identified up to a maximum tested size of 105 bp in clinical samples. In silico simulated ITDs of sizes up to 500 bp were generated at a variety of MAFs as low as 5%, and were all detected via the combined system. At extremely large size, ITDHunter outperformed both Abra and Pindel. Conclusions: FLT3 ITDs are a critical target that requires highly sensitive detection methods. NGS systems may only replace traditional fragment length assays when equivalent or superior performance is attained. Existing published software have weaknesses for detection when ITDs reach very large sizes (>100 bp). ITDHunter fills this performance gap by flagging the existence of large ITDs beyond this threshold, allowing for highly sensitive detection at even extreme ITD size. Introduction: Detection of large indels by NGS is often challenging. Different methods have been adopted to improve indel detection. Here, we demonstrate bioinformatics pipeline enhancements for a TSM assay that lead to reliable detection of a 52 bp CALR deletion. Methods: 1) Update to the alignment tool: The Novoalign aligner (Novocraft Technologies) was upgraded to version 3.04.04. Novoalign version 3.02.07 had required that a deletion of 52 bases be at least ~52 bases from either end of a read to avoid soft-clipping. The CALR deletion is ~ 40 bp from the end of the read based on the Illumina assay design. In the updated Novoalign version, the distance requirement from the end of the read is cut in half, resulting in successful alignment of deletion-containing reads. 2) Update to the variant caller: Freebayes version 0.9.21-19-gc003c1e was used for calling insertions and deletions. Standard Freebayes parameters were modified to lower the minimum base quality requirement, resulting in a more accurate calculation of the variant allele frequency (VAF) of the 52 bp deletion. 3) Updates to the validated filter criteria: The initial TSM Panel used a custom filter that required evidence of an insertion or deletion in both forward and reverse reads if there were overlapping reads in the region of interest. 4) Validation of the improved pipeline: The data from 12 deidentified patients was run through the improved pipeline to evaluate the detection of the CALR 52 bp deletion. Results: In the previously validated TruSight Myeloid pipeline, the CALR 52 bp deletion was not detected. After pipeline updates, the CALR 52 deletion was reliably detected in multiple samples known to exhibit the deletion. Lowering the minimum base quality requirement for indel calling as well as applying less stringent filter criteria may reduce overall assay specificity. To minimize this risk, we evaluated the effect of each of these changes individually and found that most added indel calls lie in non-coding regions and therefore have minimal impact on our clinical reporting, which focuses on coding regions and splicing sites. Conclusions: Large indels may render a drastic change in targeted sequence resulting in soft-clipping by the alignment software or filtering based on quality criteria. Optimization of the match reward feature in Novoalign v.3.04.04 allows for improved alignment with the reference sequence. Optimization of the indel caller Freebayes yields a more accurate determination of the variant allele frequency (VAF). Removing the requirement that the deletion be present in both forward and reverse reads allows the 52 bp CALR deletion to pass preset validated filters for clinical reporting. The combined dataset and all individual institutions had a lower 95% confidence of greater than 99% for both PPA and PPV for all classes of SNV variants examined. Short deletions (< 10 bp) present at a VAF > 20% had a lower 95% confidence interval of 95% for PPA and PPV while short deletions present at a VAF < 20% had a lower 95% confidence interval of 90% for PPA and PPV. There were 11 identified large deletions (> 10 bp) in the combined dataset so analytical performance was not assessed. Short insertions (< 10 bp) present at a VAF > 20% had a point estimate for PPA and PPV of greater than 95%, but the lower 95% confidence interval was determined to be 80% based on the number of samples. There were not enough variants identified for short insertions with a VAF < 20% or larger insertions (> 10 bp) to determine performance. Intermediate precision was determined to be greater than 95% for all variant classes examined containing sample replicates. Conclusions: Analytical performance of the CHP was found to be high for all variant classes examined and met or exceeded requirements as described in the MolDx analytical performance specifications. Performance was found to be concordant across multiple institutions running this panel. This indicates that the CHP can meet stringent analytical performance criteria in a clinical setting such as that set out by the MolDx program for SNVs and short INDELs. Introduction: The identification of disease-associated genes is an important step towards understanding disease mechanisms, diagnosis, and therapy for the future. However, due to the complex and distributed nature of the problem, current scientific knowledge is spread out over several overlapping databases maintained by independent groups. It is unclear how to rank gene-disease research associations due to the distributed and dispersed nature of our knowledge. To fill this gap, we developed Information Genetic Content (IGC), a comprehensive knowledgebase and discovery tool for human genes and genetic disorders research use. IGC is unique in two aspects. First, it integrates data from multiple databases into one system. Second, it provides an unbiased scoring algorithm to rank gene-disease research association at any level of the disease ontology hierarchy. Methods: IGC comprises three components: the Disease-Association Database (DAD), the Gene Scoring Algorithm (GSA), and the Virtual Panel Library (VPL). The DAD module contains over 400,000 associations between over 17,000 genes and 15,000 Mendelian and complex diseases from both expert-curated and text-mined data. The DAD module also features a hierarchical organization of human diseases using a UMLS-controlled vocabulary, permitting queries at any level of the disease ontology hierarchy. The GSA module aims to prioritize genes for a specific disease of interest. This gene scoring algorithm is distinctive in the way it combines the strength of association and the number of associated diseases to provide an unbiased score for each gene. In conjunction with the DAD module, the GSA module is able to produce a list of ranked genes for one or more diseases at any level of the disease hierarchy. The VPL module generates optimal gene grouping by disease classification using hierarchicalclustering-based network analysis. Genes that are involved in the same pathological pathways are grouped into the same cluster. Results: We have successfully applied IGC to select research genes for a variety of diseases, including auto-immune disorders, cleft palate, and autism. Our automated approach achieved more than 70% overlap with the genes prioritized by years of expert selections, providing a crucial verification of our database and algorithms. Moreover, our unbiased selection engine identified disease-related research genes that were missed by subjective The Journal of Molecular Diagnostics ■ jmd.amjpathol.org expert reviews. Conclusions: Taken together, IGC provides both an integrated knowledgebase and a discovery platform that makes it easy for basic and translational researchers to find the most informative genes for their diseases of interest. V.S. Williamson, V.S. Gadepalli, A. Kusmerik 1 , A. Popa, R. Ren, M. SabatoCharreun, C.N. Vlangos, C.I. Dumur Virginia Commonwealth University Health System, Richmond, VA. Introduction: A crucial step in analyzing targeted amplicon sequencing (TAS) data for oncology is the accurate determination of clinically relevant somatic variants. The process is challenging, confounded by tumor heterogeneity and stromal admixture; and many laboratories risk reporting false positives by not sequencing matched tumor-normal pairs. To address this issue, we have developed Onco-Somatik, a program which applies a linear mixed-effects model and expectation maximization (EM)-based clustering of TAS datasets to determine variant origin. Here, we describe the algorithm underlying Onco-Somatik, and present results from a study testing its performance in patient samples and a comparison to variant classifications from the Cancer Genome Atlas (TCGA) dataset. Methods: The algorithm underlying Onco-Somatik is a two-step process. First, purity and ploidy levels are modeled alongside each variant's observed copy number (CN) and allele frequency (AF). CN is calculated using overlapping windows measuring 100 bp which centers on each variant. AF is calculated using the equation where FAO and FRO refer to flow space alternate allele and reference allele counts. Model strength and fit is evaluated using Akaike information criterion scores and a "leave-one-variant-out" cross-validation approach. Then, after model selection is complete, all variant AFs are recalculated and groups are determined using EM-based soft clustering with randomized selection of prior probabilities. To test Onco-Somatik, we examined predictions from twenty patients with melanoma, colon, and lung cancer. All samples were sequenced using the OncogenomicDxOnePLUS assay on the Ion Torrent PGM. To test accuracy, we compared Onco-Somatik predictions to those found in the TCGA dataset for matched normal solid-tumor samples of similar diseases. Results: A total of five hundred eighteen variants in 50 genes were evaluated using the software; predictions directly related to known hotspot variants comprised 23.8% of patient variants. The call rate for all variants tested was 92.3% with 94.4% of the variants in patient samples matching TCGA classifications. Allele frequencies recalculated after model selection correlate strongly (Pearson: r = 0.978) with observed values, suggesting that computed purity and ploidy levels are appropriate. A key parameter in the estimation of group membership appears to the variability of CN as its variance is directly related to model performance. Conclusions: Onco-Somatik offers a potential alternative for determining the status of novel variants without a tumor-normal match sample. EM clustering of variant allele frequency and copy number enables users to estimate statistically the probability of group membership. Introduction: Whole-exome sequencing (WES) has been increasingly adopted for the diagnosis of rare genetic disorders. A bioinformatics pipeline that translates raw sequence reads into biologically meaningful information is a key component of the workflow. Bioinformatics tools are constantly being upgraded by the community and commercial entities. Additionally, clinical labs continue to accumulate data that can be used to refine specific quality metrics and related filters in the clinical practice. Therefore, it is essential for clinical labs to review and update their bioinformatics pipelines routinely. Here we present our experience with a major upgrade of our WES pipeline, validation strategy, and defining variant quality criteria that can eliminate unnecessary Sanger confirmation. Methods: Over 200 clinical cases have been signed out since the launch of the Medical Exome Test at CHOP in 2014. To capitalize on the data and to take advantage of latest tools, we have developed CWES 2.0. The new pipeline is set out to improve our indels calling by using GATK HaplotypeCaller and reduce the false de novo variants by joint calling on trio samples. Validation of the pipeline was performed using HapMap sample NA12878 and the Genome in a Bottle dataset. The sample was sequenced in multiple runs to assess inter-and intra-run reproducibility. Additionally, using our Sanger sequenced variants, we have derived quality metrics for highly confident SNVs that do not require Sanger confirmation based on a combination of variant quality metrics. Results: CWES 2.0 achieved sensitivity of 99.7% for SNVs and 98% for indels <20bp. Additionally, a set of quality features including Quality by Depth (QD), strand bias metric of FS (phred-scaled p-value using Fisher's exact test) and allele ratio (AF) were identified to be the most effective features to define variants requiring Sanger confirmation. In contrast, variant quality (QUAL), Read depth (DP) did not separate true variants from false positives. Our experiment also showed GATK VQSR score is not suitable as the score range varies depending on cohort used for variant calling. Our data suggest that a SNV needs to be Sanger confirmed if it meets ANY of these criteria: QD < 13.3; FS > 2.5; AF: Hom <67%, Het < 33%. These criteria captured all confirmed variants and will reduce unnecessary Sanger sequencing by more than 50%. Conclusion: In summary, we presented the evolution of our exome sequence analysis pipeline. In addition, we have also presented the methodology to derive criteria for variants to be excluded from Sanger confirmation. Our strategy and the proposed framework may be beneficial to clinical labs to improve efficiency of their variant detection and analysis workflows. M. Syed, A. Zehir, M. Arcila, J. Casanova, T. Baldi, M. Haque, A. Razumova, I. Kiecka, R. Benayed Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: The Molecular Diagnostics Service in the Department of Pathology at MSKCC processes thousands of samples per year for multiple next-generation sequencing (NGS) and non-NGS assays. Seamless interaction with upstream and downstream systems, interfacing with instruments and storage systems, and capturing important information from sample receiving to sample analysis are critical requirements of high-throughput laboratories. We have achieved significant performance improvement by implementing different LIMS workflows and automating interaction with upstream and downstream systems. Methods: We use Exemplar LIMS to handle sample tracking and processing needs of the laboratory. LIMS receives patient clinical data from another database in a process that is monitored to ensure that all data are received and delays are avoided. Several workflows have been implemented for a seamless integration: 1) Sample annotation: users can add annotations such as tumor type and purity for the samples; 2) DNA extraction: allows for sample batching and receives QC data from instruments; 3) various assay specific workflows: these workflows allow batching samples and automatically calculates necessary amounts for DNA and other reagents for the experiment. For NGS assays, sample sheets are also created automatically; 4) Sample storage: interfaces with an automated sample storage system to track samples stored and aids in sample retrieval. Custom plug-ins were written using the JAVA APIs for data transformation and other helpful utilities. Every data point captured in the LIMS is available to downstream systems through web services. Built a dashboard outside Exemplar LIMS, which provides complex data reports, turnaround times, and several user-friendly tools to explore data captured in LIMS. Results: LIMS workflows are used to manage sample registration and annotation, automated DNA extraction and to track several custom NGS assays including MSK-IMPACT as well as non-NGS assays. Collective display of experimental results provides important quality control (QC) information. Providing this information in timely manner using automated APIs, saves time if a sample needs to be reprocessed, often with important consequences for the patient's diagnosis and treatment. For NGS assays, time for gathering information and setting up a batch decreased from 6 hours to less than an hour. Automatic generation of sample sheets eliminated user errors for the sequencing experiments. Conclusions: Our implementation of LIMS has improved performance significantly. Reducing manual typing and adding data checks has decreased errors throughout processes. Seamless communication with upstream and downstream systems has resulted in minimum downtime and faster access to the data. A. O'Hara, A. O'Hara, R. Keshavan, W. Sherman, V. Wasnikar, Z. Che BioDiscovery, El Segundo, CA. Introduction: Targeted panel sequencing has been a popular method to achieve high depth of coverage for certain regions of interest at an affordable cost compared to whole genome sequencing. Shallow whole genome sequencing, where average read-depth can be as low as 0.1x, provides a cost savings-approach for identification of large copy number variant (CNV) events; it has been utilized in various application areas, including oncology. Methods: Here, we introduce a new approach (dCNVSeq) that extends the Pooled Reference algorithm to function with shallow, as well as targeted, sequencing data by introducing a novel dynamic binning approach. The approach uses a Hidden Markov Model to segment the genome into areas forming the "backbone" using the off-target reads and additional areas where targeted reads are present. dCNVSeq uses coarse binning in the backbone area providing base line copy number as well as detection of large CNV events and fine binning in targeted areas to provide high resolution CNV detection in targeted regions. Results: Depending on genomic coverage (whole genome vs. whole exome vs. targeted panel) and sample type (constitutional vs. cancer), different algorithmic approaches may prove to be more or less ideal for estimating copy number from next-generation sequencing (NGS) results. Some approaches require matched normal samples (e.g., ngCGH) while others use the characteristics of the sample itself for normalization (e.g., Wisecondor and QDNASeq). Other methods use a pool of reference samples (e.g., Pooled Reference, xHMM, and CoNFIER). We present a comparison of copy number estimation results from ngCGH, QDNASeq and dCNVSeq using WES, targeted panels, and shallow sequencing data. Whereas each method has strengths for individual samples, only dCNVSeq is functional and reliable across the full data set. Conclusions: dCNVSeq uses a novel dynamic binning approach with a Hidden Markov Model to segment the genome into areas forming the "backbone" using the off-target reads and additional areas where targeted reads are present, providing base line copy number, detection of large CNV events, and high resolution CNV detection in targeted regions within an individual sample. V. Williamson, C.I. Dumur, R. Ren, M. SabatoCharreun, A. Kusmirek, V.S. Gadepalli, A. Ferreira-Gonzalez Virginia Commonwealth University Health System, Richmond, VA. Introduction: The use of targeted amplicon sequencing as a diagnostic tool is increasing and thus creating a need for standardized annotation of variants, scalable, repeatable interpretation and reporting. Qiagen Clinical Insight (QCI) provides a solution for these challenges but must be validated according to guidelines and laboratory's reporting policies. QCI is a web-based clinical decision support platform that provides comprehensive literature references, reported clinical case information, targeted therapy and trial information in support of a lab's annotation, interpretation and reporting of variants. We conducted a QCI platform validation study to determine the concordance rate of variant classifications with cases previously reported by VCU and to assess the relevance and actionability of targeted therapy. Methods: We present the results of a variant classification concordance study comparing the variants from 22 samples with advanced Non-Small Cell Carcinoma, Melanoma, Colorectal Cancer and Thyroid Malignancies analyzed with the QCI platform. Sequencing of samples and primary and secondary analysis of samples were made on the Ion Torrent PGM with the Oncogenomic DX One Assay and patient information and variant calls were manually reviewed for quality prior to upload to the QCI platform where samples were interpreted. The parameters for this study include: 1) the relevance and/or accuracy of pre-computed variant classification before and after QCI rules optimization by an expert user, 2) the consistency with which de-identified patient demographics are uploaded to the QCI annotation pipeline to the final reporting stage. To gauge the impact of the QCI platform training on variant calling, we examined the number of variants for which the results from other subjects were noted. Results: A total of 358 variants encompassing all ACMG categories were analyzed by QCI. A weighted Cohen's Kappa static comparing QCI to our existing knowledgebase dashboard for both preeducation (Kappa = 0.81, Lower CI: 0.75 and Upper CI: 0.87) and post-education (Kappa = 1, lower CI: 1, and upper CI: 1) stages (p <0.05) suggests meaningful improvement of the QCI software after system training. Further, a significant number of variants (Student's t-test p = 0.002, two-tailed) were influenced by the calls of other subjects, suggesting that the system's ability to retain training contributes to increased accuracy. Conclusions: Our results suggest that the QCI platform is a robust tool for variant annotation, classification and report generation that allows more scalable and repeatable guidelines interpretation, creation of reports containing insightful and actionable information. In clinical practice and research it is a common challenge to identify disease type or disease subtype based on various types of clinical and biological data. In complex diseases such as cancer where information patterns are complicated due to inherent heterogeneity, involvement of highly complex interaction of biological pathways, continuous multi-level changes with disease progression, this is a challenging task. The challenge with molecular data appears even harder due to very high dimensionality compared to the number of samples, however, unsupervised learning tools may help to quantitatively and more accurately describe a patient's tumor. Methods: We consider quantitative molecular data (such as gene expression, or gene methylation levels) and proceed with feature (e.g., gene or transcript) pre-selection procedure to characterize a given dataset. Next, we apply a variance-based method to identify characteristic distinct subgroups in a given set of samples using an algorithm for optimal partition selection. We applied our method on Level III RNASeq gene expression data from 20531 genes in 889 tumor samples of renal cell carcinoma (RCC) across three subtypes -clear cell renal cell carcinoma, papillary renal cell carcinoma, chromophobe carcinoma, and 129 normal samples from The Cancer Genome Atlas. Results: Our results show close correspondence to known 9 molecular subtypes in RCC. Moreover, we are able to discern novel subsubtypes of interesting patterns, which bear corresponding representative sets of genes. In addition to genes known to play a role in RCC our analysis uncovers important genes not previously associated with renal cell carcinoma. Finally, using pathway enrichment analysis we identified distinctive activation of various pathways across subtypes. Conclusions: In our study, we clearly demonstrated the effectiveness of our computational framework and obtained novel biological characteristics of RCC. Our framework is generic and can be applied in combination with other types of molecular and clinical data. The growth in next-generation sequencing (NGS) and bioinformatics are synchronous. Amplicon based sequencing is one approach to target a small number of genetic regions of interest and can be utilized for a limited number of clinical samples. Primer cross-talk, the finding of overlapping primers generating short reads, can go unnoticed during the initial design, and are found as samples get processed through the bioinformatics pipeline. Thus, the informatics for an ampliconbased bioinformatics pipeline must recognize the insertion artifacts generated by primer cross-talk. In addition, it is necessary to align all the primer concordant reads without removing the primer sequences to detect the end-of-read indels, which may extend into primer sites, which would go undetected otherwise. Methods: An inhouse bioinformatics pipeline was used to map the demultiplexed paired-end primertrimmed reads (fastqs) to human reference genome (hg19), filter the off-target and poor quality reads, detect variants and annotate them. The entire pipeline was coded using Unix-Shell and Python scripts which used the open source tools such as Novoalign, SamTools, Picard, BTrim, GATK, Annovar, SnpEff and Alamut. The pipeline also included in house built programs such as "Absolute Var" and "GarbagePicker" to detect SNVs and Indels at as low as 5% allele frequency. Results: False insertions called in our variant call files were investigated thoroughly by reviewing the aligned files (bam) in IGV tool. These insertions were found to be the part of primer sequences that belonged to a different amplicon but bound to the target site of a given amplicon and yeilded shorter amplicon products. These reads also contributed towards a bias in read coverage across positions in a given target region. To address these scenarios Unix shell script was written to identify the reads containing non-concordant primers (i.e. primer cross talk pairs from different amplicons) from the primer-trim-summary files and filter them out from the primertrimmed fastq files. This eliminated the calling of false insertions present in target sites in the bams (i.e. output of alignment of primer-trimmed fastq files with hg19 reference genome). Additionally, a logic was included in the same shell script which filtered all the primer discordant reads from the un-primer-trimmed fastqs and aligned those fastqs with hg19 reference genome. This allowed the detection of indels at the end of target sequence. Conclusions: Removal of primer cross talk insertion artifacts and detection of end-of-the-read indels through the bioinformatics pipeline implemented for Amplicon based Next-generation sequencing assays. The ERG fusion is a key genomic event in prostate cancer development occurring in 40% to 70% of PSA screened prostate cancers. FISH and IHC are the two gold standard ERG-fusion detection methods that are diagnostically available at several centers as stand-alone methods. With the increase use of highresolution microarrays for research or clinical use, predicting ERG fusion using microarray platforms is of valuable clinical importance that hasn't been implemented as part of microarray-based genomic tests. Methods: We previously developed and validated a highly accurate Random-Forest model (m-ERG) to predict ERG FISH status in a multi-institutional radical prostatectomy cohorts using the Decipher assay that is using Human Exon 1.0 ST Arrays. Here, we further validate it in terms on accuracy on genome-wide data generated from Decipher assay on 354 FFPE tissue from RP samples with IHC ERG annotations from John Hopkins Medical Institutions' prostate cancer biorepository. Further, precision and reproducibility of the model was tested in technical replicates. Finally, the model was then evaluated on 3,699 prospective RP tissues from the Decipher GRID program. Results: The m-ERG model was locked and independently validated in 354 RP samples achieving an AUC of 95.5% (93-98%), 95.8% sensitivity and 96.7% specificity. To analytically validate the robustness of the model, the model scores were generated for 5 different samples, each with 5 replicates. The m-ERG model had 100% robustness in terms of agreement on the ERG status within each sample. The model scores were highly reproducible with Intraclass correlation coefficient of 0.97 [0.9 to 0.99]. The m-ERG model was also applied on 65 samples from LNCaP cell lines (ERG-) run on the same platform showing 100% sensitivity. These results support implementing it in the Decipher assay; a CLIA test using the Human Exon 1.0 ST Array. Based on the m-ERG model, 41.5% of the prospective cohort samples are predicted to be ERG+. Conclusions: m-ERG model, which is based on the expression of probesets spanning the ERG locus is predictive of the IHC and FISH status. These results support developing and implementing ERG-fusion prediction model in a clinically adopted genomic test pipeline. Despite the lack of impact of ERG fusion on post-RP prognosis, we anticipate that incorporating m-ERG model into a clinically available prognostic assay has several areas of potential near term clinical utility beyond routine pathology such as evaluating multi clonality, clinical trial The Journal of Molecular Diagnostics ■ jmd.amjpathol.org design, evaluating ERG in AS setting. D. Brinza 1 , R. Chen 1 , J. Schageman 2 , E. Ballesteros-Villagrana 2 , R. Chaudhary 1 , J. Gu 2 , V. Bagai 2 , P. Kshatriya 2 , Y. Li 2 , D. Dhingra 1 , J. Au-Young 1 , F. Hyland 1 , K. Bramlett 2 1 Thermo Fisher Scientific, South San Francisco, CA; 2 Thermo Fisher Scientific, Austin, TX. Introduction: Detection and characterization of tumor DNA from circulating cell-free DNA (cfDNA) for the purpose of tracking tumor recurrence and resistance of tumors may improve outcomes in the future. Research studies suggest that somatic DNA mutations in tumor clones can serve as biomarkers that can be tracked in plasma from blood. Tumor DNA comes from different tumor clones, and its abundance in plasma can be very low requiring ability to detect mutation biomarkers at very low frequency from cfDNA. For an efficient low input DNA solution it is necessary to have highly accurate and low-loss interrogation of cfDNA fragments. Here we describe computational and analytical methods used to develop such assay, including a variant calling method that utilizes molecular tagging and enables accurate detection of variants at frequency above 0.05%. Methods: We present computational methods used in development of a multiplex next-generation sequencing assay to optimize cfDNA fragments capturing efficiency, on-target rate, amplification uniformity, primerdimer interactions, molecular tag structure, and number of systematic errors. Followed by an analysis algorithm that models errors accumulated during amplification and sequencing, and accurately reconstructs sequence of tagged DNA molecules based on multiple next-generation sequencing reads. We then developed a variant calling method that uses accurately reconstructed sequences, models offtarget amplification, and PCR polymerase errors to enable sensitive and specific detection of somatic mutations to 0.05% allele ratio. The limit of detection is sample specific and it is dynamically determined by the number of captured DNA molecules. We used our design methods to develop a next-generation sequencing assay that allows interrogation of~150 biomarkers relevant in lung from COSMIC and Oncomine databases, and de-novo variant detection at ~1,700 genomic positions in 11 genes implicated in non-small cell lung cancer. Results: We demonstrate assay performance and accuracy of variant analysis in control and archived cfDNA research samples. The assay delivers >95% on target reads and highly uniform amplification across targeted cfDNA molecules with input DNA of 1ng or higher, and has a fast turnaround time from extracted DNA to variants of less than 24 hr. Observed limits of detection are 0.5%, 0.1%, and 0.05% for amounts of input cfDNA of 1ng, 20ng, and 50ng respectively. This limits are defined for >90% sensitive and >99.5% specific variant calling. For doubled input DNA amount we observe >95% sensitivity and >99.8% specificity. Conclusions: Described computational and analytical methods for assay design and variant calling may facilitate researchers to study relevant biomarkers at 0.05% frequency in cfDNA. University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: HGVS nomenclature is a de facto method in clinical molecular laboratories and critical component of clinical reports. HGVS representation of variants is a requirement in the CAP accreditation checklist. Conversion of variant representation from genomic to cDNA and protein coordinates using HGVS recommendations for insertion and deletion (indel) variants is particularly challenging. The large number of variants detected in NGS assays requires automation of this conversion process. A python software package, hgvs, was authored by Hart et al (PMID: 25273102) for automated generation of HGVS variant representation. We present our experience with clinical validation of this automated approach. Methods: A total of 50 indel variants across 12 genes from routine clinical somatic mutation testing were evaluated for validation. The indels mapped to both coding and non-coding regions of the above genes. Variants were called by Torrent Variant Caller (TVC) v4.4.3. The hgvs v0.4.5 was used for generating the HGVS nomenclature strings. Custom modules were developed in Python to convert variants in the VCF file to the genomic variant representation (g.) required as an input by the hgvs package. The g. variants were then parsed by hgvs and mapped to coding DNA (c.) and protein (p.) representations with a database of specific RefSeq transcript identifiers and a local instance of Universal Transcript Archive (UTA) v20150827. For each variant the HGVS nomenclature was generated manually by a pathologist by reviewing the sequence pileups in Integrated Genomic Viewer (IGV, Broad Institute) per 2016 HGVS recommendations. Any discrepancies in the automated and manually generated strings were noted. Results: Overall HGVS annotations for 44 of the 50 variants were completely concordant. Six HGVS variant annotations revealed differences between the automated and manually generated c./p. HGVS strings. These differences included miscalculation of the position of the termination codon for frameshift indels, misalignment by 1-2 nucleotides for c. annotation for in-frame indels, and use of abbreviated HGVS format for c., particularly for non-coding indels. In addition, incorrect generation of the input g. annotation is a potential bottleneck and a source of downstream errors for generation of HGVS annotations. Conclusions: Validation of automated HGVS annotation for high volume NGS testing is necessary for optimizing the downstream workflow. Whereas the majority of the annotation using the hgvs package were concordant, specific variants revealed discrepancies in HGVS string. It is desirable to have the ability to accept native VCF variant representation to minimize creating of custom software modules in clinical laboratories with slim bioinformatics support. Introduction:Tumor molecular profiling is rapidly becoming the standard clinical test for selecting targeted therapies in refractory cancer patients. DNA extracted from patient samples is enriched for cancer genes and sequenced to identify actionable somatic mutations therein. A major challenge arises when tumor-derived data is analyzed in the absence of normal tissue data, as it is common in clinical scenarios. The distinction between somatic and germline variants become difficult, leaving clinicians to resort to crude heuristic filtering. Methods: We present here a variant calling software, developed under quality system regulation protocols, capable of accurately identifying somatic mutations from targeted next-generation sequencing data. A novel Bayesian Network approach models the distribution of reads harboring germline and somatic mutations, estimates the contamination from normal tissue in the sample, scores somatic mutations, and imputes germline variants, without matching normal tissue data. This approach also allows joint analysis of multiple specimens from the same patient (e.g., FFPE and ctDNA), when available, improving the limit of detection. To improve specificity, our caller can also utilize prior information from different databases including somatic mutations, germline variation, and healthy controls data, in a principled fashion. Results: We validated our method by analyzing data from the TOMA OS-Seq 131 cancer gene panel using the Illumina platform. Sample inputs ranging from 2-600ng of DNA were sequenced to a depth of Through adaptors with molecular barcodes we measured a median duplicate rate <2. We analyzed somatic mutations simulated at various variant allele fractions on a background of data from reference samples from the Genome-in-a-Bottle consortium, data on a dilution series from two reference samples, and several commercial control and clinical samples, including matched FFPE, PBMC, and ctDNA specimens. In the absence of normal tissue, our method scores each variant with respect to their likelihood of being somatic or germline. We show that, as compared to other commonly used methods, our algorithm can achieve a higher true positive rate whilst controlling a false discovery rate of 1%. We also show that jointly analyzing serial samples (e.g., ctDNA), we can improve sensitivity of shared variants. Conclusions: In conclusion, in contrast to currently used academic software developed for research projects, we observe that our caller outperforms these software and is particularly well suited for the clinical use cases. Introduction: Amicrobial pustulosis (AP) is an aseptic neutrophilic dermatoses characterized by the presence of sterile pustular eruptions affecting cutaneous folds with head and neck involvement. AP may occur as a sporadic disease, or more often, associated with autoimmune diseases (APAD) such as systemic lupus erythematosus. One of the main causes of these cutaneous lesions is an unbalance in the synthesis of inflammatory cytokines, determinant for the development and extension of these lesions. Cytokines are regulating proteins produced by lymphocytes and macrophages, also mediating cell growth, migration, immunity, differentiation and inflammation. For these cellular processes to occur, cytokines need to be transitorily synthesized and tightly regulated. Although this group of diseases have an immunological basis, little is known specifically about their Th1/Th17 cytokine profiles. The aim of this study is to compare the mRNA expression profiles of Th1 (IFN---18, CXCL9), and Th17 (IL-17, IL-23,FOXP3) of AP, APAD as well as skin lesions of systemic erithematous lupus (SEL) and discoid lupus erythematosus (DLE). Methods: We retrospectively reviewed the archives of the Department of Anatomic Pathology of the Instituto Nacional de Ciencias Medicas y Nutrición from 2003 to 2015. We retrieved 5 patients with AP, 7 with APAD, 3 with LED and 5 skin biopsies of healthy individuals (control tissue). Extraction of mRNA was performed with RNA FFPE extraction kit (Promega) and cDNA synthetized with Omniscript Reverse Transcriptase (Qiagen). Primers were design to amplify IL-17, TNF--18, IL-23 e IFN-and FOXP3 using the program Primer Express, based on sequences from ensemble.org. Assessment of mRNA expression using Quanti Tect Master Mix Kit SYBR Green jmd.amjpathol.org ■ The Journal of Molecular Diagnostics (Qiagen) and copy number/ul determined using standard curves. Mean and standard deviation were determined and the results compared using T-Student test. Results: APAD, SLE and DLE show a Th1 cytokine profile. AP show a statistically significant increase in IL-17 (Average -Standard deviation) (8.86 copies/ul -3.63) and IL-23 (38.5copies/ul -6.3) mRNA expression with respect to all other groups, as well as a moderate increase in IL-18 (10 copies/ul -2.82) and loss of FOXP3 (0.23 copies/ul). Conversely, APAD show statistically significant increase in FOXP3 (12 copies/ul) mRNA, and decreased levels of IL-17 (10.76 copies/ul -1.20) compared to AP. Conclusions: These findings indicate different immunological mechanisms subjacent to the pathogenesis of apparently similar morphologic lesions (PA versus APAD), and suggest a possible diagnostic role of Th17 profiling in the discrimination of these conditions. Introduction: Next-generation sequencing (NGS) has quickly become part of standard clinical care and an increasingly important tool in oncology personalized medicine. In 2013, Penn Medicine's Center for Personalized Diagnostics (CPD), a CAP/CLIA laboratory, was an early adopter of NGS technology and needed to navigate transitioning an ever-changing field into a very rigid structure. Starting small and then rapidly becoming a staple to our oncology group, takes good planning, flexibility and learning from mistakes. In an effort to aid and support the molecular diagnostics community, we describe the lessons learned from the establishment and growth of an oncology based NGS laboratory. Methods: Using samples tested via orthogonal methodologies during validation allowed for the establishment of good NGS laboratory practices including defining specimen characteristics, library prep quality control statistics, and true positives for mutation detection. A custom bioinformatics pipeline and integrated database was built to generate and maintain the data. Assays were invented to analyze difficult to sequence or capture genes (e.g., CEBPA) and low quantity or poor quality specimens (e.g., RNA extracted from FFPE tissues). With a growing test menu and increasing sample volume, logistical adjustments were implemented to accommodate the service burden placed on all aspects of the laboratory. Results: Based on thresholds in pre-validation analyses, minimum sample requirements were established at >10% tumor, 100 ng to 250 ng input, and >50% intact DNA. NGS quality controls thresholds were designated at an amplicon minimum depth of 250 reads, average mean coverage of 2000X to 5000X, and minimum reporting allele frequency of 5%. In total, 785 clinical samples were received in year one, 1482 in year two, and 2407 in year three, while staffing went from 8, to 6, to 10 personnel and the total number of assays (library preps and extractions) went from 4, to 7, to 12 respectively. Conclusions: Instituting an oncology NGS center, from inception to a functional clinical laboratory, is complex and multifactorial. The decision to begin the CPD with a limited number of highly specialized personnel and focus on two hotspot oncology panels facilitated the establishment of a robust workflow and clinical reporting. Additional personnel enabled the growth of testing, leading to different iterations of the initial panels, creating RNA based assays, and developing a reflex test to capture low quantity/quality samples. With continued success, the CPD is expanding in personnel, computing power, and clinical utility. Like NGS, the center is changing and evolving, even in a clinical setting, and hopefully lessons learned from our experience may be helpful to others in the molecular community. Introduction: Hepatitis B virus real-time PCR test is a method of quantitative PCR. The quality control limit of HBV real-time PCR was less than 1 log (IU/mL). Considering HBV treatment roadmap, more strict control limit was required. We developed a novel double X bar chart to establish acceptable control limits for a realtime quantitative PCR test. Methods: Two control materials (high and low control) and HBV PCR kits were donated by the Roche Diagnostics. The high and low control material were composed of three kinds of lot number. We tested each control materials two times a week for 6 months. We transformed total 456 results with the Box-Cox, then confirmed the normal distribution. We analyzed the transformed data by X-bar chart, MA (Moving average) chart, EWMA (Exponentially weighted MA) chart, CUSUM (Cumulative sum) chart and a novel double X-bar chart. To create a novel double X-bar chart, we used the jackknife (one-leave out) method and applied the X-bar chart algorithm to each Lower Limit and Upper Limit dataset. Result: The MA, EWMA, CUSUM chart showed good error detection rates, but they could not be utilized as a quality control method due to a narrow control limit. A novel double Xbar chart showed more wide range of control limits than X-bar chart with 3SD control limits. The value of quality rates for X bar chart is 99.7% and were fixed. However those of double X bar chart are 99.96% (n=25) and 99.90% (n=100), so these values depend on the number of data. Conclusions: Double x bar control chart has acceptable control limits compared with original x bar control chart. Therefore, we expect that this method could be utilized for the quality control of a real-time quantitative PCR. Introduction: Minimally invasive diagnostic procedures, such as core biopsies or fine needle aspiration are often the specimen of choice for genomic analysis of actionable variants. There are situations however, where cytologic smears are the only specimen type available for ancillary molecular testing. Next-generation sequencing (NGS) is able to interrogate multiple variants simultaneously from specimens with limited amount of neoplastic cells. The aim of our study was to investigate the minimum cellularity needed on cytologic smears for successful NGS. Methods: Thirty cases from October 2013 to June 2015 of resection-proven primary lung adenocarcinomas previously used for molecular analyses were identified in our institutional archives. Twenty-five cases were known to harbor clinically relevant variants and 5 were negative controls. One Diff-Quik-stained (DQ) and one Papanicolaou-stained (Pap) slide were selected from each case. Based on neoplastic cellularity, cases were categorized as containing < 100 tumor cells, 100-500 tumor cells, or >500 tumor cells. Tumor percentage was estimated as the relative amount of tumor cells vs non-neoplastic nucleated cells. NGS was performed on the Ion Torrent Personal Genome Machine using the Ampliseq cancer hotspot panel v2. Results: Paired DQ and Pap smears from each case showed similar cellularity and cases that differed in cellularity were within one category of each other. Two DQ showed <100 cells, 18 showed 100 to 500 tumor cells, and 10 showed > 500 tumor cells. Four Pap showed <100, 15 showed 100 to 500 tumor cells, and 11 showed >500 tumor cells. Tumor percentages in DQ and Pap were comparable with 46% and 47% respectively. The number of cases with >100 tumor cells had a 93% success rate which was significantly different from, cases with <100 tumor cells that were successfully sequenced only 67% of the time (p= 0.002). All of the DQ cases with <100 tumor cells failed whereas 3 of the 4 Pap cases with <100 tumors cells were successfully sequenced. Overall, NGS was successful in 80% of DQ and 87% of Pap with a total of 31variants identified in DQ and 33 in Pap. There were 3 cases where both DQ and Pap failed NGS. DQ smears failed to identify variants that were successfully identified in 3 cases on Pap whereas there was only 1 case where NGS on Pap failed to identify a variant successfully identified in DQ. Conclusions: There is a significantly higher likelihood of successful NGS using cytologic smears with >100 tumor cells. In cases with <100 tumor cells, Overall, there was a trend for both a higher NGS success rate and a higher likelihood of successful variant identification in Pap smears vs DQ. The advantage of Pap specimens over DQ warrants further investigation. Introduction: Recent technological advancements have reduced the turnaround time and cost of molecular testing and increased access to genetic testing. These changes have enhanced the role genetic tests play in the diagnosis and management of a number of genetic and neoplastic diseases. These advancements have led to an increased demand for genetic testing, which in turn has led to considerable increases in reference laboratory costs. Our institution has not been immune to these cost increases and to address them we previously formed a test utilization review team, consolidated reference laboratory testing and internalized testing. These actions led to a dramatic reduction in annual reference laboratory costs of approximately 2.7 million dollars between FY 2012 and FY 2014. In this report we share the progress we have made to further manage these costs. Methods: The multidisciplinary team consisting of two PhD laboratory medical directors, a PhD clinical chemistry fellow, a rotating pathology resident and a MD medical geneticist review all genetic test requests. The initial review is done by pathology resident or clinical chemistry fellow in consultation with a PhD medical director. The review process incorporates the use of an online genetic test/laboratory selection search engine allowing for price and gene panel composition comparison between reference laboratories. The search engine also enables the retrospective review of orders to look for additional savings. Complex cases or large gene panels elicit a second review by the medical geneticist. Appropriate requests are tiered to cover the most probable genes or directed to the more appropriate panel based upon clinical presentation and family history. The patient's medical record is also reviewed to prevent duplicate testing. All genetic test requests are reviewed and a team member proactively offers consultative services to clinicians to guide test selection based on clinical presentation and a thorough review of the genes associated with the suspected condition. We continue to internalize genetic test panels that are frequently ordered at our institution. Results: Using the review methods highlighted above we have further reduced our costs by a monthly average of $25,000. We have also recently validated a 54 gene myeloid panel that we have used to internalize the sequencing of CALR, JAK2 and MPL. Internalizing these three genes via the myeloid panel has reduced our costs by a monthly average The Journal of Molecular Diagnostics ■ jmd.amjpathol.org $20,000. Combined these approaches account for an average savings of ~$45,000 per month. Conclusions: We have made substantial progress in managing the increased cost of molecular laboratory testing and are on target to reduce our molecular sendout costs by over $1 million this year. Introduction: ACL Laboratories pathologists have been proactively working to improve patient outcomes by helping providers determine the right test at the right time for the right patient. They recognize the importance of staying on top of advances in medicine and improving the ongoing communications with clinical colleagues. Recently, utilization of genetic risk information, particularly Factor V Leiden, to influence patient management in the absence of supporting evidence related to health outcomes has been heavily scrutinized. Current evidence suggests that the Factor V Leiden mutation is not a determinant of the risk of thrombosis recurrence. The risk of recurrence for heterozygous Factor V Leiden was minimal, but significant for homozygotes. Furthermore, compound heterozygotes involving Factor V Leiden and another thrombophilia also meaningfully increases risk. Additionally, since anticoagulation reduces recurrence regardless of mutation status, its presence doesn't change management. To assess our physician usage of Factor V Leiden, the Clinical Effectiveness Utilization Committee at Advocate Health Care System, in conjunction with ACL laboratories, analyzed inpatient orders for Factor V Leiden. Methods: Data from ACL Laboratory "Data Warehouse" was accessed via data collection and analytical software; IBM Cognos Business Intelligence version 10.1.1 from IBM (International Business Machine) and Cube Functions in Microsoft Excel 2010 from Microsoft. Clinical Effectiveness Utilization Committee at Advocate Health Care System implemented system wide testing algorithm, which restricted ordering Factor V as a standing alone order and implemented APC with reflex to Factor V test as the only option. Results: Advocate's baseline positivity rates for wild-type, heterozygous, and homozygous mutant results were 94.7%, 5.2% and 0.1% respectively. Comparison to national positivity rates (89.1% wild-type, 10.5% heterozygous, 0.4% homozygous mutant), suggested an over utilization of the test at Advocate Health Care. With education and communication efforts in conjunction with minor CPOE interventions, physicians were led to order the more appropriate screening test Activated Protein C Resistance (APC-R). Following the changes, Factor V Leiden genetic ordering was successfully reduced by 82%. This resulted in a $50,000 savings after the first 5 months of year 2016 (projected savings for 2106 are ~$120,000). Post implementation Factor V positivity rates were as follow (49.2% wild-type, 50.4% heterozygous, 0.4% homozygous mutant). Conclusions: Our experience illustrates that in a healthcare system designed to encourage active intervention, and more precise patient care, physicians can do a better job of ordering the right test at the right time. This had become a tractable approach for the interrogation of disease progression and immune response due to recent throughput and read length improvements in next-generation sequencing (NGS) technologies. However, structural and sequence complexities of antibody genes have made reliable targeting approaches challenging. Methods: We have developed and optimized a method for accurate sequencing of full-length immune gene repertoires of B-cells and T-cells. The method uses a unique barcoding scheme specifically designed to tag every mRNA molecule with a unique molecule ID (UMI) so that all PCR copies of each mRNA fragment can be collapsed into a single consensus sequence, making the assay extremely accurate by resolving PCR bias and sequencing errors, as well as allowing quantitative assessment of clonal antibody evolution in longitudinal studies in biological samples where disease has triggered an immune response. Results: Immune sequencing libraries were generated from total RNA extracted from PBMCs in duplicate from a single patient. The use of UMIs enabled absolute quantification of starting RNA molecules present in the original sample and therefore accurate ranking of the antibody clone abundance by avoiding the bias incorporated by PCR or sequencing when total reads only were measured. Using the same sequencing method, tumor samples were analyzed for abundance of expanded clones via grouping clones by V gene, J gene and CDR3 similarity and ranking by mRNA abundance. Additionally, the use of isotype-specific primers (IgM, IgD, IgG, IgA and IgE) enabled measurement of the heavy chain isotype proportions within the samples. Further, alignment of full-length heavy chain antibody sequences generated using this method to germline genes from reference databases enabled quantitation of the mutation level of each antibody sequence, thereby providing information on the overall maturity and mutational profile of the sample repertoire. Conclusion: This novel method allows for exhaustive somatic mutation profiling across complete V, D and J segments, full isotype information analysis, and the possibility for synthesis and expression of complete antibody chains for downstream immunological assays. Introduction: With enormous growth in the field of molecular pathology, reporting of results gleaned from this testing is essential to guide patient care. In this study, we examine molecular reports for BRAF mutation testing from labs participating in the College of American Pathologists (CAP) proficiency testing for required elements to convey molecular lab test results to clinicians and patients. This study expands on previously presented data. Methods: Molecular labs participating in the CAP proficiency testing program for BRAF mutation analysis were solicited to submit examples of final reports from two separate proficiency testing reporting cycles. The reports were de-identified at CAP and the data were abstracted by two members of the CAP Molecular Oncology Committee. Reports were reviewed for presence or absence of relevant components based on CAP checklist requirements and published guidelines on molecular reporting (Gulley et al. Arch Pathol Lab Med 2007; 131:852-863) . Results: A total of 116 reports were received (62 demonstrating a positive result for the BRAF V600E mutation and 54 negative). The majority of lab reports (75%) included basic reporting information consisting of lab identity, dates and pertinent specimen information. The test name as the header of the results was present in 98% of reports. Methods for BRAF testing varied, with 95% adequately describing their assay methods and 87% of labs adequately describing the target(s) of their assays. Information on the analytical sensitivity of the assay was present in 74% or reports and 83% reported at least one assay limitation, though only 34% reported on variants not detected by their assays. Analytical and clinical interpretive comments were included in 99% and 90% of reports, respectively. Of participants that perform a laboratory developed test (LDT) only, 88% included language addressing the development of the assay and 62% included all required elements for LDT reporting. Conclusions: Labs participating in BRAF proficiency testing through the CAP are including most of the required reporting elements to unambiguously convey molecular results. Labs should continue to strive to report these results in a concise and comprehensive manner. Close attention should be paid to the required elements for reporting, particularly a clear description of the methods and targets, limitations of the assays, and LDT reporting requirements. Limitations of this study include a relatively small sample size and the fact that it evaluated only one molecular test. Responder bias may have contributed to the overall high rates of reporting compliance, while compliance in some categories may have been underestimated due to report formatting and the nature of proficiency testing samples. Introduction: This tool is born of two frustrations: lost time confirming annotation of insertion and deletion variants from sequencing results and lack of a simple method to demonstrate base and codon changes, transcript variants, and nomenclature to students, residents, and fellows. This tool uses a common spreadsheet program to visualize reference sequence as individual bases in cells numbered according to the given refseq identification number, automatically translates the sequence to numbered codons, allows one to make changes to a parellel copy of the reference sequence, translates the new sequence via macro, and allows one to compare differences between the original and variant sequence. Methods: Microsoft Excel is used as the base platform. An embedded link allows for retrieval of the given FASTA text for the desired NCBI refseq ID as input. Embedded formulae convert the FASTA text into a simple text string and deconcatenates the characters into numbered bases. Translation is performed using a vlookup function and embedded protein coding chart. Re-translation of the manipulated variant sequence is performed using a visual basic macro as is resetting of the sequence if an error is made and the user would like to start over. A formula is used to located sequence of interest which is gleaned from the output of the sequencing data. Search functions can be used to locate base positions, protein codon positions, and stop codons rapidly. A resident with little molecular pathology experience (never previously worked up a sequence variant) was given the "c." coordinates of two variants (an insertion and deletion, both frameshift) and given the task of identifying the "p." information and nomenclature of the changes either manually or with the assistance of the spreadsheet. The methods were staggered to avoid bias. A brief primer on each method, manual and spreadsheet assisted, was given along with printed instructions for each method. A timed trial was also conducted for a molecular faculty member familiar with the manual and spreadsheet methods. Results: The trainee showed manual solution times of 35 and 23 minutes and assisted times of 9 and 9 minutes for problems 1 and 2 respectively. The faculty member demonstrated manual solution times of 13 and 11 and assisted times of 3.3 and 3 minutes. Alternative transcripts from 16 genes were examined to verify that sequences matched reference annotation to validate the translation function of the spreadsheet. Conclusions: This tool can significantly cut down time in verifying annotation of insertions and deletions. It is also useful as a teaching tool when instructing trainees about sequence interpretation and gene structure. It provides visual and interactive feedback to reinforce concepts. Introduction: Circulating cell free (ccf) DNA is a challenging material to work with, the DNA is fragmented and can cause issues in molecular testing. There is also an issue with stability where we want to extract the DNA as soon after collection as possible, otherwise increasing the issues with this challenging material. Here we are testing the feasibility of using the PAXgene Blood ccfDNA Tube to increase the stability of ccfDNA when transporting plasma samples across state to our central lab for up to 5 days, in combination with the Therascreen EGFR RGQ PCR Kit, to ensure a valid result. Methods: Collection of multiple blood samples into EDTA and PAXgene Blood ccfDNA Tube, Day 0, and then subsequent isolation of plasma and DNA extraction using the QIAamp DSP Circulating NA Kit at Day 1, Day 3 and Day 5. Followed by testing with the Therascreen EGFR RGQ PCR Kit for DNA assessment using the control assay amplifying exon 2 of the EGFR gene. Results: With all samples collected in the PAXgene Blood ccfDNA, we were able to obtain a valid Ct, as determined by the Therascreen EGFR Kit, at all time points for this study, showing that amplifiable DNA is isolated. Whereas with the samples collected in EDTA, and the plasma isolation and DNA extraction performed at the respective time point showed a decrease in the amount of amplifiable DNA present, with a worsening Ct at each time point tested. Conclusions: Our study found that the PAXgene Blood ccfDNA Tube can be used for collection of blood and subsequent EGFR testing for samples that take an extended period of time, 5 days, to reach our central lab for testing. Compared to blood collecting in EDTA, which would result in an invalid result being reported, and new sample needed to be requested. Introduction: Amplification of the MET gene or increased MET gene copy number has been reported in a subset of non-small cell lung carcinoma. MET amplification has been suggested as a mechanism for resistance to EGFR tyrosine kinase inhibitor-based treatment. Lung cancer patients with MET amplification, defined as MET/CEP7 However, heterogeneous MET amplification in lung tumors has not been examined. Next-generation sequencing (NGS) technology can be used to detect copy number alterations in cancer. The objectives of this study were to assess heterogeneous MET amplification in lung tumors, to compare results of capture-based NGS assay to FISH for detecting MET amplification, and to identify any additional gene alterations in MET amplification-or gene copy number-positive cases. Methods: FISH reports of 696 de-identified lung cancer samples were reviewed retrospectively for MET amplification. At least fifty cells were scored for CEP7 and MET probe signals in each case. FISH data were analyzed as follows: positive for MET amplification (MET/CEP7 MET gene copy number (MET/CEP7 ratio of <2.2, average MET MET amplification (presence of 5-<50% of cells with MET/CEP7 conducted with a targeted lung panel (25 genes) in ten cases. Results: MET amplification was detected in 3.5% (n=24) of cases, whereas 9.2% (n=64) showed increased MET gene copy number. Of the 24 MET-amplified cases, 12.5% (n=6) exhibited heterogeneous MET amplification. The MET/CEP7 ratio in these six cases samples that were MET amplification-positive by FISH were subjected to NGS. MET amplification was detected by NGS in four samples in which the MET/CEP7 MET/CEP7 -<3.0. Four samples that were positive for an increased MET gene copy number were MET amplification-negative by NGS. Alterations in the TP53 gene were also detected in 7/10 cases. Pathogenic variants in lung cancer driver genes such as EGFR, KRAS, and PTEN were detected in tumors with an increased MET gene copy number. Conclusions: This study provides evidence for heterogeneous MET amplification in lung tumors and is associated with the tumors exhibiting MET/CEP7 -<3.0. Our NGS assay detected MET amplification in all cases with a MET/CEP7 for detecting MET amplification-positive patients for crizotinib treatment. Driver mutations in EGFR, KRAS, and PTEN genes appear to play a key role in the pathogenesis of lung cancers with an increased MET gene copy number. Introduction: Castration-resistant prostate cancer (CRPC) patients are often treated with drugs that target the androgen receptor (AR) ligand-binding domain (e.g., enzalutamide and abiraterone). Constitutively-active AR splice variant 7 (AR-V7) lacks the ligand-binding domain and, if detected in circulating tumor cells (CTCs), is associated with resistance to these agents. Methods: We validated the AR-V7 assay in a CLIA-certified laboratory. CTCs were isolated from blood by immunomagnetic enrichment, and mRNA reverse transcribed into cDNA. Real-time PCR amplification of reference genes (ACTB and GAPDH), "prostate-specific" genes (PSMA, PSA and AR-full length (AR-FL)) and AR-V7 was performed. Specimens for validation included an AR-V7 expressing prostate cancer cell line (LNCaP95), 38 peripheral blood controls (14 males, 24 females), and 21 blood samples from CRPC patients. Results: The assay detected as few as five LNCaP95 cells spiked into peripheral blood, demonstrating high analytic sensitivity. Multiple inter-and intra-run replicates of LNCaP95 cell line experiments yielded similar Cq values for all genes, showing high analytic precision (AR-V7 Cq CV of 0.67%). All 38 healthy control samples were negative for AR-V7, demonstrating high diagnostic specificity (100%). All healthy control samples were negative for PSA and PSMA expression. The diagnostic accuracy was confirmed by concurrent testing of 21 CRPC samples in the research laboratory and the clinical diagnostic laboratory: concordance in AR-V7 status was achieved in all cases (positive in 4, negative in 17) (100% accuracy). Since, our laboratory is the first CLIA-certified laboratory performing AR-V7 testing, inter-lab proficiency is not possible. In the interim, the clinical samples are subsequently tested by the research laboratory as an additional quality control. Seventeen patient samples were subsequently tested by the research lab. Overall 16 of 17 patients were categorized identically between the two labs (94.11% concurrency). There was one discrepancy between the labs, where the clinical lab detected AR-V7 transcript at a very low level (Cq of 38.10) in one of the two duplicates, which was not detected by the research lab. Conclusions: This first validated clinical assay detects CTC-derived AR-V7 with high analytic sensitivity, analytic precision, diagnostic specificity and diagnostic accuracy. It is currently in use for targeted therapy trials. Within this pathway, BRAF, KRAS, and NRAS as amongst the most frequently mutated genes, with mutation rates as high as 50% and 30% for BRAF and NRAS, respectively, in malignant melanoma, and 40% for KRAS in metastatic colorectal cancer (mCRC). To robustly, rapidly, and reliably detect these mutations across indications, Roche has developed two allele-specific PCR, research use only assays for the detection of mutations in the BRAF, KRAS, and NRAS genes from DNA isolated from formalin-fixed, paraffin-embedded tissue (FFPET) samples. The BRAF/NRAS Mutation Test (life science research only; LSR), covers 36 unique mutations in exons 11 and 15 in the BRAF gene and exons 2, 3 and 4 of NRAS gene; similarly, the KRAS Mutation Test v2 (LSR) detects 28 unique mutations in exons 2, 3, and 4 of the KRAS genes. The assays are designed for use on the userdefined workflow (UDF) of the Cobas z 480 analyzer. To further assess the performance of these assays, 50 FFPET samples of either melanoma, non-small cell lung cancer (NSCLC), or colorectal cancer origin previously characterized with Next-Generation Sequencing were retested with the two assays to determine concordance. Methods: DNA from 50 specimens was isolated using the manual Cobas DNA Sample Preparation Kit and tested by an in-house NGS gene panel. The DNA was tested at Carolinas Medical Center (Charlotte, NC) using the BRAF/NRAS Mutation Test (LSR) and KRAS Mutation Test v2 (LSR) assays and compared to the initial NGS results to determine concordance. Results: The frequency of BRAF, NRAS, and KRAS mutations was 30%, 28%, and 12% respectively using the LSR Tests. Fifteen BRAF mutations were detected -ten in melanoma and five in mCRC. Fourteen NRAS mutations were detected in melanoma. Six KRAS mutations were detected -five in CRC and one in NSCLC. One sample was invalid by KRAS LSR. Another sample was discordant between LSR and NGS for BRAF. The overall agreement between the two methods was 98.0% (48/49). The positive percent agreement (PPA) was 100% (34/34), and negative percent agreement (NPA) was 93.3% (14/15). The overall call rate with all samples was 98% (48/50). Conclusions: We showed excellent concordance between the Roche BRAF/NRAS Mutation Test (LSR) and KRAS Mutation Test v2 (LSR) to NGS. These LSR assays allow for highly sensitive and specific detection of 64 unique BRAF, KRAS, and NRAS mutations in a comparatively simple, streamlined, and rapid workflow. Carolinas Medical Center will be testing an additional 50 mCRC and melanoma samples previously characterized by NGS to further assess concordance. Introduction: Low-grade fibromyxoid sarcoma (LGFMS) and sclerosing epithelioid fibrosarcoma (SEF) were thought to be related before the characteristic FUS-CREB3L2 chimeric gene was identified in both lesions. Once identified, fluorescent in-situ hybridization (FISH) analysis was adopted to detect the t(7;16) and t(11;16) translocations involving the FUS gene in LGFMS. SEF can exist in hybrid or pure forms, with hybrid lesions containing areas of morphologic overlap with LGFMS. Genetically, hybrid tumors have been found to carry the FUS rearrangement whereas pure lesions more frequently show EWSR1rearrangements. We present a case of hybrid SEF that was negative for FUS translocation by FISH. Archer FusionPlex testing revealed an unexpected EWSR1-CREB3L1 fusion that was subsequently confirmed by EWSR1 FISH rearrangement. Methods: FUS and EWSR1 FISH testing was performed via breakapart probe using the LSI FUS 16p11 (Abbott Molecular, Des Plaines, IL) and LSI EWSR1 22q12 probes (Abbott Molecular, Des Plaines, IL). Two hundred cells were counted with a threshold of positivity of 15% (FUS) or 25% (EWSR1) of cells. Fusion detection was performed using the Archer FusionPlex Sarcoma Kit (ArcherDx, Boulder, CO), sequencing was performed on the Miseq (Illumina, San Diego, CA), and final analysis utilized the Archer analysis pipeline. Results: Histology revealed areas consistent with both SEF and LGFMS (hybrid tumor). Immunohistochemical stains were positive for CD99 and MUC4 only. RT-PCR performed at an outside institution for EWSR1-FLI1 was negative. FUS FISH was negative. The Archer analysis identified a strong evidence EWSR1-CREB3L1 (t11;22) fusion supported by 24 unique reads. Follow up testing by FISH for EWSR1 by break-apart probe was positive in 72% of cells. Conclusions: FISH can detect fusion events without previous evidence of the fusion partner; but is hindered by the need to select the appropriate target gene. Nextgeneration sequencing (NGS) does not require previous knowledge of potential fusions, and can be cost effective and more rapid than FISH when multiple studies are needed. The Archer FusionPlex Sarcoma kit uses multiplex anchored-PCR followed by NGS to simultaneously detect 26 different genes and reports any translocation partner, such as in our case detecting EWSR1-CREB3L1. In scenarios with unusual or promiscuous rearrangement, the Archer FusionPlex system offers a way to quickly and affordably identify potential clinically relevant translocations. Inc) . TST 26 provides valuable mutational information to patients, however approximately 40 % of patient samples are referred to pyrosequencing because of poor DNA quality and / or insufficient quantity of DNA. To circumvent this problem, Illumina recently developed the TruSight tumor 15 gene panel (TST 15), which uses next-generation sequencing (NGS) to assess clinically relevant and most prevalent mutations in 15 genes in solid tumors. TST 15 designed to detect low-frequency variants from 20 ng of DNA from formalin-fixed, paraffin embedded (FFPE) tumor tissue. TST 15 will be offered as a single assay for accurate, economical, and rapid analysis of solid tumors. Moffitt Cancer Center was one of the institutions selected by Illumina for a pre-production beta testing of TST 15. Methods: For beta testing of TST15, all the reagents including one control sample with known mutations and the pertinent software were provided by Illumina. The TST15 contains two separate pools of tagged oligonucleotide primers. These pools are used in multiplex PCR to target and amplify regions for a select set of genes. Using adapter primers, templates are indexed and further amplified in a second PCR and then combined into a single tube for paired-end sequencing on the MiSeq Instrument. Data are analyzed by using the on-instrument reporting software. We performed three runs using MiSeq with 8 samples per run. A total of 19 patient samples previously sequenced either by TST 26 gene panel or by pyrosequencing, control sample and horizon diagnostic reference standards were tested. Results: The TST15 demonstrated a remarkably high tolerance to poor DNA quality and / or insufficient quantity when compared with TST 26. All nine patient samples rejected for mutation analysis by TST 26 because of insufficient amount of DNA, were successfully analyzed by TST 15. Mutation profiles obtained for all samples tested were 100% concordant between TST15 and TST 26 and pyrosequencing. TST15 provided uniform coverage of target regions, identifying somatic mutations at frequencies as low as 5% with minimum coverage of 500X. Conclusions: TST15 accurately detected low frequency variants from 20 ng of DNA. The time required for library preparation was approximately 7 h versus 24 h for TST 26. TST 15 offers a rapid, economical and reliable NGS work flow solution that can easily be implemented into laboratories that perform NGS. (SFT) is a mesenchymal tumor rarely seen in the clinic. Conventional treatment for an SFT is a surgical resection and a poor prognosis is frequently observed once postoperative recurrence occurs. However, molecular-targeted therapy provides a new treatment strategy with higher efficiency and lower toxicity. Targeted drugs act on specific gene mutations and their treatment efficiency largely relies on the presence of drug-sensitive mutations within the tumor tissue. Methods: This report presents the study of a 50-year-old female with an SFT who was being treated at the Third Affiliated Hospital of Harbin Medical University (Heilongjiang Cancer Hospital) in China. Formalin-fixed paraffin-embedded (FFPE) samples of the SFT tissue were got form the patient prior to pazopanib treatment. Next-generation sequencing (NGS) was performed on the Illumina X platform to analyze a total of 483 cancer-related genes. Results: The data of NGS showed that the tumor had mutations in BRCA1 p.Y856H and FGFR2 p.N549K. Based on the results of gene analysis and the clinical evidence of the randomized, placebocontrolled PALETTE trial (Graaf, et al, Lancet 2012) , the patient in this study started taking pazopanib. The size of the lesion continuously decreased from 130.36 × 157.64 mm to 107.5 × 146 mm after two months of treatment. Conclusions: To date, no clear biomarkers have been identified for the targeted treatment of SFT. By combining cancer gene mutations with the clinical research progress of pazopanib treatment, this case report provides new insights into SFT-targeted therapy in terms of clinical research and medication. Since patients with SFT in the liver are rare, a comprehensive genetic analysis offers clinical evidence and directions for further research of the pathogenesis of SFT. Introduction: Identifying actionable mutations is an important aspect of cancer patient management. With the increasing use of fine needle aspiration (FNA) biopsies, it is critical to validate FNA specimens for targeted next-generation sequencing (NGS). In this study, we retrospectively examined NGS performed on FNA samples at a single institution to evaluate their adequacy and utility. Methods: Our database was searched for FNA specimens (between 1/1/2015 to 12/31/2015) with requests for our institutional NGS panel (198 genes) . Clinical data, pathological features, and molecular profiling results were collected and analyzed. Formal histologic review of all cell blocks/core biopsies to evaluate tumor percentage was performed by two independent reviewers. Results: Of the 2752 FNAs performed in 2015, 154 specimens were sent for NGS (5.6%). These included specimens from various anatomical sites, with the majority from lung (51.9%) and lymph node (23.4%). In 49% of the cases, the sample was obtained from the primary tumor. Lung adenocarcinoma was the predominant diagnosis (70.8%, including primary and metastatic lesions), with the remainder representing a spectrum of gastrointestinal, pancreatobiliary, and gynecological carcinomas, large/small cell neuroendocrine carcinomas, and malignant melanoma. There are 21 cases with tumor percentage less than 20% (13.6%). In six cases (3.9%), NGS was cancelled due to insufficient DNA or low DNA quality. Six specimens were also sent out for sequencing and showed concordant mutational profiles. Repeat NGS was performed in subsequent resections in two cases, with both demonstrating identical pathogenic/likely pathogenic variants compared with the primary FNA samples. In total, 642 variants were reported with 171 (26.6%) annotated as pathogenic variants, 51 (0.8%) as likely pathogenic variants, and 420 (65.4%) as variants with unknown significance. For pathogenic/likely pathogenic variants, 46 were identified inEGFR, 81 in TP53, 41 in KRAS, and 6 in PIK3CA. Conclusions: FNA samples are reliable and robust for detecting clinically actionable mutations. The NGS data generated from FNA specimens are used to guide targeted therapy, enroll patients into clinical trials, and distinguish a second primary tumor versus metastasis. LGSCs are indolent and evolve via Type I pathway from serous cystadenoma to typical serous borderline tumor (TSBT) and micropapillary serous borderline tumor (MPSBT). KRAS and other oncogenes mutations are frequent in LGSC. In contrast, HGSCs are aggressive, lack KRAS mutations, and arise via Type II pathway in all or none patterns. Exceptionally, rare HGSCs harboring KRAS mutations have beeb reported to arise via Type I pathway. Methods: To contribute to this scant pool of evidence, we present one such case of a 54 years old woman. She was diagnosed with endometrial carcinoma and underwent laparoscopic hysterectomy with bilateral jmd.amjpathol.org ■ The Journal of Molecular Diagnostics salpingo-oophorectomies. Gross and microscopic findings are described. Tissue microdissections were conducted using laser capture microdissection (LCM) for serous cystadenoma, and manually for TSBT, MPSBT, LGSC, and HGSC. DNA was extracted by Life Technology's Arcturus PicoPure DNA Extraction Kit and TrimGen's WaxFree Paraffin DNA Extraction kit from LCM and manual microdissections, respectively. Codon 12/13 of the KRAS gene and flanking DNA sequences were PCR amplified followed by pyrosequencing on Qiagen's PyroMark Q24 pyrosequencing system. Results: In addition to the endometrial adenocarcinoma, a left ovarian cyst with papillary excrescences was identified. Microscopically, the ovarian tumor is comprised of an admixture of serous cystadenoma, TSBT and MPSBT with minimal invasion associated to each, LGSC and LGSC in transformation to HGSC. The HGSC portion shows brisk mitoses (16 mitoses/10HPF) and high grade histologic features characterized by nuclear pleomorphism, prominent nucleoli with multinucleation. The same KRAS gene c.38G>A, p.Gly13Asp mutation was detected by pyrosequencing in all of the sampled areas, i.e., serous cystadenoma, TSBT, MPSBT, LGSC, and HGSC. Conclusions: Given the wide variety of KRAS mutations, this same codon mutation serves as a de facto biomarker to illustrate the clonal essence of tumor evolution. The findings that the same KRAS mutation carried from serous cystadenoma, TSBT, MPSBT, LGSC, to HGSC in one single case are novel. The results support the existence of HGSCs evolving via Type I pathway and hence improve our understanding on mechanisms behind tumorigenesis of ovarian serous carcinoma. In clinical settings, preanalytical variables greatly affect downstream technical analysis by next-generation sequencing (NGS). This is pronounced in developing countries lacking standardized protocols. To understand the effect of type and time of fixation on DNA and RNA NGS results, human cell lines and tissues were fixed with commonly available anatomical pathology fixatives for three timepoints. Methods: All experiments were conducted with a developed RNase minimized protocol. Human cell lines A549, MCF-7 and Caco-2 were grown to a confluence of 2x10^6, harvested, pelleted and snap frozen in liquid nitrogen (LiN2). Cell pellets were fixed with neutral buffered formalin (NBF), 100% ethanol (EtOH), 100% methanol (MeOH), zinc buffered formalin (ZBF) or Bouin's Fluid (BF) for 8, 24 or 72 hrs. LiN2 frozen human lung, breast and colon resections were chosen based upon adenocarcinoma subtype, >90% tumor volume, and RIN >8.5. Tissue resections were fixed with NBF, EtOH, MeOH, ZBF and BF for 8, 24, or 48 hrs. Cell pellets and tissue were processed with an 8 hour tissue processing protocol, embedded in paraffin, and cut into 4 and 7μm sections. 4 μm sections were adhered to glass slides and H+E stained for visual comparison. DNA and RNA (NA) were extracted from the 7 μm sections using the RecoverAll Total NA Isolation Kit for FFPE. Quality control (QC) was performed on the extracted NA by fluorometric quantitation and qPCR. NGS was performed on NA using the Ion AmpliSeq Cancer Hotspot Panel v2 and RNA Cancer Panel upon the Ion PGM. NGS success criteria were NA yield, library amplification, low clonality, depth of coverage and number of reads. Results: For fixed and processed (FP) cell lines and tissues, no significant difference was seen between cell type or 8 hour to 24 hour fixation; however NBF, ZBF and BF performed worse with NGS success criteria than EtOH and MeOH samples after 72 hrs. Conversely EtOH and MeOH produced worse morphology compared to NBF, ZBF and BF at all timepoints. Conclusions: Alcohol containing compounds are excellent cell and solid tissue NA fixatives but do not preserve cellular architecture. Aqueous formaldehyde containing fixatives, primarily NBF, should be used for 24 hours due to its ability to generate NGS results as well as histomorphology. A. Ordobazari, M. Sharma, Z. Wang Univeristy of Florida College of Medicine Jacksonville, Jacksonville, FL. Introduction: Lung biomarkers testing on patients with non-small cell lung carcinoma (NSCLC) have become routine laboratory practice as companion tests for targeted therapies. EGFR mutations and ALK rearrangement assays are the standard tests as recommended. Although not recommended as a sole determinant of EGFR tyrosine kinase inhibitor (TKI) therapy, KRAS mutations have been determined as mutually exclusive with EGFR mutations and intrinsically resistant to TKI therapy. However, given that negative KRAS mutation results are not helpful, to test or not to test remains a dilemma. This study reports our experience in adopting a modified testing algorithm to ensure testing efficiency by taking advantage of a slew of known parameters that are associated with KRAS mutations or EGFR mutations. Most important parameters include patients' smoking history and mucinous vs. papillary histology of the tumor. Methods: The cohort consists of 111 consecutive cases of surgically resected or biopsied NSCLCs from the past 1.5 years. KRAS codon 12/13 mutation assay was indicated if the patient is a smoker and the tumor is mucinous. Non-smokers, relatively young female patients with papillary type tumor are not indicated. Laboratory work flow, sample volume, turn-around time, patients' clinical conditions, and specimen cellularity are also considered. KRAS codon 12/13 mutation detection was conducted using the pyrosequencing technology with a low limit of detection of 5% mutant alleles. Results: Among the 111 cases of NSCLCs, 64% (71 cases) were tested for KRAS codon 12/13 mutations. Mutations were detected in 26 cases, accounting for 37% of the cohort. Of the 26 positive cases, 24 were codon 12 mutations and two were codon 13 mutations. Conclusions: Our cohort of cases represent a North Florida patient population with most being African Americans. Given that the overall documented frequency of KRAS mutations in such ethnic group ranges from 15%-25%, achieving a 37% detection frequency is significant. The results support and promote KRAS mutation assay for NSCLCs by an efficient testing algorithm. Establishment of one such algorithm eliminates the need to test for EGFR and ALK which are more costly, time consuming, and labor intensive. Introduction: KRAS mutant non-small cell lung cancer (NSCLC) is remarkable for its divergent biological and clinical behavior. Distinct KRAS variants have been identified that engage different downstream effector pathways and vary in drug sensitivity patterns. In particular, G12C and G12V subtypes depend to a greater extent on the RAS/RAF/MEK/ERK signaling cascade and may benefit from the addition of a MEK inhibitor to standard chemotherapeutic agent. The impact of coexisting genetic mutations on treatment responsiveness in KRASmutant NSCLC has not been fully explored. Preclinical studies indicated that concomitant loss of TP53 or LKB1/STK11 tumor suppressor genes impaired the response of KRAS induced lung cancer to docetaxel. The addition of selumetinib provided the greatest benefit for mice bearing tumors with KRAS mutation only, and lesser yet substantial benefit in KRAS tumors with TP53 co-mutation. Conversely, the concurrent LKB1/STK11 loss resulted in primary resistance of KRAS mutant lung cancer to docetaxel and selumetinib therapy. In this study we investigate TP53 and LKB1/STK11 mutation co-occurrence with major subtypes of KRAS as potential biomarkers for efficacy of MEK inhibitors in NSCLC patients within our institution. Methods: A total of 591 NSCLC samples were screened for somatic mutations using the 50 gene AmpliSeq Cancer Hotspot Panel v2 (CHPv2), and samples positive for variants in the KRAS gene (N=198) were included in further analysis. Fisher's exact test was used for statistics. Results: The overall mutation rate in the KRAS gene was 33.5%. The MEK-dependent G12C and G12V KRAS variants were the most common and occurred with a combined frequency of 61%. Across all KRAS subtypes the LKB1/STK11 and TP53 mutations were detected in 26 and 61 cases, respectively. Whereas the TP53 mutations were distributed evenly in-between G12C+V and non-G12C+V cancers (31 vs 30), the LKB1/STK11 mutations clustered with MEK dependent subtypes of KRAS (22 vs 4, p=0.009). As LKB1/STK11and TP53 mutations were found to be mutually exclusive, 3 different subgroups of G12C+V tumors were identified: G12C+V (56%), G12C+V with co-mutated TP53 (26%), and G12C+V with co-mutated LKB1/STK11 (18%). Conclusions: Over a half of KRAS-NSCLCs in our cohort carries G12C or G12V mutation with potential susceptibility to MEK inhibition. As TP53 and LKB1/STK11 co-mutations may impact efficacy of MEK inhibitors, we define 3 subpopulations of patients with predicted: best (G12C+V), moderate (G12C+V and TP53), and minimal (G12C+V and LKB1/STK11) response to such treatment. The concomitant LKB1/STK11 loss in a subset of G12C+V KRASdriven lung cancer may indicate an aggressive phenotype with the need for alternative therapeutic approach. Response to targeted therapies such as Imiatinib, a tyrosine kinase inhibitor, depends on the specific mutation present. A small percentage of GISTs are wild-type for KIT and PDGFRA, but show deficiencies in the succinate dehydrogenase enzyme complex. This study represents an attempt to better characterize the mutations present in GISTs as well as to categorize tumors into prognostic treatment categories. Methods: Histologically diagnosed GISTs were selected from specimens processed at the Dartmouth-Hitchcock Medical Center (DHMC) from 2014-2016. Genomic DNA was extracted from a total of ten formalinfixed and paraffin-embedded (FFPE) tissues. Samples were multiplexed using the AmpliSeq Cancer Hotspot Panel (CHPv2) and sequenced with the Ion Torrent 318v2 chips using the Ion PGM TM System. Variants were identified using the Variant Caller Plugin (v.4.0). Results: Of the ten GIST specimens analyzed, nine had adequate DNA after extraction. The nine sequenced specimens were from the stomach (n=4, 44%), small bowel (n=3, 33%), esophagus (n=1, 11%), and shoulder (metastatic, n=1, 11%). Histologic morphologies were spindled (n=5, 56%), epithelioid (n=1, 11%), and mixed (n=3, 33%). KIT was the most commonly mutated gene (n=6, 67%); five of those mutations were in exon 11 and one was in exon 13. One specimen contained a PDGFRA mutation (11%) and two were wild-type for the sequenced genes (22%). The following KIT mutations were identified: c.1674_1676delGGT, p.K558fs; c.1671_1676delGAAGGT, p.W557_V559delinsC; c.1961T>C, p.V654A; c.1667_1674delAGTGGA, p.Q556_K558delinsQ; c.1676T>A, p.V559D; c.1729_1730insCTTATG, p.Y578_D579insAY. Two patients positive for KIT mutations also had mutations in the MET and IDH1 genes .Sample positive for the PDGFRA mutation (V561D), also had a mutation in the JAK3 gene. Conclusions: As expected, the most common mutation site was in exon 11 of the KIT gene; these tumors generally have an excellent response to Imatinib therapy. One specimen contained a mutation in exon 13 of the KIT gene, which generally show partial response to imatinib. This study demonstrates the importance of somatic mutation detection for multiple genes to better characterize the molecular profile of GISTs. Additional studies are required to elucidate other mutations present in our wild-type GIST cases. Introduction: Clinicians are increasingly relying on tumor profiling via NGS to offer targeted therapies to patients. Quality controls are fundamental for developing and monitoring these assays. However, use of cell lines, either patient-derived or generated through targeted mutation of the genome, has short-comings: the degree of multiplexing and control over allele frequencies are limited, and the background genetic material is not well characterized. We developed a unique technology for generating highly multiplexed, cellular controls for both RNA and DNA sequencing, and demonstrated their compatibility with popular tumor profiling assays. Methods: A biosynthetic construct was generated that contained portions of fifteen (15) genes with clinically actionable somatic mutations. A second construct had twelve (12) clinically actionable RNA fusions, including fusions of ALK, RET and ROS1, as well as rare fusion events. The biosynthetic DNA or RNA was introduced into GM24385 reference cell line. The engineered cells were then characterized, normalized, fixed in formalin, and embedded in paraffin for use as whole-process controls. The DNA reference material was tested by NGS using Ion Ampliseq Cancer Hotspot Panel v2. All 15 mutations were detected. The allele frequencies of mutations could be modulated by mixing engineered cells with wild-type GM24385 cells; a dilution series generated 15%, 7%, and 4% target allele frequencies (AF). Mutation frequencies for the 15% target AF material ranged from 9.6% for ERBB2 p.A775_G776insYVMA to 15.8% for EGFR p.T790M. The RNA reference material was tested by ArcherDx FusionPlex Lung Thyroid Panel, and all twelve fusions were detected as "strong evidence" fusions (per Archer Analysis Software v3.3). There was a ten-fold range observed in the detection levels: SLC34A2-ROS1 had only 34 unique, spanning reads, whereas FGFR3-TACC3 had 328. These differences in breakpoint-spanning reads on NGS were very consistent from one lot of the reference material to the next, and were also observed for digital PCR analysis. One possible explanation is sequence-specific damage caused by formalin fixation. Testing was also successful on the Thermo Fisher OncoMine Focus Assay and Ion Ampliseq RNA Fusion Lung Cancer Research Panel. Conclusions: Highly multiplexed, cellular reference materials with a well characterized background (GM24385) are needed to monitor the whole process of both RNA and DNA NGS tumor profiling assays and to aid in assay optimization and verification of lower detection limits. The ability to create multiple synthetic DNA mutation or RNA fusion targets enables clinical laboratories to ensure their assays perform appropriately even when the variants are rarely observed in the laboratory. The histologic grading of appendiceal mucinous neoplasms into low grade and high grade groups identifies patients with distinct prognosis. The molecular alterations that define these unique subsets, however, are less wellknown. Previous studies were controversial as to the role of GNAS mutations and loss of SMAD4 by immunohistochemistry in correlating with tumor grade. We explored characteristic molecular alterations between two histological groups of appendiceal mucinous neoplasm and their association with overall survival. Methods: Seventy-four surgical resections of primary appendiceal mucinous neoplasms underwent mutation analysis by next-generation sequencing between 2012 and 2015 at MD Anderson Cancer Center. These tumors, classified by degree of differentiation following the WHO grading system, consisted of 21 low grade tumors and 53 high-grade tumors. Substitutions as well small insertions and deletions in 50 cancer-associated genes were reviewed using next-generation sequencing-based AmpliSeq Cancer Hotspot panel v2. We analyzed frequently identified somatic mutations in appendiceal mucinous neoplasms and their association with tumor grade. Results: Tumor grade was significantly associated with GNAS mutations, KRAS and GNAS co-mutation, and TP53 mutations. Low grade appendiceal mucinous neoplasms harbored more frequent GNAS mutations (71%), either alone or in combination with KRAS (67%) than high-grade tumors (30% and 21%, respectively; p<0.01). In contrast, high-grade appendiceal mucinous neoplasms had more frequent TP53 mutations (34%) than low-grade tumors (0%; p=0.002). SMAD4 mutations were found in seven high-grade tumor (13%) and three low-grade tumors (14%); the frequencies of SMAD 4 mutations was not associated with tumor grade (p=1.0). Patients with KRAS and GNAS co-mutation showed a better overall survival than patients without these co-mutations (log-rank test; hazard ratio 0.164; 95% confidence interval, 0.0627 0.8765; p=0.04). Conclusions: Low grade appendiceal mucinous neoplasms have different molecular alterations such as frequent GNAS mutations and co-mutation with KRAS compared to high grade tumors, which harbor frequent TP53 mutations. Identification of these mutations in appendiceal mucinous neoplasms may be helpful as useful diagnostic markers to aid in the distinguishing of these two clinically important subsets. Introduction: NGS is used extensively in research for the molecular characterization of cancer. This has primarily focused on formalin-fixed, paraffin-embedded samples, the use of which is limited by surgical feasibility, tissue availability and patient preference. An emerging field of study has focused on circulating-free DNA (cfDNA), sometimes termed "Liquid Biopsy." These fragments of DNA are released by the tumor mass into the circulatory system, and their mutational information can often shed light on tumor characteristics. Although accepting it has the potential to be a valuable source of clinical information during diagnosis, treatment and monitoring for disease recurrence, the liquid biopsy field is still nascent, and lacks an integrated NGS workflow solution with proven performance. Methods: QIAGEN is an established expert in the field of sample preparation for liquid biopsy. The combination of the QIAamp Circulating Nucleic Acid Kit with the GeneReader NGS System and the GeneRead QIAact Actionable Insights Tumor Panel provides the first true Sample to Insight workflow solution for analyzing cfDNA. A set of 16 liquid biopsy samples from non-small cell lung cancer patients and cfDNA reference standards (Horizon), with allelic frequency at or above 1%, were used for this study. DNA was extracted from the liquid biopsy samples using the QIAamp Circulating Nucleic Acid Kit. The reference standards did not require extraction and were used to confirm detection of low frequency variants (typical of cfDNA samples). cfDNA from the liquid biopsy samples and reference standards were run through the complete QIAGEN GeneReader workflow, including a new QIAGEN Clinical Insight (QCI ) Analyze bioinformatics pipeline developed specifically for liquid biopsy applications. Results: Analysis of the 16 liquid biopsy samples for EGFR-associated mutations, designed to assess system performance on biological samples, identified seven to be EGFR mutation positive (either L858R or deletion mutations) that were verifiable using PCR based technologies. Testing of the fully characterized cfDNA reference standards identified all mutations covered by the Actionable Insights Tumor Panel, with 100% consistence with those reported in ddPCR studies. Both single nucleotide and insertion-deletion variants were correctly identified down to a 1% allele frequency threshold. Conclusions: These data show that high sensitivity and high performance consistency have been achieved with both cfDNA reference standards and liquid biopsy samples with the complete QIAGEN GeneReader workflow for liquid biopsy. This is the first study of its kind to systematically demonstrate the accuracy of a liquid biopsy assay combined with the robustness of a fully integrated NGS system. Introduction: Pancreatic ductal adenocarcinoma (PDAC) is an aggressive disease, often presenting with metastasis at diagnosis. Several genetic mutations as well as alterations of miRNA expression have been described in PDAC, however it is unclear how these alterations contribute to metastatic disease. Our goal in this study was to analyze and compare miRNA expression and NGS sequencing results in metastatic and non-metastatic PDAC. Methods: PDAC samples from twenty-four patients were analyzed. Five cases were excluded because of inadequate tissue. Eleven cases were from patients with metastatic disease (M:F = 2:9; age 60.4±8.1 years). Eight cases were from patients without metastasis (M:F = 1:1; age 73.1±6.9 years). For each sample, seven unstained FFPE tissue sections were used for miRNA extraction using the miRNeasy FFPE Kit (Qiagen). Expression of miRs was evaluated using the TaqMan MicroRNA Reverse Transcription Kit (Applied jmd.amjpathol.org ■ The Journal of Molecular Diagnostics Biosystems) with primers specific for RNU6B (internal control), miR-10b, miR-21, miR-148a, miR-196a and miR-217. Real-time PCR was performed on the AB 7500 Fast Real-Time PCR System. The expression level of each miRNA was calculated using the formula log2 (2 -). For NGS, DNA was extracted from eight FFPE slides, and quantified using the PicoGreen method. Sequencing was then performed using the IonTorrent PGM on 318v2 chips. Variants were annotated using GoldenHelix SVS software 8.3.4. Results: Metastatic sites included lung (4 cases), liver (3 cases), abdominal wall/peritoneum (3 cases), and bone (1 case). The miRs in both metastatic and non-metastatic groups had similar expression patterns: miR-10b (underexpressed), miR-21 (overexpressed), miR-148a (inconsistent), miR-196a (underexpressed), and miR-217 (underexpressed). Seven cases failed QC for NGS, three due to DNA concentration and/or quality, and four failed qPCR. Like miR expression, NGS results were similar among metastatic and non-metastatic groups. Tumors from both groups contained KRAS (5 cases), TP53 (4 cases), CDKN2A (2 cases), and APC (2 cases). One case of non-metastatic disease showed a RET mutation. Of metastatic disease, one case each showed a SMAD4 mutation, and a JAK3 mutation. Three cases were wild-type. Conclusions: Expression patterns of miR-10b, miR-21, miR-148a, miR-196a, and miR-217 in PDAC were not predictive of distant metastasis. Similarly, NGS results revealed a variety of variants in both patients with and without metastasis. No correlation was seen between variant(s) and miRNA expression pattern. Studies evaluating the genetic basis for metastasis may lead to a better understanding of the pathogenesis and prognosis of pancreatic cancer. Conformation Analysis (CE-SSCA) has been used as a clinical diagnostic assay at the Royal Marsden Hospital (London, UK) for melanoma patients in over 6,000 specimens since 2011. The aim of the study was to assess the performance of the BRAF/NRAS Mutation Test in unselected archived samples and to compare agreement between the two methods. Methods: BRAF exon 15 and NRAS exons 2 and 3 were amplified in a multiplex fluorescent PCR followed by CE-SSCA using an ABI Genetic Analyzer 3130XL. This method has shown a sensitivity of >96% for the common BRAF and NRAS mutations described in malignant melanomas and has a turn-around-time of 4.5 working days for up to 93 samples with a limit of detection between 5-15%, depending on the specific variant. 233 of the 297 samples tested using the BRAF/NRAS Mutation Test were prepared on a Hamilton Starlet liquid handler. The BRAF/NRAS Mutation Test is based on allele-specific, real-time PCR and has been developed for formalin-fixed paraffin-embedded (FFPE) tissue -around-time is about 3 hours for 30 samples. Results: DNA from 304 samples was analyzed by both the BRAF/NRAS Mutation Test and CE-SSCA. Discordant specimens were retested using Illumina's TruSeq protocol. The frequency f BRAF and NRAS mutations was 28.6% and 25.7%, respectively, using the BRAF/NRAS Mutation Test. Seven samples had rare mutations detected by CE-SSCA but were out of scope for the BRAF/NRAS Mutation Test. Seven samples were invalid by the BRAF/NRAS Mutation Test and nineteen by CE-SSCA. The overall agreement between the two methods was 98.6% (273/277). The positive percent agreement (PPA) was 99.4% (165/166), and negative percent agreement (NPA) was 97.3% (108/111). The overall call rate with all samples was 93.3% (277/297). Using Cohen's kappa coefficient the agreement was 0.885. Conclusions: We showed excellent concordance between the Roche BRAF/NRAS Mutation Test (LSR) and SSCA. The BRAF/NRAS Mutation Test offers a sample-to-answer solution, including sample preparation, in 8 hours or less. Furthermore, a cloud-based analysis tool provides a user-friendly data analysis experience and delivers straightforward results in minutes. The mutational status of colorectal carcinoma tumors can provide vital information concerning the prognosis, clinical behavior, and treatment response to specific chemotherapeutic agents. The KRAS and NRAS genes, which are associated with the RAS/RAF/MEK/ERK signaling pathway, are considered consistent predictive biomarkers for monoclonal antibody therapies targeting the epidermal growth factor receptor (EGFR). Traditionally, the efficacy of anti-EFGR therapy was thought to be dependent upon the presence of a wild-type KRAS. However, recent studies have shown that the specific KRAS mutation present may play a more important role in predicting tumor response to EGFR targeted treatments. The value of extended NRAS mutational analysis has not yet been established. The goal of this retrospective study was to identify the specific mutational composition of the RAS gene family (KRAS, NRAS, and HRAS) in primary and metastatic colorectal carcinoma. Methods: Tumor specimens from patients diagnosed with colorectal carcinoma between May 2013 and December 2015 were selected for analysis. Genomic DNA was extracted from formalin-fixed, paraffin-embedded (FFPE) tissues processed at the Dartmouth-Hitchcock Medical Center (DHMC). DNA specimens were multiplexed and sequenced on the Ion Torrent 318v2 chips using the Ion PGM System. Next-generation sequencing was performed using a 50 gene hotspot panel (AmpliSeq Cancer Hotspot Panel v2, Life Technologies) which includes the KRAS, NRAS, and HRAS oncogenes. Variants were identified using the Variant Caller Plugin (v.4.0), and annotation and functional predictions were performed using Golden Helix SVS (v.8.3.4) . Results: We identified a total of 155 colorectal carcinoma tumors harboring a RAS mutation. The most commonly mutated gene was KRAS 93%, followed by NRAS 8% and HRAS 0.6%. KRAS mutations were most prevalent in exons 12 (G12D 24%, G12V 23%) and 13 (G13D 17%). The most common NRAS mutations were identified in exons 12 (G12D 23%) and 61 (Q61H 23%). Only one HRAS mutated tumor was identified. Conclusions: Our study indicates that significant variability exists between the RAS mutational profiles of colorectal carcinoma tumors. Further studies to determine the clinical significance of these molecular features are warranted. Introduction: Approximately 8% of non-small cell lung cancer cases feature oncogenic fusions or alternative splicing of cancer-related genes which are amenable to on-market or emerging targeted therapies. Established methods to detect these variants have complex multi-step workflows, restricted target coverage or resolution, and/or limited sensitivity. A new product, the QuantideX NGS RNA Lung Cancer Kit, has potential to address these gaps by covering 110 known oncogenic gene fusions, MET exon 14 skipping, 23 expression markers, and 5 imbalance markers using a streamlined sample-to-answer workflow. In this study, we report consistent results obtained using this technology with well-annotated sample sets in an independent evaluation. Methods: A cohort of 30 study samples were prepared from a collection of total nucleic acid (TNA) isolates derived from formalinfixed, paraffin-embedded (FFPE) residual tumor biopsies and cancer celllines. Target enrichment was performed at Jewish General Hospital and Asuragen on de-identified samples using the QuantideX NGS RNA Lung Kit reagents (Asuragen) and controls, and sequenced on the MiSeq platform (Illumina). Analyses were conducted using QuantideX Reporter (Asuragen), a software suite that includes a FASTQ processing pipeline and incorporates pre-analytical QC information with a fusion-caller algorithm and reporting tool. Results: Targeted RNA-Seq results demonstrated 100% agreement for fusion and splice variant calling across 11 ALK fusions, 1 ROS1 fusion, 1 FGFR3 fusion, 1 MET exon 14 skipping and 16 negative samples. The detection of 3'/5' ALK expression imbalances in known fusion-positive samples indicated potential utility for the detection of fusions not explicitly targeted by the panel. ALK rearrangements were detected down to an input of 400 referencegene amplifiable copies, equivalent to <10 ng of TNA. mRNA expression profiles for the complete set of expression markers covered by the panel were highly correlated (R2>0.98) between intra-and inter-site replicates. Bench procedures from reverse transcription to sequence-ready libraries could be completed in just over 9 hrs. Conclusions: The accuracy of a novel targeted NGS assay for RNA fusions in lung cancer was demonstrated in an independent laboratory evaluation using clinicallyrelevant specimens and low inputs of TNA. The ability of the panel to detect both common and rare gene fusion transcripts and exon splicing events within an integrated wet-bench workflow and companion bioinformatic software provides a foundation for the reliable detection of oncogenic RNA fusions and aberrant splicing events that respond to current and emerging targeted therapies. Introduction: Colorectal carcinomas can be separated into two broad categories; 1) those that are microsatellite unstable and associated with hypermutation, and 2) those that are microsatellite stable, chromosomally stable, and associated with nonhypermutation. Amongst the non-hypermutated tumors, studies have reported a striking similarity in the mutational profiles of colon and rectal primary tumors. For this reason, these tumors are typically grouped into a single entity, colorectal The Journal of Molecular Diagnostics ■ jmd.amjpathol.org carcinoma, for molecular purposes. The Cancer Genome Atlas Network identified the most frequently mutated genes in colorectal carcinoma: APC (81%), TP53 (60%), KRAS (43%), TTN (31%), PIK3CA (18%), FBXW7 (11%), SMAD4 (10%), NRAS (9%), and TCF7L2 (9%). Despite their genetic similarities, the clinical management of colon and rectal carcinomas differ due to anatomical location and the higher incidence of local recurrence in rectal tumors. The goal of this retrospective study was to evaluate primary rectal carcinomas and identify mutational characteristics that may differentiate them from colon carcinomas. Methods: Tumor specimens from patients diagnosed with rectal carcinoma between the years of 2013 and 2016 were selected for analysis. Genomic DNA was extracted from formalin-fixed, paraffin-embedded (FFPE) tissues. DNA quantification and quality were assessed using dsDNA Qubit and KAPA hgDNA Quantification and QC, respectively. Barcoded libraries were multiplexed and sequenced on Ion Torrent 318v2 chips using the Ion PGM System. Variants were identified using the Variant Caller Plugin (v.4.0), and annotation and functional predictions were performed using Golden Helix SVS (v.8.3.4) . Results: A total of 66 rectal carcinomas were identified for evaluation. Mutations were most commonly identified in the following genes: TP53 (74%), APC (52%), KRAS (32%), FBXW7 (13%), MET (6%), and PIK3CA (5%). Of the 20 KRAS mutations identified, the most prevalent were in exons 12 (G12D 15%, G12V 15%) and 13 (G13D 20%). Conclusions: Our study indicates that the primary rectal carcinomas encountered at our institution have a mutational profile similar to that previously described in colorectal carcinoma. However, in primary rectal carcinoma, the frequency of mutations in the APC and TP53 genes are inversely proportional to that reported in the combined classification of colorectal carcinomas (p<0.001). Further investigation into the possible clinical significance of this finding is warranted. Y. Lin 1 , R. Yu 2 , W. Hsu 1 , Y. Liu 3 , K. Chiu 2 , Y. Yeh 1 1 General Biologicals Corporation, Hsinchu, Taiwan; 2 Genetics Development Corporation, Lake Bluff, IL; 3 Curiemed Corporation, Hsinchu, Taiwan. Introduction: Determination of HER2 status before administrating therapeutic regimens plays an essential role in breast cancer. Immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) are currently working assays for HER2 testing. With the advantages of cost-effectiveness and high accuracy for IHC and FISH respectively, these two assays have been recommended as the routinely-used diagnostic tool. However, several studies have been reported that false signals and lab-to-lab variations are major unsolved issues for HER2 testing. Moreover, both IHC and FISH assays do not have the internal specimen quality control mechanism, which might be the root-causes of the issues that we're facing. Methods: In this study, we developed and evaluated a quantitative HER2 testing assay with built-in internal specimen quality control mechanisms by the principle of multiplex real-time PCR. HER2 messenger RNA copy number could be quantified in real-time using external standard curve. The expression level of reference genes were utilized for monitoring the quality of specimens. Eighty FFPE tissues of breast invasive ductal carcinoma were collected and analyzed for the assay validation. Results: The dynamic range of the assay for detecting HER2 mRNA is 5.2E+06 ~ 5.2E+01 copies/mL and showed excellent linearity (R2=0.98). Also, the precision of the assay was demonstrated with inter-operator coefficient of variance (CV) <10%. Most importantly, the validation result showed that the agreement between the developed assay and FDA certified FISH assay is over 90%, suggesting the developed realtime PCR is suitable and reliable for HER2 testing. Furthermore, the quality of FFPE specimens can be distinguished simultaneously with an established cut-off value, which can prevent false signals or lab-to-lab variations that resulted from poor qualified samples. Conclusions: Taken together, the clinical utility of the developed HER2 molecular diagnostic assay shows great FFPE specimen quality control ability and high analytical consistency with current assays for HER2 testing. The ease of use and broad dynamic range makes the developed assay a viable tool for implementation into clinical labs, which may complement IHC tests. Anaplastic Lymphoma Kinase (ALK) represents the only 'druggable' target with a mutation rate of 9-10% in hNB cases. ALK has been implicated as a driver oncogene in a variety of cancer types. As an oncofetal antigen, ALK is an ideal therapeutic target with minimal side effects and it has been simple to target with small molecules. However, resistance to ALK-targeted therapy in other cancers have already been noted. To proactively identify and develop strategies to counter these resistance mechanisms to ALK inhibitors, the present study investigated putative resistance mechanisms identified from our unpublished CRISPR-based genomewide knockout (KO) or overexpression (OV) screens in human NB cell lines. Methods: CRISPR-based whole-genome screening libraries (GeCKO for KO and SAM for OV) were obtained from Zhang Lab at MIT. CRISPR viral libraries were then transduced into SHSY-5Y (ALK+ human NB line) cells after which the cells were cultured in the presence of ALK inhibitors (AP26113 or LDK-378) for 13 days. Enriched CRISPR constructs were identified via HiSeq (RapidRun 100 cycles) of the PCR amplified sgRNAs using custom NGS primers. Resulting FASTQ files were processed and analyzed with MaGECK to identify specific genes/pathways that may induce resistance to ALK inhibitors. Functional validations on each 'hit' identified from the screens were performed in 20 genetically distinct NB cell lines. Results: The KO screen identified 39 candidate genes whereas the OV screen identified 36 candidate genes capable of inducing resistance to ALK inhibitors. The candidate lists included previously reported mechanisms such as RAS pathway genes and mTOR pathway genes. But the lists mostly included novel candidates and Kaplan-Meier analysis (5-year event-free survival) based on publicly available NB patient expression array datasets (n>600) revealed many of those novel candidates (AASDHPPT, CRTC2, TAF12, ETV1, TXLNG, etc.) to be significant (p<0.05) prognostic markers in NB. Some of the putative resistance mechanisms have provided novel therapeutic strategies as PIM1 inhibitors (AZD-1208) produced highly synergistic inhibitory effects against NB cells when applied with ALK inhibitors at low concentrations. Conclusions: The present study produced a list of genes capable of inducing resistance to ALK inhibitors in vitro. As indicated above, this list may be useful as a novel prognostic marker panel for hNB patients and also as a guide to novel therapeutic strategies to ultimately extend the survival periods for the hNB patients. A.M. McDonald, N. Fadra, J.I. Davila, A.A. Nair, J. Jen, R.B. Jenkins, B.R. Kipp, J.L. Winters, B.R. Crusan, W.E. Highsmith, K. Rumilla, J.S. Voss, X. Wang, E.W. Klee, R.N. Wehrs, K.C. Halling Mayo Clinic, Rochester, MN. Introduction: RNASeq is a powerful method for detecting fusions in tumors and can be useful in establishing a diagnosis, determining prognosis and guiding targeted therapy. In this study we describe the verification of an RNASeq assay for the detection of fusions in solid and hematologic tumors. Methods: Total RNA was extracted from solid tumor tissue using a Qiagen miRNeasy Micro kit or whole blood using a Qiagen miRNeasy Mini kit. RIN values were determined using an Agilent 2100 Bioanalyzer and RNA quantity was determined using a Qubit 2.0 fluorometer. Poly-A mRNA selection and cDNA synthesis were performed on a Janus Automated Workstation (Perkin Elmer) and NGS libraries were prepared with a Biomek FXp Liquid Handler (Beckman Coulter) using custom-built protocols and a TruSeq RNASample Preparation v2 Kit (Illumina). Paired-end, 101 bp-read sequencing was performed on a HiSeq 2500 (Illumina) in Rapid Run mode. Data analysis utilized a custom suite of alignment, fusion detection, filtering, and annotation programs to detect fusions in 558 genes. Three known fusion negative tumors and 16 tumors with 15 unique fusions (previously identified by FISH, RT-PCR, or RNA Seq) and 7 normal tissue specimens were analyzed to assess the sensitivity and specificity of the assay. Four samples were analyzed for intra-and inter-assay reproducibility. Seven tumors were used to assess the assay's analytical sensitivity based on the total input mRNA, tumor percentage and total reads per sample. In addition, the impact of RNA degradation on the ability to detect fusions was assessed. Results: Gene fusions were detected in 14/16 tumors with known fusions and 0/10 normal or known fusion negative tumor tissues for a sensitivity of 88% and specificity of 100%. Intra-and inter-assay reproducibility studies revealed 100% concordance for the presence or absence of a fusion in all replicates. Limit of detection thresholds were found to be 62.5ng of input RNA, 25% tumor and 50 million total reads to yield successful fusion detection. Receiver operator curve analysis revealed that 5 or more supporting reads provided the optimal test sensitivity and specificity. We observed that RNA degradation is an important factor found to reduce the ability to detect fusions with a higher impact as the distance of the fusion from the 3' end of the mRNA increases. Conclusions: These studies provide valuable information on the performance characteristics of RNA Seq for the detection of fusions in solid and hematologic tumors. An understanding of the strengths and limitations of this assay will aid in the appropriate interpretation of RNA Seq results obtained during clinical testing. R. Luthra, R.R. Singh, A.J. Lazar, W. Wang, J.M. Meis, K.P. Patel, M.J. Routbort, D.D. Panditi, J. Yau, D.Z. Hatfield, J. Medeiros, R. Luthra University of Texas MD Anderson Cancer Center, Houston, TX. Introduction: Sarcomas are a rare heterogeneous group of mesenchymal tumors, which can be challenging to diagnose. Fortunately, a sizeable portion of sarcomas are driven by gene translocations (30% to 40%) and can be useful in classifying sarcomas. A sensitive assay that can identify gene translocations in sarcoma in a timely fashion is of high diagnostic value. In a clinical diagnostic laboratory, we have designed and validated a nanofluidics-based high-throughput RT-PCR platform to screen for fusions in 12 genes using a single input of FFPE RNA from sarcoma tumor specimens. Methods: Taqman-based real-time PCR assays (ThermoFisher Scientific) were designed to cover 38 exonic fusions in 12 genes. GAPDH was used as an RNA-expression control. Synthetic control plasmids with the fusion sequence jmd.amjpathol.org ■ The Journal of Molecular Diagnostics interrogated by the 38 assays were custom ordered (Integrated DNA Technologies) and used to optimize the assays. Sarcoma specimens (n=35), two cancer cell lines (A4573 and RD-ES) and 3 normal samples were used. The tumors were positive for fusions involving genes such as SS18, EWSR1, NAB2, HEY1, PAX3 as tested by RT-PCR, FISH or predicted by the tumor sub-type. RNA was extracted from the FFPE tissue slides using the AllPrep DNA/RNA FFPE kit (Qiagen) and cDNA was prepared using SuperScript VILO kit, (ThermoFisher Scientific). On each Nanofluidics PCR 48X48 Fluidigm chips, 15 samples (14 tumors and A4573 positive control) were simultaneously tested for 38 fusions in triplicates on BioMark HD instrument. Limit-of-detection and reproducibility studies were performed using sequentially diluted FFPE tumor and A4573 cell line RNAs into normal FFPE RNA. Results: The 38 assays were tested and optimized using the synthetic plasmid controls. Out of the 35 samples tested, expected fusions were detected in 32 samples and the two cell lines. In the remaining 3 samples, two were sub-optimal specimens (low RNA and/or decalcified). No fusions were detected in the normal samples. Considering the results from all positive samples, 250ng RNA input for analysis and a Ct value of 20 or lower was found to be optimal cutoff for positive fusion calls. RNA dilution studies demonstrated robust detection sensitivity and intertech and inter-run reproducibility. Conclusions: We have optimized and validated a nanofluidics-based PCR platform which is highly suited for routine high-throughput screening of multiple gene fusions in sarcomas with a single input of FFPE RNA in a clinical diagnostic laboratory. T. Sakamoto 1,2 , S. Matsumoto 2 , T. Saito 2,3 , A. Tsuyada 3 , K. Tsuchihara 2 1 Tottori University, Yonago, Tottori, Japan; 2 National Cancer Center, Kashiwanoha, Kashiwa Chiba, Japan; 3 Riken Genesis Co., Ltd., Tsurumi, Yokohama Kanagawa, Japan. Introduction: Detection of driver gene alterations including gene rearrangements is an essential step for a treatment decision of advanced non-small cell lung cancer. However, genetic testings generally require invasive biopsy. Liquid biopsy is a promising technology which can detect gene alterations with minimal invasions. Circulating cell-free tumor DNA (ctDNA) has been revealed to be useful to detect cancer gene alterations. However, it is difficult to accurately detect gene rearrangements by ctDNA based liquid biopsy even though various methods have been tried. On the other hand, there is cell-free tumor RNA in circulation (ctRNA) as is the case with ctDNA. Therefore, we aimed to develop high sensitivity detection methods for EML4-ALK fusion using ctRNA. Methods: We developed two different high sensitivity detection methods of EML4-ALK fusion based on RT-PCR. The first was multiplexed RT-PCR with fusion specific hydrolysis probes combined with preamplification (OneTube assay). The second was detection of 3' ALK mRNA with nested RT-PCR (NestedPCR assay). Blood samples were collected from patients with EML4-ALK positive lung adenocarcinoma with EDTA blood collection tubes and then centrifuged. Blood plasma samples were stored at -80°C until use. RNA extracted from cell-free culture supernatant of H2228 cell line was used as positive control. RNA was extracted from 1mL stored plasma with Plasma/Serum RNA Purification Midi Kit (Norgen). Results: OneTube assay accurately detected EML4-ALK fusion from diluted samples which contained approximately 2.5 copies of cellfree culture supernatant RNA. OneTube assay was considered to have the ability to detect 1 copy of fusion mRNA considering sample concentration inhomogeneity. NestedPCR assay had equivalent sensitivity with OneTube assay. Both methods could detect EML4-ALK fusion from culture supernatant ctRNA although they could not detect from six clinical plasma samples. Conclusions: We succeeded in development of 2 different high sensitivity detection methods of EML4-ALK fusion. However, we could not detect EML4-ALK from plasma ctRNA. It is expected that there is no or very little amount of ctRNA in plasma samples. Development of a novel method enriching ctRNA is needed to detect gene rearrangements by liquid biopsy with plasma ctRNA. Introduction Several guidelines for metastatic colorectal cancer currently recommend genotyping for RAS (KRAS/NRAS) and BRAF mutations to select for patients eligible to anti-EGFR antibody therapy. PIK3CA mutations have been associated with resistance to anti-EGFR therapy in several studies. The frequencies of mutation estimated for KRAS (around 40%), NRAS (2% to 15%), PIK3CA (14.5%) and BRAF (5% to 10%) vary according to methodology and population. In Mexico, colorectal cancer represents the third most common cause of cancer with an estimated incidence of 3.3x100,000 individuals, around 30% presenting as stage IV disease at diagnosis.No data published currently exist to document the status of relevant biomarkers for colorectal cancer therapy in Mexican population.The aim of the present study was to determine the status of these biomarkers in a cohort of patients referred to the Cancer Center of the American-British Cowdray Medical Center in Mexico City. Methods: We performed a retrospective search on the archives of the Pathology Department for cases diagnosed with colorectal adenocarcinoma with KRAS/NRAS/BRAF determination from January 2013 to May 2016. DNA was extracted using QIAamp DNA FFPE kit (Qiagen) and biomarker assessment was performed with Therascreen KRAS, BRAF and PI3KCA kits (Qiagen) using Rotor-Gene Q Thermocycler. NRAS determination was done with Sanger sequencing using consensus primers to amplify regions adjacent to codons 12, 13 and 61 in a 3500 Genetic Analyzer (Applied Biosystems). Results: We retrieved 99 cases, all of them with KRAS determination (59% wild-type/41% mutated). Mutation frequency by type is as follows: c.35G>A(p.G12D)14%, c.35G>T(p.G12V)9%, c.38G>A(p.G13D)7%, c.34G>T(p.G12C)4%, c.34G>A(p.G12S)3%, c.35G>C(p.G12A) and c.34G>A (p.G12R)2% respectively. Of the whole population, 33.3% (n=33) had NRAS assessment. No NRAS mutations were identified. Twenty-four percent (n=24) of the cases had additional BRAF mutation determination, only two cases (2% overall) had a c.1799T>A(V600E) mutation, one of them with a concurrent KRAS mutation c.35G>T(p.G12V). PI3KCA mutation assessment was performed in 14 cases with sufficient DNA to study, 28.6% were mutated, 2 cases with exon 20 mutation c.3140A>G (H1047R)(2% overall) and 2 with exon 9 mutation c.1624G>A (E542K)(2% overall). Conclusions: In our study, KRAS mutation profile of colorectal adenocarcinoma reflects what has been found in other populations around the world. The stepwise-approach of sequentially extend mutation assessment (KRAS-NRAS-BRAF) (based on being mutually exclusive) as well as biased biomarker selection (NRAS versus BRAF), may not fully reflect mutation prevalence. International consensus guidelines to extensively address colorectal predictive biomarker testing are needed. A. Bergamaschi, K.M. Clark-Langone, D.A. Eberhard, J. Han, R. Loverro, P. Harrington, C. Ku, Y. Ma, W. Gibb, A. Dei Rossi, L. Shen, A.D. Goddard, K.M. Clark-Langone, J.M. Anderson, G. Alexander Genomic Health, Inc., Redwood City, CA. Introduction: There is increasing interest in the field of molecular diagnostics to perform liquid biopsies rather than tissue biopsies when tissue is unavailable, or when invasive procedures are associated with high risk to the patient. The 17-Gene Liquid Biopsy Mutation Panel is a Next-generation sequencing (NGS) assay, designed to detect clinically actionable variants in circulating tumor DNA (ctDNA) in blood. Prior to clinical use, analytic performance must be demonstrated to meet the intended use of the assay, and thus requires using clinically relevant variants at levels representative of ctDNA. Methods: Analytical specificity and sensitivity (collectively representing analytical accuracy) were characterized through determination of Limit of Blank (LoB) and Limit of Detection (LoD), respectively. LoB was determined using 73 cell-free DNA (cfDNA) samples from 60 healthy donors, whereas LoD (defined as minimum allelic fraction (AF)/ copy number (CN) required for a 95% detection rate) was determined using a model system of titrated cell line DNA harboring clinically actionable variants. Within and between assay runs, reproducibility was assessed across replicate assays of cfDNA pools derived from cancer patients. The impact of high molecular weight DNA (HWW-DNA) as a potential interfering substance was assessed using 2 approaches; stored blood from healthy controls and HMW-DNA spiked into variant positive samples. Results: Detection thresholds were set above the LoB corresponding to >99% per sample specificity. Mean LoDs on a per sample basis were as follows; single nucleotide variants, 0.56% AF; insertions/deletions, 0.19% AF; fusions, 0.37% AF, and copy number gain, 2.7 copies. The assay was highly robust, demonstrated by detection of >95% of variants near the LoD in 2 cfDNA pools. Increasing amounts of HMW-DNA proportionally impacted intermediate analytical metrics, however, sensitivity of detection of SNVs, indels and SVs was relatively unaffected. For CNV, a decrease in specificity was apparent at longer blood storage times and with increasing amounts of spiked HMW-DNA. Conclusions: Here we report the analytic performance of the 17-Gene Liquid Biopsy Mutation Panel on a per sample basis. In addition to being highly sensitive and specific, we demonstrated that the 17-Gene Liquid Biopsy Mutation Panel yields reproducible results. Finally, we illustrate the importance of characterizing the impact of HMW-DNA resulting from blood storage time on the detection of clinically actionable variants. D. Wieczorek, S. Hermanson, C. Knox, J. Mook, D. Horejsh, E. Vincent, D. Storts, T. Schagat Promega, Madison, WI. Introduction: Quantity and quality of DNA from formalin-fixed, paraffin-embedded (FFPE) tumor tissue samples is highly variable, with degradation and crosslinking due to the fixation process leading to issues with amplification and difficulty in NGS analysis. An alternative to FFPE is circulating cell-free DNA (ccfDNA) from plasma or other biological fluids. Compared to gDNA, ccfDNA yields are typically low, with tumor cell present at significantly lower frequencies. Due to the inherent variability of FFPE and ccfDNA, knowing the quantity of DNA is not in itself reliably predictive of downstream NGS success. In this poster, we describe novel methods for predicting sequencing result quality utilizing a multiplexed qPCR assay. Methods: DNA was purified from four matching tumor and normal FFPE tissue types as well as ccfDNA from plasma samples using multiple methods. DNA quantity was measured via single-target qPCR and used for downstream NGS library construction with a 56 gene oncology panel and subsequent data analysis. Discrepancies between quantity The Journal of Molecular Diagnostics ■ jmd.amjpathol.org of DNA input into library preparation and expected library yield and sequencing coverage uniformity were noted. To investigate if downstream library yield and sequencing quality could be better predicted, a multiplexed qPCR assay was designed that included three different amplicon sizes (75, 150, and 300bp) . The quantitative differences between the increasingly larger amplicon sizes were calculated as a ratio to determine the level of degradation of the DNA from FFPE samples. ccfDNA fragments cluster around 170bp; thus, the 150bp target closely estimates ccfDNA concentration. Since gDNA is expected to be much larger in size, the ratio of 150bp to 300bp targets can help predict the ratio of ccfDNA to gDNA. FFPE samples with high 75/300bp ratios are indicative of highly degraded samples, and ccfDNA samples with low 150/300bp ratios are indicative of gDNA contamination. Results: Retroactive testing with the multiplexed qPCR assay showed a strong correlation of degraded DNA to low library yield and low coverage uniformity. Samples with equal concentration and library input performed much better when degradation ratios were low. Conclusions: Data derived from a multisize target qPCR assay can be very effective in predicting downstream NGS success. Using such a QC method can drive researchers to triage samples and make informed decisions about what downstream library method to use. Concentrating on less complex panels or ddPCR vs. highly multiplexed panels or whole exome sequencing for degraded samples can ensure getting the most useful information out of an individual sample, thus saving time, cost, and loss of information about precious samples. A. Olar, R. Broaddus, M. Routbort, S. Roy-Chowdhuri, K.B. Hodges, R.R. Singh, A.J. Lazar, A. Rashid, H. Chen, A. Yemelyanova, J.L. Medeiros, S.R. Hamilton, R. Luthra, K.P. Patel University of Texas MD Anderson Cancer Center, Houston, TX. Introduction: Broader genomic sequencing by NGS panels pose a unique challenge for oncologic somatic mutation analysis due to identification of variants for which the somatic versus germline origin cannot be definitively determined. The higher number of calls resulting from broader coverage further increases the complexity of data analysis. Bioinformatics tools and public databases can help to a certain extent in identification of germline variants, but, are not perfect. In comparison, a paired normal sample allows definitive identification of variants of germline origin, but, poses challenges in collection, analysis and throughput. We evaluated the value of paired tumor-normal sample analysis in solid tumor genotyping. Methods: NGSbased targeted genotyping of 134 genes (Oncomine Comprehensive Assay, ThermoFisher) was performed on paired tumor-normal (T/N) samples from 103 solid tumors. Sequencing data analysis was performed by TorrentSuite software. Variant calls and coverage information were evaluated using an in-house developed genomic analysis and reporting tool (Oncoseek) and Integrative Genome Viewer (Broad Institute). A combination of variety of database-level and matched T/N filters were applied to identify optimal analysis workflow. We compared the analysis and reporting workflow for the same set of samples under tumor-only (T) vs. paired T/N scenarios.Results: Average number of total variants called per sample was 214.5 (range=141-432;median=208). In a tumor only scenario, database-level filters reduced the average number of calls to 8.6 (range=2-21, median=8). In paired T/N scenario, the germline filter alone reduced the average number of variant calls to 30.6 (range=8-227; median=23) (p<0.0001). The number of germline variants that were filtered out ranged from 130 to 249 (mean=183.85; median=183). A combination of germline and database-level filters in paired T/N setting significantly reduced the average calls for review per sample to 3.6 (range=0-17, median=3) in comparison to T-only or T/N with only germline filters. The final reported somatic mutations averaged 2.8 per sample (range=0-14; median=2) after pathologist's review. As expected, polymorphisms were common in BRCA1 and BRCA2; however, 8 of 118 variants were determined to be of somatic origin, which may be relevant for ongoing clinical trials. Conclusions: Database-level (population and laboratory specific) filters are efficient in identification of majority of artifacts, polymorphisms, non-coding and synonymous variants. The addition of paired normal analysis further improves the somatic mutation analysis significantly by eliminating ambiguity in reporting of calls where the determination of germline versus somatic origin in not feasible through informatics tools alone. Introduction: We previously defined 3 molecular subgroups of High Grade Serous Ovarian Cancer (HGSOC), using gene expression data from 265 FFPE samples obtained from treatment naive patients who received platinum based treatment following surgical resection. The 3 molecular subgroups were Angio: characterised by upregulation of angiogenesis genes; Immune: characterised by upregulation of immune genes and Angio_Immune: characterised by upregulation of angiogenesis and immune genes. Patients within these 3 subgroups respond differently to standard of care treatment with the Immune subgroup having best prognosis and Angio and Angio_Immune groups having similar worse prognosis. A weighted gene signature to identify each of the molecular subgroups was developed. This dataset and associated subgroup designation was used to investigate the effect of chemotherapy on molecular subgroup designation. Methods: To investigate the effect of chemotherapy on predefined molecular subgroups we analysed 35 matched pre-and post-chemotherapy samples by gene expression. The samples were scored with each of the 3 gene signatures and assigned into molecular subgroups. Novel cisplatin resistant HGSOC cell lines were generated to study mechanisms of acquired resistance. Results: Forty-four percent of the treatment naive samples were aligned with the Angio_Immune subgroup and this increased to 71% postchemotherapy. This was especially evident in the Immune group where 67% of cases associated with this good prognosis group pre-chemotherapy, were associated with the bad prognosis Angio_Immune group, post-chemotherapy. Hence platinum chemotherapy selects for the Angio_Immune group, suggesting that this subgroup represents tumours which are innately platinum resistant but also provides a mechanism of acquired resistance. Additionally we demonstrate that the Angio_Immune group is driven by activation of the MAPK pathway and show that cisplatin resistant HGSOC cell lines are specifically sensitive to MEK inhibitors using trametinib suggesting an approach to treating this patient group. Conclusions: The MAPK pathway is a mechanism of innate and acquired platinum resistance in HGSOC. Furthermore the data suggests that original pre-treatment surgical/biopsy samples may fall within a different molecular subgroup to samples taken postplatinum therapy. Introduction: Next-generation sequencing is replacing traditional molecular methods in clinical laboratories to simultaneously detect multiple genetic alterations. At the same time, bioinformatics challenges for handling raw sequence data, through to clinical decision support, must be overcome. We evaluated the Illumina TruSeq Amplicon Cancer Panel (TSACP), along with bioinformatics softwares from Illumina and Philips in our clinical laboratory. Although we found similar sensitivity and specificity to traditional methods, with little difference between the bioinformatics pipelines for detection of most variants, we encountered lower concordance for larger (>15bp) InDels between both analysis pipelines and ground truth. Post variant detection, we have implemented a decision support workflow to streamline variant filtering and annotation, provide assistance with on-and out-of-indication therapy, reporting and clinical trial matching. Although we have not formalized the evaluation of the clinical decision support workflow, our initial experience leads us to believe it will also improve overall laboratory efficiency and patient experience. Methods: In this study, 55 clinical specimens (formalin-fixed, paraffin-embedded; peripheral blood; or bone marrow), 7 human cancer cell lines and HapMap DNA NA12878 samples were used for TSCAP validation on an Illumina MiseqDx. All of these samples were well characterized by traditional methods with mutations found in 7 genes (KRAS, BRAF, EGFR, JAK2, NPM1, FLT3 and IDH1) . All samples successfully processed with the TSACP and sequenced in 4 sequencing runs. The data were analyzed by both VariantStudio and Philips pipeline. The Philips pipeline resulted from indepth comparison of four different variant callers. Results: TSACP was highly repeatable and reproducible with 5% analytic sensitivity. Except for 10 amplicons, all 212 amplicons had >X500 average coverage in all runs. Both VariantStudio and Philips pipeline showed 100% concordance with clinically validated results for point mutation and smaller InDels (<15bp). For larger InDels (>15bp), the detection rates were 56%, 62% and 81% for VariantStudio, Philips pipeline and both methods, respectively. Conclusions: TSACP is a sensitive, reliable pan-cancer mutation panel for sequencing of cancer hot-spot mutations and could be used in clinical laboratory to replace previous method. All bioinformatics softwares demonstrated excellent performance for point mutations and smaller InDels. Howerver, large InDels are much more difficult to be detected. A combination of different data analysis softwares may increase the detection rate for large InDels. In non-small cell lung cancer (NSCLC), the canonical EML4-ALK inversion results in a fusion protein with a constitutively active ALK kinase, and is associated with response to targeted inhibitors such as crizotinib. Clinically, anti-ALK therapy is initiated based on evidence of an ALK rearrangement, detected in tissue jmd.amjpathol.org ■ The Journal of Molecular Diagnostics sections by fluorescent in situ hybridization (FISH) of interphase cells or by anti-ALK immunohistochemistry (IHC). However, FISH cannot differentiate between different ALK partner genes or non-canonical breakpoints, raising the possibility that not all detected ALK fusions may be responsive to targeted inhibitors. Indeed, "ALK positive" NSCLC has only a 60% objective response rate to ALK inhibitors; we sought to determine whether next-generation sequencing (NGS) of targeted DNA regions and/or RNA could refine the diagnosis of ALK rearranged NSCLC. Methods: From a set of 1532 cases at our institution evaluated by interphase FISH for ALK rearrangement, we identified 20 positive cases and sequenced them using a targeted, capture-based clinical NGS panel that included ALK introns 16-22 among other other cancer-related genes. DNA level rearrangements were identified using the Breakdancer and ClusterFast software packages. Strand-specific RNA-seq libraries were created from FFPE-derived RNA on 12 cases with remaining tissue and analyzed for fusion transcripts using the Tophat fusion package. Immunohistochemistry (IHC) was performed on 18 cases with the D5F3 antibody. Results: Of the 20 cases positive for ALK rearrangement by FISH, 17 (85%) showed evidence of rearrangement by DNA sequencing. These included four different canonical rearrangements: three EML4-ALK variants and a KIF5B-ALK variant. Non-canonical breakpoints and rearrangement partners were identified in 6 cases (35%). RNA-seq results were largely consistent with DNA sequencing, failing to detect expected fusion transcripts in 2 of 12 cases. In contrast, ALK IHC was positive in 16 of 18 cases, including 3 cases in which neither DNA nor RNA sequencing provided evidence of a fusion transcript capable of producing protein. Conclusions: Current FISH and IHC testing for "ALK positive" lung adenocarcinoma underestimate the biologic and genomic heterogeneity among ALK driven NSCLCs. In contrast, NGS-based testing can resolve novel partner genes and non-canonical breakpoints. The genomic heterogeneity observed in this study may account for differences in therapeutic response to targeted ALK inhibitors; however, future clinical trials will be needed to ascertain the exact significance of non-canonical and non-expressed ALK rearrangements in NSCLC. Introduction: The use of targeted gene panels in cancer to identify somatic mutations has the potential to inform prognosis and guide treatment decisions. One challenge is that selection of genes for such panels and the designation of actionable findings will continue to evolve as new therapies are developed and new mechanisms of tumor progression are discovered. In this regard, ongoing genomic profiling efforts have recently identified recurrent mutations in the exonuclease POLE) that are associated with high mutational burden in several cancer types. Whereas mismatch repair status has been shown to predict immunotherapy response in clinical trials, tumors with POLE mutations have yet to be broadly included for evaluation in such trials. Here, we describe two cases of colon cancer found to harbor POLE mutations after initial testing revealed an unusually high mutational burden in the setting of microsatellite stability and intact mismatch repair protein expression. Methods: Extended genomic profiling was performed on two cases of microsatellite stable colorectal carcinoma found to have high mutational burden. DNA was extracted from formalin-fixed tumor samples and used to prepare hybrid-capture libraries for genomic profiling by next-generation sequencing. A clinically validated capture panel of 198 cancer-related genes and a research panel of 142 genes undergoing clinical validation were used. Data analysis was performed using an internally developed and clinically validated bioinformatics pipeline followed by expert review and annotation. All testing was performed in a CLIA-certified, CAP-accredited laboratory. Results: Extended evaluation of these two cases of hypermutated, microsatellite-stable colon cancer revealed nonsynonymous changes in the POLE exonuclease domain at known mutational hotspots (p.P286R and p.V411L, respectively). This testing demonstrated high reproducibility in regards to variant identification and allele frequency between two distinct but overlapping capture panels. Notably, these findings led to discussion at a molecular tumor board and initiation of the approval process for immune checkpoint therapy on a compassionate use basis for these patients with POLE mutations and high mutational burden. Conclusions: These cases exemplify the impact of molecular testing in clinical decision-making and further support the inclusion of POLE on targeted cancer panels. Evaluation of patients with POLE mutated tumors in immunotherapy trials will further guide the best use of such testing for treatment decisions. These results highlight the evolving nature of actionable findings in cancer profiling and underscore the importance of thoughtful interpretation and design when working with targeted gene panels. B. Das, S. Bhaumik SRL Ltd, Mumbai, Maharashtra, India. Introduction: Lung cancer is the leading causes of cancer related death worldwide. In India, the incidence of lung cancer is rising at alarming rates which constitutes around 14.4% of all cancer. Molecular markers such as EGFR, KRAS, BRAF, HER2, and PTEN play a considerable role in lung carcinoma through their involvement in cell proliferation, apoptosis, and assist in therapeutic decision. Unlike EGFR, KRAS mutations are more common males, smokers and frequently found in Western populations. BRAF, HER2, PTEN gene mutation is found in a small proportion of lung cancers with activating mutations in 1% to 4% of NSCLC. In the current study, we evaluate the frequency, distributional pattern of these gene mutations and their association with the clinicopathological characteristics in Indian NSCLC patients. Methods: The present study was performed on 400 formalin-fixed, paraffin-embedded (FFPE) tumor samples. Extracted genomic DNA was amplified by nested PCR followed by direct sequencing. Results: The frequency of EGFR, KRAS, BRAF, HER2 mutations: 29% (116/400), 6.4 % (14/204), 1.5% (3/204), 1.5% (3/204), respectively. There were 3 different BRAF mutations were detected (L584P, V600E, K601E). The presence of HER2 insertion and deletion were observed in 3 (1.5%) cases, whereas remaining 201 cases confirmed wild-type alleles. HER2 mutations were preponderant in exon 20. Sequencing analysis revealed two c.2324-2325ins12 (ATACGTGATGGCduplication, p.Ala775_Gly776insTyrValMetAla) and one c.2305delC in HER2 gene. Interestingly, the study cohort demonstrated wildtype PTEN allele for all EGFR negative cases. Conclusions: To date, present study consist the largest number of Indian NSCLC patients. Several novel variations added new insights into the genetic heterogeneity of NSCLC patients. Further exploration with clinical follow up will strengthen the association of these mutations with disease outcome in Indian ethnic group. N.E. Stachowicz, N.E. Stachowicz, M.B. Campion, K.C. Halling, E.W. Highsmith, B.R. Kipp, B. Crusan, L.A. Kottschade, S.N. Markovic, C.J. Nelson, J.M. Cunningham, J.S. Voss, K.E. Hasselkorn, A. Rajeswari, D. Walker, M.C. Liu Mayo Clinic, Rochester, MN. Introduction: The proposed assay was validated for the detection of BRAF V600E and K/R mutations in cfDNA from double spun, platelet poor plasma using the RainDrop Digital PCR system. This blood-based assay is intended to be utilized as a substitute for invasive tissue biopsies. Reliable mutation detection in melanoma is critical, as patients with BRAF V600 mutant positive tumors benefit significantly from small molecule BRAF and MEK inhibitors. Methods: Clinical peripheral blood samples were collected in 10 mL Streck Cell-Free DNA BCT tubes from 20 healthy volunteers and 47 patients with metastatic melanoma of known BRAF mutation status. cfDNA was extracted from plasma using the Qiagen QIAamp Circulating Nucleic Acid Kit. cfDNA was processed with the RainDrop Digital PCR System using Taqman probes designed to detect wild-type BRAF and BRAF V600E mutations. Results: Cross hybridization of the V600E probe to V600K and V600R mutated DNA allows for the detection of all three mutations; however, the close proximity of V600K to V600R renders them indistinguishable. considered positive for a BRAF V600E or K/R mutation. The LOD for a BRAF V600E and K/R mutation is 5% at 1 ng input DNA. Results were 100% concordant for two variant calls among 3 samples (within and across runs) for the reproducibility study. BRAF RainDrop and tumor biopsy results were compared to determine the agreement between the two assays. Among the 47 melanoma patients, 20 were found to have a BRAF mutation in their tissue specimen, and 27 were not. The sensitivity, specificity, positive and negative predictive value of the RainDrop BRAF assay when compared to biopsy results was 9/20 (45%), 27/27 (100%), 9/9 (100%), and 27/38 (71%), respectively. Notably, the 11 patients with a negative cfDNA BRAF result but a positive tissue result had either well-controlled, low burden metastatic disease or an isolated local recurrence at the time of the blood draw. Conclusion: The cfDNA BRAF Mutation Detection assay was developed to assess for clinically relevant V600 mutations in melanoma. The test was found to have high specificity and positive predictive value such that patients with a positive cfDNA result may avoid invasive tumor biopsies for the determination of BRAF mutation status. However, patients with a negative BRAF result may still require a tissue biopsy for the determination of BRAF mutation status if clinically indicated. This blood-based assay may serve as a reliable substitute for determining BRAF status when tissue biopsies are not feasible or practical, and it may provide an effective means of serial monitoring for patients receiving relevant targeted agents. S. Kongkarnka, P.E. Oberstein, S. Hsiao, A.T. Turk, A.N. Sireci, B.M. Kurtis, H. Varma, A.I. Neugut, M. Mansukhani, A.R. Sepulveda Columbia University, New York, NY. Introduction: Molecular testing of colorectal cancer (CRC) for mutations in exons 2, 3, and 4 of KRAS and NRAS, and BRAF p.600 is recommended as predictive markers of response to anti-EGFR therapy and as a marker of prognosis in advanced disease, respectively. DNA mismatch repair (MMR)/MSI testing is emerging as a predictive marker of response to immune checkpoint therapy for CRC. Next-generation sequencing (NGS) multiplex cancer panels allow simultaneous detection of multiple mutations. We evaluated the clinical utility of molecular testing of all CRC cases received in our Pathology Department as a standing order, whether received as initial biopsy or resection specimens of primary or metastatic disease, tested with a CRC molecular panel and by IHC for MMR and/or MSI testing. Methods: Ninety CRC samples and patients who had CRC molecular panel testing at Columbia University Medical Center were identified. FFPE sections were macro or microdissected, NGS was performed with the TruSeq cancer panel (Illumina) and extended RAS and BRAF p.600 and PIK3CA mutation status were reported. Sequencing data was analyzed with NextGene software (SoftGenetics). IHC for MMR proteins and/or MSI testing were also performed. Results: Patients' mean age was 63 years (24 to 92) years; 45 patients were male (50%). Seven of 90 (7.8%) cases were tested from a metastatic lesion and the remaining 83, from biopsy or resection of primary tumor. RAS mutation was identified in 48 cases (53.3%). 3 cases had NRAS mutation, including one case with KRAS and NRAS mutations. RAS mutations other than KRAS exon 2 were observed in 6 cases (12.5%). PIK3CA mutations were detected in 12 cases (13.3%). BRAF mutation was present in 11 cases, with concurrent RAS mutation in 2 cases. Deficient DNA MMR or MSI-High were present in 12 of 88 cases tested (13.6%). Thirty-two of 90 cases (35.5%) had advanced metastatic or recurrent disease. RAS testing provided useful information for therapy selection in 29 of 90 (32.2%) cases and MMR in 3 cases, for biological therapies. The remaining 58 patients were tested on primary site biopsy or resection and received adjuvant therapy, were early stage disease, were treated elsewhere or were not eligible for therapy. Conclusions: Extended RAS testing increased mutation detection rate by 12.5%. Molecular testing of CRC cases performed as a standing order in all initial biopsy or resection specimens of primary or metastatic CRC, tested for RAS/BRAF mutations with a TruSeq cancer panel and for dMMR/ MSI, showed frequent clinical utility (greater than one third of cases tested) in the selection of patients for biological targeted therapy. For early stage disease at presentation, mutational data may still be of utility at the time of recurrence or metastasis. E.M. Azzato, M. Rusch, S. Shurtleff, J. Nakitandwe, Z. Zhang, D. Hedges, S. Newman, M. Edmonson, A. Thrasher, J. Becksfort, C. Kesserwan, S. Raimondi, K.E. Nichols, D.W. Ellison, J. Zhang, J. Downing St. Jude Children's Research Hospital, Memphis, TN. Introduction: As part of a prospective clinical research study, Genomes for Kids, we are performing a clinical feasibility study of three-platform next-generation sequencing (NGS) for molecular diagnostics, utilizing an integrative whole-genome (WGS), whole-exome (WES) and transcriptome (RNA-Seq) analysis on fresh/frozen tumor samples from pediatric oncology patients. Methods: Eligible patients receive parallel standard-of-care cytogenetic and molecular evaluations prior to consent. Matched tumor/normal samples undergo WGS and WES at 45x coverage and RNA-Seq at 20-30x coverage; the assay pipeline integrates genetic lesions detected by all three NGS platforms to cross-validate and characterize single nucleotide variations (SNVs), short insertions and deletions (indels), structural variations (SVs) including fusions, karyotypes, focal copy number alterations (CNVs), and copy neutral loss of heterozygosity (CN-LOH). After calls are manually confirmed, a multidisciplinary committee determines variant categorization and reportability. An integrative report is generated through Molecular Pathology and placed in the medical record. Results: Over the first 10 months, we analyzed 73 tumor samples from 68 patients, which included 35 leukemia, 19 brain tumor and 19 solid tumor samples. Seventy-one specimens (97.3%) had reportable pathologic or likely pathologic lesions, whereas two tumor samples (2.7%) had no reportable alterations. We identified 36 SVs, 53 focal CNVs, 81 SNVs/Indels, 53 ploidy alterations and 17 instances of CN-LOH of clinical relevance. Fifty-five of these tumors (75%) had cytogenetic or molecular testing performed as part of their routine diagnostic workup; 107 clinically significant alterations, including 22 (61.1%) SVs, 36 (44.4%) SNV/Indels, and 49 (92.5%) focal CNVs, had not been detected by traditional evaluation. This included 10 novel SVs that resulted in in-frame fusion transcripts or aberrant expression. We observed perfect correlations between NGS and fusion detection by RT-PCR, single gene sequencing and FISH. Karyotyping was performed on 33 leukemia samples; hyperdiploidy and hypodiploidy were properly categorized in all cases (n=13). Minor karyotype discrepancies were observed in 12 samples, the majority of which were explainable by marker chromosomes, subclonality or confirmed by additional FISH analyses.Conclusions: Use of an integrative, multi-platform NGS approach is feasible in the clinical setting, improving detection of clinically relevant alterations, while matching performance with gold-standard diagnostic assays. Optimizing laboratory, computational and review processes will improve turn-around-time, allow real-time reporting and ideally streamline multiple laboratory approaches into one comprehensive platform. Introduction: The concept of exclusive mutation of certain driving genes during oncogenesis is facing challenge due to the availability of more and more next gene sequencing data. Although there is usually one critical mutation within one oncogene on majority of the tumors sequenced in our institute, the exception of such general observation exists. Here, we reported that even within one oncogenic gene, KRAS, in the colonic adenocarcinoma, multiple point mutations were present. Methods: The colonic adenocarcinoma was diagnosed by staff at University of Miami and confirmed by second pathologist. The adenocarcinoma was macro dissected and DNA was extracted using QIAamp DNA kit. Illumina truseq cancer panel was employed to perform the next gene sequence. Unusual results such as multiple mutation in same gene were confirmed by pyro-sequencing. Results: After nextgeneration sequencing of 432 colonic adenocarcinoma, 214 tumors were found to carry KRAS mutation (214/432). We found 5 adenocarcinoma cases had multiple point mutations (5/214). The mutation sites varied among cases. Pyro-sequencing data confirmed that multiple mutations were present in the oncogenic gene KRAS within all five cases (5/5). Conclusions: Although it is rare event, multiple mutations could be present in the single oncogenic gene KRAS in colonic adenocarcinoma. Whether it represents multiple mutations with different sites in one allele or in different alleles in the tumor cell or in different tumor cells needs to be elucidated in the future. L. Ritterhouse, B.E. Howitt, V. Rojas-Rujilla, F.C. Kuo, L.M. Sholl Brigham and Women's Hospital, Boston, MA. Introduction: A subset of endometrial carcinomas (EMC) is characterized by hypermutation. Mismatch repair (MMR) defects are a well-recognized mechanism of hypermutation in EMC and may be due to germline or somatic alterations in MMR POLE) have been associated with an ultramutated phenotype in ~10% of EMC. Both MMR and POLE-related EMC have characteristic mutational signatures, and other EMC signatures have been described but are poorly characterized. This study sought to further characterize hypermutated EMC. Methods: Targeted massively parallel sequencing of 275 or 300 known cancer genes was performed on tumor tissue using an Illumina HiSeq 2500 sequencer. Hypermutated tumors were defined as those having > mean number of mutations (21.9 per Mb) present in the EMC cohort. Hotspot genotyping of POLE (exons 9/13) was performed by PCR and Sanger sequencing. Mismatch repair immunohistochemistry (IHC) and MLH1 promoter methylation was performed in a subset of cases. Mutational signatures were defined by the substitution class and sequence context of the mutated base. Results: Of 170 total EMCs sequenced on the targeted panel, 35 (21%) met the criteria for hypermutation, and included 12 grade 1, 11 grade 2, and 7 grade 3 endometrioid, as well as 1 mixed endometrioid/serous and 4 undifferentiated carcinomas. DNA was available for POLE sequencing in 34/35, of which POLE mutations were identified in 7 (20%), including 3 P286R and 4 V411L mutations. Mutational signature analysis identified three distinct patterns in hypermutated EMC: 1) POLE EDM (n=8) (Tp(C>T)>50% and (C>A)pT>20%), 2) Signature 12 "S12" (n=3) (Np(T>C)>40%), and 3) MMR (n=18) (% of frameshift>20% or Np(A>G)>20% or Gp(C>T)+(C>T)pG>50% and frameshift>9% and neither of above). 6 cases were not classifiable based on mutational signature. All 7 confirmed POLE-mutated cases had the POLE signature and intact MMR expression by IHC. 9 cases with an MMR signature had IHC available with abnormal MMR protein expression in 100% (7 MLH1/PMS2 loss (confirmed MLH1 promoter methylation), 1 MSH6 loss (MSH6 germline), 1 MSH2/MSH6 loss (MSH2 germline)). An additional 2 MMR signature cases were known to have MSH6 germline mutations. The S12 signature cases exhibited unusual subclonal changes in MMR protein expression. Conclusions: A subset of EMC exhibit a hypermutated molecular profile, of which ~20% exhibit POLE mutations and a characteristic mutational signature. In addition, ~10% of EMC have a unique signature (S12) with unusual MMR IHC patterns. The remaining hypermutated EMC include those with germline or somatic defects in MMR, as well as those with an unclassifiable signature. J. Saab, S. Mathew, P. Zhang, H. Zia, M.J. Kluk, H. Fernandes NewYork-Presbyterian Hospital, Weill Cornell Medicine, New York, NY. Introduction: Oncogenic driver mutations in KRAS have been identified in a third of lung adenocarcinomas and predict a lack of response to EGFR tyrosine kinase inhibitors. Approximately 5% to 15% of lung adenocarcinomas show amplification of the KRAS gene and 37% of these tumors harbor somatic KRAS variants. The impact of KRAS amplification or polysomy on patient survival is not clear with some studies suggesting a potential tumorigenic effect, particularly with concomitant KRAS mutations. The objective of this study was to evaluate for KRAS amplification in lung adenocarcinomas demonstrating increased relative coverage of the KRAS gene as determined by next-generation sequencing (NGS). Methods: NGS results for 140 lung adenocarcinomas with clinically relevant variants were examined for evidence suggestive of KRAS copy number alterations (CNA). NGS was performed on the Ion Torrent platform using the AmpliSeq Cancer Hotspot Panel v2 and sequencing data was analyzed using the Variant Caller v4.4 (Thermofisher, CA). Tumors were suspected of having CNA when the relative coverage of KRAS amplicons showed a fold increase >1.3 after normalization to total coverage of pooled normal samples. The presence of KRAS variants with concomitant amplification of the KRAS gene in lung adenocarcinoma is not infrequent. KRAS copy number alterations including amplifications and polysomy could potentially be flagged using an algorithm that utilizes the relative amplicon coverage. FISH can further differentiate KRAS amplification from polysomy. The clinical relevance in terms of disease progression and outcome in patients with KRAS variants with concomitant KRAS CNA needs to be further explored. Extended RAS Profiling Due to Insufficient DNA Material H. Shojaei, B. Sundman, L. Zhou University Hospitals Case Medical Center, Case Western Reserve University, Cleveland, OH. Introduction: RAS molecular profiling has become important companion test to guide the use of anti-EGFR antibody treatment in metastatic colorectal cancer patients. Although sequencing offers better coverage of the coding area and identifies specific locations of genetic alterations in the gene, the mutant/wild-type allele ratio present in the tumor affects its diagnostic sensitivity and makes the detection of point mutations particularly challenging. Real-time polymerase chain reaction (PCR) based methods are more sensitive but do not test for less frequently occurring mutations with uncertain clinical implications. Sequencing is not successful in all cases, the aim of this study is to evaluate the overall failure rate of testing caused by "insufficiency" by sequencing method and reevaluating PCR results in these specimens. Methods: At our center, a series of 121 patients who had been enrolled in extended RAS gene profiling between 2014 to 2015 was evaluated. All samples were sent out to a molecular reference laboratory for extended RAS analysis (Next-generation sequencing). Sixteen of these samples were returned to our institution due to insufficient DNA material for analysis. These samples were reevaluated in our lab by a real-time qualitative PCR assay (Therascreen KRAS RGQ PCR Kit, used on the Rotor-Gene Q MDx instrument) for the detection of seven somatic mutations in the human KRAS oncogene (Gly12Ala, Gly12Asp, Gly12Arg, Gly12Cys, Gly12Ser, Gly12Val, and Gly13Asp) . This assay has an analytical sensitivity of 1% to 5%. Results: The overall success rate for PCR was 75% (12/16), in returned samples due to insufficient DNA material. PCR evaluation for KRAS mutations in sixteen cases, revealed eight patients with KRAS mutations (one Gly12Ser, four Gly12Val, two Gly12Asp, and one Gly12Cys), three patients with no KRAS mutations and one case had indeterminate result. Four remaining samples did not have sufficient DNA material for PCR evaluation. Extended RAS testing was successful in 105/121 (86.7%) of send out specimens. There were 45 cases with positive results for KRAS mutations (six Gly12Val, two Gly12Ser, four Gly12Cys, three Gly12Ala, sixteen Gly12Asp, eleven Gly13Asp, two Gln61His and one Gln61Lys) and 4 cases had NRAS mutations (two Gly12Asp, one Gly61Leu and one Gln61Lys). Conclusions: The majority of patient samples submitted for extended RAS molecular cancer genomic profiling have successful results reported. Reevaluating specimens for KRAS hotspot point mutations by realtime PCR, who had insufficient DNA material for extended RAS analysis may improve the success rate of molecular profiling. Introduction: Serous, endometrioid, and clear cell subtypes of ovarian and endometrial carcinomas have similar embryological origins and share common molecular abnormalities. Distinction between these subtypes has prognostic and potentially therapeutic implications, but is often challenging based on traditional histopathological examination. Different DNA methylation profiles are associated with each subtype. We sought to identify a panel of DNA methylation biomarkers that could be used to complement histopathology and improve diagnostic accuracy. Methods: Genome-scale DNA methylation data for 196 serous, 49 endometrioid, and 17 clear cell subtypes of ovarian carcinomas were retrieved from a previous study, and from 210 serous, 40 endometrioid, and 14 mixed subtypes of endometrial cancers from The Cancer Genome Atlas (TCGA). These data sets were screened to identify markers that could distinguish each subtype based on their methylation status. The selected biomarkers were further tested on formalin-fixed, archival tissue sections from 18 serous, 16 endometrioid, and 10 clear cell uncontentious ovarian carcinomas using real-time PCR-based MethyLight analysis. The tissue sections were macrodissected to ensure a cancer cell content of at least 50%. We built a three-group classification model using Lasso regression and 10-fold cross validation. Predicted probabilities of each subtype were calculated on a set of 12 tumors, and the percentage of true calls/correct classification summarized. Results: We identified 21 potential subtype specific DNA methylation markers, of which 9 were further verified on independent ovarian carcinoma samples. Four of the 9 markers tested (OvaCA1, OvaCA3, OvaCA4 and OvaCA6) correctly predicted the serous tumor subtype, 3 (OvaCA6, OvaCA8 and OvaCa10) predicted the endometrioid subtype, and 2 (OvaCA10 and OvaCA11) predicted the clear cell subtype with probabilities of 100%, 94% and 90% respectively. Two markers with poor predictive probabilities were excluded from further analysis. Conclusions: Our panel of 7 biomarkers can distinguish between ovarian serous, endometrioid and clear cell carcinomas based on DNA methylation profile. The performance of this panel to help distinguishing between subtypes of endometrial carcinomas has not yet been evaluated. This panel may be useful in establishing a molecular classification that may complement histopathological diagnosis of ovarian and endometrial carcinomas and increase our ability to predict clinical outcome and therapeutic responsiveness in these cancers. D. Dunaway, C. Merritt, J. Jung, P. Webster, C. Ngouenet, G. Ong, I. Sprague, C. Warren, G. Geiss, S. Warren, A. Cesano, J. Beechem NanoString Technologies, Seattle, WA. Introduction: As intra-tumoral heterogeneity has emerged as a challenge in development of targeted cancer therapeutics, the tissue context of biomarker levels has become an increasingly important aspect of patient classification and stratification. Historically, immunohistochemistry and in situ hybridization have been used to assess spatial heterogeneity of proteins and RNA transcripts, respectively, in clinical samples. These approaches, however, have limited multiplexing capacity and dynamic range. Here, we report the development of a spatially resolved approach for quantifying up to 800 protein or RNA targets with over 5 logs of dynamic range in a single formalin-fixed, paraffin-embedded (FFPE) section. Methods: An automated prototype capable of imaging and sample collection was developed by modifying a standard microscope. For protein detection, a multiplexed cocktail of 30+ primary antibodies, each with a unique, photocleavable oligo tag, and 1-2 fluorescently labeled antibodies was applied to a slide-mounted FFPE tissue section. Regions of interest, selected based on a fluorescence imaging scan of the entire tissue, were illuminated sequentially with focused UV light to release the oligos. Following each illumination cycle, eluent was collected from the local region, moved to a microtiter plate, hybridized to NanoString barcodes, and subsequently analyzed with an nCounter Analysis System. The resulting digital counts corresponded to the abundance of each targeted protein in the regions of interest. For RNA detection, a cocktail of multiple UV-cleavable in situ hybridization probes were used in a similar manner. Results: In control samples, we found expected levels of protein and RNA targets, including HER2 and CD3. We demonstrate multiplexed detection from discrete regions within a tumor and adjacent normal tissue, enabling systematic interrogation of a heterogeneous tumor microenvironment. We also demonstrate a linear correlation (R 2 > 0.99) between observed counts and area of UV illumination with a resolution limit of 500 μm^2, or approximately four T-cells. We are testing methods for analyzing target abundance from individual cells and from noncontiguous cells with the same phenotype. Conclusions: With further development, our novel approach to capture the spatial context of protein and RNA levels will have many applications in biomarker and translational research. The ability to digitally measure RNA and protein at up to 800-plex from FFPE tissues could facilitate drug mechanism-of-action and resistance studies within the tumor microenvironment. Quantitative, high-plex data should also greatly accelerate the discovery of immune biomarkers in tumors and the development of companion diagnostics for targeted therapies. Introduction: RAS gene mutations play a significant prognostic and predictive role in colorectal cancer (CRC) patients. RAS-mutated CRC patients exhibit a poor response to anti-EGFR therapy resulting in adverse outcomes. In the current study, we sought to investigate the incidence and patterns of RASmutations in a cohort of histopathologically proven CRC patients from India. Methods: The study included formalin fixed paraffin embedded tissue samples of CRC patients for KRAS and NRAS mutational testing. The extracted DNA from the samples were further analyzed to investigate the mutational status of KRAS, covering exons 2, 3, and 4 (codons 12, 13, 61,117, and 146) andNRAS exon 2, 3, and 4 (codon 12, 13, 61, and 146) using a real-time PCR-based assay. Results: A total of 1109 and 1010 CRC patient samples were screened for KRAS and NRAS mutations, respectively. 388 (35%) of the cases (Male: Female = 1.4:1; 20 years to 85 years) presented with a mutation of the KRAS gene whereas 26 (2.6%; Male: Female = 0.8:1; 25 years to 80 years) patient samples harbored an NRAS mutation. For the KRAS gene, 88% of the mutations were found in exon 2, 27% in exon 3, and 4.4% in exon 4. For NRAS, 16% of the mutations were observed in exon 2, 55% in exon 3, and 28% in exon 4. Majority of the KRAS mutations were detected in CRC patients above 40 years of age. G12D was the most common mutation observed in the KRAS gene whereas Q61H mutation was predominant in NRAS positive CRC cases. Conclusions: This is the largest study to report the frequency of the KRAS and NRAS mutations in the Indian population. In this study, the KRAS mutation pattern in Indian CRC patients was comparable to that reported from other parts of the world. This is in contrast to the previously reported Indian studies where the incidence is lower. We attribute this to the use of a more sensitive assay with a broader coverage of the two genes. A high incidence of G12D and Q61H mutations were found in the KRAS and NRAS gene respectively. Further analysis with clinical correlation would help better understand the role of RAS mutations in the CRC patient treatment and outcome. F. Ahmad, B.R. Das SRL Ltd, Mumbai, Maharashtra, India. Introduction: Mutations in exon 9 and 20 of the PIK3CA gene represent to most common genetic alterations in breast carcinoma. These mutations are reported in around 20% to 40% of the breast cancer patients. Recent studies suggest that PIK3CA mutations are associated with predicting prognosis in breast carcinoma. In the current study we explored frequency of PIK3CA mutation in Indian breast cancer patients. Methods: The study was performed on 190 breast cancer patients using direct sequencing. Results: Sequencing analysis revealed PIK3CA mutations in 22.6% (43/190) of breast tumor samples. Exon 20 mutations were more common in comparison to exon 9 mutations. Mutations were predominantly spotted between nucleotide positions 1624 to 1636 or between 3129 to 3140. Overall four different missense mutations (E542K, E545K, E545A and E545G) were detected in the helical domain, and two different amino acid substitutions at codon 1047 (H1047R and H1047L) in the kinase domain. PIK3CA mutations were predominantly seen in older age patients with grade II ductal carcinomas. Furthermore, PIK3CA mutations were common in hormone positive cases and a comparativel lesser in triplenegative tumors. Conclusions: This is one of the largest study evaluating PIK3CA mutation in breast cancer from India. This study confirms that PIK3CA mutations are quite prevalent in Indian subcontinent. Furthermore, identification of prognostic utility in Indian scenario is the way forward and warrants further evaluation. To evaluate whether a single targeted next-generation sequencing-based (NGS) assay yields sufficient molecular information necessary to classify LGG, we retrospectively surveyed 50 cases of LGG and reclassified them based on the updated criteria. Methods: A targeted, hybridization capturebased, clinical NGS assay was used to assess at least 25 genes and somatic copy number alterations (SCNA) in 50 LGG cases, consisting of 36 mixed oligoastrocytomas (MOA), 11 oligodendrogliomas (OD), 2 astrocytomas (AC), and 1 low-grade glioma NOS (LGG NOS), all originally classified by the 2007 CNS WHO criteria. SCNA detection was performed following alignment using CopywriteR, an off-the-shelf, open source tool that utilizes off-target reads in hybridization capture-based sequencing data, including data from limited gene panels. The NGS results were compared with immunohistochemistry (IHC) and fluorescence in-situ hybridization (FISH) and also used to reclassify the tumors based on the updated 2016 CNS WHO classification scheme. Results: Our NGS-based approach was able to re-classify the 36 MOAs into 30 AC (20 IDH-mutant and 10 IDH-wildtype) and 6 OD (IDH-mutant and 1p/19q codeleted), and 1 LGG NOS into an AC (IDHwildtype). 9 of the 11 OD remained in their original category; one of the remaining 2 ODs was reclassified as an IDH-mutant AC, the other had findings consistent with a "pediatric" OD. The 2 ACs remained in their original categories, both yielding an integrated diagnosis of IDH-wildtype AC. IDH1-R132H IHC was performed on all 50 cases and was concordant with NGS when the IHC was positive. Additionally, IDH1/2 NGS molecular results yielded an additional 5 cases of non-R132H IDH1 mutations and one case with IDH2 p.R172K. FISH and NGS based testing for 1p/19q deletions were concordant for 45 of the 48 cases with informative FISH results (93.75%). NGS-based SCNA detection revealed segmental, partial losses of 1p and or 19q in the regions of the FISH probes in three discordant cases, which would not meet the 2016 CNS WHO criteria for whole arm deletions. Conclusions: Here we show a single targeted NGS assay can serve as the sole molecular testing modality necessary to categorize LGG by the updated 2016 CNS WHO classification scheme. This increased diagnostic efficiency enables the potential for greater accuracy and cost-efficiency while reducing the specimen tissue requirements compared to multimodal approaches. C. Huang, R. Vemula, P. Kamineni, J. Wu, B. Anekella SeraCare Life Sciences, Gaithersburg, MD. Introduction: Developing, optimizing, and running somatic mutation detection assays is challenging due to variability in tumor samples and NGS workflows. Monitoring these assays using appropriate reference materials is critical for laboratory quality control. A mix of purified DNA containing well characterized, clinically actionable tumor-specific somatic mutations at three different allele frequencies was developed to assess detection of a wide range of single nucleotide variants (SNVs), insertion/deletion mutations (indels), mutations within homopolymers and structural variants (SVs) to fulfill unmet needs for accurate, multiplexed quality controls. Methods: Human genomic DNA was extracted from the well-characterized GM24385 cell line and blended with biosynthetic constructs bearing forty (40) cancer-relevant mutations. Blending was performed to achieve three levels of allele frequencies within the mix: approximately one third of the forty targets were at 10% allele frequency (AF), approximately one third at 7% AF, and the remaining targets were at 4% AF. These levels make it possible to detect mutations well above many assays' stated limit of detection (LOD), right above the LOD, and just below the LOD, within the same quality control material. Allelespecific digital PCR was used to verify the frequency of each mutation. Testing was also performed using Ion Ampliseq Cancer Hotspot Panel v2 and the Illumina TruSeq Amplicon Cancer Panel. Results: Average allele frequency by dPCR was 10.43% (range 9.33% to 11.53%) for the AF10 tier of mutations, 6.97% (range 6.07% to 7.88%) for the AF7 tier, and 3.93% (range 3.52 to 4.34) for the AF4 tier. Variant allele frequencies detected by NGS were generally consistent with dPCR results, and all discordances could be explained by limitations of commonly used NGS panels. For example, TruSeq Amplicon Cancer Panel did not detect PIK3CA c.3204_3205insA, ATM c.1058_1059delGT, and ERBB2 c.2324_2325ins12 mutations at the expected frequency, as these mutations are either under a primer used for amplification of target regions, or are at the very end of a sequencing read such that the strand bias filter prevents positive detection. Conclusions: We developed a highly multiplexed and accurate reference material that allows simultaneous monitoring of a broad range of somatic mutation types at a large number of loci. This material may aid in optimization and verification of detection limits for NGS-based oncology tests and provide laboratories greater assurance in their ability to correctly call various types of mutations on a daily run basis. H.J. Dubbink, P.N. Atmodimedjo, R. van Marion, P.H. Riegman, J.M. Kros, M.J. van den Bent, W.N. Dinjens Erasmus MC Cancer Institute, Rotterdam, Netherlands. Introduction: Cancer cells are genomic unstable and accumulate tumor typespecific molecular aberrations, which may represent hallmarks for predicting prognosis and targets for therapy. Co-deletion of chromosomes 1p and 19q marks gliomas with an oligodendroglioma component and predicts a better prognosis and response to chemotherapy. Methods: In the current study, we present a novel method to detect chromosome 1p/19q co-deletion or loss of heterozygosity (LOH) in a diagnostic setting, based on single nucleotide polymorphism (SNP) analysis and next-generation sequencing (NGS) on an Ion Torrent platform. We selected highly polymorphic SNPs evenly distributed over both chromosome arms. To experimentally determine the sensitivity and specificity of targeted SNP analysis, we used DNAs extracted from 49 routine formalin-fixed, paraffin-embedded (FFPE) glioma tissues and compared the outcome with diagnostic microsatellite-based LOH analysis and calculated estimates. Results: We show that targeted SNP analysis by NGS allows reliable detection of 1p and/or 19q deletion in a background of 70% normal cells according to calculated outcomes, is more sensitive than microsatellitebased LOH analysis, and requires much less DNA. Conclusion: This specific and sensitive SNP assay is broadly applicable for simultaneous allelic imbalance analysis of multiple genomic regions and can easily be incorporated in NGS mutation analyses. The combined mutation and chromosomal imbalance analysis in a single NGS assay is perfectly suited for routine glioma diagnostics and other diagnostic molecular pathology applications. J Mol Diagn 2016, 18:775-786. One of the most commonly mutated genes in GBMs is the epidermal growth factor receptor (EGFR), and one of the most common mutations in EGFR is EGFR variant III (EGFRvIII). EGFR is a transmembrane receptor tyrosine kinase, and the EGFRvIII mutant is characterized by a deletion of 267 amino acids in the extracellular domain, resulting in a novel residue at the newly formed junction and leading to ligand-independent constitutive activation. The University of Pennsylvania initiated a pilot study of autologous T cells re-directed to the EGFRvIII mutation using a lentiviral vector encoding a chimeric antigen receptor (CAR). Methods: One hundred and ninety-nine patient samples with newly diagnosed or recurrent GBMs were screened for genetic abnormalities, including EGFRvIII, MGMT methylation, and/or next-generation sequencing analysis of 47 genes. EGFRvIII was detected from RNA extracted from FFPE tissue by a custom library jmd.amjpathol.org ■ The Journal of Molecular Diagnostics methodology and sequenced on the Illumina MiSeq. Correlation analysis was performed using a custom program in R to detect positive and negative association of co-occurrence of mutations. Results: One hundred and ninety tumor specimens from patients with histologically confirmed GBMs were tested for EGFRvIII at our institution as standard-of-care; of these, 37 tested as high positive (>30% reads from EGFRvIII/total EGFR) for EGFRvIII (19.5%). Eleven patients were infused and 5 of these 11 had post infusion resections, which were analyzed for EGFRvIII, gene mutations, and amplifications. Loss of some or all of EGFRvIII was detected in 3 of 5 specimens, with the remaining 2 not having shown expansion of the CAR T-EGFR cells in the peripheral blood post-infusion. Comparison of pre-and post-infusion studies in patients with peripheral expansion showed complete or partial loss of EGFRvIII; however other mutations including EGFR amplification remained, suggesting tumor heterogeneity. Patients with EGFR amplification and/or EGFRvIII show significant overlap, and often harbor additional EGFRmutations. The most significant positive co-mutation correlations were between EGFR amplification and EGFR mutations, and the strongest negative co-occurrence was between IDH1 mutations and EGFR amplification. Conclusions: The complexity observed in our clinical specimens suggests that late stage GBMs are heterogeneous tumors with EGFRvIII representing a late mutation in the development of disease. We also observe that amplification and mutation of EGFR tends to occur in tumors without prognostically favorable mutations in IDH1, supporting that EGFR-driven tumors constitute a more aggressive subset of GBMs. M.N. Nikiforova, M. Durso, K.M. Callenberg, A. Wald, Y.E. Nikiforov University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: Whole transcriptome (RNA-Seq) analysis is mostly used as a discovery tool for detection of gene fusions and gene expression, typically in freshfrozen tissue samples. Its use for clinical samples with limited quantity and quality of RNA has not been previously demonstrated. We have developed and validated an RNA-Seq approach that can be effectively used for detection of novel driver gene fusions and gene expression in clinical thyroid FNA and fixed tissue samples. Methods: RNA-Seq was performed on 41 thyroid samples with sequencing parameters optimized for use in clinical samples with limited quantity and quality of RNA (thyroid fine needle aspirates (FNA) and formalin-fixed paraffin-embedded (FFPE) tissues). As low as 30 ng of RNA was used for library generation with the TruSeq Stranded Total RNA kit (Illumina) and sequencing on the HiSeq 2500 (Illumina). A custom bioinformatics pipeline was developed that uses both existing bioinformatics tools (TopHat, Chimerascan, FusionCatcher, SOAPfuse) and also inhouse developed scripts and filters for accurate detection of driver gene fusions and gene expression analysis. All samples were previously analyzed by ThyroSeq v2 amplification-based targeted NGS panel. Results: Initially, clinical RNA-Seq analysis was performed in a set of 20 thyroid tumors with known fusion types and showed 100% sensitivity for fusion detection. Next, we performed RNA-Seq on 21 clinical samples (9 FNA, 12 FFPE tissue) that showed suspicious differential expression profile (n=10) and/or absence of all known driver mutations (n=11) by ThyroSeq v2. Novel driver fusions or new breakpoints of the known fusions were detected in 15 (71%) cases (7 FNA, 8 FFPE). They included 2 novel RET fusions, 2 NTRK3, 1 PPARG, 2 NTRK1 novel fusion breakpoints, 3 novel gene partners for BRAF, two for THADA, and three novel fusions involving genes not previously reported in thyroid cancer. This included identification of fusions in 8 (80%) of samples that showed suspicious expression profile on ThyroSeq v2 consistent with upregulation of a known oncogene, but no known fusions types. Each fusion demonstrated 3 to 300 reads spanning a break point. Functional driver status of fusions was confirmed by gene expression profile. All samples positive for gene fusions were papillary carcinomas and a follicular carcinoma on a surgical follow up. Conclusions: Clinically modified whole transcriptome (RNA-Seq) analysis can be successfully used to detect oncogenic fusions in small thyroid FNA samples and in fixed tissue specimens. Our results demonstrate that novel gene fusions account for a large proportion of thyroid carcinomas lacking all currently known driver mutations. Introduction: Advances in Next-Generation Sequencing (NGS) have facilitated lower input sample volumes, and sample material. These advances are not without the challenges of library bias and library complexity increasingly affecting data quality. Accel-NGS 2S DNA Library kits (Swift Biosciences) have been designed to overcome these challenges providing uniform and more complete coverage, resulting in improved sequencing efficiency and reduced costs. Additional cost and time savings can be found in automating library preparation, leading to increased throughput while maintaining high-quality data and reproducibility with little to no variation across wells. Methods: In this study, we demonstrate high-throughput automated NGS library preparation for targeted capture and WGS from 10ng FFPE and 10ng cfDNA on the PerkinElmer Sciclone NGSx with Accel-NGS 2S DNA Library kits. An Alu 247/115 repeat assay was used to assess contamination of high molecular weight DNA, quantity and integrity of cfDNA and FFPE prior to library preparation. Reproducibility and quality were determined by qPCR, LabChip GX analysis and subsequent sequencing on the Illumina HiSeq X10. Results: As DNA extracted from FFPE exhibits varying degrees of damage, the Alu repeat assay is an accurate way to assess and quantify usable DNA. Our results show a DNA integrity cfDNA, sufficient for 10ng input into the prep. Library yields were sufficient for sequencing, and LabChip analysis validated the targeted insert size of 350bp for FFPE and 165bp for cfDNA. Sequencing of manual versus automated library preparations yielded similar results: ultra-low duplication, no adapter dimer formation, median insert size in-line with the LabChip results and library complexity as expected for 10ng human gDNA. Conclusions: Accel-NGS 2S DNA Library kits on the Sciclone NGSx provide sequence-ready libraries, with equal or better reproducibility and quality to manually prepared libraries. Highly efficient library preparation driven by end repair of both 3' and 5' DNA termini delivers a more complex library requiring less sequencing, enabling comprehensive analysis of low input DNA samples such as FFPE and cfDNA. Significant cost savings and improvement in lab efficiencies were achieved combining the use of PerkinElmer automation for higher throughput library prep with Swift's highly efficient library preparation. Introduction: Accumulating evidence suggests the relevance of PARP inhibitors and BRCA-like phenotype and PARP inhibitors have become a part of clinical practice in the ovarian cancer patients with BRCA1/2 germline mutation. Germline and somatic sequence variants of DNA repair genes are thought to be predictors of BRCA-like phenotype and recent advances in sequencing technology enable us to effectively detect variants in DNA repair genes. We are trying to identify germline and somatic sequence variants of DNA repair genes in female cancer patients. Methods: Genomic sequencing of investigative biomarkers was prospectively offered to patients with cancers. Normal and tumor DNA libraries were prepared separately from peripheral blood and a retrieved archival FFPE tumor sample from each patient. Relevant targets were enriched by custom designed Agilent SureSelect hybrid capture baits using standard protocols. Samples were sequenced on Illumina platforms. We analyzed germline and somatic pathogenic variants of DNA repair genes using Freebayes (Cornell), Oncotator (Broad Institute) and UNC-created pipelines. Results: Four hundred and ninety-six patients' samples were analyzed with 201 breast, 201 uterine endometrial, 68 ovarian, and 26 colorectal cancers. Pathogenic germline mutations were detected in 91 (18.3%) patients and somatic mutations were present in 151 (30.4%) patients with both germline and somatic mutations present in 24 (4.8%) patients. In total, 218 (44.0%) out of 496 female cancer patients had germline and/or somatic mutations in DNA repair genes. Genes with pathogenic germline variants included ATM, ATR, BARD1, BLM, BRCA1, BRCA2, BRIP1, CHEK1, CHEK2, ERCC2, FANCD2, FANCG, MLH1, MSH6, NBN, PMS2 , and POLE. Somatic variants were observed in genes such as ATM, ATR, BLM, BRCA1, BRCA2, BRIP1, CHEK1, CHEK2, ERCC2, FANCA, FANCB, FANCD2, FANCE, FANCG, FANCI, FLCN, MLH1, MRE11A, MSH2, MSH6, NBN, PALB2, PMS2, PTEN, RAD50, SLX4, and SMARCA4 . Conclusions: We demonstrate that almost half of female cancers are characterized by germline and/or somatic mutations in DNA repair genes, which could make these patients potential candidates for PARP inhibitor trials in a clinic or research setting. Targeted sequencing using normal and tumor samples with focus on DNA repair genes could be a useful strategy to identify patients with BRCA-like phenotype in female patients. S. Gunn 1 , 2 , C. Sims 1 , 2 , S. Govender 1 , A. Khurana 1 , M. Moore 1 , P. Cotter 1 1 ResearchDx/PacificDx, Irvine, CA; 2 Targeted Genomics, San Antonio, TX. Introduction: In breast cancer with amplification of the CEP17 probe region and equivocal HER2 results by FISH, alternative HER2 probe sets (representing the 17p13, 17p12, and/or 17q21 regions) have been proposed as a method for establishing a stable denominator to reveal the tumor's "true" HER2 gene copy number status. However, the chromosome 17 regions represented by these alternative probes can themselves contain genomic loci that are commonly rearranged in cancer (TP53 loss at 17p13), (RARA amplification at 17q21), and inherited polyneuropathy (CMT duplications, deletions, or translocations at 17p12) resulting in false skewing of the HER2 gene/alternative probe ratio towards negative, positive, or equivocal. Methods: In the current study, DNA was extracted from IHCtargeted HER2 receptor "hot-spots" representing 25 FFPE breast tumors previously characterized by FISH for HER2 gene copy number. Tumor DNA (test) and human genomic DNA (reference with known diploid HER2 gene copy number) (Promega, The Journal of Molecular Diagnostics ■ jmd.amjpathol.org Madison WI) were differentially labeled with Alexa Fluor dyes, and competitively hybridized to a custom-designed oligonucleotide genomic DNA microarray with highdensity probe coverage of the HER2 amplicon on chromosome 17 (Agilent Technologies, Santa Clara CA). The array design includes over 4,600 chromosome 17 probes representing the p arm, q arm, telomeric and centromeric regions with 66 tiling probes over the HER2 (ERBB2) gene. Following hybridization, average HER2 gene copy number was calculated for each tumor sample by converting mean log2 signal intensity ratio value into genomic region copy number adjusted for % clonal fraction and experimentally established log2 ratio compression of the assay. Results: All samples yielded adequate DNA for array CGH analysis and HER2 gene copy number was scored as: HER2-Negative, < 4 copies, HER2-Low, 4-6 copies, HER2-Positive, > 6 copies. In 25/25 (100%) of cases, HER2 gene copy number results were consistent with FISH results: Conclusions: Calculation of average HER2 gene copy number using array CGH derived mean log2 ratios with reference DNA of known diploid HER2 gene copy number circumvents ratio skewing associated with unstable chromosome 17 reference probes, and provides an alternative high-resolution method for determining HER2 status in FISH equivocal breast cancer. R. Ruiz-Cordero, A. Khanna, G. Lyons, R. Bassett, M. Guo, J. Heymach, R. Luthra, S. Roy Chowdhuri MD Anderson Cancer Center, Houston, TX. Introduction: Comprehensive molecular profiling of lung adenocarcinomas (LADC) has identified three different molecular subtypes: terminal respiratory unit (TRU) based on EGFR, proximal-inflammatory (PI) based on TP53, and proximalproliferative (PP) based on KRAS mutations. Herein we correlate results of FISH and Next-Generation Sequencing (NGS) in LADC cytology samples to stratify by these molecular subtypes. Methods: LADC patients with fine needle aspirations between May 2010 and October 2015 for diagnosis and staging were identified to collect demographic, clinical, FISH (ALK, ROS1, MET) and NGS data. Frequency tables with number of patients and proportions for categorical variables. To assess relationship between the molecular subtypes and age, sex, ethnicity, smoking, vital and aneuploidy statuses, cross-tabulations and Chi-square or Fisher's exact tests were used. Results: 283 LADC patients were included (mean age=65y, female=53%, Caucasian=76%, Asian=9%, other=14%, current/former smokers=77%, Alive=65%). 82% of cases were run on NGS, 14.5% on Sanger or Pyrosequencing platforms and 3.5% did not have molecular data. NGS yielded positive mutations in 93% of cases with TP53 (43%), KRAS (34%), and EGFR (25%) being the most frequently mutated genes. Molecular results showed TRU=14%, PI=16% and PP=16%, EGFR/TP53 and TP53/KRAS co-mutation=19%, other genes mutated=12%, no mutations detected=18%. In keeping with the Cancer Genome Atlas Research Network (TCGA) results, we also found that TRU subtype is more frequent in Asian women (p=0.01) and never-smokers (p<0.0001), while least associated with ALK/ROS1/MET positive tumors (p=0.0089). The PI subtype did not show any significant differences by ethnicity (p=0.90), gender (p=0.18), smoking (p=0.17), vital status (p=0.74), or FISH results (p=0.12); however this group had a significantly higher number of cases with aneuploidy and ALK/ROS1/MET positivity (p=0.0012). The PP subtype was significantly associated with Caucasian women (p=0.003) and smokers (p=0.0002), but was not associated with ALK/ROS1/MET positivity (p=0.0161) or aneuploidy (p=0.61). Among patients with mutations detected by NGS, about 25% were deceased, compared with 76% of patients with no mutations detected by our NGS panel (p=0.0135). Conclusion: Our study confirms that cytology specimens are suitable for FISH and molecular testing. Whereas the clinical significance of aneuploid cells found in FISH in cases of LADC is not completely understood, aneuploidy seems to be associated to the PI molecular subtype. Our results confirm previous associations identified by TCGA as well as new additional information. Further studies evaluating the possible worse scenario of NGS-mutation-negative patients are in progress to corroborate this finding. Skipping Using nanoString Digital nCounter Technology in a Clinical Setting L. Borsu, K. Mullaney, R. Benayed, P. Chi, K. Busam, P. Paik, A. Drilon, M. Ladanyi Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: There is an increasing need to screen patient tumors for certain oncogenic alterations that are manifested exclusively or most clearly at the RNA level, such as ALK Alternative Transcription Initiation and mutations causing MET exon 14 skipping. A novel isoform of ALK gene initiating from a de novoalternative transcription initiation (ATI), ALK ATI , has recently been shown to be oncogenic in melanomas and other cancers (PMID: 26444240). Lung cancers harboring a wide variety of mutations causing MET exon 14 skipping have shown excellent responses to MET inhibitors (PMID: 25971939). Here, we report our clinical validation of the nanoString nCounter RNA detection technology for RNA-based detection of these alterations and its benchmarking against targeted RNA and DNA NGS assays. Methods: To evaluate the digital nCounter RNA detection technology, 8 cell lines, 12 melanoma and 5 lung adenocarcinoma FFPE specimens also tested by Archer targeted RNA-Seq sequencing were selected to evaluate ALK ATI transcript expression. Twelve FFPE lung carcinoma specimens previously testedby MSK-IMPACT NGS assay (PMID: 25801821) were selected to assess MET transcripts lacking exon 14. RNAs were extracted from unstained FFPE sections. The digital nCounter RNA detection 250ng RNA with barcoded Elements Reporter, biotin-labeled Capture Tags and two target-specific probes. After hybridization, excess tags and probes were removed while hybridized complexes bound to streptavidin were counted for RNA acid target to calculate fold-changes in gene expression. Results: Archer targeted RNA-Seq data for 25 ALK ATI samples (TAT 3.5 days) and targeted DNA NGS data for 12 MET Exon 14 samples (TAT 21 days) were compared to digital RNA detection results for the same cases (TAT 2 days). The expected ALK ATI and MET Exon 14 results were found in 23 (92%) and 11 (91.5%) samples using the digital RNA detection technology, respectively. In addition, 8 (17%) ALK ATI FFPE RNA samples that failed quality control metrics for the Archer RNA-Seq assay, nonetheless passed quality thresholds for nanoString and generated satisfactory data by digital nCounter detection. Conclusions: The digital RNA detection technology identified the correct variant transcripts in 34 (92%) samples tested by other clinical laboratory methods. This platform does not require cDNA synthesis, amplification, or library preparation thereby decreasing technical variation and enabling greater tolerance of RNA degradation and inhibitors associated with the formalin fixation process. Our results show that the digital nCounter expression technology can rapidly and accurately detect and quantify aberrant, oncogenic transcripts in FFPE RNA in a clinical setting. M. Zhu, B. Sundman, M. Atchley, D. Winner, N. Sadri University Hospitals Case Medical Center, Case Western Reserve University, Cleveland, OH. Introduction: About 40% to 50% of metastatic melanomas harbor a mutation in BRAF, with V600E and V600K being the most common mutations which are targeted by several first line FDA-approved therapies. In addition, non-V600E/K mutations in BRAF and mutations in KIT and NRAS have been shown to respond to targeted therapy in case studies and are currently being better characterized in clinical trials. Although allele-specific PCR can be used for detection of BRAF V600E mutation, some clinicians have more recently advocated for use of targeted-NGS in patients with advanced melanoma. Methods: To help guide the testing on metastatic melanoma cases in our institution, we analyzed all samples sent to UHCMC molecular lab for BRAF V600E PCR and targeted-NGS testing during 2015 and 2016. Test results, turnaround times, and costs were compared between the two tests. Samples which were negative for BRAF V600E by allele-specific PCR were tested by NGS using the Ion Torrent PGM and Ampliseq Cancer Hotspot Panel v2 to look for additional mutations. Results: Lab-developed allele-specific BRAF V600E PCR is less expensive with shorter turnaround time, 3.2 +/-0.3 days as compared to targeted-NGS assay that is roughly 4x more expensive with longer turnaround time, 6.6 +/-1.6 days. In the samples sent for testing by allele-specific PCR, 25% are positive for V600E. In samples sent for targeted-NGS, 24% are V600E positive, 16% harbor other actionable BRAF mutations (ie. V600K, complex mutations involving V600, and L597R), and 16% harbor other potentially actionable mutations in other genes (ie. NRAS, GNAQ). Of the mutations detected by NGS testing, 57% of potentially actionable mutations would not be detected by allele-specific PCR alone. Similarly, of the samples negative for BRAF V600E by allele-specific PCR, 72% harbor potentially relevant mutations when analyzed by targeted-NGS (12% harbor another V600 BRAF mutations and 60% harbor mutations in KIT or NRAS), all of which have either an FDA-approved therapy or an available clinical trial. Conclusions: Although allele-specific BRAF V600E PCR is less expensive with a shorter turnaround time, it does not detect over two-third of potentially actionable mutations in metastatic melanoma. These findings need be viewed in the context of the institution's participation in clinical trials and oncologist preference for targeted therapy, immunotherapy, and off-label drug use. Within our institution, we recommend screening patients with allele-specific PCR and reflexing negative cases for targeted-NGS, or using targeted-NGS upfront to maximize patient benefit. This study provides guidance on choosing appropriate molecular test for metastatic melanoma in a tertiary medical center harboring on-site NGS testing. Introduction: KIAA1549-BRAF rearrangements occur in approximately two thirds of pilocytic astrocytomas. A duplication within 7q34 encompassing a portion of both genes results in fused copies of the 5' end of KIAA1549 and the 3' end of BRAF and the constitutive activation of BRAF kinase activity. The detection of this rearrangement can be useful in the diagnostic distinction of pilocytic astrocytoma from diffuse astrocytoma and other low-grade gliomas. We sought to compare two methods -fluorescence in situ hybridization (FISH) and reverse transcription PCR (RT-PCR) -for the clinical detection of BRAF rearrangements in pilocytic astrocytoma. Methods: We designed an RT-PCR assay to detect the three most common KIAA1549-BRAF fusion transcripts -exon 16/exon 9, exon 15/exon 9, and exon 16/exon 11 -as well as the more unusual exon 15/exon 11 transcript. Together these encompass approximately 98% of KIAA1549-BRAF fusions. In addition, we developed a FISH assay with probes flanking all breakpoints described jmd.amjpathol.org ■ The Journal of Molecular Diagnostics in BRAF rearrangements. A total of 31 cases of pilocytic astrocytoma were evaluated by both methods, 9 of which were previously found to have a BRAF rearrangement at another institution. Results: BRAF rearrangements were identified by RT-PCR and/or FISH in 18/22 (81.8%) of pilocytic astrocytomas without a previously identified rearrangement and in all 9 cases found to have a rearrangement at an outside institution. One of 31 cases failed both RT-PCR and FISH. Among the remaining cases, 3 cases were not scorable by FISH but were positive by RT-PCR (with a weak control product) and one case was positive by FISH but was inadequate by RT-PCR due to a very weak control. Among cases positive by RT-PCR, 20/25 (80%) demonstrated an exon 16/exon 9 fusion, 4/25 (16%) an exon 15/exon 9 fusion, and 1/25 (4%) an exon 15/exon 11 fusion. No exon 16/exon 11 fusions were identified; however, evaluation of a plasmid control confirmed the ability of RT-PCR to identify this transcript. Concordance was observed between FISH and RT-PCR for 25/26 cases evaluable by both techniques. One case was negative by RT-PCR but positive by FISH with the typical doublet pattern, suggestive of an uncommon KIAA1549-BRAF fusion transcript not detectable by the RT-PCR primer design. Conclusions: FISH and RT-PCR both represent accurate and complimentary methods for detection of BRAF rearrangements which are useful in the diagnosis of pilocytic astrocytoma. Whereas RT-PCR is more specific for KIAA1549-BRAF fusions and had slightly fewer failures, FISH is more comprehensive and may be useful for the detection of rare breakpoints, less common BRAF fusion partners, and BRAFrearrangements in other tumor types. H. Mellert, L. Jackson, W. Hahn, N. Dupuis, A. Weaver, J. Greer, G. Pestano Biodesix Inc., Boulder, CO. Introduction: Blood-derived proteomic signatures have clinical utility, with one example, VeriStrat (VS), measuring acute phase reactant proteins in serum. Patients with a POOR (VS-P) prognostic classification have worse clinical outcomes relative to those with a GOOD (VS-G) status. A VS-P result limits standard of care therapeutic options, thus clinical trials may be a good option. A global Phase II study, FOCAL is being conducted that examines the addition of ficlatuzumab to erlotinib for first line EGFR sensitizing mutation positive, VS-P patients. In contrast, treatment options for patients with NSCLC and EGFR wild-type (wt), VS-P status are limited. For this study, we first profiled serum using MALDI-ToF for VeriStrat-POOR classification in donor specimens. Circulating-free DNA (cfDNA) was then profiled in matched plasma specimens with a targeted next-generation sequencing (NGS) panel to identify actionable somatic variant mutations. Methods: Plasma samples determined to be EGFR wt by droplet digital PCR and VS-POOR by MALDI-ToF were analyzed by NGS with a targeted 15-gene panel. Screening of cfDNA isolated from the plasma of 11 VS-P patients with advanced stages of NSCLC are presented. Results: As expected, EGFR sensitizing variants were detected only in the positive control but not among any of the EGFR wt pre-screened VS-P cohort. Six of the 11 plasma samples contained TP53 mutations, which included missense, splice-site and frame-shifts. Although targeted therapies for p53 are not currently approved by the FDA, several pre-clinical and clinical trials are underway with either compounds such as the p53 re-activating agent APR-246 (a PRIMA-1 analogue), or with wt p53 gene-therapy such as SGT-53. Conclusions: Targeted profiling using NGS on cfDNA from plasma can identify actionable mutations in patient samples with a poor prognosis as determined the VeriStrat test. It is likely that other targets associated with current therapy may be uncovered with additional profiling. Expanded profiling of EGFR wt/VS-P patients with a 35-gene NGS panel that measures additional somatic variants, indels and amplifications is underway. We expect that these studies will yield targets that could provide options for patients with advanced stages of NSCLC. A. Cheng, N. Hernandez, R. Sunnadeniya, D. Meza, J. Whiting, W. Roberson, M. Carter, X. Fang Thermo Fisher Scientific, Austin, TX. Introduction: As personalized cancer care evolves, the patient's nucleic acid becomes ever so important to provide valuable information regarding their genetic makeup and disease state. Common sample types for these analyses include biopsies, which can be very limited in material making the downstream measurement of more than one analyte rather difficult. Obtaining another biopsy, using a different section or splitting the sample can be problematic because of tumor heterogeneity. Even adjacent areas of the same tumor tissue can result in different RNA/DNA profiles so the ability to isolate multiple analytes from the same sample offer a number of benefits, which include preserving samples and data consistency eliminating any sample to sample variation. As more tests are developed to simultaneously monitor genetic alterations, there is a strong need to efficiently isolate both DNA and RNA from the same starting sample in a format compatible with highthroughput processing. Methods: Here we describe a novel chemistry using magnetic beads and robust workflow that will eliminate extraction variabilities and process a large number of samples with consistency and ease. We've developed both manual and automated protocols on the KingFisher Flex and Duo Prime purification systems. We have extracted from a number of different FFPE cancer tissues including breast, lung, colorectal, and melanoma from tumor resections, core needle biopsies (CNBs) and fine needle aspirates (FNAs) for various downstream applications such as real-time PCR and targeted sequencing with different panels. Results: High sequencing metrics were met from tumor resections aged 5 to 25 years old, and from CNBs and FNAs. DNA sequencing metrics include >95% uniformity, >90% end to end reads and >95% no strand bias. RNA sequencing metrics include expression of all 5 endogenous controls and high mean read lengths >80bp. FFPE samples with known variants or fusions were extracted and analysis for variant calling or fusion detection confirmed the corresponding variants. Extractions processed manually or automated on the KingFisher yielded a high concordance of hotspot variants identified. We have also titrated tissue inputs from 200mm 2 down to 5mm 2 showing sensitivity of variant detection is not affected when working with small pieces of tissue or low amounts of tissue. Conclusions: In summary, we have a robust workflow that allow for the extraction of DNA and RNA from the same FFPE sample. Samples can be processed manually or with automation on the KingFisher purification systems with concordant variant detection. From tiny amounts of material such as CNB's or FNA's to bigger tissues such as tumor resections, samples generated quality sequencing data meeting all of our core sequencing metrics. The KRAS oncogene is mutated in approximately 35% to 45% of colorectal cancers and NRAS in 1% to 6%. Studies have shown that EGFR monoclonal antibody therapy is unlikely to be beneficial in tumors with any KRAS or NRAS mutations. NCCN guidelines have recommended that cetuximab and panitumumab for the first-line treatment be given only to patients who are negative for both KRAS and NRAS mutations. Our KRAS, NRAS extended gene mutation analysis assays will detect mutations in exons 2 (codons 12, 13), 3 (codons 59, 61) and 4 (codons 117, 146), allowing the determination of drug response. In this study, we have evaluated the clinical and analytical performance features of this assay. Methods: Genomic DNA was isolated from FFPE tumor specimens. Exons 2, 3 and 4 of the KRAS and NRAS genes were subjected to SNaPshot multiplex PCR and primer extension for mutation detection. DNA from colorectal cancer FFPE specimens and synthetic oligonucleotides were used to evaluate accuracy, repeatability, reproducibility and analytical sensitivity of the assays. Results: Of the specimens tested during validation, 21 specimens with known mutations in KRAS and 14 in NRAS had SNaPshot analysis results that were 100% concordant. The 21 synthetic oligonucleotides positive for mutations in KRAS and 40 in NRAS also showed 100% concordance. Repeatability and reproducibility were 100% concordant. This assay can detect 5% of mutant DNA in a background of wildtype genomic DNA when the input DNA is 25ng or more. The KRAS and NRAS assays have been offered as clinical tests based on the successful performance features. From the 1272 KRAS specimens tested for the past year, 55.58% had no mutation detected, 39.31% had a KRAS mutation detected. Among the 500 specimens with a KRAS mutation, 87% were found in Exon 2, 6.8% in Exon 3 and 5.6% in Exon 4. Three specimens (0.6%) had double mutations that were most likely due to tumor heterogeneity. From the 1496 NRAS specimens tested for the past year, 91.24% had no mutation detected, 4.55% had a mutation detected. Among the 68 specimens with an NRASmutation, 41.18% were found in Exon 2, 58.82% in Exon 3 and none in Exon 4. In this data set, 236 specimens were tested for both KRAS and NRAS, 50% had no mutation, 39.41% had a KRAS mutation and 5.93% had an NRAS mutation. The mutation distribution of our data set is in agreement with other published studies. There were 1.63% specimens that had partial results and 3% that had no results due to FFPE specimen degradation or low DNA yield. Conclusions: The KRAS and NRAS extended gene mutation detection assay using multiplex SNaPshot method is a robust, reproducible, sensitive, and fast assay for molecular diagnostic utilization on targeted therapies in colorectal cancer. Introduction: The PathVysion, FDA-approved dual probe HER2 fluorescence insitu hybridization (FISH) assay provides the ratio of HER2 to CEP17, a centromeric enumeration probe for chromosome 17. Several reports have suggested that CEP17 copies are increased in many breast cancers which were traditionally labeled as "polysomy" cases. However, recent data questioned the existence of true polysomy in breast cancer and increased CEP17 signals were thought to be due to pericentromeric amplication which might then skew the ratio of HER2/CEP17 underestimating the HER2 status. The aim of the current study was to analyze the utility of an alternative chromosome 17 reference locus (D17S122) to assess HER2 gene status accurately in cases with unusual FISH signal patterns, viz. CEP17 aneusomy, average HER2 th borderline HER2/CEP17 ratio. Methods: We retrospectively reevaluated the HER2 FISH status of 25 invasive breast cancers accessioned in the year 2016 in the Division of Molecular Pathology, The Journal of Molecular Diagnostics ■ jmd.amjpathol.org Tata Memorial Centre, which displayed a mean CEP17 copy number greater than 3. -<6 copy numbers or with HER2/CEP17 ratio >1.8-2.2 by PathVysion (Abbott Molecular Inc., Des Plaines, IL, USA) HER2 DNA Probe Kit. The cases were reflex tested using alternate, non centromeric FISH probe by ZytoLightSPEC/D17S122 (Zytovysion, Bremerhaven, Germany). HER2 status was interpreted in accordance with the ASCO/CAP 2013 guidelines. Results: Twenty five cases comprised of mean CEP17 -<6 copy numbers (n=8), and HER2/CEP17 ratio >1.8 -2.2 (n=8). Of 9cases of polysomy with mean CEP17 centromeric probe. Two of nine polysomy cases displayed an unusual HER2/CEP17 colocalization with coamplication pattern. Though the amplified HER2 status was retained in the latter cases, the colocalization pattern was not seen with the alternate probe.Out of 8 cases with mean HER2 cases HER2 gene status was upgraded from equivocal to amplified. In rest of the cases (n=11), the results were concordant. Conclusions: Our results support the finding that polysomy of chromosome 17 is indeed a rare event in breast cancers and increased CEP17 signals indicates pericentromeric amplification. Our data also highlight the limitations of currently used centromeric probe (CEP17) for HER2 FISH testing and demonstrates that the usage of non-centromeric chromosome 17 reference probe alters the HER2 status thereby increasing the eligibility for antiHER2 based therapy in a significant proportion of patients. We thus recommend the use of non centromeric chromosome 17 reference probe in cases with CEP17 aneusomy and equivocal HER2 status. Introduction: Human papillomavirus (HPV) is known to be associated with squamous cell carcinomas of the head and neck (HNSCC), especially oropharynx. Whereas patients whose tumors are HPV positive tend to have a better response to treatment, most of the clinical tests for HPV detection are DNA-based and use chromogenic in situ hybridization (CISH) as the preferred modality. Whereas HPV DNA test (CISH) has good sensitivity, its specificity is poor and it does not correlate with biologically active infection. The APTIMA HPV Assay uses transcription mediated amplification (TMA), thereby targeting the E6/E7 mRNA of 14 high-risk HPV types and has shown good sensitivity and specificity in liquid-based cytology specimens. Therefore, the objective of this study was to evaluate the performance of this assay on formalin-fixed paraffin embedded (FFPE) samples from patients with HNSCC. Methods: Twenty-two FFPE samples of HNSCC were tested. Two non-SCC cases and 10 papillomas that were secondary to low-risk HPV were also evaluated. Macrodissection was performed on all the specimens followed by RNA extraction using the Promega LEV FFPE RNA kit. Purity and concentrations were measured using Nanodrop and QuantiFluor on Quantus. All samples were run in duplicate with an input of at least 40ng/tube using the APTIMA HPV Assay on Panther instrument (Hologic). In addition to sensitivity and specificity, samples were also tested for accuracy, precision, tech-to-tech and instrument-to instrument variability. Results were compared to a national reference laboratory's HPV high-risk DNA ISH assay. Results: Most common site of involvement was tonsil (n=8). Majority of the tumors were moderate to poorly differentiated and non-keratinizing. 21 of the 22 cases were positive for high-risk HPV by APTIMA assay. Both non-SCC and all papillomas were negative for high-risk HPV by APTIMA. In contrast, the HPV DNA ISH assay was positive in 12 of the 22 samples, with 4 negative and 6 indeterminate results. One case was negative by both ISH and APTIMA assay. Serial dilutions revealed a minimum RNA input of 2.5ng/tube and a viable tumor cut-off of at least 10% to achieve a positive result by APTIMA assay. Thus, sensitivity and specificity of APTIMA was 95.5% and 100% respectively. Tech-to-tech and instrument-to-instrument variability was 100% at 40ng/tube input, however at lower inputs there were frequent occurrences of variability between replicates. Conclusions: APTIMA assay is highly sensitive and specific for detection of highrisk HPV subtypes associated with HNSCC. It offers improved specificity compared to ISH and can be used reliably even with FPPE tissues. It also correlates with biologically active infection, consequently improving clinical management of patients. and PIK3CA in Pulmonary Adenocarcinoma Using iPLEX HS, a New Highly Sensitive Assay for MassARRAY R.T. Birse 1 , D. Irwin 2 1 Agena Bioscience, San Diego, CA; 2 Agena Bioscience, Herston, Queensland, Australia. Introduction: Increased early detection and personalized therapy for lung cancer have coincided with greater use of minimally invasive sampling techniques such as endobronchial ultrasound and needle core biopsies. These procedures offer lower risk alternatives for tissue diagnosis. However, they also generate analytical challenges such as a requirement for robust detection of low level somatic mutations, particularly when the starting sample is small or has sparse tumor cells. In this study we assessed 43 clinical cases of pulmonary adenocarcinoma (PA) previously tested for EGFR, KRAS, NRAS, and BRAF mutations using a novel multiplexed analytic approach that reduces wild-type signal and allows for detection of low mutation load approaching 1%, the iPLEX HS for MassARRAY (Agena Bioscience, San Diego, CA). Methods: Archived frozen deoxyribonucleic acid (DNA) samples were searched for PA cases previously tested for EGFR, KRAS, NRAS and BRAF mutations using the OncoFOCUS Panel v1.0 or v2.0 and the MassARRAY system. Specimens were deidentified prior to entry into the study. DNA from formalin-fixed, paraffin-embedded (FFPE) PA tissue samples. All histologic diagnoses were confirmed by a pathologist and minimum tumor cellularity for analysis was 20%. DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Boston, MA). Prior to repeat testing, all specimens were assessed for DNA integrity using the iPLEX Pro Sample ID, and all specimens with adequate amplifiable DNA were then interrogated with a new high sensitivity, single PCR reaction iPLEX HS panel that includes more than 34 common mutations in BRAF, EGFR, KRAS, NRAS and PIK3CA, both using the MassARRAY platform. Results: In 43 samples, mutations in KRAS (n=11; 11/43= 26%), BRAF (n=1; 1/43 = 2%), EGFR (n=6; 6/43 = 14%) and NRAS (n=1; 1/43 = 2%) were detected using iPLEX HS. When compared to previous results from the OncoFOCUS assay (sensitivity of 5% to 10% mutated allele burden), 2 additional EGFRmutations (L858R, G719A), 2 additional KRAS mutations (G12A, G12D) and one more BRAF mutation (V600E) were identified. The majority of these cases had smaller, more limited tissue specimens (three needle core biopsies, one pleural fluid cytology cell block, and one excision specimen). Previously undetected mutations in PIK3CA were identified in 4 cases (4/43 = 9%). Conclusions: 1) Use of a highly sensitive assay such as iPLEX HS and MassARRAY increases detection of clinically significant mutations in challenging PA samples: 12% more KRAS, EGFR and BRAF mutations were identified in this pilot study of 43 cases. This study is currently being expanded to include 185 archived patient samples. 2) PIK3CA mutations were detected in 9% of PA cases, one of which also had a mutation in NRAS (A59D). E.M. Hissong 1 , P. Zhang 1 , J. Shia 2 , R.K. Yantiss 1 , H. Fernandes New York, NY; 2 Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: The TCGA data identified four molecular signatures among gastric cancers: EBV positivity (EBV+), microsatellite instability (MSI), genomic stability, and chromosome instability. However, these data are heavily skewed toward signatures of common morphologic variants. Gastric carcinoma with lymphoid stroma (GCLS) is an uncommon subtype that shows a solid, or medullary, growth pattern with a dense lymphocytic infiltrate. Our aim is to characterize the molecular alterations in GCLS. The study group consisted of formalin fixed, paraffin embedded tissue sections from 30 GCLS resection specimens. All cases were subjected to in situ hybridization for EBER and immunohistochemistry (IHC) for HER2. DNA mismatch repair status was assessed using IHC (MLH1, MSH2, PMS2, MSH6) and PCR analysis for MSI. DNA was extracted from tumor and matched-normal specimens and sequenced on the Ion Proton using Comprehensive Cancer Panel (ThermoFisher) that targets 409 cancer-related genes. Data were analyzed using the Variant caller 4.4 software and Ion Reporter (ThermoFisher). Results: Five (17%) cases were EBV+ and microsatellite stable (EBV+), 12 were EBV-negative with MSI (MSI), and 13 were EBV-negative and MSS (EBV-/MSS). PIK3CA variants within the kinase domain were common in EBV+ tumors (43%), whereas 36% of MSI tumors showed helical domain variants. ERBB2 variants were more prevalent in EBV+ (43%) and MSS (27%) groups. Somatic alterations in KRAS were detected in 45% MSI cases. EBV-/MSS tumors often harbored TP53 variants (>80%). and 3 (23%) were genetically unstable with numerous alterations. Copy Number Alterations (CNA) were more common among MSS and EBV+ tumors than EBV-/MSS cancers, with frequent amplification of PIK3CA, JAK2, and MYC in EBV+ cases and amplifications in EGFR, JAK2 and BAI3 in the MSS group. Several tumors in all groups had amplifications and deletions in APC. Amplification of ERBB2 was lacking in GCLS, and only one case (3%) showed HER-2 staining by IHC. Conclusions: Molecular changes in GCLS are heterogeneous, but seem to cluster in three major groups: EBV+/MSS, EBV-/MSI, and EBV-/MSS. EBV+ tumors commonly show PIK3CA variants and high-frequency CNA, raising the possibility that affected patients may benefit from PI3-kinase inhibitors, or have a better prognosis than patients with EBV-/MSS tumors. R.N. Ptashkin, C. Pagan, S. Middha, R. Yaeger, M. Ladanyi, M. Berger, A. Zehir, J.F. Hechtman Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: Recurrent copy number gains of 20q have been observed in several malignancies. Despite several studies investigating 20q gain on a functional level, 20q gain in colorectal carcinoma (CRC) has not been methodically investigated in relation to clinicopathologic and molecular features including prognosis and response to treatment. In this study, we aimed to elucidate the molecular and clinical features of CRC harboring 20q gain. Methods: Tumor and matched normal DNA from patients with advanced CRC were sequenced using the MSK-IMPACT assay which detects somatic alterations in 410 cancer genes (PMID: 25801821). jmd.amjpathol.org ■ The Journal of Molecular Diagnostics Microsatellite instability (MSI) status was calculated using MSIsensor, a clinically validated assay that assesses microsatellite regions in next-generation sequencing (NGS) data. We calculated the Tumor/Normal log2 ratio (lr) for all 20q genes in the (0.45 -0.45 < lr < 0.45). Associations with clinical and molecular data were assessed by Fisher's test with multiple hypothesis testing correction. A Cox proportional hazard model was used to assess survival. Results: Of the 413 CRC samples from 401 patients, 150 (36.3%) had 20q gain, and 30 samples (7.3%) had 20q amplification. CRCs with 20q gain or amplification were more often KRAS WT (67% vs. 42%, p=<0.0001), BRAF WT (94% vs. 80%, p=0.001), TP53 mutant (86% vs. 61%, p<0.0001), arising as left sided primaries, (78% vs. 52%, p<0.0001) and microsatellite stable (>99% vs. 84%, p<0.001) in comparison to CRC with diploid 20q. For 390 patients for which survival data were available, 237 (60.6%) presented with stage IV metastatic disease and 117 (30%) progressed to stage IV during the course of study (median time to progression = 15.9mo). 20q amplification or gain was associated with longer survival from date of diagnosis with metastatic disease (Hazard Ratio = 0.52, p=0.004) on multivariate analysis controlling for age at diagnosis, RAS/ RAF and MSI status. Conclusions: Our data show that 20q gain or amplification is frequent, occurring in approximately 40% of CRC. This molecular event defines a specific subgroup with distinctive clinicopathologic features including left sided primary tumors, microsatellite stability, RAS/ RAF WT status, and better overall survival in the setting of metastatic disease. Introduction: Molecular profiling of DNA alterations and mutations has offered unprecedented insight into the causes and treatments of cancer. However, a comprehensive understanding of how these DNA alterations affect the expression of tumor-related genes remains unknown. Methods: We analyzed the expression of 315 cancer related genes across 7 different tumor types using RNA-seq data obtained from The Cancer Genome Atlas (TCGA) database. We compared the expression data to single nucleotide variant (SNV) and structural variants including copy number alterations (CNA) and translocations detected in the genomes of the same samples. Results: We found that, among the cancer-related genes analyzed, 5% of the time when a SNV is detected in the DNA, there is no detectable expression of that allele in the RNA. We further analyzed the correlation between somatic gene amplifications to the expression levels of the same gene. We found that, overall, CNA status was a very weak predictor of gene expression (r-squared = .00012) with CNA amplification only associated with expression in 8 genes correlated with CNA amplification at a level of r-squared >0.3. Finally, we examined the expression of ALK, a known oncogene involved in an intrachromosomal fusion (EML4-ALK) in 4-6% of lung adenocarcinomas that results in overexpression of ALK, and is associated with increased response to ALK-inhibitors. We confirmed that the EML4-ALK fusion resulted in ALK overexpression in lung cancer compared to cases with non-rearranged ALK. However, we found that in a subset of patients, particularly in breast cancer, ALK was overexpressed to the same degree (Z-score>10) as seen in lung cancer EML4-ALK fusions. Conclusions: Our analysis indicates that not all somatic DNA mutations are expressed in tumors, suggesting that concurrent RNA expression profiling would provide additional clinically-actionable information. We further conclude that increased ALK expression may be seen in cancer cases that do not harbor EML4-ALK fusions, raising the possibility of treating such tumors with ALK-inhibitors. R. Birse 1 , D. Irwin 2 1 Agena Bioscience, San Diego, CA; 2 Agena Bioscience, Herston, Queensland, Australia. Introduction: The 2016 National Comprehensive Cancer Network treatment guidelines (NCCN Guidelines) recommend that patients with metastatic colorectal adenocarcinomas (mCRC) have tumor tissue genotyped for RAS (KRAS and NRAS) mutations, and suggest BRAF V600E testing be done as well. In the United States, clinical laboratories offer mutation testing using many different platforms with a wide range of analytical sensitivity. Despite considerable progress, analytical challenges remain to be resolved such as the need for reliable detection of low abundance somatic mutations, particularly in small specimens with a low percentage of tumor cells. In this study we assessed 46 clinical CRC cases previously tested for EGFR, KRAS, NRAS, and BRAF mutations using a novel multiplexed analytic approach that reduces wild-type signal and allows for detection of low mutation load approaching 1%, the iPLEX HS for MassARRAY (Agena Bioscience, San Diego, CA). Methods: Archived frozen DNA samples were searched for CRC cases previously tested for EGFR, KRAS, NRAS and BRAF mutations using the OncoFOCUS Panel v1.0 or v2.0 and the MassARRAY system. Specimens were deidentified prior to entry into the study. DNA originated from formalin fixed paraffin embedded (FFPE) CRC tissue samples and all histologic diagnoses were confirmed by a pathologist. DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Boston, MA). Prior to repeat testing, all specimens were assessed for DNA integrity using the iPLEX Pro Sample ID and all specimens with adequate amplifiable DNA were then interrogated with a new high sensitivity, single PCR reaction iPLEX HS panel that includes more than 34 common mutations in BRAF, EGFR, KRAS, NRAS and PIK3CA, both using the MassARRAY platform. Results: The OncoFOCUS assay has a mutation detection limit of 5% to 10% mutant allele burden, whereas the iPLEX HS is more sensitive at approximately 1%. In this pilot study of 46 patient samples, the iPLEX HS panel confirmed all previously identified CRC mutations in KRAS (n=10; 10/46=22%), NRAS (n=1; 1/46=2%), and BRAF (n=8; 8/46 = 17%). Previously undetected mutations in PIK3CA were also identified (n=10; 10/46=22%), and 5 of the PIK3CA mutations coexisted with other driver mutations, including KRAS G12D and G12C; NRAS A59T; and BRAF V600E (2 cases). Conclusions: 1) In this small sample set, lowering the assay sensitivity from 5% to about 1% mutant alleles detected all previously known mutations in KRAS, NRAS and BRAF and 10 previously undetected mutations in Pi3KCA in CRC samples. This study is currently being expanded to include 286 archived patient samples. 2) PIKC3A mutations may coexist with other driver mutations in RAS or BRAF in about 11% of CRC cases. Histologically, these entities may present a diagnostic challenge and the usage of molecular methods such as FISH, RT-PCR/next-generation sequencing to detect specific translocations is currently a routine part of diagnosis. Methods: In this study, we evaluated the presence of EWSR1 translocation with FISH and RTnext-generation sequencing using the anchored-multiplex PCR technology (AMP). Results: Among the 18 SBRCTs where there was strong histologic/immunehistochemical evidence suggesting the presence of a EWSR1 translocation, concordant positivity was seen by FISH and AMP in 13 cases (72%). In the 5 negative EWSR1 FISH cases, EWSR1 translocations were detected by AMP in 4 cases. Additional orthogonal methods such as partner FISH and/or RT-PCR also confirmed the presence of EWSR1 rearrangements. Conclusions: Consistent with the reported sensitivity of 82% to 97% of EWSR1 FISH in the literature, our results demonstrate the potential for false negativity when using EWSR1 FISH as the only molecular diagnostic method. In cases with classic morphology and immunehistochemical profile, additional molecular testing such as AMP should be considered. Introduction: Premalignant epithelial lesions are commonly diagnosed yet current medical practice is limited in its ability to stratify them for risk of progression to carcinoma. We cataloged the genomic mutation spectrum in lesions adjacent to invasive gastric adenocarcinoma to explore how frequently premalignant lesions shared mutation signatures in common with those found in the malignancy. Methods: Macrodissected cancers and 30 premalignant lesions from 25 gastric carcinoma patients' paraffin embedded tissue weresequenced across hotspots in 26 human cancer genes (Illumina TruSight Tumor 26 reagents on a MiSeq). Also sequenced were an additional 4 premalignant gastric lesions from 4 patients without subsequent cancer on long-term follow-up. Non-synonymous mutations and small indels at allele frequency >5% with population frequency < 1% were cataloged. Results: 24/25 invasive carcinomas had a detectable mutation (range 1 to 4 mutations per tumor) in TP53, KRAS, APC, PIK3CA, FBXW7, CDH1, SMAD4, PTEN, MSH6, MET, or ALK. All 30 dissected premalignant lesions yielded adequate read depth (>500x) for interpretation of hotspot mutations. Whereas 23 premalignant lesions had no detectable mutation, 7 (23%) had at least one mutation (range 1 to 2 mutations of TP53, APC, MET, or CDH1) that nearly always matched mutation(s) present in the adjacent invasive cancer. Average mutant allele fraction in these 7 premalignant lesions was 31 (range 8 to 53%). In two patients there was evidence of clonal evolution whereby the premalignant lesion harbored only one of two mutations detected in the carcinoma. Only 1/30 premalignant lesions had a mutation that was absent in the matched cancer, suggesting that variants in these 26 cancer-related genes may be uncommon unless the patient has cancer or cancer precursor lesions. A non-canonical NRASmutation was also identified in an intestinal metaplasia (IM) lesion from one of four IM patients who did not progress to cancer on long term follow-up. Conclusion: Successful sequencing of dissected premalignant mucosal lesions is feasible. Genomic findings reflect intrinsic biology of precursor lesions and suggest they can be clonally related to cancer tissue of the study provides evidence that real-time PCR assays enhanced by ARMS, Scorpions or PANClamp technologies despite occasional lack of reproducibility (in more challenging samples) could serve well as a more sensitive testing option in EGFR analysis. O.A. Shetty, M.R. Ramadwar, M.Y. Gurav, T.D. Pai, V. Kamble, G. Chinnaswamy, T. Vora, S. Qureshi, S. Dhanavade, S. Tambe, S.B. Desai Tata Memorial Hospital, Mumbai, Maharshtra, India. Introduction: Neuroblastoma (NB) is a primitive neuroectodermal tumor and the most common neoplasm of childhood presenting with genetic instability in which prognosis and response to therapy varies greatly. Genomic amplification of MYCN oncogene has been used to predict outcome in NB for over 30 years, however clinical behaviour of MYCN non-amplified cases are ambiguous. Recent advances enables detailed analysis of NB genome, leading to the identification of new prognostic markers such as loss of heterozygosity at chromosome arms 1p, 3p, and 11q, unbalanced gain of 1q, 17q and numerous mutations in key genes such as ALK and DDX1. Analysis of these will in turn help in better patient stratification.The present study is aimed at characterizing the genetic abnormalities in MYCN non amplified NBs along with clinical and histological parameters. Methods: The present study was conducted on 26 cases of NB reported in Tata Memorial Hospital during 2014 to 2016. MYCN amplification was assessed by fluorescence in situ hybridization (FISH) using Vysis LSI N-myc dual colour probe (Abbott, USA). Multiplex Ligation Probe Amplification assay (MLPA) was used to assess the genetic profile of cases using probe mixes P251, P252 and P253 (MRC Holland, Amsterdam). MLPA is used to study segmental chromosomal aberrations at chromosomes 1p, 3p, 14q and 11q, unbalanced gain of 1q, 11p and 17q and genomic aberrations in ALK, DDX1. Coffalyser software was used for data analysis and SIOPEN guidelines were used for nomenclature. Results: The study included patient's in the age group between 1 month to 2 years with a median age of 1 year and male to female ratio of 2.5:1. Interpretable MLPA results were obtained in 22 cases of which 18 (81%) were MYCN non amplified and 4 (19%) were amplified. Among the MYCN non amplified (n=18), 11q loss in 5 (28%) cases, 17q amplification in 7 cases (39%), 17q gains were observed in 4 (22%) cases, followed by 3p loss in 5 (28%) cases, 1p loss in 3 (17%) cases. ALK gene gains were seen in 7 (39%) cases.17q amplification was found to be mutually exclusive with MYCN non amplification in this series (p=0.014). 17q amplification was found to be correlated with unfavourable histology, high MKI index and poor clinical response to treatment though statistically not significant. Conclusion: 17q amplification was associated with aggressive behaviour with poor clinical outcome in MYCN non amplified NB. This indicates the need to study the genetic profile of MYCN non amplified cases to predict prognosis and clinical management of patients. MLPA analysis hence becomes an important tool which can be used effectively and simultaneously to detect multiple genomic imbalances and these changes are being utilized to classify NB by prognostic subtypes. E. Golomb 1 , I.B. Achache 1 , R. Beeri 1 , T. Grenader 1 , I. Haviv 2 1 Shaare Zedek Medical Center, Jerusalem, Israel; 2 Bar Ilan University, Safed, Israel. Introduction: Differences in the prevalence of RAS/BRAF somatic mutations in colon cancers among ethnic groups have been suggested, raising the possibility that epigenetic environmental and genetic factors govern the selection of somatic mutations. The population examined in Jerusalem consists of two main populations, Jews and Arabs. We compared the spectrum of RAS/BRAF mutation in metastatic colorectal cancers between these populations. Methods: During the period between 2014 and May 2016, 244 extended RAS / BRAF analyses were performed at Shaare Zedek Medical Center in Jerusalem, Israel. Seventy of these patients (25.6%) were Palestinian Arabs. The rest of the patients were Jews and tourists from Eastern Europe. KRAS codons 12/13and BRAF V600E mutations were analyzed by allele discrimination assay. The other mutations were detected by Sanger sequencing. Results: The overall incidence of WT versus mutated RAS/BRAF was similar between the groups. However, the Arab population of Jerusalem showed a specifically high prevalence of KRAS mutations in exons 3 or 4 (15.7%) compared to the Jewish / non Arab population (2.9%, p<0.01 by Fisher exact test). This was balanced by a lower rate of NRAS and KRAS exon 2 mutations in Arab patients. Conclusion: Prevalence of somatic driver mutations may differ between ethnic groups, so that a mutation that is considered rare in one population may be common in others. The genetic / epigenetic basis for these differences may shed light on the mechanism of carcinogenesis. This may stem from a different genetic background among populations, or from different environmental factors, such as dietary and smoking habits. Introduction: Next-generation sequencing has enabled large-scale studies of cancer genomes to identify mutations. These tumor-specific somatic changes offer great potential as biomarkers to inform clinical practice, including detection, personalized medicine and disease monitoring. The ability to detect the mutational landscape of tumor cells using non-invasive approaches (liquid biopsies) could enable rapid, personalized variant screening and efficient monitoring of therapy. Unfortunately, limitations of current standard technologies include lack of sensitivity and the scalability for a rapid analysis, especially in the case of small amounts of circulating cell-free DNA (cfDNA) from blood plasma samples. Methods: Here we present a streamlined, DNA library preparation workflow that enables sample multiplexing for Illumina next-generation sequencing with high sensitivity and reproducibility for variant detection using minuscule amounts of circulating plasma DNA. Mixed plasma from healthy female and male donors were used to generate different ratios of the X and Y chromosomes for determining sensitivity of copy number detection. Exome enrichment of the libraries was done using Nimblegen's VCRome kit for in-depth analysis of single nucleotide variations. Results: With as little as 500 pg of cfDNA we were able to generate sequencing libraries for whole genome (WGS) and whole exome (WES) with good GC coverage. All samples with 10ng input DNA have >99.5% reads mapping back to the reference genome and >99% have unique sequences. Even at the lowest input of 500pg >98% of reads map and 99% have unique sequences. We accurately determined the ratio of sex chromosomes in a mixed sample down to 5% of Y chromosome input. Exome target enrichment allowed recovery of known variants (>80%), showed good correlation with expected allele frequencies, and detection was consistent and reproducible across different starting concentrations of DNA. Conclusions: The simple and rapid workflow allows a quick and sensitive approach to detect genomic variations. This method could facilitate biomarker identification to supplement diagnostic, prognostic and therapeutic applications. We also achieved sufficient library complexity for highconfidence copy number detection and variant recovery. J. Tull 1 , J. Yang 2 , W. Song 2 , S. Zhang 2 1 Upstate Medical University Hospital, Syracuse, NY; 2 Upstate University Hospital, Syracuse, NY. Introduction: Activating mutations in the epidermal growth factor receptor (EGFR) gene in non-small-cell lung cancers (NSCLC) are predictive of response to treatment with tyrosine kinase inhibitors. Therefore a sensitive assay to detect EGFR mutations is recommended in clinical practice. The real-time PCR based Therascreen EGFR RGQ kit is FDA approved for the detection of the most frequent EGFR alternations. During validation in our laboratory, up to 25% of clinical samples could not pass the kit's DNA sample assessment assay for amplifiable DNA. Obtaining amplifiable DNA presents challenging issues for mutation detection. These include the poor quality of DNA in formalin fixed paraffin embedded (FFPE) tissue and the low quantity of DNA as most tissue are from needle biopsies. In this study, we modified the DNA isolation method to obtain more high-quality DNA optimizing the Therascreen EGFR mutation assay. Methods: DNA was extracted in a total of 152 cases using QIAamp DSP DNA FFPE Tissue Kit. The isolation method was modified with pre-digestion heat treatment, overnight proteinase-K digestion, and double elution of the provided column. The Therascreen EGFR PCR kit was used to test the presence of amplifiable DNA using the kit's assessment assay or a control reaction. The control reaction mix uses a Scorpion primer and an unlabeled primer to amplify a short sequence of exon 2 of EGFR. The cycle threshold (Ct) values were used to assess the amount of amplifiable DNA obtained (the lower the Ct value, the more amplifiable DNA input). Three groups were included in this study: non-modification control group (n=48), pre-heat treatment with overnight digestion group (n=67), and combination of pre-heat treatment, overnight digestion and double elution group (n= 37). In addition the success rate, defined as % cases passing the DNA assessment, was compared between non-modification (n=48) and post-modification groups (n=104). The data was analyzed by repeated-measures ANOVA followed by Neuman-Keuls post hoc test. Results: A higher PCR success rate was obtained after method modifications (91.13%) compared to the control group (66.25%). There was a significant difference in Ct values between the three groups (p=0.0036). The control group had significantly higher Ct values than the other two groups (Newman-Keuls post hoc test, control versus overnight, p <0.05; control versus overnight with double elution, p <0.05). The lowest Ct values were obtained from the group with the combination of pre-heat treatment, overnight proteinase K digestion and double elution (mean Ct=27.12). Conclusions: Our modifications in DNA isolation from FFPE samples have significantly optimized Therascreen EGFR mutation assay in NSCLC patients. S. Jain, M.A. Melan, A.I. Wald, R. Hamilton, M.N. Nikiforova University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: Methylation status of MGMT (O 6 -methylguanine-DNA methyltransferase) is used as a prognostic/predictive marker for glioma patients treated with alkylating agents, such as temozolomide (TMZ). Conventional methylation specific PCR (MSP) methods provide evaluation of CpGs in the MGMTpromoter region; however, some patients with methylated MGMT status show poor survival and response to TMZ, likely due to aberrant MGMT expression not evaluated by these assays. Using targeted NGS panel (GlioSeq), we analyzed MGMT mRNA expression and correlated it with the MGMT promoter methylation status to determine if MGMT expression plays role in mediating tumor sensitivity to alkylating agents. Methods: GlioSeq, a targeted in-house developed NGS panel was used for assessing MGMT mRNA expression in addition to SNVs, copy number changes and gene fusions in 53 glioblastoma (GBM) patients and in 10 normal temporal lobe brain biopsies. MGMT promoter methylation was determined by methylation-specific PCR and gel electrophoresis (MSP-G) and MethyLight qPCR (MLqPCR). MGMT expression was compared to expression of GUSB, HPRT1 and PGK1 housekeeping genes and correlated with MGMT methylation status. Results: Normal brain tissue showed a high level of MGMT expression (mean=51.73). Thirty GBMs negative for methylation showed MGMT mRNA expression by GlioSeq that ranged from 5.02 to 35.27 (mean=16.32). Twenty-three MGMT methylation positive GBMs showed mRNA expression levels that ranged from 0.26 to 18.77 with a mean of 6.81. Using an mRNA MGMT expression cut off >10 for high level of expression and <10 for low level of expression, the MGMT promoter methylation status showed correlations with MGMT mRNA expression in 42/53 (79%) of cases, including 78% concordance in the methylated GBM group and 83% in the unmethylated GBMs. Conclusions: GlioSeq targeted NGS panel allows to detect MGMT mRNA expression in fixed brain tissue specimens during routine NGS analysis for most common genetic alterations in brain tumors. MGMT mRNA expression was concordant with the MGMT methylation status in 79% of GBMs tested. Discordant findings may indicate methylation-independent pathways for MGMT expression regulation and explain aberrant responses to temozolomide therapy. C.L. Sims 1 , S. Govender 1 , A. Khurana 1 , M. Moore 2 , P. Cotter 2 , S. Gunn 2 1 PacificDx, Irvine, CA; 2 ResearchDx, Irvine, CA. Introduction: DNA-based microarray technology is useful for providing information about gene copy-number and chromosomal aneuploidy. Formalin-fixed, paraffinembedded (FFPE) tissue is an important yet difficult sample type to work with in molecular assays. The chemical fixation process intended to preserve specimens for long term storage causes genetic information to be lost through DNA/protein cross linking and the creation of apurinic sites. Further compounding the problem, longterm storage of blocks before molecular analysis contributes to nucleic acid degradation. Here we describe a method for reference size matching, amplification and fluorescent labeling of FFPE specimens that yields high quality material for downstream DNA-based microarray analysis of this difficult sample type. Methods: In the current study, nucleic acid extraction was performed for 50 FFPE samples using the QIAmp DNA FFPE Tissue Kit either manually or on the QIAcube, followed by fragmentation using an Episonic 2000 sonicator to a mean library size of 500bp (with a +/-100bp cutoff interval) tapering off to 2000bp. Mean library size was determined electrophoretically using a Bioanalyzer 2100 (Agilent); samples were matched to an appropriately fragmented reference. Whole genome amplification and incorporation of 5-(3 aminoallyl)-dUTP was performed using the GenomePlex Whole Genome Amplification kit (Sigma-Aldrich). To allow for size selection of DNA libraries, while maintaining the mean library size, amplification products were cleaned up using Agencourt Ampure XP DNA magnetic beads (Beckman Coulter). Fluorophore (NHS Ester) coupling was performed under basic conditions. Final labeled products were cleaned up using the NucleoSpin Gel and PCR clean up kit (Machery Nagel). Degree of labeling (DOL), dye molecules per 100 bases, and total recovery was measured spectrophotometrically using a Nanodrop 2000 instrument (Thermo Scientific). Results: All samples yielded labeled DNA of sufficient quality and quantity for downstream analysis on the Agilent SureScan G4900DA using the following QC metrics: optimal DOL of 2.5, and at least 1000ng to 2000ng of labeled DNA required for an Agilent 8x60k microarray. Across the 50 samples tested, mean DOL was 2.2 (acceptable range 1.5 to 3.5). Mean labeled DNA yield was 3000ng (acceptable range 2000ng to 4000ng). Conclusions: As technology and our ability to ask and answer genetic questions about solid tumors improve, FFPE solid tumor specimens will become central to the molecular diagnosis, treatment and monitoring of disease. Given the use of commercially available kits and easily sourced reagents, the method described could be optimized for use on any microarray based analysis platform. Introduction: Salivary duct carcinoma (SDC) is a rare type of salivary gland cancer (SGC), typically refractory to treatment and with a particularly aggressive clinical course. Whereas specific gene fusions have been conclusively associated with other types of salivary gland malignancies, it remains to be determined whether genetic rearrangements are a characteristic feature of SDC. In the current study, we sought to explore this possibility. Methods: Nucleic acid from 42 SDC archival specimens was evaluated for structural rearrangements by a targeted gene-fusion assay based on anchored multiplex PCR and next-generation sequencing (AMP-NGS). In addition, tumor genotyping, focused on a panel of clinically-relevant genes, was performed using a DNA-based AMP-NGS platform. HER2 gene amplification was assessed using fluorescence in situ hybridization (FISH). Results: Oncogenic fusions were detected in 8 (19%) tumors. Four SDC harbored RET gene rearrangements, including 3 NCOA4-RET fusions and one case with a TRIM33-RET fusion. Additional translocations included one case each of: CHCHD7-PLAG1 and FGFR2-KIAA1598 fusions, and previously unreported CCDC6-EGFR and ETV6-MET rearrangements. Out of these translocations, only two had been previously associated with salivary gland neoplasms: CHCHD7-PLAG1 fusions have been reported in pleomorphic adenomas of the salivary gland, and ETV6 rearrangements (with NTRK3, not MET) have been associated with mammary analogue secretory carcinoma. By contrast, similar rearrangements involving the RET oncogene have been described in papillary thyroid tumors and in lung adenocarcinoma, and FGFR2-KIAA1598 fusions have been reported in intrahepatic cholangiocarcinoma, but were never associated with salivary gland tumors. Targeted genotyping identified additional mutations in TP53 (63%), PIK3CA (17%), HRAS (11%), CDH1 (7%), PIK3R1 (7%), and BRAF (5%). HER2 gene amplification was observed in 31% of the cases. With the exception of mutations in TP53, oncogenic rearrangements in SDC did not overlap with other genetic abnormalities. Conclusions: Our study demonstrates, for the first time, that genetic rearrangements are prevalent in SDC. Molecular analysis detected clinically-relevant alterations in 86% of the cohort and implicated novel genetic drivers in the pathogenesis of SDC. Importantly, due to the availability of targeted inhibitors, EGFR, MET and FGFR fusions may have therapeutic potential. Our findings highlight the utility of using multiple testing modalities in the discovery of actionable genetic alterations in SDC. The high prevalence of potential targets identified in our study may also expand the therapeutic strategies for these patients. Introduction: Next-generation sequencing (NGS) has become a critical technology in guiding patient treatment in clinical oncology. As laboratories are increasingly challenged to reduce testing time while managing increased sample volumes, there is a high demand for targeted panels that offer rapid library preparation and the ability to highly multiplex patient samples. Here we evaluate the Pillar SLIMamp Lung and Colon Hot Spots Panel and compare the results to the Ion Torrent Cancer Hotspot Panel v2 (CHPv2). Methods: A total of 15 samples were included in this evaluation: six non-small cell lung carcinoma (NSCLC) and nine colon adenocarcinoma. All samples had DNA concentration higher than 50 ng/μL and high DNA quality (Q129bp/Q41bp: 0.8 to 0.92) according to the KAPA hgDNA Quantification and QC Kit. Library preparation was performed using 50 ng and 5 ng of gDNA of each sample. A total of 30 samples were normalized using Qubit, pooled and sequenced on the v3 cartridge on the Illumina's MiSeq system. For data analysis, FASTq files were uploaded to the Pillar, where sequence alignment, annotation, and variant classification were performed. Variant calls within genomic regions covered by both panels were compared. Results: For the 15 FFPE samples, there was a high degree of concordance between the SLIMamp Lung and Colon Hot Spots Panel and CHPv2 variant calls (90.0%, 27/30 variants). Three variants that were called by the Pillar panel were not called using the CHPv2 (two single base-pair deletions and one-point mutation). In addition, variant calls for the Pillar panel were highly reproducible using both 50 ng and 5 ng of input material (100.0% concordance, 30/30 variants). Allelic frequencies for the variants detected in the 50 ng and 5 ng replicates were also highly reproducible (average deviation of 1.5% between replicates). Conclusions: As NGS tumor profiling becomes an increasingly integral component in determining patient treatment, clinical laboratories will need to accommodate high sample volumes and variable specimen quality. The Pillar SLIMamp Lung and Colon Hot Spots sequencing panel allows laboratories to perform accurate, highly-multiplexed, targeted NGS using benchtop instruments. In addition, this panel demonstrates a high degree of reproducibility in variant calls using both average and extremely low FFPE DNA inputs. jmd.amjpathol.org ■ The Journal of Molecular Diagnostics Introduction: The PI3K/AKT/mTOR pathway is frequently activated in ovarian cancer. Mutations and copy number alterations (CNAs) involving genes within this pathway may have clinical significance in predicting chemoresistance in ovarian cancer. Further, a number of inhibitors are currently in development that target PI3K pathway deregulation. To evaluate whether a single targeted next-generation sequencing-based (NGS) assay can fully characterize the spectrum of mutations and CNAs that may deregulate the PI3K pathway in ovarian cancer, we retrospectively analyzed a series of ovarian carcinomas. Methods: Clinical next-generation sequencing (NGS) was used to assess the coding regions of 131 genes and select introns in a series of 47 ovarian carcinoma specimens. For analysis, Varscan 2 and Genome Analysis Toolkit (GATK 1.2) were used to call single nucleotide variants (SNVs) and small indel variants, respectively. Pindel (0.2.4.d) was used to call larger indels. Genome-wide CNA prediction was extracted using the open source CNV detection tool, CNVkit. Fifteen non-tumor cases lacking CNAs were used as a reference, averaged, control for normalization. PI3K pathway gene CNAs of interest to this study included gains in PIK3CA and AKT1/2/3, and PTEN loss. Results: Of the 47 ovarian carcinomas, 41 were high grade serous carcinomas (HGSC), 3 were clear cell carcinomas (CC), 1 was a high grade carcinoma, mixed epithelial type with both HGSC and CC histologies, 1 case was a malignant mixed mullerian tumor (MMMT) of the ovary, and 1 case was a low grade serous carcinoma (LGSC). The genome-wide CNA landscape in all of the cases was suggestive of aneuploidy. Twenty-two of the 41 HGSC (53%) harbored PIK3CA copy number (CN) gains, 11 of which also harbored other co-existing PI3K deregulating CNAs: 7 harbored AKT1/2/3 gains, 1 harbored aPTEN loss, and 3 harbored both AKT1/2/3 gains and a PTEN loss. An activating PIK3CA mutation was identified in 1 of the HGSC cases that was negative for PI3K CNAs. None of the 3 pure CC harbored PI3K CNAs. The mixed HGSC/CC case harbored an isolated AKT1 gain and the MMMT showed an isolated PTEN loss. Forty-two of 47 total cases (89%) harbored a TP53 mutation. Cases without a TP53 mutation included 2 CC, 1 LGSC, and 2 HGSC. None of these 5 cases harbored any PI3K pathway CNAs; however, one (CC) harbored an activating PIK3CA mutation. Conclusions: Ovarian cancers with genomic alterations affecting the PI3K pathway, including CNAs, can be identified from a single clinical, targeted NGS assay. Such assays increase diagnostic efficiency and are particularly useful in cases where limited material is available for multiple diagnostic tests. Introduction: Cancer is a complex disease caused by the accumulation of genomic alterations including mutations, CNVs, and gene rearrangements. Integrated approaches are required to generate comprehensive genomic profiles to meet the needs of precision cancer care. This is extremely important in childhood cancers due to the higher long-term survival rate of these patients. Targeted-panel NGS enables testing of many cancer-related genes and different genomic alterations simultaneously and has become the method of choice for clinical cancer genomic profiling. Methods: Four large custom designed DNA-or RNA-based NGS panels (HeredPanel, HemePanel, SolidPanel and FusionPanel) were developed to detect different types of mutations (SNV, indel, CNV, LOH, and gene fusion) associated with hereditary and/or somatic cancers. These panels interrogate 402 genes, 5506 exons, 232 intronic regions, and 586 known fusions and can potentially detect many more novel fusions. Results: The performance of the three DNA panels were evaluated using six cancer cell lines, one HapMap cell line, and three normal controls. The results showed 100% sensitivity, specificity, and reproducibility for mutation detection. The 1% detection limit was established using serially diluted tumor cell lines. The average sequencing depth for the targeted regions was >650X for HerePanel and >1600X for HemePanel and SolidPanel. Thirty-three RNA samples were tested blindly to assess the performance of the FusionPanel. The panel correctly identified all 32 (100%) known fusions and two new fusions. The detection limit of FusionPanel is 1% (1 tumor cell in 99 normal cells). These panels were further validated using 28 discarded clinical samples including blood; bone marrow; and fresh, frozen, or FFPE tissues and showed concordant results with those of Sanger, FISH, microarray, and RT-qPCR. These panels have been offered as clinical services at the Children's Hospital of Philadelphia and have provided genomic data important for clinical decision-making in over 80% of patients tested. Conclusions: We reported the development, validation, and assessment of the performance and clinical utility of four large NGS panels for childhood cancers. Our experience demonstrated that these NGS-based panels provide comprehensive genomic information of pediatric cancers, which plays a significant role in clinical cancer management, and can be incorporated as part of routine clinical oncology practice. J.L. Dillon, J.D. Marotti, H. Jung, J.A. Lefferts, L.J. Tafe Dartmouth-Hitchcock Medical Center, Lebanon, NH. Introduction: Using the updated 2013 ASCO/CAP HER2 testing and scoring guidelines, we and other institutions identified an increased detection in the number of breast cancer (BC) cases equivocal for HER2 (ERBB2) by FISH. By definition, these cases have a HER2 copy number of 4 to less than 6 copies per cell (equivocal range) with a HER2:CEP17 ratio less than 2.0. Some laboratories utilize alternative chromosome 17 gene probes in instances of HER2 equivocal by FISH. Here, we evaluate the utility of the OncoScan SNP array to evaluate chromosome copy number (CN) as an alternative assessment inHER2 FISH equivocal BC. Methods: Our laboratory performs primary HER2 FISH testing on all breast cancer cases with reflex to immunohistochemistry (IHC) for equivocal cases. Ten cases of BC that were HER2 equivocal by FISH (PathVysion) were included; all cases also had HER2 IHC results. DNA was extracted from 8 unstained, 4 micron FFPE slides with the QIAamp DNA FFPE Tissue Kit (Qiagen). The samples were processed using the Affymetrix OncoScan FFPE Array according to the manufacturer's protocol and analyzed using Nexus Express and Affymetrix software packages. Results: For all 10 cases, the HER2:CEP17 FISH ratio was <2.0 and the average number of HER2 signals per cell by FISH ranged from 4.1 to 5.4. The SNP array identified aneuploidy of the HER2 locus in all cases ranging from a CN of 3 to 6. In addition, multiple additional chromosomal CN abnormalities were seen in a majority of tumors including amplifications of CCND1(2 cases), FGFR1(4 cases), GNAS (4 cases) or MYC (1 case). Conclusions: Low CN gains at the HER2 locus were able to explain our equivocal FISH results in 9 of 10 cases. In one case (EQ8) a HER2 CN of 3 was detected. However, the SNP array provides a summary result of all cells tested so this sample may consist of a heterogenous population with an average CN 3. Our findings support the utility of SNP array analysis to further evaluate the CN state of BC that is HER2 equivocal by FISH and in the process, detect additional potentially meaningful CN alterations. Case The proto-oncogene MET is known to harbor donor/acceptor splice site mutations that can give rise to transcripts lacking exon 14 (METex14). Although this "skipping" event occurs at frequencies comparable to ALK rearrangements in lung adenocarcinomas, current METex14 detection technologies are underdeveloped and burdened by a combination of high costs, low resolution or complex workflows. Herein we report the development of a novel, accurate, and accessible 2-channel droplet digital PCR (ddPCR) assay that can quantify the fraction of skipped METex14 transcripts in patient samples. Successful identification ofthis event correlates with MET inhibitor responsiveness, potentially benefiting treatment decisions and outcomes. Methods: A METex14 RT-ddPCR assay was developed using optimized reverse transcription and PCR reagents, with an optional pre-amplification step incorporated for samples with low nucleic acid quantity or quality. Gene-specific primers and junction-spanning hydrolysis probes were designed for MET exons 13/14 or 14/15 (wildtype (WT), HEX labeled) and 13/15 (METex14, FAM labeled), respectively. Droplets were generated and run on the QX200 ddPCR platform (Bio-Rad) with a 4-bin output. Samples consisted of cell lines and a set of 125 adenocarcinomas (MD Anderson), which were additionally sequenced for RNA fusions and METex14 skipping using the QuantideX NGS RNA Lung Cancer Kit (Asuragen). Results: MET exon 14 skipping status was determined with 100% accuracy in representative cell lines from as little as 2 ng RNA inputs.Known copy number titrations demonstrated analytical sensitivity down to 1% METex14 skipped/WT transcripts, exceeding levels typically achieved by NGSbased approaches. METex14 positive and negative samples were correctly identified from previously characterized FFPE specimens, including two FFPE samples that showed 27% and 95% METex14 skipping levels, respectively. Compatibility of the RT-ddPCR assay for both RNA and total nucleic acid inputs was also demonstrated. Conclusions: The prevalence of METex14 skipping events in NSCLC, combined with reports of patient responses to MET inhibitors such as crizotinib and cabozantinib, makes the lesion a high-priority diagnostic target for precision medicine. The described assay provides absolute copy number jmd.amjpathol.org ■ The Journal of Molecular Diagnostics and other solid tumors. Quantification of RNA variants can complement DNA analyses to capture the dynamic molecular state of tumor cell populations and inform treatment strategies. We describe a novel targeted next-generation sequencing (NGS) technology that offers both modularity and systems-level integration to achieve accurate RNA and DNA profiling from challenging clinical specimens. Methods: Formalin-fixed, paraffin-embedded (FFPE) or fine-needle aspirate (FNA) specimens from three cohorts, including the BATTLE-2 clinical trial, were collected at MD Anderson Cancer Center. Pre-analytical QC was performed using novel qPCR assays that quantify distinct populations of amplifiable RNA and DNA from total nucleic acid. RNA or DNA was enriched by PCR using the QuantideX NGS RNA Lung Cancer Kit, or a prototype DNA panel with common non-primer-based reagents, and sequenced on a MiSeq System. Data analysis was achieved using QuantideX NGS Reporter, an analysis suite that directly incorporates pre-analytical QC data into the variant calling algorithm to improve the identification of DNA mutations and RNA fusions. Results: We evaluated the QuantideX NGS RNA Lung Cancer Kit, which targets 110 gene fusions, MET exon 14 skipping, 23 mRNA expression markers, and 3'/5' imbalance markers for ALK, RET, ROS1, NTRK1 and PGDFRA. A total of 224 NSCLC FFPE specimens were characterized to reveal multiple fusions involving ALK, RET, ROS1 along with evidence of 3' expression imbalances in ALK and RET. Cases of partially and fully skipped MET exon 14 were also identified and quantitatively confirmed by digital RT-PCR. A majority of the specimens were also evaluated with a DNA panel that targets 20 genes associated with NSCLC. Mutations were observed in 77% of samples with profiles that were qualitatively consistent with previous reports but with quantitative molecular features that reflected cohort-specific clinical annotations. In addition, the DNA panel identified mutations in blinded BATTLE-2 clinical trial FNA specimens with 95% agreement to an independent NGS method. Conclusions: We describe an efficient and accurate workflow that reports prognostic and theranostic markers in lung cancer and leverages the integration of common wet and dry bench components for targeted RNA-and DNA-Seq. The flexibility of these components means that new NGS panels can be rapidly developed and verified. We also show that a "sampleaware " approach can unify the analysis of SNVs, indels, RNA fusions, and aberrant transcripts within a single NGS run to support comprehensive and streamlined characterization of tumor biopsies. Traditionally, gene fusions are detected by fluorescence in situ hybridization (FISH) and reverse transcription PCR (RT-PCR). Both methods require prior knowledge of the fusion partners and specific probes or primer sets for different fusions, have no or limited multiplex capacity, and lack the ability to identify novel fusions. We evaluated the performance of a next-generation sequencing (NGS)-based assay utilizing Anchored Multiplex PCR technology for detection of gene fusions from formalin-fixed, paraffin-embedded (FFPE) lung and thyroid tumors. Methods: Total RNA was extracted from 13 cases of microdissected FFPE tissues including adenocarcinoma of the lung (9), PTC (2) and FC (2) Introduction: There is increased demand for clinical genome-wide copy number assessment of pediatric solid tumors, given the established clinical significance of certain chromosomal aberrations and its use for prognostic classification and treatment of these patients. Neuroblastomas characterized by whole chromosome changes tend to act favorably, whereas tumors with MYCN gene amplification and/or segmental chromosomal aberrations, including loss of heterozygosity (LOH), exhibit poor outcomes. Given the clinical relevance, our aim was to verify the accuracy of OncoScan FFPE SNP Array (Affymetrix) in assessing genome-wide copy number changes and LOH in tumor samples, for use in the clinical laboratory. Methods: Thirty-six neuroblastic tumors diagnosed in our pediatric hospital were analyzed using the OncoScan FFPE SNP Array. Tumor genomic DNA was isolated and arrays hybridized according to manufacturer's protocol. Data was analyzed using the OncoScan Console and viewed using Chromosome Analysis Suite (ChAS) and Nexus Copy Number 7.5. Copy number results determined by SNP array were compared to ploidy results obtained by karyotype analysis and/or flow cytometry. Additionally, chromosome 2, 12, 22, X and Y aneuploidy data by FISH were used as a surrogate ploidy marker in cases where cytogenetics and flow cytometry were unavailable. MYCN and ALK FISH were performed for gene copy number assessment and 1p36 and 11q23 LOH were assessed by PCR. Results: The results for MYCN and ALK gene amplification both showed 100% concordance (33/33 and 8/8 cases respectively). The 1p36 and 11q23 LOH also showed 100% (3/3) and 67% (2/3) concordance, in selected cases. Karyotype and FISH for chromosome aneuploidy showed the highest concordance for assessing tumor ploidy. After initial analysis, 2 cases showed discordance with karyotype, flow cytometry or FISH results (2/36 cases or 6%). Both specimens were confirmed as near tetraploid by FISH and/or flow cytometry and required the OncoScan baseline to be adjusted appropriately. Conclusions: Our results demonstrate that the OncoScan SNP array is useful in the clinical laboratory for accurate assessment of gene amplification, LOH and genome-wide copy number. In 2 cases, a second method was necessary to correctly assign tumor ploidy baseline, despite complex algorithms used by the OncoScan software to assign copy number calls. The complex chromosomal copy The Journal of Molecular Diagnostics ■ jmd.amjpathol.org number changes present in tumors can challenge the ability of the software to correctly assign ploidy state. Therefore, we conclude that a second independent method may be necessary in complex tumor cases to correctly assign ploidy that truly reflects tumor biology and that may be necessary for correct patient management. Introduction: Genomic medicine has been advancing rapidly. The complexity of delivering precision medicine to oncology patients across a university health system suggested the need for a Molecular Tumor Board (MTB). Unlike MTBs reported elsewhere, we decided to discuss the cases before molecular tumor profiling for appropriateness of the test. When the results were available, further treatment options were discussed in the MTB. The objective of this study was to analyze our progress over the first two years. Methods: Patients were reviewed in the MTB for appropriateness prior to the submission for comprehensive next-generation sequencing (NGS) cancer gene testing. A set of criteria were in place to determine appropriateness, in brief, 1) the patient has treatment resistant or recurrent measurable tumor, which has failed at least one standard therapy or no standard therapy options are available, 2) the patient has a good performance status and 3) the cases with a driver gene mutation that can be found by routine single gene testing such as those in lung adenocarcinoma, melanoma and colon cancer, were excluded. In this study, we analyzed the cases discussed and sent for testing with their results and outcomes. Results: 138 cases were discussed at the MTB, 92 of them were approved for testing and 46 were not approved. The most common reason for rejection was the availability of a standard therapy option (n=8). 22 cases did not have adequate tissue for testing. Most common tumor types tested were gynecologic (43%), central nervous system (15%) and breast (11%). Among the available results (n=88) (4 cases did not have enough DNA), 29 cases (31.5%) had driver gene mutations associated with an active targeted therapeutic agent, includingBRAF, PIK3CA, IDH1, KRAS, and BRCA1. An additional 40 cases (43.5%) had cancer-related mutations. 19 cases (20.7%) did not have any clinically significant alterations. Alterations in over 40 different genes were identified, TP53 alteration being the highest in incidence (54.5%). One third of the patients with an actionable mutation benefited from the test with a good or partial response to targeted therapy. Conclusion: The MTB has provided the opportunity to offer precision medicine for our cancer patients. The application of precision medicine and molecular genetic testing for cancer patients remains a continuous educational process. Since our MTB patient population excluded lung adenocarcinoma, melanoma and colon cancer, the fact that the rate of finding actionable mutations is comparable to other studies suggests the case selection through the MTB is appropriate. A well-designed MTB will evolve along with the technology to ensure that patients receive the best possible treatment without unnecessary costs or risks. Introduction: Mutation detection in the EGFR, KRAS, NRAS and BRAF genes is standard of care for targeted therapy of metastatic non-small cell lung cancers (NSCLC) and colorectal cancers (CRC). Mutations within these genes are usually mutually exclusive since they are in the same signal transduction pathway, though a low incidence of such coexisting mutations has been reported. In the clinical diagnostic setting, encounter of unexpected coexisting mutations raises a concern for quality assurance. Methods: We studied 510 NSCLC and 304 CRC consecutive cases submitted to a clinical molecular diagnostic laboratory for mutation profiling by a next-generation sequencing platform. Results: We identified coexisting EGFR p.G917A mutation and KRAS p.G12C mutations in one NSCLC, co-existing activating KRAS and NRAS mutations in one NSCLC and 3 CRC, and two activating KRAS mutations in one NSCLC and one CRC. Although kinase-deficient BRAF mutation can be seen in specimens with an activating RAS mutation (Zheng, et al., BMC Cancer 2015) , no coexisting kinase-activated BRAF p.V600E and activating RAS mutations were observed. Retrospective tissue identification testing and repeated NGS assay of multiple subareas confirmed that these coexisting mutations are valid and may occur in a single tumor population or separately in tumor subpopulations. One mistaken case of dual KRAS and NRAS mutations was also identified. Conclusions: This retrospective study for quality assessment confirmed true coexisting mutations in the EGFR, KRAS and NRAS genes, which may also be caused by laboratory errors. A standard operating protocol is proposed to confirm unexpected coexisting mutation within the same signal transduction pathway in the clinical diagnostic setting. Analysis R.T. Taylor, J. Hill, A. Fuentes, C. Heiser, Z. Abdulateef, J.R. Hill Spot On Sciences, Inc., Austin, TX. Introduction: Robust preservation of tissue biospecimens for molecular analysis is essential for personalized disease diagnosis. Current methods for the preservation of tissue, such as formalin-fixed, paraffin-embedded (FFPE) or flash freezing in liquid nitrogen, are labor intensive and present challenges for collection and maintenance. Additionally, analyte integrity can become compromised due to cross-linking and freeze/thaw degradation which poses challenges to downstream molecular analysis. We demonstrate an alternative method that is more streamlined and robust. Leveraging proven dried blood spot technology to preserve tissue homogenate by drying on HemaSpot-HF collection devices. We seek to identify the optimal homogenization and extraction buffers as well as demonstrate suitable recovery for common molecular analytes including proteins and nucleotides. The simple procedure provides long-term analyte stability and recovery with minimal infrastructure requirements. Methods: Small samples (50 mg) of Wistar-Hannover Rat tissue were added to a Dounce homogenizer in the presence of 0.7 mL of stabilizing buffer. 88 uL of homogenate was applied to a HemaSpot-HF sample storage collection device and stored at room temperature. Aliquots of homogenate were frozen and stored at -20°C. Dried and wet homogenates were extracted using common methods and representative protein (BCA protein assay kit), Beta-galactosidase activity (LC-MS/MS), DNA (PicoGreen assay), and RNA (Taqman) markers were quantitated. Results: High recovery (>85%) of total proteins from dried rat liver, lung, spleen, and kidney homogenates were observed compared to frozen homogenates. Beta-galactosidase activities from dried liver homogenates were comparable (>90% activity) to corresponding wet, frozen homogenate samples. The presence of Triton X-100 in the homogenization buffer is shown to inhibit DNA recovery while protease inhibitors are shown to improve total protein recovery. PicoGreen assay results demonstrate high recovery (75%) of double stranded DNA extracted from dried liver homogenates compared to frozen samples. Levels of mRNA (AUF1, LCN2, b-actin, p53 and TNF-a) were equivalent or slightly higher for dried samples as compared to wet samples. Conclusions: These results demonstrate feasibility for solid tissue preservation as a dried homogenate on filter paper at ambient temperatures. Subsequent quantitation of protein and nucleotide biomarkers without significant degradation or loss of activity is observed. This method offers an efficient and cost-effective alternative for tissue biospecimen preservation and molecular analysis. They are almost exclusively seen in pediatric high grade midline tumors with the most frequent mutation being the K27M mutation whereas the G34R mutation is less frequent. They have been suggested as sensitive markers of HGGs. We sought to i) determine the frequency of these mutations in our cohort of pediatric gliomas and to ii) explore the utility of a single pyrosequencing dispensation protocol for the detection of both K27M and G34R mutations. Methods: We analyzed fifty two (52) pediatric brain tumors including 29 pilocytic astrocytomas (PA), grade I, 3 anaplastic PAs, grade III, 3 diffuse astrocytomas (DA) grade II, anaplastic DA, grade III, 10 glioblastoma, grade IV, 2 dysembryoplastic neuroepithelial tumor (DNET), grade I, 1 ganglioglioma (GG), grade II and 1 anaplastic GG, grade III. 27 adult tumors including 23 glioblastomas, 3 anaplastic DAs, grade III and 1 DA, grade II were also analyzed. Results: Using appropriate and known controls as reference, it was possible to use a modified pyrosequencing protocol that allowed for the detection of the respective mutations in a single assay. All of the adult tumors were negative for the K27M and G34R mutations. A total of 5 pediatric tumors were positive for mutation including 2/10 glioblastoma, 1/3 anaplastic PA and 2/3 DA. The 2 DA and 1 GBM were thalamic masses. The other positive GBM was a right frontal lobe mass. The 2 positive thalamic masses that were initially diagnosed as DA, grade II were worrisome for higher grade due to elevated proliferation index and may have been inadequately biopsied. We demonstrated H3F3A mutations in pediatric high grade tumors only. This finding is consistent with a correlation between the presence of an H3F3A mutation and high grade histology in the pediatric age group. We recommend analysis of histologically low grade appearing midline diffuse astrocytomas for H3F3A mutations as biopsies may inadequately represent the true biology of such tumors. We also confirm the utility of a previously described single dispensation pyrosequencing protocol that allows for the simultaneous identification of both K27M and G34R mutations in a single assay. RNA-seq Using Unique in vitro Transcripts K. Lawrence, R. Leonard, D.R. Hout, S. Qadri, B.L. Schweitzer Insight Genetics Inc, Nashville, TN. Introduction: The potential for technical errors such as cross-contamination and/or misidentification of individual samples can occur while performing high-throughput next-generation RNA-seq analysis. To mitigate this risk, a method was developed to add individually bar-coded in vitro transcripts (IVTs) to samples prior to library construction. Therefore, any such errors can be identified after processing, thereby improving the quality of reported data. This IVT set may be incorporated into highthroughput RNA-seq workflows to serve as a quality control check on sample identification and cross-contamination. The IVTs are compatible with multiple library preparation methods including both positive and negative enrichment strategies as well as methods used for preparation of formalin-fixed paraffin embedded (FFPE)derived nucleic acid. Methods: Bar-coded IVT RNA was spiked into 1 microgram of FFPE RNA extracted from archival surgical resection tissue. To establish a proper IVT input, various quantities (2x10 11 to 2x10 7 copies) were spiked prior to mRNA enrichment and library generation. The RNA was processed using ScriptSeq Complete Gold H/M/R Low input kit and ScriptSeq v2 library prep reagents. The standard protocol for degraded samples was utilized, which involves no RNA fragmentation prior to cDNA synthesis, and 15 cycles of library amplification. To quantify the IVT present in the NGS data, the sample fastq files were aligned to the IVT sequence to calculate the number of reads that align to the IVT sequence. To quantify the relative IVT transcript abundance in the IVT libraries, three Taqmanbased qPCR assays were designed to target the unique sequence in the IVT. Results: FFPE RNA-seq libraries were successfully constructed without noticeable interference from the addition of IVT as evidenced by the fragment distribution of IVT-spiked samples being indistinguishable from non-spiked replicates. We observed a consistent relationship between IVT spike amounts and relative transcript quantity detected by qPCR, and between IVT spike amounts and the number of IVT-aligned reads per total reads in both FFPE RNA-seq libraries and poly-A enriched libraries. Conclusions: The feasibility of an IVT spike as a quality control measure in FFPE RNA-seq protocols was demonstrated by the lack of interference in standard RNA-seq protocols, the successful detection and qualification with post-sequencing bioinformatics tools, and the correlation of IVTaligned read counts with qPCR results. The resulting IVT design could be utilized as a novel QC to monitor technical errors in RNA-seq based studies. A. Iyer, R. Haigis, M. Porter, D. Lee, S. Smith, S. Dhillon, T. Dunn, S. Murillo, N. Udar Illumina, Inc., San Diego, CA. Introduction: Next-generation sequencing (NGS) assays allow for the simultaneous detection of numerous somatic mutations in multiple samples in a single run. Ensuring that an assay is performing correctly and optimally is of critical importance in any setting. accurate identification is therefore of utmost importance with regard to therapeutic decision making in non-small cell lung carcinoma (NSCLC). Here we assess the differential abilities of a variety of clinical assays, to identify and appropriately annotate these complex variants in lung cancer. Methods: A systematic review of clinical lung cancer cases, which were tested by next-generation sequencing using AmpliSeq HotSpot Cancer Panel v2, was undertaken at two major medical centers. Samples that demonstrated complex indels in exon 19 were further examined. Sequence alignments were visualized in Integrated Genomic Viewer (IGV) to evaluate the presence of possible complex variation. Cases with identifiable complex indels were then subjected to a multi-platform review including alternative pipelines, Sanger sequencing, PCR-CE, and an FDA approved assay for the detection of EGFR exon 19 variants. Results: Of 531 NSCLC tested, 11% (n=57) demonstrated EGFR exon 19 deletions and 22% of these (n=13) were found on alignment visualization to actually represent complex indels. PCR followed by fragment analysis using capillary electrophoresis, confirmed variable product sizes in EGFR exon 19 in all 13 cases. Current pipeline with validated settings, failed to identify any feature whatsoever in 15% of complex indels. The FDA approved Therascreen assay had a clinical sensitivity of 78% for these complex indels. Pipelines optimized for complex indel detection did identify the presence of some form of variant in all samples, however, the left alignment normalization strategy they employed for annotation typically produced several separate variant calls as a result of microhomology within the inserted sequence. The best performing pipeline produced an HGVS acceptable indel annotation for only 38.5% of these complex variants. Conclusions: Complex indels within exon 19 of EGFR may be more common than previously documented. Several factors affect proper identification and annotation of these variants and even a FDA approved assay demonstrates limitations in identification of all possible indels. Localized microhomology of the inserted sequence may lead many NGS pipelines to treat complex mutations as a series of separate events (deletions and SNPs). Recognition and visualization of entirely phased variants should prompt careful review to assess the need for a more accurate annotation of these clinically actionable complex indels within the EGFR gene. Conclusions: Evaluation of MLH1 promoter methylation by MethyLight is highly sensitive, specific, reproducible, and robust. Our data support that MLH1 promoter methylation testing may reduce the number of unnecessary cases sent for germline sequencing given the appropriate clinical context. Z. Shajani-Yi, L.N. Nguyen, W.F. Hickey, S.J. Deharvengt, F. Blumental de Abreu, J.D. Peterson, T.L. Gallagher, A.J. Erskine, J.A. Lefferts, G. Tsongalis Dartmouth-Hitchcock Medical Center, Lebanon, NH. Introduction: Gliomas are composed of a heterogeneous group of neoplasms with respect to malignancy grade, histological subtype, invasiveness and treatment response. Histology is the gold standard for grading and typing of gliomas; however, molecular biomarkers such as MGMT hypermethylation, co-deletion of the chromosome arms 1p and 19q (1p/19q co-deletion) and IDH1/IDH2 mutations are generally recognized to be associated with a more favorable prognosis. In high grade gliomas, these markers have also been indicated in tumors that have an improved response to alkylating chemotherapy and information on the molecular status of the tumor could be used to direct patient care. Here we retrospectively examined patient tumor samples to determine the frequency of co-occurrence of these biomarkers in our patient population. Introduction: Microarray analysis can accurately assess genome-wide copy number variations in DNA extracted from FFPE tissues. This type of testing may be beneficial in cases with both neoplastic and non-neoplastic disease in the differential diagnosis. For example, a 7-year-old male presented with several months of worsening headaches. He had had a history of severe head trauma, and shunts had been placed for bilateral subdural fluid collections at 4 months of age. MRI identified a large complex collection; however, the artifact from the shunt device made interpretation difficult. The histologic appearance of surgical biopsy specimens was atypical with spindled and pleomorphic cells in a myxoid matrix background. Epithelioid cells with eccentric nuclei and eosinophilic cytoplasm were observed. Scattered multinucleated neoplastic cells and abundant necrosis were also identified. Reactive subdural membranes, however, are also known to have an atypical histologic phenotype; therefore, it was difficult to differentiate a neoplastic process from a reactive process. The differential included an unclassifiable myxoid sarcoma and reactive subdural membranes with organized clot adjacent to the shunt site. An accurate diagnosis is critical because of the different treatment approaches. We performed genome-wide detection of alterations in copy number using the Affymetrix OncoScan FFPE assay. Methods: Tissue samples from the lesion were collected through a craniotomy. Cytogenetic analyses were performed at a reference laboratory. DNA was extracted from unstained tissue sections using the QIAamp DNA FFPE Tissue kit. The sample was analyzed using the OncoScan assay kit following the manufacturer's protocol. The TuScan algorithm employed in the assay was used for copy number variation analysis. Results: The OncoScan assay demonstrated multiple chromosomal copy number gains in chromosomes 8 and 14q, and copy number losses in chromosomes 1p, 3, 4p, 5, 6, 10q, 12, 13, 15, 17, 18, 21, 22 , X, and Y. Additionally, chromosomes 11 and 20 contained multiple gains and losses. These alterations appeared to be relative to a state of tetraploidy and were consistent with a neoplastic process. FISH analysis with NR4A3 (9q31.1) gene probes ruled out a diagnosis of extraskeletal myxoid chondrosarcomas (no rearrangements/fusion detected) and also supported the suspected tetraploidy with 3 to approximately 5 copies of the gene in 62% (31/50) of the nuclei. The OncoScan assay generated the discernible diagnosis between the differentials. Conclusions: This case represents a rare diagnostic challenge for pathologists. The OncoScan assay confirmed a neoplastic lesion and helped rule out a reactive process. The OncoScan microarray may be used for confirmatory diagnosis of rare tumor types. The molecular analysis of cell-free DNA (cfDNA) has gained increasing attention during recent years. cfDNA is an easily accessible source for disease monitoring at the molecular level, particularly in patients with disseminated metastatic cancer who are receiving molecular targeted therapies. The underling mechanisms by which cfDNA are released in blood under normal and pathologic conditions are not yet fully understood. However, multiple studies hypothesize that pathologic conditions like cancer will yield a higher amount of cfDNA in the circulation compared to the amount of cfDNA obtained from healthy individuals. There is significant variability in the amount of cfDNA obtained from human plasma. It is uncertain if this difference is related to the variation among patients, preanalytic conditions and/or different isolation methods. In this study, we isolated cfDNA from plasma samples from patients with lung adenocarcinomas using two different methodologies. Next, we PCR amplified a region of the EGFR gene exon 20 harboring T790 codon. Sequencing analysis was then performed on Ion Torrent next-generation sequencing platform. Methods: Plasma from EDTA-drawn samples was collected after two centrifugation steps. The column-based system from Qiagen (QIAmp Circulating Nucleic Acid Kit) was used in 16 samples and the Maxwell 16 instrument (Promega Circulating DNA Purification kit) was used in 7 samples, using protocols supplied by both manufacturers. Cell-free DNA concentrations were measure using Qubit Fluorometer. The sizing and quantification of cfDNA were analyzed by High Sensitivity DNA chip (2100 Agilent Bioanalyser). Ten samples using Qiagen and 3 samples using Promega with >1ng/μL DNA were selected, for further PCR amplification and sequencing. Results: Three out of 13 samples with >1 ng/μL DNA revealed T790M mutation at low allele frequency <1% detectable only visually on integrated genomic viewer (IGV). The depth of reads in all 13 samples was >23,000 with no strand bias. The allele frequency in three T790M harboring samples were as follow: 0.14% (33 out of 23411 reads), 0.17% (42 out of 24983), and 0.085% (21 out of 24836) respectively. Conclusions: The results of our study shows that successful PCR amplification can be achieved even with very low cfDNA concentration yield. Several preanalytic factors impact the cfDNA yield including the type of collection tube, age of sample and plasma volumes used for DNA isolation. Given the extremely low level of positive reads that are typically bellow the detection sensitivity; target enrichment such as digital PCR appears to be an essential step in successful mutation detection in clinical settings. A. Ras, S. Helm, K. Kelly, V. Spotlow, H.V. Reddi The Jackson Laboratory, Farmington, CT. Introduction: FFPE tissues have long been considered a valuable resource in the study of solid tumors. The extraction of nucleic acids from FFPE specimens is challenging due to cross-linking and the damaging effects of the paraffin embedding process to genomic DNA. The quality of purified DNA related to degree of fragmentation and yield depends highly on the sample type, age, archiving process and storage conditions of the FFPE tissue, but the method of extraction can also have a significant impact. Selection of the proper FFPE DNA extraction technique is critical to generate DNA adequate for genetic profiling. Methods: In this study, DNA was purified from five 5μm tissue scrolls, each from 5 FFPE blocks, using 8 commercially available FFPE DNA extraction kits. All purifications were executed manually according to the manufacturers' protocol. DNA was also subjected to subsequent removal of any residual PCR inhibitors. Aspects of purification including tissue deparaffinization with various solvents, enzymatic digestion time and extraction technique were assessed based on DNA quality and quantity using the Nanodrop and Qubit. Results: Final DNA extraction results differ significantly between kits in terms of DNA quantity and purity (OD 260/280 ratio). Whereas all 8 kits produced measurable amounts of DNA, only 5 recovered DNA of sufficient yield and purity to proceed effectively to library preparation for next-generation sequencing. The 3 FFPE kits that performed the best were QIAamp, Bio-Chain, and E.Z.N.A, which yielded a minimum of 700ng and a purity of at least 1.4. The ReliaPrep and Nucleospin extractions generated satisfactory DNA quality, but with lower yields. The purity obtained with MasterPure DNA, UltraRapid and GenElute kits were considerably lower, suggesting highly impure DNA, perhaps caused by insufficient digestion time allowing for residual protein contamination. Lysis time has a significant impact on DNA purity with enzymatic digestions less than 2 hours resulting in inadequate purity. Also, utilization of xylene as a paraffin dissolver has harmful effects on the tissue, resulting in lower yield and purity. Conclusions: In summary, all kits evaluated allowed for isolation of quantifiable DNA. However, only 5 recovered sufficient DNA yields for downstream applications. The performance of the QIAamp kit was superior, producing the highest average DNA yield within acceptable purity range and turnaround time. In addition, purification kits using nonxylene paraffin solvents and those having lysis time more than 2 hours yielded significantly better results than those that did not, suggesting that these steps have significant impact on the outcome of the purification process. Prostatectomy: Initial Results from the Decipher GRID D. Dolginow, J. Chelliserry, M. Alshalalfa, N. Erho, H. Al-Deen Ashbab, M. Takhar, P. Wood, L. Lam, E. Davicioni GenomeDx, Vancouver, British Columbia, Canada. Introduction: Prostate cancer patient management has been enhanced with several commercial genomic prognostic tests such as the Decipher prostate cancer classifier. Unlike other tests, Decipher generates genome-wide expression data for each patient. This data has been anonymized and made available for research in the Decipher GRID. Here we report an initial analysis further characterizing the genomic landscape of localized prostate cancer patients most at risk for recurrence after radical prostatectomy. Methods: We conducted a literature search to identify genes and signatures of potential clinical relevance to prostate cancer. The expression and distribution of 697 genes and 31 prostate cancer disease signatures for metastasis risk, proliferation, luminal/basal, small cell and AR signalling were examined across 2,978 radical prostatectomy tumors with adverse pathology tested with Decipher and available in the GRID. For the 697 genes, expression distribution was characterized and high and low expression were defined using thresholds based on median +/-1.5*1.48*MAD (median absolute deviation). Genes from 31 published prostate cancer signatures were adapted to the Decipher platform and scores were calculated. Results: For various druggable targets including immune checkpoint inhibitors (PD1, PDL1) and growth receptors (c-MET, EGFR, HER), between 2-11% of patients had high expression above right threshold. Clustering of scores for 31 signatures revealed several clear groupings of patients. About 20% of patients consistently had high scores for all the metastasis risk signatures and low-average AR signalling scores. About 10% of patients had high scores for the proliferation signatures but low metastasis risk scores. Tumors with low AR signalling scores were enriched with high basal and small cell signature scores whereas most luminal tumors had higher AR signalling. Conclusions: Since every patient who has received the Decipher test also has a genome-wide expression profile, the Decipher GRID allows researchers to evaluate on a systematic, population-level the expression of genes and signatures that may guide therapies. Such information may be useful for selection of optimal systemic therapy and inclusion into clinical trials of novel targeted agents. This rich genomic resource is being made available on a research use only basis to prostate cancer researchers and to clinicians seeking to better understand prostate cancer to advance precision medicine. V. Thodima, A. Guttapalli, R. Padmanabhan, S. Kamalakaran, J. Houldsworth, B. Gowrishankar Cancer Genetics Incorporated, Rutherford, NJ. Introduction: There is a growing body of evidence in the literature showing that mutations, copy number changes and a few single nucleotide polymorphisms (SNPs) consistently correlate with prognosis and therapy response in clear cell renal cell carcinoma (ccRCC). Here, we describe analytical validation of a custom targeted next-generation sequencing (NGS) assay for assessing genomic alterations relevant in ccRCC. The assay has been validated in fresh-frozen/formalin-fixed paraffin embedded (FFPE) resected specimens and core needle biopsies. Methods: The targeted hybrid capture-based NGS panel (FOCUS::Renal) comprised of 76 genes mutated in ccRCC (including 7 targets for FDA-approved drugs), 15 SNPs of prognostic relevance, and a 3Mb-spaced SNP backbone for copy number assessment (Nimblegen, 2400 targets). Sequencing was performed (MiSeq/NextSeq, Illumina) and variants identified using CLCbio Biomedical Genomics Workbench 2.1 (Qiagen). Genomic gain/loss estimated using Nexus Copy Number Algorithm 8.0 (BioDiscovery) and CNVkit (https://github.com/etal/cnvkit). Performance metrics were set on 152 ccRCC (128 fresh-frozen and 24 FFPE) specimens. An additional set of 50 ccRCC were used for analytical validation of the panel. Diluted Jurkat cell line was used as positive control. Results: For quality control, each sample was expected to achieve >95% of targets with at least 60X coverage for 95% of each target. Among targets, ten were identified as lowperforming and excluded from analysis. About 100 ng (fresh-frozen) and 250ng (FFPE) DNA were used as input routinely, whereas as low as 25 ng (fresh-frozen) and 50 ng (FFPE) yielded reproducible results. Average read depth of 750X and 1143X was achieved across fresh-frozen and FFPE samples respectively. Accuracy of variant detection (identified by Sanger sequencing or CLIA-approved NGS assay) and gain/loss (detected by array-CGH) were evaluated in 15 and 25 specimens respectively and >95% concordance was observed for both. For variant detection, analytical sensitivity was established using Jurkat cell line and assay limit of detection (LOD) was found to be 5% allele variant frequency (AVF) at >200X read depth and 10% AVF at 60-200X read depth. LOD was also confirmed using diluted clinical specimens. For gain/loss detection, analytical sensitivity was 40% for single copy gains/losses using diluted A498 cell line. Reproducibility/precision analysis also showed high intra-and inter-run concordance. Conclusions: Overall, the Focus::Renal NGS assay developed by us has undergone rigorous validation to detect genomic alterations relevant in ccRCC and thereby assist in the overall clinical management of ccRCC patients. H. Fernandes, H. Zia, H. Rennert, M. Kluk Weill Cornell Medicine, New York, NY. Introduction: NGS assays for detection of tumor variants have replaced conventional methodologies at several major medical centers. However, due to the lack of consensus guidelines in variant classification and reporting of assays, the data presented in NGS-based diagnostic reports varies considerably among laboratories. We compared the clinical NGS reports from an academic laboratory at Weill Cornell Medicine (WCM) with a commercially derived report from QIAGEN's Clinical Insight Interpret (QCI) platform generated from the same variant call format (VCF) file. Methods: 50 FFPE lung and colorectal cancer specimens that were sequenced using the Ion Torrent platform (Life Technologies), were selected for the comparison. The VCF files from the specimens were found to harbor single nucleotide variants (SNVs) and indels with COSMIC ID's that were annotated at WCM, using the Torrent suite variant caller software v4.4 (Life Technologies). The same VCF files were subsequently parsed through QCI software for decision support of clinically relevant variants. Variants from 5 additional lung adenocarcinoma specimens with complex EGFR variants including indels, were also compared. Data present in the molecular pathology reports generated for clinical purposes at WCM, were compared to the reports created by QCI. Results: At WCM, the validity of positive variant calls is based on assigned QC metrics and the classification of variants into 3 tiers is based on relevance to therapeutic actionability, diagnostic and prognostic significance. Using QCI, the rationale for reporting a variant and classification based on pathogenicity is user assigned. Using comparable metrics 83 variants in 12 genes, all having COSMIC ID's were evaluated in 50 specimens, by WCM and QCI. Among these were several clinically relevant cancer-related hotspot variants pertinent in lung adenocarcinoma and colorectal cancer including EGFR exon 19 deletions and exon 20 insertions. WCM reported the variant allelic frequency, coverage and COSMIC ID for each variant while QCI provided detailed information on approved therapies and current clinical trials related to the variant in the tumor. Only QCI was able to annotate targetable complex indels present in exons 19 and 20 of the EGFR gene from 5 lung adenocarcinoma specimens. The NGS-based reports generated using laboratory developed tools may have limitations in providing supplemental information on the clinical report. QCI reports provided an evidence-based approach to targeted therapy and trial matching. The dedicated resources in a commercial setting offer clinical evidence to enhance Target with known fusions from colon, brain, bladder and lung tumors, 13 previously uncharacterized tumor samples of thyroid carcinoma, prostate carcinoma, and sarcoma were screened. Methods: The libraries were prepared using the Archer Universal RNA Reagent Kits v2 and the FusionPlex Solid Panel probe mix, quantified using KAPA Biosystems Library Quantification Kit and then sequenced on the Illumina MiSeq using v2 Reagent Kit/300 cycles. Data were analyzed using the Archer Analysis 4.0 software followed by Sanger sequencing for confirmation of detected fusions. The fusion status of 13 clinical cases was determined using RNA extracted from formalin-fixed paraffin-embedded tissues (FFPE). Samples tested included 5 papillary thyroid carcinomas (PTC), 4 anaplastic thyroid carcinomas (ATC), 3 prostate carcinomas, and one sarcoma. Results: All expected fusions including CCDC6-RET (thyroid), ELM4-ALK (lung), NPM1-ALK (lymph node), TPM3-NTRK1 (colon), FGFR3-TACC3 (bladder), GOPC-ROS1 (brain) were successfully detected during the technical validation. Of the 13 previously uncharacterized tumor samples, two PTC cases were positive for the common CCD6-RET fusion, one was positive for the NCOA4-RET fusion with a novel breakpoint in NCOA4 gene, and one was positive for a novel fusion PEX14-BRAF. One ATC case was positive for a novel fusion, WHSC1L1-NUTM1. The sarcoma case was positive for the EWSR1-CREB3L2 fusion. Two of the cases ofprostate carcinoma cases were positive for the TMPRSS2-ERG fusion. Conclusions: NGS fusion assays enable simultaneous screening and detection of known gene fusions using FFPE tissue, and allows detection of novel translocations which would be missed by targeted FISH assays. This multigene NGS assay can be implemented in routine clinical practice at a lower cost than comparable FISH panels, utilizes tissue more efficiently and allows detection of rare and/or novel breakpoints. Barcoded libraries were prepared using 10ng of DNA; sequencing was performed on the Ion Torrent PGM (318 chip). Variants were identified using the Ion Torrent Variant Caller Plugin and reference genome hg19. Golden Helix's SVS software was used for annotation and prediction of the significance of the variants. Results: Seven samples from six patients (1.2%) were positive for IDH1 (p.R132C and p.132L in two patients each) or, IDH2 mutations (p.R140W and p.R172S in one patient each). All patients' tumors had adenocarcinoma histology. The patients' age ranged from 72 years to 83 years and two of the patients were female (33.3%). All patients were current (2) or former (4) smokers. Four patients presented with stage IV disease and two with stage I disease. Three are deceased, two are alive with disease and one is lost to follow up. In five instances (83%), the tumors also had a KRAS mutation (p.G12D (2), p.G12V, p.G12C, p.Q61H). In the sixth patient, KIT p.M541L and APC p.T175fs mutations were also detected. Conclusions: Known cancer associated IDH1/IDH2 mutations are rarely identified in a subset of NSCLCs. In our population, these were associated with patient age greater than 70 years, a history of smoking and KRAS mutations. Additional studies are needed to understand the role of IDH1/IDH2 mutations in the development of NSCLC but such patients appear to be of more advanced age and may be eligible for IDH1/IDH2 targeted therapies such as the multi-kinase inhibitor dasatinib. Introduction: Human papillomavirus (HPV) is known to be associated with squamous cell carcinomas of the head and neck (HNSCC), especially oropharynx. Patients whose tumors are HPV positive tend to have a significantly better response to treatment and better overall survival. Also, transcriptionally-active HPV-related HNSCC is clinically, biologically, morphologically and molecularly distinct form of HNSCC. Therefore it is critical to choose the best testing modality for an accurate diagnosis of high-risk HPV in HNSCC. Both, p16 IHC and HPV DNA ISH, lack specificity and do not correlate with biologically active disease. We have validated the APTIMA HPV assay which targets E6/E7 mRNA of 14 high-risk HPV on FFPE samples with excellent sensitivity and specificity. Objective of this study was to correlate these 3 assays on formalin-fixed paraffin embedded (FFPE) to identify single best testing modality. Methods: Twenty two (22) FFPE samples of HNSCC were tested. One mucoepidermoid carcinoma (MC) was also included. P16 IHC was performed on Leica Bond III per manufacturer protocol. P16 slides were independently reviewed by 2 authors who were blinded to the HPV ISH and APTIMA jmd.amjpathol.org ■ The Journal of Molecular Diagnostics HPV results. HPV DNA ISH results were available from a national reference laboratory. For in-house molecular testing (APTIMA HPV Assay), macrodissection was performed on all the specimens followed by RNA extraction using the Promega LEV FFPE RNA kit. Purity and concentrations were measured using Nanodrop and QuantiFluor on Quantus. All samples were run in duplicate with an input of at least 40ng/tube using the APTIMA HPV Assay on Panther instrument (Hologic). Results: Most common site of involvement was tonsil (n=8). Majority of the tumors were moderate to poorly differentiated and non-keratinizing. P16 IHC was positive in all the cases, including MC, with percentage positivity ranging from 20% up to 100% (median 60%). HPV DNA ISH was positive in 12 of the 22 samples, with 4 negative and 6 indeterminate results. Twenty one (21) of the 22 cases were positive for highrisk HPV by APTIMA assay. One HNSCC was negative by both ISH and APTIMA assay but positive by p16 (40%). ME was negative by both DNA ISH and APTIMA assay. There was no significant correlation between histologic subtype, keratinization and/or percent p16 positivity with the presence of high-risk HPV as seen by the APTIMA assay. Conclusions: APTIMA HPV mRNA assay is highly sensitive and specific for detection of high-risk HPV subtypes associated with HNSCC. Since the presence of transcriptionally active HPV in HNSCC is highly and independently predictive of better patient survival, we propose that the mRNA based assays, preferably using PCR, should be used as the sole testing modality for an accurate determination of HPV status in HNSCC. The ability to construct a library is critically dependent on the preanalytical tissue selection as well as the quality and quantity of nucleic acid extracted from FFPE material. Whereas there are general guidelines for NGS, most of the preanalytical and sample preparation steps are set by individual institutions based on a validation process. Here we reviewed clinically tested FFPE samples submitted for solid tumor NGS testing to determine factors that contribute to a successfully constructed library and sequencing. Methods: We performed a retrospective review of 592 FFPE samples that were submitted for solid tumor by NGS clinical testing. Our primary objective was to identify pre-analytic factors that contribute to the overall success of extracting DNA and preparing a sequenceable library from FFPE tissue for next-generation sequencing. Pre-analytic factors reviewed were sample type (resections, biopsies, and cytology cell blocks), tumor percentage, tissue fixation, percent necrosis, percent mucin, and tumor heterogeneity. Additionally, extracted samples were quantified for DNA concentration with the Qubit fluorometer using the dsDNA BR Assay. In a subset of the samples, quality of the extracted DNA was assessed with the QuantideX DNA Assay to amplifiable DNA copy number. Results: There was significant correlation (p < 0.001) for successful library construction in four of the observed variables: extracted DNA concentration, quality of DNA, tumor percentage, and sample type. Samples with an extraction concentration > 10ng/μL had a 98% success rate whereas 61.9% respectively. In assessing the quality of the extracted DNA, libraries above a cut-off of 300 amplifiable copies per μL were five times more likely to produce a successful library than those below this cut-off. Samples with a tumor percentage > produced a successful library 84.4%. Lastly, the rate of successful libraries produced for resection samples was 93.1%, for biopsies was 83.6%, and for cytology cell block samples was 90.9%. Understanding the underlying factors as they relate to sample preparation of the NGS workflow enabled us to construct tree models for troubleshooting clinical samples in the future. Conclusion: The results highlight the value of understanding pre-analytical and early sample preparation factors that contribute to the ability to extract DNA, construct successful libraries and subsequent sequencing. Furthermore, this study has allowed us to implement a review process clinically to predict whether sample failure may be due to technical issues or sample integrity thus guiding repeat testing and improving overall patient care. Introduction: Next-generation sequencing panels are becoming increasingly important in cancer diagnosis for determining treatment as well as prognosis for the patient. Assays to detect point mutations in oncogenes are common, but detection of translocations and copy number variation may also be critical. In collaboration with oncologists and pathology medical directors, a comprehensive solid tumor NGS panel was developed, consisting of actionable and prognostic genes covering SNVs, translocations, and copy number changes. Methods: DNA was isolated from blood, FFPE tissue, or FFPE cell lines. Libraries were prepared using 10 ng to 50 ng of DNA and the KAPA Hyper Prep Kit, and then sequenced on the Illumina NextSeq500 instrument. Analysis for single nucleotide variants (SNVs) used LoFreq, translocations used Delly, copy number used a modified Contra/DNAcopy, and LOH used B-allele frequencies. This panel contained full coding exon coverage of 126 genes for SNV detection. For translocation detection, 18 genes had selected intronic coverage. The 79 CNV genes had exon as well as intronic coverage over select known population SNVs; to gain B-allele frequency data for LOH detection, as well as providing additional copy number data. Samples consisting of three sets of tumor/normal pairs for CNV analysis, four cell lines with ALK, RET, ROS1, and NTRK1 translocations, and the Horizon FFPE Quantitative Multiplex reference standard were tested for proof of principle to determine if multiple types of mutations could be detecting using this NGS assay design. Results: All three types of variations (SNVs, translocations, and copy number changes) were successfully detected. The Horizon standard had all SNVs called as expected above a 3% variant allele frequency. The NGS copy number and B-allele frequency data correlated well with results using the Oncoscan microarray data. The translocations were identified in all cell lines. Conclusions: This comprehensive cancer panel is successfully able to capture regions of interest for somatic cancer and can identify translocation, copy number, LOH and SNVs in the same NGS assay. 1 1 Thermo Fisher Scientific, South San Francisco, CA; 2 Thermo Fisher Scientific, Carlsbad, CA; 3 CosmosID, Rockville, MD. Introduction: At this time next-generation sequencing (NGS) is hindered by slow and often manual workflow procedures. Decreasing overall workflow times is critical for the widespread adoption of targeted and whole genome sequencing (WGS) for many time-sensitive applications, in particular for infectious disease analysis. To this end, we describe improvements to the four main steps of the NGS workflow: i) library preparation; ii) template preparation, iii) sequencing; and iv) data analysis. Together, these advances dramatically decrease the overall turnaround times. Methods: The new rapid workflow innovations were applied to two different library preparation protocols: i) targeted libraries created using a highly-multiplexed PCR approach consisting of 1200 amplicons targeting the 16S rRNA gene as well as speciesspecific identification targets and antimicrobial resistance determinants; and ii) an unbiased WGS approach using a MuA transposon-based library preparation method. To assess the accuracy of detection with the improved workflows, nucleic acid from six bacterial cultures (A. baumannii, E. cloacae, E. faecium, K. pneumoniae, P. aeruginosa, and S. aureus) were extracted as input for targeted and WGS sequencing. Targeted libraries were optimized for speed by reducing amplification and incubation times as well as substitution of the polymerase used in standard library protocols. The MuSeek-based WGS library preparation times were improved by eliminating a cleanup step. Both targeted and WGS libraries were clonally amplified (template preparation) using an isothermal amplification approach that saves 3 hours over the standard methods. Sequencing times were improved by reducing nucleotide flow times and the total number of flows. The implementation of On-Instrument Analysis (OIA) enabled near real-time base calling reducing the primary analysis time. Results: Targeted and WGS libraries were generated, sequenced, and analyzed in 6.5 hours with sequencing and analysis taking 50 minutes compared to 2.5 hours and 1 hour for standard sequencing and analysis, respectively. Analysis of sequencing accuracy revealed a raw read accuracy >99.5%, comparable to data from the standard workflow. The read length distribution for the targeted libraries was unaffected by speed improvements with a distribution similar to the standard workflow. Further, 100% specificity for species identification and antimicrobial resistance determinants was observed for targeted and WGS libraries indicating rapid sequencing without compromising detection accuracy. Conclusions: For the targeted and WGS methods described, the total turnaround time from isolated nucleic acid to sequencing data could be completed in a typical workday. Introduction: ChimerMarker (SoftGenetics) is a commercially available software package designed to integrate analysis, genotyping, and chimerism calculations by short-tandem repeat (STR) polymerase chain reaction (PCR). We present the evaluation, validation, and implementation of ChimerMarker into the workflow of a clinical laboratory for post-transplant chimerism analysis by STR PCR. Methods: Thirteen cases ranging from 0% to 100% recipient were selected for software validation with comparison to a manual, single locus calculation method. PCR was performed using Powerplex 16 HS System (Promega), followed by capillary gel electrophoresis on a 3500 (Applied Biosystems). Pre-transplant projects were created using a novel algorithm developed to include or exclude informative loci based on stutter positions and signal background. Special features, such as deconvolution were incorporated as needed. Post-transplant samples were analyzed using the selected loci from the pre-transplant project. Data from 12 recipient/donor pairs were gathered to calculate the percent stutter at each locus and allele in the cohort (n-1, n-2, and n+1 positions), which were then used to set marker-specific stutter filtering and adjustment. Varied DNA input was also tested in 9 patients to identify optimal assay conditions. An additional 120 patients were examined to compare the program's calculated % chimerism result with the manual, single locus method as well as to determine cut-offs for the lower limit of detection. Criteria for interpreting 100% donor and 100% recipient were developed. Results: On average, 4 loci were used (range 2 to 12) per sample for calculation in ChimerMarker. Correlation with the manual single locus calculation method was strong (R 2 =0.99) across all levels of engraftment. DNA inputs of 2 ng were more reliable, with lower % CV, less bias, and less allelic imbalance or allele dropout compared to 1 ng. The average stutter percentage varied at each locus from 1% to 10%, with highest % in the n-1 position. Since the stutter adjustment chimerism calculation performed by the software only corrects for n-1 stutter, manually excluding loci with n-1, n-2, and n+1 stutter from the analysis markedly improved the accuracy of the results. The analytic sensitivity of the software reliably detected donor or recipient DNA down to at least 1%. Conclusions: We developed a novel algorithm for the selection of loci in the ChimerMarker software that minimizes stutter positions and background, providing a reliable, sensitive, and reproducible approach to post-transplant chimerism analysis. These alterations empower the software to detect complete engraftment equal to or better than the manual, single locus method. H. Achi, F. Abbas, R. Abdel Khalek, R. Mahfouz American University of Beirut Medical Center, Beirut, Lebanon. Introduction: Quantitative measurements of HIV viremia in the peripheral blood have shown that higher virus levels may be correlated with increased risk of clinical progression of HIV disease, and reduction in viremia level is associated with a decreased risk. Virus levels can be quantified by measurement of the HIV p24 antigen, by quantitative culture of HIV, or by direct measurement of viral RNA using nucleic acid amplification technologies, such as the polymerase chain reaction (PCR) that achieve high sensitivity and dynamic range for the quantitative detection of HIV RNA, to assess prognosis and monitor the effects of antiretroviral therapy. This pilot study compares the performance and the results of two HIV qPCR platforms, COBAS Ampliprep/COBAS TaqMan (Roche Molecular Diagnostics) and GeneXpert (CEPHEID), for a total of 34 patients referred for HIV testing at a major center. Methods: The HIV-1 Quant Assay kit, constitutes a readyto-use system for the rapid quantification of HIV-1 group M, N and O, in infected individuals using polymerase chain reaction on GeneXpert Instrument. The system automates sample purification, nucleic acid amplification, and detection of the target sequence using real-time reverse transcriptase PCR. The process require the use of single-use disposable GeneXpert cartridges that hold the reagents and host the PCR. It includes reagents, 2 internal controls for quantitation of HIV-1 RNA and for identifying possible PCR inhibitors, and a Probe Check Control that verifies the probe integrity.The COBAS AmpliPrep/COBAS TaqMan HIV-1 Test permits automated specimen preparation followed by PCR amplification, detection of HIV-1 target RNA and HIV-1 Quantitation Standard (QS) RNA for group M. The Master Mix reagent contains primers and probes specific for both HIV-1 RNA and HIV-1 QS RNA. The detection of amplified DNA is performed using target-specific and QSspecific dual-labeled oligonucleotide probes that permit independent identification of HIV-1 amplicon and HIV-1 QS amplicon. The quantitation of HIV-1 viral RNA is performed using the HIV-1 QS. It compensates for effects of inhibition and controls the preparation and amplification processes, allowing a more accurate quantitation of HIV RNA in each specimen. The results of 32 patients run using both kits were compared using the EP evaluator software. Results: The difference between the 2 methods was within the allowable error for 28 out of 32 specimens (87.5%), with a correlation coefficient R=0.85. Conclusion: The HIV-1 GeneXpert PCR Kit and the COBAS AmpliPrep/COBAS TaqMan HIV-1 kit are 2 acceptable assays that can be used for the sensitive detection of HIV in a wide variety of clinical specimens. (QIAGEN) and the platform used by the College Of American Pathologists on respiratory samples for a total of 9 patients at AUBMC, using two extraction methods. Methods: The principle of the FTD respiratory pathogens 33 RG PCR Kit is based on the transcription of viral RNA into cDNA using a specific primer mediated reverse transcription step followed by amplification of the DNA of different pathogens. The presence of specific pathogen sequences in the reaction is detected by an increase in fluorescence observed from the relevant dual-labeled probe, and is reported as a cycle threshold value (Ct) by the Real-Time thermocycler. The assay uses Equine arteritis virus as an internal control, which is introduced into each sample and the negative control at the extraction process. Nucleic acid material was manually extracted from the sample using QIAamp UltraSens Virus kit for 4 of the patients and QIAamp Viral RNA Mini kit for the others. We studied overall 18 samples and the results of 14 were compared to the College of American Pathologists results using the EP evaluator software. The remaining 4 were compared to the meridian platform results, and awaiting the CAP results. Results: The samples extracted using QIAamp UltraSens gave "no signal" result on Rotor Gene. When retested using QIAamp Viral RNA mini kit, the results of the 14 patients were identical to those of the CAP. As for the 4 pending patients we had similar results between Rotor Gene and Meridian platforms. Difference between the methods was within the allowable error for 18 out of 18 specimens (100%), with a correlation coefficient R=1. Conclusion: The FTD respiratory pathogens 33 RG PCR Kit is an acceptable assay that can be used for the sensitive detection of a wide variety of respiratory pathogens using, as recommended, the QIAamp Viral RNA mini kit for nuclei acid extraction. Introduction: Next-generation sequencing is a powerful technique capable of detecting a variety of genetic alterations in multiple genes simultaneously. Focused cancer hotspot panels are widely used and can detect single nucleotide variants and small insertions/deletions (in/dels) that are recurrent in cancer. Although focused ampliseq panels are not typically used to detect copy number alterations, we sought to determine if the data from such an assay could be used to infer copy number alteration status of specific genes in glioblastoma multiforme (GBM). Methods: A total of 41 GBM cases were tested to date. DNA was extracted from the tumorenriched area of FFPE sections and was sequenced on the Ion Torrent Personal Genome Machine (PGM) after library preparation and enrichment using AmpliSeq Cancer Hotspot Panel v2 (Thermo Fisher Scientific). Copy number alteration status was determined by an algorithm to calculate the Log2 values of the fractional coverage of EGFR, CDKN2A, PDGFRA and PTEN for each tumor after normalization with pooled normal samples. The copy number status from the ampliseq panel was then compared to results from other platforms, i.e., whole exome sequencing (WES), targeted hybrid-capture panels, EGFR chromogenic in situ hybridization (CISH) and digital droplet PCR (ddPCR). Results: Thirty-seven percent of cases (15/41) were positive for EGFR amplification (log2 ratio >3), consistent with the reported prevalence of EGFR amplification in GBM. Of these, 100% (15/15) were confirmed to show EGFR amplification by other methods (CISH, WES, ddPCR). 63% of cases (26/41) were negative for EGFR amplification by ampliseq; of these, 11/26 cases had data available from other methods and 100% (11/11) were confirmed to be negative for EGFR amplification. 39% of cases (16/41) showed CDKN2A deletion (log 2 ratio <-2) and 100% (12/12) cases with data available from other methods were confirmed for CDKN2A deletion. PDGFRA amplifications (Log2 ratio >2) were found in 5% of cases (2/41) and were confirmed by orthogonal testing. Lastly, PTEN deletion (Log2 ratio <-2) was observed in 20% of cases (8/41). Conclusions: These findings demonstrate that selective copy number alteration status can be successfully determined in glioblastoma multiforme using the data acquired during the routine use of a targeted cancer hotspot panel, thereby broadening the scope of genetic alterations that can be detected by this panel. Introduction: Next-generation sequencing (NGS) has been rapidly adopted in clinical laboratories as highly complex tests that can substantially vary in both their design and application, bringing challenges that can be associated with sample handling, library preparation, data analysis, and reporting. To be able to monitor the multi-step NGS workflow and consistently provide accurate results, quality control materials should be included both during the validation phase, as well as during routine clinical tests. The goal of this multi-laboratory study was to evaluate the Seraseq AF10 and AF20 reference materials by verifying their performance as a quality control material, and by evaluating their ability to measure quality metrics vital to a clinical test. These characteristics were assessed within and between the cooperating laboratories. Methods: Six CLIA certified clinical laboratories and one research laboratory participated in this study. The Seraseq AF10 and AF20 reference materials consist of 26 biosynthetic DNA plasmids, each containing a specific variant or Mutation of Interest (MOI) and an Internal Quality Marker (IQM). Both reference materials were included in their respective routine clinical NGS pipelines over an 8-week period. Six of the seven laboratories ran a target amplicon based assay, and four of these laboratories used the Ion Torrent PGM instrument, whereas the other two laboratories used the Illumina MiSeq. One laboratory used a target hybrid capture based assay on the Illumina NexSeq500. Five out of six NGS panels used by laboratories during this study covered all 26 variants present in Seraseq AF10 and AF20 reference materials. Results: Both reference materials generated reproducible results when performed by different operators, using different sequencing panels and platforms. Although the laboratories identified most of MOI and their corresponding IQM present in the reference materials, some discrepancies involving either the MOIs (MiSeq platform only) or IQMs (MiSeq and Ion Torrent PGM platforms) were observed. No significant difference was observed between expected and observed allelic frequency in the Seraseq AF10 and AF20 reference materials. When comparing allelic frequency in each variant type (deletion, insertion, SNVs in homopolymer regions, and SNVs) for each laboratory, no significant differences were observed. Allelic frequency was also used to indicate reagent stability over the study period. Conclusion: The Seraseq AF10 and AF20 reference materials showed characteristics of an ideal quality control material (high DNA quality, high stability, high flexibility, and genomic complexity) that could be used during validation and routine clinical sequencing processes of NGS cancer panels. Introduction: Fusion proteins created by gene rearrangements are common in cancer and are often attractive drug targets. However, the clinical detection of gene rearrangements can be challenging, particularly if performed via next-generation sequencing (NGS). This becomes even more difficult in cases of complex or noncanonical rearrangement events and requires sophisticated bioinformatic approaches. Here we demonstrate a clinical case of a difficult to detect EML4-ALK fusion in which the genomic breakpoint occurred within exon 20 of ALK and the resulting fusion transcript included intronic sequence fromEML4. Methods: The rearrangement was initially detected by break-apart fluorescence in-situ hybridization (FISH). Anchored multiplex PCR (AMP) was performed on total nucleic acid via the Archer FusionPlex Solid Tumor kit. Two separate bioinformatic approaches were used for analysis of AMP NGS data: an alignment based fusion detection method and a de novo assembly strategy. The de novo assembly method took advantage of the anchored nature of the AMP assay to allow for anchored clustering and assembly. Findings from the AMP assay were confirmed via reverse-transcriptase PCR. The genomic breakpoint was confirmed via a DNA-based capture NGS assay. Results: The ALK FISH-positive lung cancer patient responded to crizotinib and then developed the well characterized ALK F1174C resistance mutation, confirming the presence of a productive ALK rearrangement. Tumor material from the patient was subsequently used for validation of the AMP assay. Initial attempts to detect the fusion using the alignment-based method were unsuccessful. However, upon adoption of the updated analysis algorithm that employed de novo assembly prior to annotation, the fusion transcript was detected, and it was revealed that the ALK breakpoint occurred within exon 20 and that the transcript included intronic sequence from EML4. The de novo assembly was required because the two breakpoints could not be detected solely via the alignment-based method. Additionally, the intron retention event created a large insertion between the two exonic regions, which was difficult to capture in a single read but could easily be assembled from a collection of anchored reads. The complex transcript and exonic breakpoint were orthogonally confirmed. Conclusions: NGS is becoming a platform of choice for detection of gene rearrangements in many laboratories. However, complex and non-canonical rearrangement events present additional challenges for bioinformatic detection. Here we report that de novo assembly of sequence reads is critical for the detection of fusion transcripts that do not conform to canonical patterns. L.Z. Hong 1 , L. Zhou 2 , R. Zou 2 , A. Chen 1 , S. Shih 1 , C. Chin 1 1 Merck Sharp & Dohme, Singapore; 2 MiRXES Pte Ltd, Singapore. Introduction: microRNA (miRNA) profiling in biofluids has the potential for identifying biomarkers that are informative for early diagnosis or predictive of treatment response. However, quantifying miRNA levels in biofluids is technically challenging due to their low abundance, the small size of mature miRNAs, and the high degree of sequence homology between family members. Using a set of reference samples, we describe a systematic evaluation on which miRNA profiling was performed using three different platforms: qPCR, NanoString, miRNA-Seq. Methods: RNA extraction from 200 μl of a Reference Serum (Ref. Serum) sample was performed using the miRNeasy Serum/Plasma Kit (Qiagen). FirstChoice Human Brain Reference RNA (Thermo Fisher Scientific) was used as a QC sample. miRNA profiling was performed on the MiRXES qPCR platform by splitting the RNA sample for reverse transcription using pools of gene-specific primers, followed by multiplex cDNA amplification. Amplified cDNA was used as template for gene-specific quantitation in a single-plex assay. The expression level of each miRNA was interpolated from a standard curve generated from a synthetic sequence. NanoString analysis was performed using the nCounter Human v3 miRNA Expression Assay Kit (NanoString). Raw counts were background subtracted using the negative controls, normalized to the positive controls, and normalized to the top 100 genes. miRNA-Seq library preparation was performed using the TruSeq Small RNA Library Prep Kit (Illumina), followed by 1 x 40 bp sequencing on a NextSeq 500. Within each sample, expression levels for each miRNA were normalized to reads per million mapped reads (RPM). Results: In Ref. Serum, the qPCR and miRNA-Seq platforms had almost perfect concordance between runs (concordance correlation coefficient, ccc = 0.99) but the NanoString platform had moderate concordance (ccc = 0.82). There were significant differences in the number of miRNAs detected above the lower limit of quantification (LLOQ)-NanoString only detected 84 miRNAs but qPCR and miRNA-Seq detected 440 and 372 miRNAs respectively. Further, 591 miRNAs were detected above the LLOQ in Ref. Serum by at least one platform, but only 48 miRNAs (8%) were detected by all three platforms. The highest inter-platform correlation was observed between miRNA-Seq and qPCR (r = 0.69), followed by NanoString and qPCR (r = 0.46) and miRNA-Seq and NanoString (r = 0.17). Fourteen novel miRNAs detected by miRNA-Seq were validated using qPCR and their expression levels were measured in six different cell lines. Conclusions: Our results suggest that using miRNA-Seq for discovery and targeted qPCR for validation is a rational strategy for miRNA biomarker development in clinical samples that involve limited amounts of biofluids. Introduction: It is known that due to adverse effects of formalin fixation and paraffin embedding on DNA quality, FFPE specimens are often not compatible with molecular genetic tests. Nevertheless, FFPE-derived DNA has been reliably used for analysing gene copy number status in a plethora of Multiplex Ligation-dependent Probe Amplification (MLPA)-based studies. MLPA probes target relatively short sequences of up to 100 base-pairs and thereby formalin treatment-induced DNA fragmentation does not hinder MLPA. However, formalin-induced DNA crosslinking and base modifications can affect MLPA if not eliminated or reduced during DNA extraction. To date, there has been no detailed investigation on the effects of tissue fixation conditions and DNA extraction methods on MLPA, so this study focused on selecting the most optimal pre-analytic conditions for MLPA. Methods: To compare tissue fixation conditions, healthy colon tissue was fixed in 1) buffered or nonbuffered formalin for 2) one hour, 12h to 24h or 48h to 60h; 3) at 4 o C or at room temperature (RT). DNA extracted from differently fixed and paraffin-embedded tissues was used for MLPA by two probemixes. To compare DNA extraction methods, four commercial kits (RecoverAll Total Nucleic Acid Isolation, QIAamp DNA FFPE tissue, Zymo Research FFPE DNA miniprep, WaxFree DNA extraction kit) and one in-house method were used to extract genomic DNA from eight different FFPE extraction in triplicate. The data were normalised against commercial genomic DNA. Results: MLPA results were greatly influenced by FFPE tissue fixation conditions, DNA extraction method and tissue type. MLPA results with lowest variability were obtained on DNA derived from tissues fixed for 12h to 24h in buffered formalin at RT. For most tissues, in-house DNA extraction method produced DNA with the lowest number of MLPA probes (1% to 27%) deviating from the normal copy number ratio range in all tissues, followed by Zymo Research FFPE DNA miniprep (1% to 41%), WaxFree DNA extraction (1% to 71%) and QIAamp DNA FFPE Tissue kit (9% to 71%). Furthermore, in-house FFPE DNA extraction method was shown to perform as efficient or superior to other methods in suitability for MLPA, DNA yield, time-, cost-efficiency and ease of performance. Conclusions: The recovery of DNA from clinical FFPE tissue samples is a challenging task. As the fixation conditions and extraction method impact copy number quantification, it is critical to use the same pre-analytic conditions for all samples, including reference samples. Our study shows that FFPE-derived DNA using optimal tissue fixation and DNA extraction methods is well-suited for MLPA analysis. The Journal of Molecular Diagnostics ■ jmd.amjpathol.org HPV Assay can be used with a Pap specimen to assess the presence of high risk HPV types. This study compares results of 3 HPV PCR platforms: Genomica microarray Hybrid Capture 2 GeneXpert, and a nucleic-acid hybridization assay, for a total of 25 patients. Methods: CLART HPV2 Genomica detects single infections or co-infections. It genotypes 35 HPV types, including the High and Low Risk based on the amplification of specific fragments of the viral genome and their hybridization with specific probes for each HPV type. It is based on the extraction of HPV DNA from swabs, cell suspensions and FFPE tissues, Followed by amplification of DNA. Then, the amplified product is visualized on CLART-strip. The detection is performed by means of a low-density microarray platform: CLART which consists of a microarray printed at the bottom of the microtiter plate. The Xpert HPV Assay for detection and differentiation of HPV, is performed on Cepheid GeneXpert Instrument. The System automate sample processing, nucleic acid amplification, and detection of the target sequences in cervical samples by real-time PCR. A disposable GeneXpert cartridges holds the reagents and carry out the processing. It contains primers and probe for the detection of HPV16 and HPV 18/45 in two distinct detection channels, and 11 other high risk types in a pooled result. Hybrid Capture 2 is a signal amplification nucleic acid hybridization assay that utilizes microplate chemiluminescent detection. It targets high risk HPV types. DNA sample hybridize with a specific HPV RNA probe cocktail. The RNA:DNA hybrids are captured onto the surface of a microplate coated with specific antibodies. Once Immobilized the hybrids react with alkaline phosphatase conjugated specific antibodies, and then detected with a chemiluminescent substrate that emits a light. The results of 25 patients run using Genomica microarray and other platforms were compared using the EP evaluator. Results: GeneXpert assigned two cases as « HPV18/45 » and other high risk types, whereas Genomica specified them differently as « 51/83 » and « 31/51/66 » respectively. As for the other patients there was no difference between the methods with a correlation coefficient R=0.92. Conclusion: The Genomica microarray, Hybrid Capture 2, GeneXpert and the nucleic acid hybridization assay are 4 acceptable assays that can be used for the detection of HPV in cervical specimens. Introduction: As a clinical core laboratory, streamlined chemistries and processes are essential to provide efficient and timely results. Standardized chemistries across assays allow for easier reagent management. Likewise, standardized processes reduce the number of workflows that must be supported in laboratory developed software and liquid handler methods necessary to process samples. Methods: Two existing TruSeq Nano (Illumina, San Diego, CA) library preparation processes supporting three assays were consolidated into one new workflow. Reagent volumes were standardized and validated across assays utilizing the existing workflows. Additionally, custom designed universal adapters and 10 base-pair (bp) index tags were validated to allow for future reagent standardization with other chemistries. Updates were made to the in-house developed LIS (Laboratory Information System), to write a driver file for Biomek FX P Liquid Handlers (Beckman Coulter, Indianapolis, IN) containing variables and transfer instructions for each process step, reducing the number of necessary robot methods from 11 down to 3. Results: Standardized reagent volumes produced concordant variant call results compared to the existing workflows. Batching of samples into single library preparation runs can now occur, allowing more efficient use of liquid handlers and technologist efforts. Combination of sample processing from the two workflows results in 8 hours of saved bench time per batched run, and opens up the use of a liquid handler for other work in the laboratory. TruSeq Nano library preparation reagent kits come with 24 index tags, limiting the amount of samples per pool on an Illumina HiSeq or MiSeq instrument. Of the custom designed 10 bp index tags, 96 individual oligos were validated, resulting in a four-fold increase of the maximum number of samples allowed per pool. LIS updates to write Biomek driver files allowed for the reduction in the amount of methods to manage, and will also reduce the inadvertent selection of the wrong method during run processing. Conclusions: Development of streamlined chemistries and processes are necessary in a core laboratory to efficiently manage the growing number of NGS assays under development. Having established core processes allows for faster run processing by sample batching, and overall process and reagent management is less time consuming. The development and implementation of custom reagents are a powerful strategy to use in a core laboratory. Custom adapters and 10 bp indexes will be utilized by other library preparation methods in the laboratory, thus eliminating issues associated with particular kits and also allowing sample pooling between library preparation methods currently using different length index tags. B. Wei, M. Kibukawa, J. Kang, M. Marton, D. Levitan Merck & Co., Inc., Rahway, NJ. Introduction: KEYNOTE-051 is a clinical trial studying Pembrolizumab (MK-3475) in pediatric patients with advanced melanoma or advanced, relapsed, or refractory PD-L1-positive solid tumors or lymphoma. One of the exploratory objectives is to perform mutation analysis of 8 genes that may be associated with pediatric tumors and potentially portend differential treatment response. Methods: To maximize amplifiable DNA extracted from the often limited pediatric tumor specimens, we selected and optimized Roche High Pure RNA Paraffin method for FFPE DNA isolation over 3 other commercial kits. We utilized NEBNext FFPE DNA Repair Mix to remove false positive calls in poor quality DNA caused by cytosine deamination, and matrix effect such as melanin contamination. A custom hotspot mutation panel based on Ion AmpliSeq targeted sequencing was designed to increase coverage for two challenging RB1 amplicons at 20 ng DNA input. For assay validation, we examined precision and analytical accuracy. We performed inter-run comparison starting from library preparation for precision evaluation with DNA extracted from 35 commercial pediatric tumor specimens, with 2 to 3 specimens per tumor type. To evaluate sensitivity and positive predictive value (PPV) for accuracy, we utilized reference material with known mutations (AcroMetrix Oncology Hotspot Control and HDx FFPE Quantitative Multiplex Reference Standard). We also compared the NGS results of FFPE DNA against digital PCR (dPCR) or commercial FoundationOne panel. Results: Panel-specific false positives such as indels caused by specific homopolymers and base substitutions caused by certain non-specific PCR priming were identified and eliminated to improve accuracy. We achieved high sensitivity and PPV (>0.98) for the AcroMetrix and HDx controls and high specificity (>0.99) for 9 HapMap DNA at 3% variant allele frequency (VAF) cutoff. dPCR confirmed the few low VAF (5-10%) somatic mutations detected in pediatric specimens and the FoundationOne panel confirmed somatic mutation calls of 4 FFPE DNA extracted from adult tumor samples. Although precision is excellent (100% concordance) for half of the pediatric FFPE DNA at 5% VAF cutoff, there were some low VAF false positive calls in poor quality DNA. To reduce the false positive calls, we decided to use consensus calls from two independent NGS libraries for mutation detection. Such practice greatly improved precision for all FFPE samples. Doubling DNA input to 40 ng removed false positives in 3 of the 7 most challenging samples. Conclusions: A robust and sensitive NGS assay combined with efficient FFPE DNA extraction and repair methods was developed and validated to study mutations that may be associated with pediatric tumors and/or treatment response to MK-3475. School of Medicine, Lebanon, NH. Introduction: DNA quantification is an important and necessary step for most DNA analyses, particularly next-generation sequencing (NGS). Amplicon-based library preparation offers the powerful option of sequencing only a targeted pool of genes. FFPE-preserved tissue presents a technical challenge to reproducible DNA extraction of sufficient quality for NGS due to chemical modification of the DNA during tissue processing. Standard DNA isolation procedures often result in low DNA yield or poor performance in downstream PCR applications. As the CGAT laboratory processes many variable FFPE tissues, the DNA concentrations can greatly vary, introducing the need for a broad range detection method. The scarcity of some samples requires an accurate DNA concentration while using as little of the sample as possible. The aim of this study is to determine which DNA quantification method allows for the most successful library construction. Methods: Genomic DNA was extracted from 60 FFPE tissues from patients using a DNA Extraction Column kit (QIAamp DNA FFPE Kit; QIAGEN) in the automated QIACube according to the manufacturer's instructions. The tissue included melanoma, glioma, lung, colon and breast tumors. The DNA concentrations were determined spectrophotometrically (Nanodrop), fluorometrically (PicoGreen, Qubit HS and Qubit Broad Range) and by amplification of a short, conserved region in the human genome (KAPA hgDNA Quantification and QC Kit). Each sample was used with the five methods of DNA quantification. Libraries were generated using Life Technology's Ion AmpliSeq Cancer Hotspot Panel v2 and quantified using the Ion Library TaqMan Quantitation Kit. Results: The Nanodrop consistently overestimated the DNA concentration compared to the other methods. The three methods using DNA binding dyes gave similar DNA concentrations with a good correlation between the library quantification and the DNA concentration. It is imperative that the determined concentration is not outside the range of the standard curve. The qPCR-based method gave in most cases similar DNA quantification as the DNA binding method, with the advantage of also measuring DNA quality, reflecting the ability of PCR amplification, necessary for amplicon-based library preparation. This method showed the best correlation between DNA concentration and library quantification. Appropriate dilution of the samples is necessary to obtain accurate DNA concentration, requiring the need to have an estimated DNA concentration range of the samples. Conclusion: For NGS purposes, the Qubit/High Sensitivity in junction with the KAPA hgDNA Quantification and QC Kit gave optimal results to obtain the expected quantity and high quality of library. jmd.amjpathol.org ■ The Journal of Molecular Diagnostics TT14. Evaluation of the Qiagen GeneReader NGS System for the Detection and Reporting of Clinically Actionable Mutations in Solid Tumors S. A. Turner, H.S. Jung, F.B. de Abreu, J.D. Peterson, G.J. Tsongalis Dartmouth Hitchcock Medical Center, Norris Cotton Cancer Center and Geisel School of Medicine, Lebanon, NH. Introduction: The identification of actionable mutations has resulted in novel treatments and improved outcomes for various solid and hematological malignancies. Although the use of next-generation sequencing (NGS) in the clinical laboratory has continued to expand, there is a growing need for integrated wet bench and analytic pipelines. The Qiagen GeneReader NGS System and the Qiagen Actionable Tumor Panel (ATP) has been designed as such, from semi-automated FFPE extraction to integrated bioinformatics and clinical insight, potentially allowing for wider clinical adoption of NGS solid tumor sequencing. Here, we compared the Qiagen GeneReader System (ATP) with our current clinical pipeline using the IonTorrent PGM (Ampliseq Tumor Hotspot Panelv2). Methods: Fifty (31 unique, 19 duplicate) FFPE samples from lung, colon, melanoma, and GIST tumors were extracted, sequenced, and analyzed on the IonTorrent PGM and the Qiagen GR system. The clinical pipeline for the Ampliseq Cancer Hotspot Panelv2 (50-gene) has been published and includes 500X coverage for 10 samples on a 318 Chip with a custom GoldenHelix bioinformatics pipeline. Sample preparation and sequencing on the GeneReader NGS system was performed according to the manufacturer's instructions using the hotspot Actionable Tumor Panel (12-gene) and 10 samples/ flowcell. Bioinformatics and clinical interpretation were performed using the included Qiagen Clinical Insight-Analyze (QCI-A) software package and Qiagen Clinical Insight-Interpret (QCI-I) online service. Results: The stock analysis for the Qiagen panel focuses on the detection of 773 unique variant positions in 12 genes. While additional bases are sequenced, this comparison was limited to clinically actionable variants. Of the 50 samples only one failed to meet established QCI-A quality metrics and was removed from further analysis. We noted high concordance (100%, 30/30 unique samples) across the actionable variants reported by QCI-I in comparison to those identified on the PGM. We also noted high reproducibility of the Qiagen GR System within duplicate samples (19/19). Greater than 99% of variant positions were covered at over 500X with KRAS p.A146 covered at >200 but <500X in 12 samples across 4 of 5 flowcells. Conclusions: The Qiagen GeneReader System is a fully-integrated NGS platform that offers the ability to identify a number of clinically relevant variants for clinical molecular oncology testing. The inclusion of a bioinformatics and clinical interpretation pipeline provides the user with up-to-date clinical information required in making clinical observations without the need for establishing additional in-house knowledge bases and maintaining additional analytics software packages. Introduction: Liquid biopsies are easier to collect and potentially offer a more comprehensive picture of the mutational landscape for more advanced solid tumors than tissue biopsies. However, the abundance of overall cell-free DNA (cfDNA) in these samples is generally low, ranging from less than 1ng to 30ng/mL of cfDNA in blood from healthy individuals. Unfortunately, this material is generally low abundance and cancer cell derived DNA (ctDNA) contributes a very small percentage to the overall material, except in more advanced disease states. Additionally, ctDNA is typically highly fragmented down to 100-300bp. Therefore, NGS-based assays to detect variants in ctDNA must be sensitive enough to detect low-frequency mutations (allele fractions <2%) from low mass inputs (10-100ng) of highly fragmented material. Methods: We modified our Anchored Multiplex PCR (AMP) method for DNA library construction to specifically enrich for low molecular weight ctDNA over larger cfDNA fragments. AMP is a target enrichment strategy for NGS that ligates molecular barcoded adapters to DNA fragments to generate random start sites for multiplex PCR. AMP is well suited to amplify small ctDNA fragments, as it only requires one intact primer-binding site within a fragment. Additionally, our method allows for capture of target regions from both strands independently. Through the use of the molecular barcoded adapters and a specific library purification step, we are able to precisely quantify the number of sequenced DNA molecules. Our ctDNA assay contains primers to enable NGS-based coverage of >3.4kb of many of the most common oncogenic driver mutations, acquired cancer drug resistance mutations, as well as full exon coverage of TP53. Results: Here, we describe our AMP-based ctDNA library preparation method, highlighting the use and advantages of molecular barcoded adapters. Molecular barcodes are used to unambiguously identify PCR duplicates, correct for PCR or sequencer-derived sequence errors, and flag run-to-run contamination. Using a set of reference ctDNA standards, we demonstrate that these benefits of molecular barcodes enabled confident variant detection at allele frequencies ~0.5% with as little as 50ng of input DNA and less than 4M reads. Furthermore, our ctDNA profiling assay demonstrated a similar power to call variants in ctDNA derived from liquid biopsies. Conclusions: We demonstrate that our modified AMP-based library construction method for ctDNA mutational profiling is a powerful tool for sensitive detection of variants in model ctDNA reference samples. The advantages observed with reference materials will likely be applicable to clinically derived specimens. TT16. Detection and Quantitation of Circulating Tumor BRAF Mutant DNA Using the BioRad Digital Droplet PCR (ddPCR) Method D. Milosevic, M.B. Campion, N. Vidal Folch, J.R. Mills, K.C. Halling, M.C. Liu, E.W. Highsmith, B.R. Kipp, J.S. Voss, E.L. Enninga, L.A. Kottschade, S.N. Markovic, S.K. Grebe Mayo Clinic and Foundation, Rochester, MN. Introduction: Defining BRAF status of melanoma patients is critical for targeted treatment decisions with BRAF inhibitors. Liquid biopsies provide a less invasive alternative to tissue biopsy for theranostics and monitoring. In this study we evaluated the performance characteristics of a quantitative multiplex assay for V600E and V600K assay using BioRad's ddPCR instrumentation. Methods: Blood samples were collected in Streck Cell-Free DNA BCT tubes from 20 healthy individuals and 47 patients with metastatic melanoma. DNA was extracted from plasma using Qiagen's QIAmp Circulating Nucleic Acid kit. Simultaneous quantitation of the wild-type and mutant BRAF V600E/K was performed using AutoDG and QX200 multi well plate system. Results: Intra-and inter-assay precision were assessed using 3 pools of BRAF V600E and V600K mutation-positive cell lines and varying percentages of mutant versus wild-type DNA (1% to 10% mutant in a background of wild-type sheared DNA). There was 100% concordance with expected results for the fractional abundance, with a copy number CV of 4.3-18%. Analytical sensitivity experiments showed a limit of detection of 1% mutant in 3ng of DNA input and 0.1% mutant in 30ng of DNA input. Serial dilutions of both V600E and V600K mutants were linear down to 1% mutant in 5ng of total DNA input. Detection specificity, assessed by testing 20 normal plasma DNA samples showed only 1 false positive droplet with the multiplex assay. No mutations were found in 27 patient plasma samples with negative BRAF tumor tissue testing (100% clinical specificity). Out of 20 patients with positive BRAF tumor tissue testing for V600E or K mutations, 9 patients were found to have the identical mutation in plasma (47% clinical sensitivity) whereas 11 patients were found to have wildtype BRAF in plasma. This could be attributed to delayed time between biopsy and blood sampling or low overall tumor burden.Conclusions: We developed a quantitative multiplex BRAF V600E/K assay for detection of circulating BRAF mutations. It is a fast, reliable and accurate method with a total analytical time of approximately 3 hours. Our findings indicate that monitoring melanoma patients' plasma for BRAF V600E/K with the BioRad ddPCR system could be an effective strategy for the detection of BRAF mutations and could allow early detection of disease recurrence which might obviate the need for BRAF analysis of tissue specimens. Additionally, this testing might play a role in assessing patient response to drug treatment. Introduction: As multiple next-generation sequencing (NGS) platforms become part of mainstream clinical testing, regulatory agencies require monitoring of various quality metrics. One important metric is the reproducibility of variant allele frequencies detected in control samples over time. We hereby describe our experience with inter-run and inter-platform control variant frequency as a quality metric. Methods: We performed retrospective analysis of sequencing data using the Ion Ampliseq Cancer Hotspot Panel V2 (Life Technologies) obtained from 6 different control samples over the past 2 years. Each control was sequenced on a MiSeq (Illumina, San Diego, CA) as the primary analysis method and on the Ion Torrent PGM (ThermoFisher Scientific, Waltham, MA) as the confirmatory method. The mutant allele frequencies from all expected variants from each control were compared run to run and between the MiSeq and PGM instruments. Result: Variant frequencies across the six different controls showed excellent correlation between the MiSeq and Ion Torrent sequencers with an R² value of 0.98. When comparing the run to run variant frequencies of the 104 variants across all controls, the MiSeq had an average standard deviation of 0.9% and coefficient of variation (CV) of 3.2%, whereas the PGM had an average standard deviation of 1.7% and CV of 7.0%. HRAS:c.81T>C and TP53:c.215C>G showed the highest variability with an average standard deviation of 1.1% and 0.8% respectively for the MiSeq, and 4.1% and 4.6% respectively for the Ion Torrent. This difference is likely due to the variants being in a homopolymer region and the nature of the Ion Torrent sequencer has more difficulty distinguishing individual bases in these regions. Conclusion: Overall, the MiSeq produces consistent variant frequencies that were slightly better than the Ion Torrent PGM over the 2 year period. In addition, there was excellent correlation of the variant frequencies between the MiSeq and PGM. These results suggest that these sequencing instruments produce reproducible and robust variant allele frequency calls across many different runs over an extended period of time. M. Vaglio Tessitore, A. Sottini, F. Serana, C. Ghidini, A. Sacco, L. Imberti ASST Spedali Civili di Brescia, Brescia, Italy. Introduction: Dried blood (DB) samples spotted on filter paper are a convenient alternative to fresh blood specimens when collection and shipment of biological materials are difficult or used for large surveys. However, the low volume of total DB could be a limiting factor in quantitative assays especially when a large representation of rare targets (i.e., a few copies/μL) is required like, for example, for The Journal of Molecular Diagnostics ■ jmd.amjpathol.org TRECs and KRECs quantification. The aim of this study was to validate if FLOQSwabs (COPAN, Brescia, Italy) can be used as an alternative for storing larger dry blood volumes for an adequate DNA recovery. Methods: Healthy donor blood samples, collected into EDTA tubes, were absorbed for 10 to 60 seconds on FLOQSwabs. Swabs were immediately placed back into their plastic tubes if equipped with active drying system or, alternatively, left to dry for 2 hours or overnight, and then stored back into their tubes. Blood was also placed on COPAN NUCLEIC-CARDS TM or directly subjected to DNA extraction. DNA from FLOQSwabs was extracted using an in-house protocol (cell lysis with ATL buffer, protein digestion overnight with proteinase K at 56°C, protein precipitation with ammonium acetate 7.5 M; DNA precipitated with isopropanol and resuspended in TE pH8). DNA from 300 μL of blood and from six 2-mm punch discs of DB (corresponding to 8.4 μL) was extracted using the Gentra blood kit. DNA concentration was estimated with the Nanodrop spectrophotometer. Percentage of recovery was obtained after calculating the expected DNA using the white blood cell count and an average DNA content of 6.6 pg/cell. Data were analyzed by ANOVA. Results: FLOQSwabs immersed in blood achieved full absorption and reached a plateau in 10 s. The average volume of absorbed blood, which was indirectly estimated by quantifying tube weight prior swab introduction and after to to 132 ng to -CARDS percentage of recovery (i.e., proportion of extracted over expected DNA) did not significantly differ in FLOQSwabs 69% (95% CI: 60% to 78%) compared to cards 74% (62% to 85%) and blood 60% (49% to 70%, p=ns). Intra-and inter-assay CV for swabs, calculated in 26 replicates, were 13.1% (95% CI: 9.9% to 19.4%) and 13.6% (5.7% to 36.7%), respectively. Conclusions: DB collected and stored on FLOQSwabs is a highly reliable and inexpensive source of DNA. This can be a valid alternative to cards when more DNA is needed to detect rare targets together with a practical solution for shipping and storing blood samples for TRECs and KRECs analysis. Introduction: Detecting mutations is becoming important for both predicting disease progression and drug responses to treatment of cancer patients. Current mutation detection methods for cancer diagnosis are mainly based on the invasive sampling technique such as a tissue biopsy, but some patients may not available for this invasive procedure. Therefore, circulating tumor DNA (ctDNA) would be a good alternative for those patients. However, testing methods for tissue biopsy sample are not applicable to ctDNA samples due to relatively lower sensitivity. A highly sensitive assay method is required for detecting mutations in liquid biopsy samples. Methods: We have developed a highly sensitive and simple method to detect somatic mutation from ctDNA in patient's plasma. This new real-time PCR-based testing method (PANAMutyper) has maximized unique properties of peptide nucleic acid (PNA). It contains a PNA clamp and PNA detection probes in the each reaction tubes. An optimized PNA clamp can tightly bind to only wild-type DNA sequences, and then suppress amplification during the PCR reaction. Meanwhile, a PNA detection probe that conjugated with a fluorescent dye and a quencher, can detect a specific target mutant-type DNA and each mutation can be genotyped by melting peak analysis. PANAMutyper is able to detect 29 different mutations in exon 2, 3 and 4 of KRAS gene with detection limits as low as 0.01%. Results: KRAS G12D, G12V, and G13D mutations are the most abundant mutations in human colorectal cancers. These mutations were tested to demonstrate performance of our new testing method. Each standard DNA was mixed with gDNA from HeLa cells (which have the wild-type KRAS gene) at the decreasing ratios, 1:1, 1:10, 1:10 2 , 1:10 3 , 1:10 4 , 2:10 4 , and 1:10 5 , respectively. The results showed 0.005% sensitivity in G12D, G12V and 0.01% sensitivity in G13D mutations. Also, we conducted spiking test using G12D and G12V standard DNA whether it is valid in real condition. Each DNA mixed with human plasma to 0.5, 5, 50 copies per microliter concentration. And then we extracted ctDNA from prepared sample. As a result, all of standard DNA was detected at 0.5 copy/ . Theses result suggest that our new method is highsensitive and applicable to ctDNA derived from patient's plasma sample. Conclusions: Therefore, PANAMutyper can be used in various clinical areas including personalized medicine and monitoring acquired mutations. J.D. Haimes, J. Covino, N. Manoj, E. Baravik, L. Johnson, L. Griffin, J. Stahl, B.P. Culver, B. Kudlow ArcherDX, Inc., Boulder, CO. Introduction: Copy number variants (CNVs) are oncogenic drivers, and impact more of the cancer genome than all other types of mutations combined. Genomic analysis using next-generation sequencing (NGS) has shown promise as a method to detect CNVs from clinical sample types. However, archival storage of formalin-fixed, paraffin-embedded (FFPE) specimens severely damages DNA by inducing proteinnucleic acid crosslinking, base modifications and strand cleavage, which results in poor sequencing coverage and limited CNV detection sensitivity. Therefore, NGSbased detection of low-level CNVs (<3-fold) and CNVs in samples with low tumor cellularity remains challenging. Anchored Multiplex PCR (AMP) is a target enrichment strategy for NGS that enables digital read counting in low-quality samples, resulting in enhanced CNV detection sensitivity. Methods: Based on AMP chemistry, we developed Archer VariantPlex targeted DNA enrichment assays for NGS to detect both low-and high-level CNVs from FFPE specimens and other lowinput clinical sample types. We used the VariantPlex Solid Tumor kit, which contains AMP primers designed to detect CNVs for 43 cancer-associated genes by NGS. To rapidly screen samples prior to library preparation and sequencing, we also developed the Archer PreSeq DNA QC Assay to determine the integrity of genomic DNA. Results: We examined over 150 FFPE tumor samples for genomic DNA integrity and CNV detection using the PreSeq and VariantPlex assays for targeted NGS. Our data show that NGS-based detection sensitivity is driven primarily by the integrity of the input genomic DNA, determined by the PreSeq Assay, which further predicts the limit of CNV detection. Using optimal input amounts of genomic DNA, we detected CNVs as low as 2-fold in FFPE samples and in samples with as low as 3% tumor cellularity with the VariantPlex Solid Tumor kit. Conclusions: These results demonstrate that Archer VariantPlex assays enable sensitive NGS-based detection of low-level CNVs from low-input clinical samples and in samples with low tumor cellularity. Furthermore, the PreSeq DNA QC Assay allows for the adjustment of DNA input to provide adequate genomic copies into the NGS workflow and maximize CNV detection sensitivity. Introduction: Short Tandem Repeat assay is a major method for monitoring chimerism status after bone marrow transplantation. We prepared a limit of detection (LOD) control by mixing 5% genomic DNA of one male with 95% of that from one female subject. LOD is tested in each run along with patient samples. 262 runs from past 2 years were analyzed with our laboratory developed software. Data indicates the difference among the markers is much greater than among runs. Result suggests an improvement in the donor percentage calculation algorithm is needed. Methods: PowerPlex 16 HS System (Promega). Genomic DNA from one male and one female subject were mixed by different ratio (2% to 98% of male DNA). LOD control is made by mixing 5% male as donor and 95% female as recipient, and used in every run. Laboratory developed software used to calculate percentage of donor. Results: 1) Makers tested: 16 markers are tested by the reagent kit, 14 of them are informative for the DNA mixture (D3S1358, D21S11, D18S51, PentaE, D5S818, D7S820, D16S539, CSF1PO, PentaD, AMEL, vWA, D8S1179, TPOX, FGA), 2 are noninformative (TH01, D13S317). 2) Reliability of the assay: a) Linearity test of mixed DNA samples (2% to 98%, 9 test points) shows good linearity of the test. Observed value vs. expected value is x=1.0024y, R 2 = 0.9952. b) Five % mixed DNA is used as LOD and 262 runs were conducted since June 17, 2014. 100% of the tests show positive result close to 5% donor. Average percent calculated with all informative markers (14) is 6.89% (stdev = 1.30%, Max = 10.93%, Min = 3.19%). 3) Variation is significant between markers: Single factor ANOVA shows the variation among markers is much greater than among different runs (F = 44.039, F critical = 1.729, p= 1.7E-105). 4) Markers performed differently: Some markers show larger variant range around median than others. Conclusions: 1) PowerPlex 16 HS System shows good linearity of percentage value in the reportable range. 2) Average percent of donor of LOD control sample is close to the expected value. 3) Variation among markers is much greater than among runs, suggests the variation may be due to the nature of each marker, rather than effects between runs. Some markers are more vulnerable to effects between runs than others. 4) Improvement of the algorithm for calculating percentage is needed and in process, which will utilize the fact that each marker is amplified differently by adding a weight to each marker and set individual control matrix range for each marker. I.S. Haque, C. Haverty, J.D. Goldberg, E.A. Evans Counsyl, South San Francisco, CA. Introduction: Although non-invasive prenatal screening (NIPS) for fetal aneuploidy has high sensitivity and specificity, prevalence varies significantly by maternal and gestational age. Variable prevalence affects the probability that a positive test indicates an affected fetus (positive predictive value, PPV). Although ACOG and SMFM direct laboratories to report PPV individualized to the particular patient, no previous work has addressed how uncertain PPV calculations are or to what precision PPV can be estimated. Methods: We introduce a statistical framework to estimate the confidence interval (CI) around NIPS PPV. We estimate the uncertainty in NIPS analytical performance by fitting beta distributions to the 95% CIs on test sensitivity (SENS) and specificity (SPEC) from large meta-analyses (Taylor-Phillips S, et al. 2016) . We sample the posterior distribution of aneuploidy prevalence by using original population data (Hecht CA and Hook HB. 1994; Snijders RJM, et al. 1999; Morris JK and Savva GM. 2008; Savva GM, et al. 2010) to sample from beta distributions or bootstrap Kaplan-Meier curves as appropriate. We estimate the portion of the CI arising from each source by holding individual parameters fixed. With this method, we evaluate the 95% CI of NIPS PPV for trisomies 13, 18, and 21 (T13/18/21) . Sampling all parameters shows that CI breadth mostly varies by maternal age (MA): for example, the CI width for T21 is +/-11% at MA=20yr down to +/-4% at MA=40yr. Low mean PPV for T13 and T18 leads to asymmetric CIs: the T13 PPV CI at MA=20yr and gestational age (GA) 12 weeks has a lower bound 13% below the mean of 20.8% but an upper bound 23% above the mean. All CIs are dominated by SPEC uncertainty: at MA=20yr, the T21 CI is +/-10.5% when sampling only over SPEC, +/-4% using only prevalence, and +/-0.1% sampling only SENS. We repeated the analysis using SENS and SPEC from a smaller (N=3,322) study (Porreco RP, et al. 2014) . As expected, at MA=20yr GA=12wk the mean T21 PPVs using parameters from the meta-analysis or the smaller study are similar, but the CI using the meta-analysis is much tighter (+/-12.6% vs +/-32.9%). Conclusions: Our results show that larger meta-analyses of NIPS specificity could narrow the PPV confidence interval two-fold for T21 or fourfold for T13 and T18 (motivating outcome data sharing among laboratories), but that additional gestational prevalence data will then be needed to further improve the precision of PPV estimates. The utility of providing PPV is well-justified even with the estimated CI. This work further demonstrates the need to use large meta-analyses to establish sensitivity and specificity rather than smaller individual studies. A. Robertson, X. Wang, G.M. Gould, J.R. Maguire, H. Kang, I. Haque, E.A. Evans Counsyl, South San Francisco, CA. Introduction: Historically, variant genotyping tests have been validated by concordance with known reference materials. However, for NGS-panel tests detecting any variant in the genes of interest, this strategy is not effective as positive controls for variants that are difficult to detect may not be readily available. Although large numbers of SNPs in genes of interest are available from reference materials, indels (especially large indels) and CNVs are quite sparse or unavailable for most clinically relevant targets. Thus to comprehensively assess the performance of an entire region of interest, in silico methods must be used to supplement reference material comparisons. We present two novel in silico techniques for validating the capability to call variants: 1) simulations to assess the ability to detect copy number variants (CNVs) and 2) a genomic reference editor to assess the ability to detect large or complex indels. Methods: The first technique, simulating CNVs, is designed to simulate realistic data with a particular CNV using a model based on the experimental performance of the assay. Then the simulated data can be analyzed via a bioinformatic pipeline to assess assay performance. The second technique, reference editing, takes advantage of the fact that sequence variants are expressed as a deviation from a reference genome. One makes a modification to the sequence of the reference genome used for variant calling, runs a negative control sample through the full assay and expects to call the inverse of the modification made to the reference. For example, it is possible to insert a 40bp sequence into the reference and re-run the bioinformatics pipeline on a negative sample to determine if the pipeline can detect a simulated 40bp deletion. Results: Using reference editing, we find that off-the-shelf NGS analysis toolchains typically fail to detect large (50bp to 150bp) deletions, and designed a specific variant caller designed to address this gap. As a demonstration, we applied both reference editing and CNV simulation to the development of a NGS assay (Counsyl Inherited Cancer Screen). In particular, CNV simulation led to a refinement of the initial assay design from the baseline design suggested by sequencing reference biological samples. This allowed us to maintain >99.7% CNV sensitivity while reducing our sample retest rate and improving our sensitivity in identified noisy regions. Conclusions: In cases where informative biological samples are not always available, these tools are valuable for validating and improving the quality of a clinical NGS pipeline to provide accurate results to patients and better inform medical management. . Each genotype-specific PNA probe, which is conjugated with a fluorescent dye and a quencher, is used as a reporter in a real-time PCR reaction. PNA probe has the length shorter than the DNA probe and Tm for each single mismatch is a big difference. Due to the above features, PNA probe is useful for multiplex-PCR to detect the target with many variant genes. Results: Test was performed to obtain DNA from 58 samples cultured in a MacConkey agar plate. After a comparison with the culture results, showed a concordance of 93.1% in 54 samples of 58 samples match. For mismatch four samples confirmed through sequencing PCR, the result of the PANA RealTyper CRE kit was correct. In addition, this was also found to be capable of detecting strain obtained from trypticase soy broth (TSB) cultured sample. Conclusion: PANA RealTyper CRE kit is easily and -lactamase gene, and it is expected to help in the inhospital infection control. B. Wu, A. McCarthy, E. Craig, W. Tahaney, A. Edmunds, S. Brahmasandra NeuMoDx Molecular Inc., Ann Arbor, MI. Introduction: Widespread use of nucleic acid based testing (NAT) for diagnostics has resulted in the routine use of PCR in molecular diagnostic (MDx) labs. Reagents for such PCR assays have typically required the need for cold or frozen storage to maintain long term viability. The absolute need for cold temperature storage offers substantial challenges to the implementation of NAT in resource limited settings as well as complicates the development of next-generation MDx instrumentation with onboard reagents storage. Development of ambient stable NAT reagents is expected to significantly alleviate these traditional drawbacks. Traditionally, lyophilization or freeze-drying has been the method of choice for producing room temperature stable active reagents such as Taq polymerase. However, lyophilization requires expensive specialized equipment & reagents, is time intensive, and is practically impossible to implement in a continuous mode operation. NeuMoDx has developed a simple, cost effective, high-throughput, and rapid drying procedure capable of producing highly effective NAT reagents using a conventional conveyer oven. Methods: Three types of molecular diagnostic reagents were dried using unique formulations of select excipients: 1) Extraction reagents including magnetic particles with a nucleic acid affinity matrix, lytic enzymes, and encapsulated DNA/RNA process controls. 2) PCR reagents, and 3) RT-PCR reagents. Magnetic particle based DNA/RNA isolation was performed in a variety of clinical matrices using these ambient stable extraction reagents. At the end of the isolation step, the eluted nucleic acids were used directly to reconstitute dried PCR or RT-PCR mix and qPCR was performed. The efficacy of these dried reagents were evaluated by assessing critical analytical performance metrics such as limit of detection, linearity etc. Accelerated aging studies were also performed with all three dried reagent formulations to assess the long term stability of these reagents. Results: The NeuMoDx ambient temperature stable reagents performed equivalently or better than the wet controls based on PCR Ct values as well as increase in fluorescent intensity. Although multiple excipient formulations were found to provide Ct values comparable to wet controls, only select combinations of the excipients were capable of producing sigmoidal amplification curves with fluorescence increase comparable to the wet controls. Introduction: Formalin-fixed, paraffin-embedded (FFPE) tissue samples are valuable resources for retrospective and prospective molecular analysis in both the clinical and research settings. Extraction of nucleic acid from FFPE samples is a critical initial step in any molecular analysis; however, many commonly used extraction methods produce inconsistent results in both quantity and quality. There are two major types of nucleic acid extraction methods, column-based and magnetic bead-based. Magnetic bead-based purification is more amenable to automation and high-throughput processes. As part of overall lab automation efforts, we decided to explore other commercial kits for high-throughput sample processing, and compared to our previous column-based extraction kit. Our laboratory assessed several methods of extraction. The following characteristics of each kit and method were evaluated: 1) availability of automated protocol, or ease of development of such a protocol; 2) ability to simultaneously extract DNA and RNA in separate eluates within a single prep; 3) flexible range for sample input amount requirement; 4) minimal upfront manual preparation time; 5) reproducibility. Methods: Three extraction kits were assessed: Thermo Fisher Scientific MagMAX FFPE DNA/RNA Ultra kit, Qiagen AllPrep FFPE DNA/RNA, and Covaris truXTRAC FFPE DNA/RNA. Other commercially available extraction kits on the market were not included in the assessment because the end-product was total nucleic acid which required the addition of DNase or RNase to obtain separate analytes. We assessed each kit's performance with the requirement listed previously. FFPE samples were tested in duplicate or triplicate with Qiagen AllPrep FFPE DNA/RNA and the MagMAX FFPE DNA/RNA Ultra kit. Results: Although Covaris truXTRAC FFPE DNA/RNA gave sufficient yield, the upfront manual preparation time was longer than the other kits, making this protocol unsuitable for high-throughput sample processing. Inter-and intra-technician testing was performed and results were compared. Extracted DNA and RNA were assessed on the Nanodrop (UV spectrophotometry), Qubit (fluorescence-based quantitation), QuantStudio (qPCR), and Bioanalyzer (capillary electrophoresis). Conclusions: Both Qiagen and MagMAX produced comparable and acceptable quality metrics across all tests for both DNA and RNA. However, for high-throughput sample processing, the MagMAX FFPE DNA/RNA Ultra kit's protocol has met all of our necessary requirements and can be integrated with liquid handling for automation. H. Leong, E. Schreiber, S. Berosik, S. Chen, W. George, A. Gerstner, J. Marks, S. Schneider Thermo Fisher Scientific, South San Francisco, CA. Introduction: Detecting minor genetic variants has become essential to cancer and infectious disease management. Many have turned to next-generation sequencing to fill this need given the common perception that Sanger sequencing can reliably detect genetic variants at allelic proportions no lower than 25%. We discovered a way to reduce this reference-based detection limit to 5% and have demonstrated detection at even lower allele frequencies. Methods: The discovery is a two-part software algorithm. The first part minimizes the noise that underlies Sanger sequencing traces. The second part detects variants, if any, in the noise-minimized traces. For noise minimization, a model of the noise is made from reference traces and subtracted from the traces of the unknowns. For variant detection, information from the noise-minimized traces of the forward and reverse sequencing reactions is combined to distinguish spurious peaks from those truly associated with variants. The performance of the algorithm was measured on samples from 22 amplicons involving eight different genes: TP53, KRAS, BRAF, EGFR, FLT3, RB1, CDH1, and ERBB2. Many of these were extracted from formalin-fixed, paraffinembedded samples. Some were commercially available reference standards, others were quantified using the RNase-P quantitative polymerase chain reaction assay and serially diluted. Allelic proportions spanned 0.6125% to 50%. These samples were amplified, sequenced, and pre-processed using standard protocols and tools for fluorescent dye terminator Sanger sequencing from Applied Biosystems. Results: A sensitivity of 95.9% and specificity of 99.8% was observed for data at 5% allele frequency; there were 785 variants and 229623 non-variant base positions spanning 704 quadruplets (forward and reverse for both reference and unknown samples). A sensitivity of 98.8% and specificity of 99.8% was observed for data at 10% allele frequency; there were 503 variants and 163037 non-variant base positions spanning 454 quadruplets. Although we have been able to detect variants at allele frequencies 0.6125%, 1%, 1.25%, 2%, and 2.5%, the algorithm did not reach 95% sensitivity and 99% specificity at these levels. Conclusions: It should now be possible to achieve a reference-based limit of detection of 5% allelic proportion with standard Sanger sequencing protocols. Existing protocols for visually reviewing the results can also be used and are enhanced because the algorithm generates results in the form of familiar electropherograms for which the noise has been substantially diminished. These two features of the algorithm may give Sanger sequencing performance and/or economic advantages in some molecular diagnostic applications that require finding minor genetic variants. However, sensitivity and specificity of these separating technologies remain suboptimal, which hinder their widespread application to the clinical practice. In this study, alternating current (AC) potential modulated microfluidic strategy coupled with amperometry was developed to separate and detect small numbers of CTC from patient blood, through the decoration of the CTC surface with an anticancer drug molecule within a short period of time (<400 seconds). Methods: Blood samples were collected from various cancer patients and were blind-tested using the proposed separation technique. The separated cells were collected from the channel and observed under the fluorescent microscope to confirm the presence of cancer cells in the sample. The carbon channel walls were modified with an organic conducting polymer followed by chemical bonding of lipid molecules which provide an affinity towards drug molecules adsorbed on the CTC and allow them to move much slower than the other cells in the microfluidic channel. The separation process was augmented with AC potential and frequency modulations and the overall size of the CTC as compared to normal blood cells. Results: The blood samples with cancer cells showed a peak corresponding to the oxidation of daunomycin (DNM) molecule. Among the patient blood samples 9 out of 15 samples showed electrocatalytic responses for the oxidation of DNM at different retention times, indicating the presence of cancer cells. These collected samples showed the presence of cells that were rather large in size as compared to RBC. When compared to the results given by morphological and molecular diagnoses, we could observe that 86.6 % agreeable results were obtained. Conclusions: Current tumor diagnosis depends on a variety of pathological tests, among which tissue biopsy is considered to be the standard diagnostic procedure. With the recent development of various devices to enrich and detect CTC, the noninvasive screening for cancer diagnosis can assist in early stage detection. The potential of these new technical platforms to improve CTC detection related to the treatment stratification warrants further study. Introduction: Buccal swab has been a convenient collection device for genetic tests. Current practice involves a dry swab that collects DNA from the cells on the inside of a person's cheek. Due to different swabbing techniques of individual patients as well as potential interfering substances, a good extraction system is crucial to harvest DNA of high yield and quality to acquire accurate results and to reduce requeue rate. Roche MagNA Pure 96 System is a high throughput instrument that performs automated nucleic acid purification. The purpose of this study is to evaluate the performance of automated DNA extraction from buccal swabs using Roche MagNA Pure 96 System compared to MagMAX DNA Multi-Sample Ultra Kit. Methods: For Epicentre Catch-All Sample Collection Swabs, 60 swabs were collected with 2 swabs per person. One set of swabs were extracted on MagNa Pure 96 System with DNA and Viral NA Small Volume Kit. The other set of 30 swabs from the same person were extracted with MagMAX DNA Multi-Sample Ultra Kit. DNA yield of all extracted DNA were tested by Qubit dsDNA HS Assay Kit. Copy number variation (CNV) of CYP2D6 and single nucleotide polymorphism (SNP) genotyping assays of 17 genes were performed on all 60 samples and the results were compared. Results: Average DNA yield acquired from MagNA Pure 96 System was 18.01±22.80ng/ul, higher than the average yield of 5.91±6.53ng/ul from the swab collected from the same person extracted using the MagMAX kit. The yield from MagNA Pure 96 varies from 1.86ng/ul to 94.4 ng/ul, and the yield from MagMAX kit ranges from 1.4ng/ul to 29.8ng/ul. This variance is largely due to the inconsistent swabbing technique between individuals. Both CNV assay and 47 genotyping assays of 17 genes achieved 100% agreement using DNA from both extraction kits. Conclusions: Roche MagNA Pure 96 is a reliable system to perform efficient DNA extraction from buccal swab specimens. It recovers DNA with a higher yield and of good quality suitable for genetic tests. Introduction: Detecting mutations is becoming important for both predicting disease progression and drug responses to treatment of cancer patients. Current mutation detection methods for cancer diagnosis are mainly based on the invasive sampling technique such as a tissue biopsy, but some patients may not available for this invasive procedure. Therefore, circulating tumor DNA (ctDNA) would be a good alternative for those patients. However, testing methods for tissue biopsy sample are not applicable to ctDNA samples due to relatively lower sensitivity. A highly sensitive assay method is required for detecting mutations in liquid biopsy samples. Methods: We have developed a highly sensitive and simple method to detect somatic mutation from ctDNA in patient's plasma. This new real-time PCR-based testing method (PANAMutyper) has maximized unique properties of peptide nucleic acid (PNA). It contains a PNA clamp and PNA detection probes in the each reaction tubes. An optimized PNA clamp can tightly bind to only wild-type DNA sequences, and then suppress amplification during the PCR reaction. Meanwhile, a PNA detection probe that conjugated with a fluorescent dye and a quencher, can detect a specific target mutant-type DNA and each mutation can be genotyped by melting peak analysis. PANAMutyper is able to detect 47 different mutations in exon 18, 19, 20 and 21 of EGFR gene with detection limits as low as 0.01%. Results: A total of 83 plasma samples from non-small cell lung cancer (NSCLC) patients were tested to demonstrate clinical efficacy of this new testing method. The results showed mutanttype DNAs in 21 patients and wild-type DNA in 62 patient's plasma samples. These results were compared with direct sequencing data from matched tissue samples. Introduction: A substantial degree of information exists in the cancer-related literature, but systematically extracting this data can be challenging. The ability to rapidly query connections between molecular aberrations, relevant therapies, cancer indications, and available clinical trials requires a comprehensive database comprising a succinct view of the extant data, but in an easily retrievable format. Methods: We have built the JAX Clinical Knowledgebase (CKB), a manually curated database that houses data related to molecular variants, relevant therapeutic drugs, cancer indications, and clinical trials, as well as an application that allows one to methodically search for links between data elements. To demonstrate utility of CKB and provide a snapshot of the data available in CKB, a side-by-side analysis was performed between actionable genes in CKB and those genes in cBioPortal. All genes in CKB with actionable variants were queried in cBioPortal and then analyzed for comparison. The top 15 genes in CKB with the most actionable variants were compared to the top 15 genes in cBioPortal with the highest mutational frequencies. Additional analyses within CKB were executed to illustrate the landscape of available treatment approaches for the top actionable genes. Results: The comparison among the 15 genes with the greatest number of actionable variants in CKB and the top 15 genes in cBioPortal with the highest mutational frequencies demonstrated a concordance of 73% (11/15). In CKB, PTEN and APC comprise the two genes with the largest number of actionable variants while in cBioPortal, TP53 and PIK3CA have the greatest mutational frequencies. The minimal difference between the two databases is primarily due to the frequency at which specific codons in TP53 and PIK3CA are mutated. Within the concordant gene list, PIK3CA demonstrates the highest number of relevant therapeutic drugs, and EGFR represents the gene with the highest number of relevant FDA-approved therapies. Conclusions: The JAX CKB is an inclusive database allowing one to retrieve pertinent information related to specific cancer genes, in a format that can be easily navigated to quickly identify utility. The retrieval analysis performed in this study demonstrates CKB's significant benefit to both the medical and scientific research communities. Y. Cheng, M. Jakubowski Cleveland Clinic, Cleveland, OH. Introduction: Lung cancer is the leading cause of cancer death in both man and women. The identification of critical gene mutations allows better decision making of targeted therapies in managing NSCLC patients. Real-time PCR and nextgeneration sequencing (NGS) have become two popular methods in detecting therapeutic actionable sequence changes. In this study, we explored the accuracy and specificity among NGS platforms and real-time PCR in detecting NSCLC mutations in a clinical laboratory. Methods: Genomic DNA was extracted from NSCLC formalin fixed paraffin embedded tissues and fine needle aspirate specimens collected in PreservCyt or CytoLyt. Targeted genomic regions in 50 oncogenes and tumor suppressor genes were multiplex amplified using Ion AmpliSeq Cancer Hotspot Panel v2 and the Ion AmpliSeq Library Kit 2.0 (ThermoFisher Scientific, Waltham, MA). Selected amplified samples were ligated with Illumina adaptors and sequenced on MiSeq. The resulted fastq files were aligned and analyzed using the NextGENe (SoftGenetics, State College, PA) bioinformatics tool. Only those variants, excluding common single nucleotide polymorphisms, residing in defined mutation hotspots with over 5% allele frequency were examined and scored. The EGFR and KRAS mutation status in the majority of these tumor samples were also determined using real-time PCR and Sanger sequencing approach. Results: Intra-and inter-assay reproducibility were determined by analyzing three specimens that possessed missense and insertion/deletion variants. These specimens were performed in triplicate in the same run or on three separate days, respectively. Inter-tech repeatability was determined by analyzing three specimens that contained missense variants by three separate technologists. All repeatability runs were 100% concordant. Additionally, among 70 previously analyzed specimens, 97% concordance was achieved. The clinical sensitivity and specificity is 97% and 100%, respectively. A multiplex DNA standard (Horizon Dx 200) containing 11 mutations at known allele concentrations, ranging from 1 % to 24.5% allele frequencies was tested six times. Six representative variants with stated allele proportions ranging from 3-25% were consistently detected. Conclusions: In general, the majority of gene mutations were consistently identified among various testing platforms being used. Our data indicate that a small discordance of testing results may be contributed from low mutation allelic frequencies and complex mutation situations. Examples of these cases and the cost effectiveness of applicable diagnostic assays are discussed. Introduction: It is difficult to identify causal agent of respiratory viral infections because of similarity of clinical signs and symptoms. Accurate identification of virus pathogen helps physician with presuming infection route and selecting effective treatment. Recently commercialized DNA probe-based multiplex real-time PCR methods were increased the accuracy of virus identification. However, this DNA probe-based technology required relatively longer probe sequence for target binding and it has its limitation that cannot distinguish sequence variations on target virus genes. Methods: A new peptide nucleic acid (PNA)-assisted melting curve analysis technique is developed. Each genotype-specific PNA probe, which is conjugated with a fluorescent dye and a quencher, is used as a reporter in a real-time PCR reaction. A PNA probe can design relatively shorter binding sequence than its DNA probe, so PNA probe can avoid sequence variation position on a gene. Furthermore, PNA probe showed bigger melting temperature difference than DNA probe when reporter probe bound to a single mismatch target. So, sequence variants on a target gene are easily distinguishable using melting curve analysis. Therefore, PNA-based reporter probe is very useful for multiplex detection in real-time PCR platform to identify a target gene with many sequence variants. Results: We have developed a simple and rapid method for detecting specific virus genes within three hours. This real-time PCR-based PNA probe-assisted fluorescent melting curve analysis assay (PANA RealTyper RV kit] can detect a total of 25 different virus genotypes (Influenza A, B, PIV1,2,3,4, RSV A, RSV B, AD, CoV-229E, CoV-OC43, CoV-HKU1, CoV-NL63, CoV-MERS, MPV, BoV, PAN-CoV, ENV, RHI). A total of 70 nasopharyngeal flocked swabs samples were tested to demonstrate clinical efficacy of this new assay method. Then, a test result of PNA-based method was compared to a result of DNAbased real-time PCR test that were performed independently by the third party. The results showed 87.1% concordance rate between two test methods (61 out of 70 samples). Meanwhile, nine discordant samples were confirmed by direct sequencing method, and results of sequencing for nine samples were agreed with the results of new PNA-based real-time PCR method. Conclusions: PANA RealTyper RV kit is simple and reliable test method for detecting pandemic respiratory virus. This new method offers a cost-effective and highly sensitive genotyping method that can be used in various clinical areas. GmbH, Hilden, Germany; 3 Medical University Graz, Austria. Introduction: One major goal of cancer clinical research is to establish new prognostic and predictive biomarkers for personalized diagnosis and therapy. Next-Generation Sequencing (NGS) has revolutionized the field of cancer genetic studies by dramatically decreasing costs and time needed for large-scale data generation and analysis. Although NGS has proven to be a very useful tool in research, major hurdles remain for its broad adoption in the clinical research setting: lack of a seamless workflow, ability to handle samples from different sources or pre-analytical treatments, and lack of actionable content to guide the interpretation of results. Here, we apply a complete sample-to-insight NGS workflow, QIAGEN´s GeneReader System, to analyze a diverse set of clinical samples. Methods: Tissue samples from different tumor types (colon or melanoma) were subject to fixation by formalin or the PAXgene system and embedded in paraffin. Donor-matched plasma samples were also analyzed when available. Sample DNA was processed and analyzed accordingly to the GeneReader Workflow: DNA was extracted using the GeneRead FFPE kit and the PreAnalytix PFPE Kit for PAXgene fixed tissue, respectively. Plasma DNA was isolated with the QIAamp Circulating Nucleic Acid Kit. The quality was multiply assessed with the QIAxpert, the QIAxcel and the QuantiMIZE Kit. Targeted amplification of cancer hotspots was performed using the Actionable Insights Tumor Panel. After sequencing with the GeneReader, variants were identified using the QCI-Analyze software and interpreted with the QCI-Interpret software. Results: Despite different preanalytical treatments, all samples showed high quality scores in the QC steps. This was followed by successful amplification of target regions and all the subsequent workflow. For all samples (n=20) more than 99% of the sequence reads were mapped to the human genome (hg19) and > 90% to the regions of interest (ROI). The majority of the samples (19 out of 20) showed more than 95% of the ROI covered with at least 500 single reads. All variants were correctly identified and interpreted, showing the robustness of the GeneReader System and its flexibility and reliability handling different sample types and sources. Conclusions: In clinical research, the development of new prognostic and predictive biomarkers is often dependent on samples from biobanks and prospective cohorts which often suffer from suboptimal conditions such as long storage period and various treatment methods. The data presented show the advantages of a fully integrated sample-to-insight workflow including initial quality control steps and prove the versatility of the GeneReader System for different types of sample materials. For research use only. Not for use in diagnostic procedures. Introduction: Cell-free DNA (cfDNA) controls for next-generation sequencing (NGS) are often derived from patient blood sample remnants. With the expansion of plasma DNA diagnostics, sufficient volumes of patient-derived materials that contain a wide number of target analytes for oncology and non-invasive prenatal screening (NIPS) assays are simply not available to make controls. Therefore we have formulated suitable stabilized biomimetic materials derived from fragmented genomic DNA (gDNA) and synthetic constructs. Stabilization is essential because fragmented DNA is less stable in blood and blood-like matrices than native cfDNA. Several different biomimetic formulations for NIPS aneuploidy controls (Trisomy 13, 18 and 21) were tested for stability by NGS at two different temperatures over nine months. Methods: Encapsulated-stabilized DNA formulations for NIPS were prepared by mixing normal gDNA with aneuploidy gDNA. The mixture was verified by dPCR for copy number and then sheared using Covaris ultrasonic shearing before incorporation into a stabilizing biomimetic. Final samples are blended into modified MatriBase, a synthetic plasma. Samples were stored at 4°C or 42°C for 1, 4, 6 and 9 months. Testing of the samples was performed using a Verinata-derived NGS assay. Results: Stressed stability studies were conducted comparing encapsulated DNA and free DNA in modified MatriBase at 4 °C and 42 °C. The encapsulated DNA samples, using Trisomy 13 reference materials, were found to be unchanged based on NIPS assay performance after 275 days at both temperatures by NGS. Free DNA samples failed library preparation at 47 days at both temperatures. Real-time stability for a Trisomy 21 aneuploidy reference material, a multi-analyte aneuploidy reference material (including Trisomy 13, 18 and 21) and a euploid (an aneuploidy negative control) remained stable for at least six months via NGS. Conclusions: We have demonstrated the ability to combine encapsulate small DNA fragments with synthetic plasma as a general means to prepare biosynthetic reference materials for molecular assays. These materials were found to be viable alternatives to patient samples for NIPS assays. As the biomimetics are readily extractable using commercially available kits, the reference materials can be inserted into existing workflows similar to centrifuged plasma samples. Moreover, the protection conferred by the encapsulation creates a stable, refrigerated reference material which will enable improved quality monitoring of diagnostic assays. Recently, exon-based enrichment strategies have been shown to be effective for not only enrichment of transcripts containing known exons, but also depletion of ribosomal content. This study examines how the Agilent Strand Specific RNA Library Prep kit can be used to generate cDNA libraries from FFPE total RNA. Enrichment was then performed on the total cDNA libraries from FFPE RNA using Agilent's V6+UTR exome DNA Enrichment kit. The effectiveness of this approach was examined by comparing the transcriptional data from these FFPE samples to RNASeq data from FF samples from the same source. Methods: Mouse xenografts derived from human breast tumors were collected. Half the material was flash frozen (FF) in OTC matrix and the other half was formalin fixed, paraffin embedded (FFPE). RNA from both FF and FFPE were extracted with standard procedures. All FF and FFPE RNA was pooled to generate the FF Xenografts (PDX) pool as well as FFPE PDX pool. The FFPE PDX pool was further treated to degrade the RNA so that the average size reflected typical archival FFPE RNA. A total of 200ng of FF RNA was processed for RNASeq according to the Agilent protocol, with poly(A) enrichment followed by cDNA synthesis and library prep. For capture only sample processing, 200ng of FF and FFPE RNA were processed with minor modifications (no ribosomal depletion or poly(A) enrichment, modifications to the fragmentation step as well as extending second strand/end repair and A tailing steps). Final libraries were sequenced on an Illumina NextSeq500 flowcell and resulting data were processed as standard RNASeq data. Results: Performing V6+UTR enrichment only on FFPE total RNA library preps was effective in depleting ribosomes. In addition, the overall alignment statistics were similar between the FFPE V6+UTR samples and FF Poly(A) RNASeq. Exome captured sample was biased toward exonic reads, with little to no coverage of deep intronic regions or intergenic regions. A Pearson's average correlation coefficient of 0.93 between the FF poly-A enriched samples and FFPE V6+UTR enrichment only samples was observed. Additionally, improved uniformity of reads across transcripts from 5' to 3' was observed for the FFPE V6+UTR enrichment only samples. Conclusions: This data indicate that using the Agilent Strand Specific RNA Library Prep kit on total RNA followed by V6+UTR exome enrichment is effective for performing RNASeq analysis on FFPE samples in the absence of a separate ribosomal depletion step. This, along with increased sensitivity to low expressers and rare transcripts and making FFPE RNA samples candidates for transcriptome sequencing, indicate an effective approach to transcriptional profiling or challenging RNA samples. C. Bissaillon, M. Young, L. Bonomi, A. Tyropolis, K. Lebel, F. Moore Baystate Health, Springfield, MA. Introduction: Herpes simplex viruses (HSV), consisting of Types 1 and 2, are ubiquitous and infectious DNA viruses in the family Herpesviridae. Like most herpesviruses, primary infection with HSV can lead to lifelong latency in cells. Sporadically, they can reactivate and cause outbreaks during which infections are spread. Though there is no cure for latent infection, reactivation rates and decrease of symptom severity can result from treatment with antiviral drugs such as acyclovir. The most common HSV-induced lesions are orolabial and genital. Diagnostic tests for HSV can help distinguish HSV infections from many other conditions that cause similar-appearing lesions. It is also important to distinguish Type 1 from Type 2, as the chance of reactivation is much greater with Type 2. The Luminex Aries Two module System (IVD) is a sample to answer system that utilizes real-time PCR performed in cassettes. The cassettes used in this system come preloaded with all the required reagents and internal controls for testing symptomatic patients. Methods: A total of 50 samples were used to test accuracy. A total of 31 residual, de-identified M4 VTM samples were tested as well as 19 Copan UTM samples obtained from Luminex for the purpose of validation. The 50 samples were comprised of 17 HSV 1 positive, 17 HSV 2 positive and 16 negative for both targets. ARIES results obtained from the M4 samples were compared to the Luminex MultiCode RTx HSV 1 & 2 real-time PCR, thermal melt assay. The ARIES results for the Copan UTM were compared to the expected result provided by Luminex. The precision of the assay was evaluated by testing one positive HSV 1 sample and one positive HSV 2 sample twice per day for 10 days, between at least two technologists for a total of 40 measures. Five HSV 1 and 5 HSV 2 samples at 2,000 cp/ml were tested to verify the established limit of detection. Results: Correlation samples show 100% (50/50) agreement with expected results for both the Copan UTM and M4 VTM samples types. Precision results show 100% (40/40) concordance for both HSV1 and HSV 2. The limit of detection was verified as 100% (10/10) samples at 2,000 cp/ml were detected. Conclusion: The Luminex Aries system has been shown to provide accurate and precise results for HSV 1 &2. The instrument is easy to use, has a sample to result process, and requires very little maintenance. Introduction: Studies on circulating cell free DNA require 1) exclusion of contaminant cellular DNA, 2) uniform sampling, 3) prevention of degradation, and 4) protection from external contamination. We compared the performance of a novel collection tube with a barrier separating cell-free from cellular material consisting of a photopolymer solidified by UV light to 1) BD EDTA tubes, 2) BD EDTA Plasma separator tubes containing a soft gel separator, and 3) Streck Cell-Free DNA BCT tubes containing a preservative to minimize cellular degradation. Methods: Peripheral blood was collected from 10 normal subjects in duplicate Plasma Separator and Photogel tubes. Following low speed centrifugation, one tube of each type was kept at -80 o C for 7 days. Cell-free DNA was isolated from samples in the remaining tubes and in the frozen samples using a Promega RSC Maxwell instrument. Amounts of amplifiable DNA recovered were tested by digital PCR and amounts of contaminating genomic DNA were tested on an Agilent TapeStation. Peripheral blood from 10 normal subjects was also collected in triplicates in EDTA, Streck and Photogel tubes, centrifuged, and cell-free DNA was extracted either without further manipulation (aliquot #1) or following a second centrifugation at 16,000 x g with or without subsequent treatment with Proteinase K. Results: Differences in the average amount of amplifiable DNA recovered from fresh and frozen Photogel plasma samples were of borderline significance (0.9+/-0.27 and 1.45+/-0.48 ng/μL respectively (p=0.09). Higher yields of amplifiable DNA obtained with plasma separator tubes and significant differences (P=0.01) between fresh and frozen samples (13.70+/-2.82 versus 34.74+/-4.45 μg/μL, P=.01) were accounted for by contaminating genomic DNA. Photogel tubes provided higher yields than Streck tubes for samples processed by low speed centrifugation only (122.3+/-24.8 pg/μL versus 63.1+/-19.1 pg/μL, P=.004). High speed centrifugation proteinase K digestion did not increase recovery from Photogel tubes, but increased recovery from EDTA or Streck tubes to levels similar to those obtained with Photogel tubes. Only trace amounts of genomic DNA contaminants were seen with either EDTA, Streck, or Photogel tubes. Conclusions: Recovery of cell-free DNA using our novel Photogel tube was high, required minimal processing, and the quality was comparable to that of material recovered from Streck tubes. The advantages of the Photogel tube over other commonly used tubes include 1) ability to standardize and preserve cell-free fractions in the primary tube, 2) ability to tolerate vigorous mixing to ensure uniform sampling, minimal processing by untrained personnel, ease of shipping, freezing tolerance, and 3) suitability for long-term storage. Ampliprep/COBAS Taqman HCV Test, v2.0 B. Lemos, C. Labaj, A. Tyropolis, C. Bissaillon, K. Lebel, F. Moore Baystate Health, Springfield, MA. Introduction: The purpose of this study is to verify the use of plasma preparation tubes (PPT) on the COBAS Ampliprep/COBAS Taqman HCV Test, v2.0. Currently, blood collected in EDTA tubes must be spun and the plasma removed within 6 hours of draw. Some collection sites cannot decant the plasma from EDTA tubes and this occasionally results in cancelled tests and recollection of the sample. Plasma preparation tubes need to be spun within 6 hours but do not need to be decanted immediately. The benefits to collecting in PPT tubes over EDTA are limiting aliquoting outside of the laboratory to reduce exposure and potential sample misidentification as well as decreasing the number of recollected specimens. Methods: For this study one PPT tube and two EDTA tubes per patient were collected for Hepatitis C viral load testing. EDTA tubes were spun and plasma separated from the red cells at the collection site. The PPT tubes were spun for 10 minutes at the collection site then transported to the laboratory. Once received in the lab, the PPT tube was briefly spun again and the plasma was removed and stored at -20°C until testing. The EDTA samples and the corresponding PPT sample were then run on the COBAS Ampliprep/COBAS Taqman HCV Test v2.0. The log viral load results from both sample types were recorded and compared using the Bland-Altman xy-scatter plot and regression line, and the Bland-Altman bias plot. Results: jmd.amjpathol.org ■ The Journal of Molecular Diagnostics Matched EDTA and PPT samples were tested on 64 patient samples. Nineteen numerical results by both tube types are plotted. Two outliers were identified and omitted from the data analysis. Linear regression analysis gives the equation y = 1.0365x -0.1813 with a correlation coefficient (R 2 ) of 0.9894. The slope of the line is not statistically different from 1 indicating no proportional bias, and the y intercept is close to zero indicating no constant systemic bias. The R 2 value was 0.9894, indicating good correlation. The mean bias is calculated at 0.12, indicating a systemic bias of 1.3 fold. Conclusions: For this study we verified that plasma preparation (PPT) tubes provide results equivalent to EDTA samples utilizing the COBAS Ampliprep/COBAS Taqman for HCV Test, v2.0. Although there were two outliers, the results show that the use of plasma preparation tubes versus EDTA plasma tubes does not show a significant clinical bias. M. Debeljak, M.C. Morrison, E. Mocci, A.P. Klein, J.R. Eshleman Johns Hopkins University, Baltimore, MD. Introduction: Fields of forensics, paternity, and hematopoietic stem cell transplantation (HSCT) testing require human identity testing. The polymorphic nature of short tandem repeats (STRs) makes them most frequently used but they are relatively insensitive. We previously demonstrated that using multiple SNPs on a short amplicon (haplotype counting) could overcome the inherently high error rate of NGS, and was used for ultrasensitive detection of human DNA mixes. In addition, we identified >4,000 additional loci in the human genome that could be used for haplotype counting. Here we present haplotypes of 45 individuals from three different populations (15 individuals/population) at five polymorphic loci. Methods: We designed primers to target five loci (HLA-A, HLA-B, CSMD1, FARP1, and MT4). We then obtained and amplified samples from 15 individuals in three populations used in 1000 Genomes Study (total of 45 individuals; CEU, JPT, YRI) at all five loci. Amplicons were pooled and sequenced. Haplotypes were determined from aligned sequencing data. Results: Comparison of heterozygous loci across populations showed few differences in informativity. Of 45 individuals sequenced, we found 41 to be heterozygous at MT4 (91%), 36 at HLA-B (80%), 34 at FARP1 (76%), 31 at HLA-A (69%), and 30 at CSMD1 (67%). One locus, MT4, was the most informative in all 3 populations with heterozygosity score of 0.89 in YRI, 0.87 in JPT, and 0.81 in CEU. Among these 45 individuals, there were 22 different alleles for MT4, 15 for HLA-A, 10 for HLA-B, 8 for CSMD1, and 7 for FARP1. Conclusions: Ultrasensitive detection of human DNA mixes can be achieved using multiple haplotypes despite the high error rates associated with NGS. Here we validated additional loci that are polymorphic and informative. These loci may be useful for detecting early relapse following bone marrow transplantation and other clinical applications. M. Thomas, T. Zhang, R. Dolatshahi, M.A. Sukhai, S. Garg, M. Misyura, T. Pugh, T.L. Stockley, S. Kamel-Reid University Health Network, Toronto, Ontario, Canada. Introduction: The performance of NGS panels and variant analysis methods need to be assessed within technically challenging regions, e.g., high GC-content and large or complex indels, to identify sub-optimal coverage of clinically relevant variants. This is particularly significant for hematological malignancies, in which clinically important genes are known with larger insertions/deletions (FLT3, CALR) or that are particularly GC-rich (CEBPA). In this study, we assessed the TruSight Myeloid Sequencing Panel (TMSP; Illumina) for gaps in NGS data and developed complementary approaches to ensure the capture of all clinically relevant regions in the above listed genes. Methods: We evaluated the TMSP for performance metrics and ability to detect known variants in actionable genes. Read alignments were reviewed to identify target region depth of coverage for all amplicons, and data quality (quality of mapped reads, read balance, base quality) for insertions/deletions. Samples with known variants were used to assess panel performance in detecting clinically actionable variants. Results: Under-performing amplicons (positions covered <500x in >90% of samples tested) were found in 7/54 genes. Although these low coverage regions did not contain known mutation hotspots, 5 of 6 amplicons of CEBPA (mutated in 8% of AML) consistently failed depth of coverage requirements. Routine analysis for CEBPA in AML cases therefore required a non-NGS (Sanger) complementary assay to ensure detection of CEBPA variants in all patients. We assessed TMSP's ability to detect clinically actionable FLT3 insertions (found in ~25% of AMLs). The default setting on MiSeq reporter (MSR) software for insertion/deletion detection was modified from 25bp to 55bp for this purpose; longer read lengths (2x250 bp) were also selected. 15/31 positive cases, including 4 cases with FLT3 insertion sizes >25bp, up to a maximum size of 33bp were detected by NGS (assay sensitivity = 48%; specificity = 100%), necessitating the use of other assays to detect larger FLT3 insertions. A custom-designed script was developed to detect the recurrent clinically actionable 52 bp deletion in CALR from NGS data. Read data from 387 samples were analyzed; all known positive cases were successfully identified. Conclusions: NGS assays may not detect variants in all clinically relevant regions required for diagnosis or management of specific patient populations. Variants in CEBPA, FLT3 and CALR required supplementation with non-NGS assays or informatics approaches to address deficiencies in performance. Laboratories need to assess low performing regions on panels to prevent false negative results, and should consider the necessary optimization as part of NGS test validation prior to implementation. K. Krishnan, E. Yigit, M. Karaca, B. Langhorst, T. Shtatland, D. Munafo, D. Rodriguez, P. Liu, L. Apone, V. Panchapakesa, K. Duggan, C. Sumner, C. Rozzi, F.J. Stewart, L. Mazzola, J. Bybee, D. Rivizzigno New England Biolabs, Ipswich, MA. Introduction: RNA-seq (RNA sequencing) has become the most popular method for transcriptome analysis. It is widely used for gene expression analysis, detection of mutations, fusion transcripts, alternative splicing, and post-transcriptional modifications. Recent improvements in next-generation sequencing technologies (NGS) and sample barcoding strategies allow analysis of multiple samples in parallel in a cost effective manner. As RNA-seq is being rapidly adopted for molecular diagnostics, the quality and reproducibility of library preparation methods are becoming more important. In addition, demand for library preparation methods that support successful NGS library construction from low input RNA or precious clinical samples is increasing. To overcome these challenges, we have developed a strandspecific RNA-seq library preparation method that retains information about which strand of DNA is transcribed from as low as 10 ng total RNA input. Strand specificity is important for the correct annotation of genes, identification of antisense transcripts with potential regulatory roles, and for correct determination of gene expression levels in the presence of antisense transcripts. Methods: poly-A mRNA from Universal Human Reference RNA (10 ng -1000 ng total RNA) was enriched to make libraries using our strand specific method. Libraries were analyzed on an Agilent Bioanalyzer, pooled at equimolar ratio and sequenced on Illumina's Nextseq 500. Paired end reads were mapped to a human reference genome (hg19) using Hisat2 and sequencing metrics were calculated using Picard's RNA-seq Metrics and RSeqC tools. Transcript abundance was measured using Salmon and the Ensembl GRCh38 CDS sequences. Results: Libraries prepared with our streamlined method using inputs that range from 10ng to 100ng show greater than 98% directionality at all input levels. Furthermore, GC content analysis, gene body coverage and gene expression correlation are similar for all inputs tested (10 ng, 100 ng, 1000 ng) even though input amount varies by three orders of magnitude. These consistent results are recapitulated with the spiked-in ERCC controls at all inputs. Conclusions: Our library preparation method is streamlined and can be used for a wide range of input RNA without any major modifications to the protocol, making it a convenient method for RNA-seq library preparation. In addition, it has increased sensitivity and specificity, especially for low-abundance transcripts, reduced PCR duplicates and sequence bias, delivering high quality strand-specific data even for low input RNA. Finally, our method is compatible with both poly A-tail enriched and ribosomal RNA depleted samples, and is amenable to large-scale library construction and automation. Introduction: A (GGGGCC)n hexanucleotide repeat expansion in an intronic region of the C9orf72 gene has been observed in the general population with a frequency of ~1/600 and is present in ~10% of all amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) cases. Fewer than 30 repeats are considered normal whereas pathogenic C9orf72 expansions have 100's to 1000's of repeats. The GCrich repeat poses formidable challenges to routine PCR-based fragment sizing methods, and currently requires analysis using multiple short-range PCR reactions for each sample followed by Southern blot (SB) analysis of expanded samples. Here we describe a two-site evaluation of a single-tube, highly streamlined PCR assay that both detects C9orf72 expansions and sizes expanded alleles with 100's of repeats. Methods: AmplideX PCR reagents were optimized for the amplification of C9orf72 hexanucleotide repeats. Amplicons were sized on a 3500xL or 3130xl Genetic Analyzer (Thermo Fisher) and/or SeaKem LE Plus agarose gels (Lonza). The assay was evaluated at both the University of Pennsylvania (Site 1) and Asuragen (Site 2). Site 1 evaluated a subset of NINDS ALS samples (n=50) and an independent set of ALS and FTD patient-derived from peripheral blood (n=100, 50 of which were expanded), saliva (n=10 expanded), and brain (n=30 expanded) specimens. Site 2 assessed a broader set of NINDS ALS samples (n=119, 72 of which were expanded), including those in common with Site 1, as part of a larger study to analyze the full collection (n=2100). Results: Across both testing sites, repeat numbers were resolved consistent with previous annotations for all normal samples and up to 145 repeats for expanded samples. Fragment sizing was limited only by the resolution limitations of capillary electrophoresis (CE). Compatibility with both blood and brain FFPE DNA at 40 ng inputs was demonstrated, and samples with size mosaicism or indels could be identified from the repeat-primed CE traces. Site 2 further demonstrated sensitivity to as little as 1 ng input of cell-line DNA. This site also showed proof-of-concept for agarose gel sizing of PCR amplicons from expansions with up to 800 repeats, which is >10-fold larger than any published report The Journal of Molecular Diagnostics ■ jmd.amjpathol.org using PCR. Conclusions: A multi-site evaluation of a long-read C9orf72 PCR technology revealed both the reliable quantification up to 145 hexanucleotide repeats at high resolution on CE and the identification of repeat expansions irrespective of size, all in a single assay. This innovative, high-throughput PCR has important implications for clinical research and emerging diagnostic, therapeutic, and screening applications for ALS and FTD, and possibly other neurodegenerative disorders. C. Christopherson, P. Sankey, C. Chan, L. Cabiling, E. Caoili, T. Fine, C. Fenech, L. Eubank, P. Parmar, O. Badrenkova, J. Wilbur, R. Woodward, M. Nelles, S. Scott CareDx, Brisbane, CA. Introduction: Circulating cell-free DNA (cfDNA) has been described as a marker to assess allograft rejection in organ transplant recipients. We developed a noninvasive targeted next-generation sequencing (NGS) assay employing 266 SNPs to quantify donor-derived cfDNA levels (dd-cfDNA) in solid organ transplant recipients. Tests such as these are complex multi-step assays that require seamless integration of bench, LIMS and bioinformatics workflows. In addition to validation of the analytical performance, software QA testing, and bioinformatics pipeline validation; it is critical to examine the entire test system as a whole in a process validation, to ensure specimens are accurately tracked and analyzed. Additionally, the potential impact of improper specimen handling should be considered and evaluated within the context of unique effects on the assay being validated. Methods: Testing was performed using cfDNA extracted from plasma from healthy volunteers collected in Streck Cell-Free DNA BCT tubes. Mock transplant specimens were created by spiking plasma from one individual into another. Before extracting cfDNA, plasma was mixed to simulate "donor" and "recipient" levels consistent with expected levels in organ transplant recipients. For process validation, commonly encountered accessioning, testing, and reporting scenarios, including repeat testing and demographic parameter changes were created to ensure that specimens could be accurately tracked through the system and reported. Interference testing was performed using purified hemoglobin, bilirubin, and triglycerides. Each substance was individually spiked into plasma and processed following standard procedures. Contamination with recipient cellular DNA present as a consequence of hemolysis during specimen handling was evaluated by spiking the separated recipient plasma with the increasing amounts hemolysate prepared from residual recipient blood cells. Results: Process validation confirmed each step in the process was accurate to final reporting, and identified areas in which future integration and workflow improvements were desirable. Testing showed that mishandled specimens may impact the accurate measurement the %dd-cfDNA due to the release of recipient cellular DNA in hemolyzed specimens. This result was also critical for interpretation of results using interferents generated from human source material. Conclusions: Process validation showed each step was accurate from specimen accession to final reporting. Hemolysis in blood collection tubes may impact accurate measurement of the %dd-cfDNA. A hemolysis rating scale is a useful tool in evaluating specimen quality to identify specimens with high levels of hemolysis that need to be excluded from testing. L. Jackson 1 , A. Wolf 2 , W. Hahn 1 , H. Mellert 1 , H. Mellert 1 , G. Pestano 1 1 Biodesix, Inc., Boulder CO; 2 Qiagen GmbH, Hilden, Germany. Introduction: Approximately 25% of patients with advanced non-small cell lung cancer (NSCLC) are not candidates for tissue biopsies and in some cases where tissue is obtained, it is not always of sufficient quantity or quality for molecular testing. Thus, the detection of circulating nucleic acids has become relevant in the Clinical Laboratory setting. Methods: In this study, we have focused on the optimization of the nucleic extraction step (manual versus automated) for a commercially-available diagnostic test system for NSCLC, GeneStrat. Criteria for evaluation included yield of circulating free (cf) nucleic acids, droplet counts for somatic mutant variants and the counterpart wild-type, as well as percentage minor allelic frequency (%MAF) as measured with validated assays. The specific assays used to evaluate performance were for KRAS G12C, KRAS G12D, EGFR T790M and EGFR del19 (A740-E746). As the reference method for evaluation of the automated QIAsymphony SP we used the previously validated "manual" method developed using the QIAamp Circulating Nucleic Acid Kit. Performance was determined by detection of recovered DNA variants by droplet digital PCR (ddPCR). For our reference and test arms, plasma from four donors with NSCLC and from four parallel processed spiked mixed allelic frequency DNA sample with known %MAF, were extracted and tested by GeneStrat. Results: Three of the clinical samples and all four of the spiked control sample extractions yielded results. Operator error accounted for the loss of one of the clinical samples during the manual extraction method. Overall, we observed that the automated method performed at least equivalently or was slightly superior to the established manual method when evaluating any of the measures in this study. The automated system yielded better results when evaluating the average percent differences for the variants and wildtype counts from both clinical samples (16% and 3%, respectively) and for the analytic samples (3% and 11%, respectively). Overall method variability as evaluated by calculated differences in %MAF between the methods were below 0.01% (0% to 0.01%). Conclusions: Our findings demonstrate that the automated and manual methods of recovering circulating nucleic from plasma perform at least equivalently for the gene read-outs and test system evaluated here. The quality and yield of cf DNA from the automated process was sufficient to meet the previously established criteria for the manual performance of the assays. Evaluations are on-going with the automated systems for larger plasma volumes and increased numbers of samples that could be processed in parallel. M.P. Powers 1 , S. Balakrishnan 2 , R. Calef 2 , P. Hartley 2 , J. Stites 2 , C. Troll 2 , N. Putnam 2 , R.E. Green 2 1 Dovetail Genomics, LLC, San Francisco, CA; 2 Dovetail Genomics, LLC, Santa Cruz, CA. Introduction: Presented herein is a novel technique for the generation of hypermegabase phased sequence data via a pre-extraction in situ DNA library preparation from formalin-fixed, paraffin-embedded (FFPE) samples. Until now, FFPE samples typically yielded sub-500bp DNA fragments confounding the identification of longrange sequence information. Long-range sequence information can only be obtained from FFPE tissue in non-repetitive loci where reads overlap unique breakpoints or other unique sequence features. Because the human genome is >50% repetitive, human FFPE samples have to date not provided adequate long-range genome information. Methods: In this first of its kind proof-of-principle study we: i) remove paraffin from FFPE treated cell lines (GM24149, GM24385; Horizon Discovery, Cambridge, United Kingdon), ii) treat the tissue fragments in situ with restriction endonuclease, iii) fill-in the resulting recessed 3' termini with a tag for purification of ligation fragments, and iv) ligate the resulting ends, creating chimeric molecules. These chimeric molecules preserve the long-range information of the starting molecule due to proximity-based ligation. Results: Sequencing of the ligated library provides data similar to those produced by Hi-C or Chicago. Preliminary assembly with a modified version of our HiRise software demonstrates that phasing and longrange information is preserved from these samples with >90% accuracy up to 1.9 Mbps. Connections greater than 1.9 MBps are also seen, but with an increase in errant connections across chromosomes. Conclusions: This technique provides a powerful new way to extract genome information for archives of tumor samples and other FFPE samples from bio-banks including the discovery of novel structural variation, copy-neutral structural variants, and other long-range sequence information including those within (or mediated by) repetitive elements. Experiments are on-going with FFPE samples from bio-banks. Construction Workflow for Use with Challenging Samples V. Panchapakesa, L. Apone, K. Duggan, P. Liu, T. Shtatland, B. Langhorst, J. Murdoch, C. Sumner, C. Rozzi, K. Krishnan, D. Rodriguez, J. Bybee, D. Rivizzigno, L. Mazzola, F. Stewart, E. Dimalanta, T. Davis New England Biolabs, Ipswich, MA. Introduction: Next-generation sequencing (NGS) plays a vital role in understanding molecular mechanisms involved in disease processes. This knowledge is important to support drug development targeted towards precision medicine. Success on NGS platforms begins with optimal sample preparation, which has become a bottleneck in NGS workflows. We have developed a robust, flexible library preparation method that integrates enzymatic fragmentation into the workflow. This method can be easily automated and is compatible with ultra low input as well as challenging samples such as Formalin Fixed Paraffin Embedded samples (FFPE). Method: Genomic DNA (gDNA) isolated from a variety of sources including FFPE were used to construct Illumina compatible, NGS libraries. gDNA ranging from 100pg to 500ng was fragmented, end repaired and dA-tailed in a single step. Adaptor ligation was performed in the same tube followed by a bead based cleanup and PCR amplification. Library quality and concentration were determined using the Agilent Bioanalyzer. Libraries were sequenced on the Illumina platform, reads were aligned to the appropriate reference genome using Bowtie 2 and quality metrics were generated using Picard tools. Results: Libraries constructed using intact gDNA and this novel library preparation method produced substantially higher yields than those generated using mechanically fragmented DNA. The greatest differences were observed with the lowest DNA inputs and most challenging samples, where a 4-5 fold increase in yield was observed with use of the novel method. Sequencing quality of libraries generated with inputs ranging from 100pg to 500ng of intact DNA show no significant difference in coverage uniformity, mappable reads, duplication rates, GC bias metrics and fragment size distribution compared to libraries constructed with mechanically sheared DNA. Conclusions: We have developed an enzymatic DNA fragmentation kit that enables the construction of high quality libraries from a range of sample types and inputs with minimal sequence bias, high yields and low duplication rates. This method eliminates the need for expensive equipment to fragment DNA as well as numerous cleanup and liquid transfer steps thereby reducing the time, cost and errors associated with library construction. These advances will permit greater use and adoption of NGS technologies in clinical and diagnostic settings. jmd.amjpathol.org ■ The Journal of Molecular Diagnostics TT48. Automation of NGS Workflow for Processing of High Sample Volume in a Clinical Laboratory S. Mercurio, A.I. Wald, K.M. Callenberg, S. Roy, Y.E. Nikiforov, M.N. Nikiforova University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: Next-generation sequencing (NGS) technologies have been successfully used in clinical laboratories. However, processing of a high sample volume is complicated due to lack of automation in multiple steps of NGS sample/library processing and data analysis. We have developed a specialized workflow that streamlines all components of testing to efficiently carry out NGS analysis in a high volume molecular laboratory. Methods: To accommodate a high increase in volume of NGS tests, we have developed a clinical NGS laboratory workflow to integrate multiple components of targeted amplification based NGS analysis, including specimen tracking, sample preparation, wet-bench automation and bioinformatics analysis. From November 2015 to April 2016, we have processed about 5,000 specimens for NGS testing (~10,000 libraries). Sequencing was performed on Ion Torrent PGM and Ion Proton platforms (Life Technologies). Results: The created NGS workflow utilizes automation and high-throughput technologies for each step of NGS testing: 1-nucleic acids isolation is performed using automated extractors (Magnapure Compact, Qiacube), 2-DNA and RNA yields are measured by high-throughput fluorometer (Promega Glomax Discover), 3multiplex PCR for custom targeted AmpliSeq library generation is set-up with the automated pipetter instruments (QIAgility), 4-library preparation, quantitation, normalization, template preparation and Ion Proton chip loading is automated using high-throughput instruments that require minimal user input (Tecan Freedom EVO 100, TapeStation 4200, Ion Chef). In addition, we developed automated bioinformatics pipelines (Variant Explorer, SeqReporter) that allows for fast variant analysis, QC monitoring, specimen tracking, functional variants categorization and reporting. Since implementation of this workflow, the average turn-around time was 8.2 days despite a 2-fold increase in volume in the past year. The workflow reduced hands-on time from 18 hours to 9 hours for library generation to sequencing of each batch of samples and significantly reduced the technical errors and repeat testing from ~5% to under 2% of all libraries. Conclusions: We have developed a special clinical NGS laboratory workflow infrastructure to integrate specimen tracking, sample preparation, wet-bench automation, sequencing analysis and reporting. This type of organization is necessary to facilitate operations in a high-throughput laboratory while maintaining appropriate turn-around time, decreasing hands-on technologist time, and ensuring assay consistency, accuracy and efficiency. Introduction: Standard FISH techniques require hybridization times of 12 hours or greater, thus amounting to turn around times (TAT) of 24 hours or greater for results reporting. To improve the TAT for the FISH tests, the new fast working Vysis IntelliFISH hybridization buffer has been recently developed by Abbott Molecular. Compared to standard LSI hybridization buffer protocols, the new fast working Vysis IntelliFISH hybridization buffer protocol decreases the required hybridization time and has the potential to improve hybridization quality. Methods: The Vysis IntelliFISH hybridization buffer was evaluated in bone marrow and formalin fixed Lymphoma tissues samples: 1) 20 pairs of matched bone marrow samples were probed with Vysis TP53/ CEP17 FISH probe kit using both the standard LSI protocol & Vysis IntelliFISH fast working protocol. 2) 10 pairs of matched formalin fixed Lymphoma tissues samples were probed with Vysis IGH/MYC/CEP8 probe using both the standard LSI protocol & Vysis IntelliFISH buffer protocol. Results: 1) The IntelliFISH hybridization buffer was clearly superior to the standard LSI hybridization buffer in analyzing bone marrow samples. Hybridization time (2 or 3 hours) for the Vysis TP53/ CEP17 probes demonstrated comparable performance in bone marrow samples when using the Vysis IntelliFISH hybridization buffer. 2) The signal intensity and specificity of Vysis IGH/MYC/CEP8 probes in Lymphoma samples was comparable using either the Vysis IntelliFISH hybridization buffer or the standard LSI buffer. Conclusions: The Vysis IntelliFISH hybridization buffer significantly reduces FISH hybridization time and simplifies the workflow of the standard overnight Vysis hybridization protocols. Signal intensity, specificity and signal to noise ratio of FISH probes were comparable to standard hybridization protocols. Regardless of Donor-Recipient Familial Relationship K. Thompson, J. Collins, C. Marchis, J. Sninsky, R.N. Woodward, M. Grskovic CareDx, Brisbane, CA. Introduction: Circulating cell-free DNA (cfDNA) is a novel biomarker for noninvasive assessment of transplanted organ injury. We developed AlloSure, a clinicalgrade NGS assay to measure the level of donor-derived cfDNA (dd-cfDNA) using a panel of 266 SNPs that does not require genotyping of donor or recipient. Previous studies validated the accuracy and linearity in genetically unrelated donor-recipient pairs. Compared to unrelated donor-recipient pairs, related donor-recipient pairs are expected to be more challenging due to the smaller number of variant alleles. In this study, we evaluate the performance of AlloSure in donor-recipient pairs with different degrees of relatedness. Methods: DNA from verified extended reference families from NIGM Biorepository was obtained from Coriell. DNA mixtures mimicking plasma cfDNA from transplant recipients with related and unrelated donors were developed. Twenty-two panels representing 22 'transplant recipients' were created across four families; and included seven unrelated, two grandparent/grandchild, five parent/child, and eight sibling donor-recipient mixtures. Each panel contained 'donor' DNA mixed into 'recipient' DNA at levels ranging from 0.1% to 12.5%. Accuracy was assessed by comparing dd-cfDNA results to trace amount spike-in levels. Limit of Detection (LOD) and Limit of Quantification (LOQ) were determined in comparison to the unrelated donor-recipient pairs. Results: There is a strong correlation between expected and AlloSure-determined values of dd-cfDNA for both related and unrelated donor-recipient pairs (R 2 0.981 and 0.993; RMSE 0.82% and 1.31%, respectively). Panels assembled from different family DNAs and replicate runs of select panels produced reproducible results with CVs <10%. LOD and LOQ are incrementally higher for the closely related 'donor-recipient' pairs compared to unrelated pairs (0.22% vs 0.16% LOD and 0.22% vs 0.2% LOQ). Conclusion: The analytical performance of a clinical-grade NGS dd-cfDNA assay was demonstrated using custom familial reference material panels mimicking both unrelated and related donor-recipient pairs. AlloSure is a sensitive, accurate, and precise non-invasive assay with application for surveillance of solid organ transplant recipients. With the exception of donor-recipient monozygotic twins, this study suggests AlloSure is applicable to single solid organ transplant recipients regardless of the degree of relatedness between donor and recipient. A. Ullius 1,3 , T. Voss 1 , S. Busche 2 , W. Hofmann 2 , D. Groelz 1 1 PreAnalytiX GmbH, Hilden, Germany; 2 LifeCodexx, Konstanz, Germany; 3 Qiagen GmbH, Hilden, Germany. Introduction: Non-invasive prenatal testing (NIPT) uses fetal circulating cell-free DNA (ccfDNA) in the maternal blood circulation. The release of maternal genomic DNA from blood cells during blood collection, transport and storage significantly reduces the sensitivity of the next-generation sequencing (NGS) based NIPT assays. Moreover, the reliable and reproducible extraction of the low abundance and short fetal ccfDNA fragments is technically challenging. Both difficulties are addressed by the PAXgene Blood ccfDNA System, consisting of the PAXgene Blood ccfDNA Tube, a plastic BD Vacutainer blood collection tube with a unique, non-crosslinking chemistry for stabilization of blood cells, and the QIAsymphony PAXgene Blood ccfDNA Kit, an automated ccfDNA extraction kit. The performance of the new system was evaluated in two research studies in comparison to K2EDTA tubes and the Streck Cell-Free DNA BCT. Methods: Study 1: Blood from 20 consented, apparently healthy donors in three replicates was collected into paired PAXgene and K2EDTA tubes and stored at different conditions to simulate sample transport and storage. ccfDNA was extracted from separated plasma using the QIAsymphony kit or the QIAGEN QIAamp Circulating Nucleic Acid Kit and analyzed by a validated quantitative real-time PCR assay targeting 18S rDNA. Study 2: Paired blood samples from 25 pregnant women were collected into PAXgene and Streck tubes, and ccfDNA was extracted using the QIAsymphony kit. Fetal ccfDNA fraction was quantified by qPCR with the LifeCodexx QuantYfeX assay. Finally, eluates were analyzed using the NGS-based LifeCodexx PraenaTest assay for chromosomal disorders. Results: Study 1: The relative mean ccfDNA yield from PAXgene tubes was 97% ± 8% in comparison to the paired K2EDTA tubes when plasma was processed immediately after blood collection. Simulation of transport and storage of the PAXgene tubes for up to 7 days at 25°C did not increase yield significantly as compared to PAXgene tubes processed immediately after blood collection (113% ± 34%). Consistent and stable performance has been demonstrated by more than 70 runs using 6 different instruments and more than 1700 samples from over 200 donors. Study 2: Relative fetal ccfDNA fractions in maternal blood in PAXgene and Streck tubes were not significantly different (11.4% versus 12.4%). All samples which passed the internal quality standards produced equivalent results with the PraenaTest assay. Conclusions: The PAXgene Blood ccfDNA System provides consistent performance after 7 days of simulated transport and storage at RT and automated ccfDNA extraction from plasma with minimal variation between replicates and highly reliable results. For research use only. Not for use in diagnostic procedures. Introduction: Detection of single nucleotide variants (SNVs) and insertion/deletions (indels) using next-generation sequencing (NGS) technology is gaining increasing importance in research into cancer development and progression. Tissue biopsies are typically archived as formalin-fixed, paraffin embedded (FFPE) blocks, which preserve tissue morphology and allow long-term storage at room temperature. However, the methods used for fixation significantly damage and compromise the quality of nucleic acids from these samples. Formalin damage can fragment the DNA jmd.amjpathol.org ■ The Journal of Molecular Diagnostics TT56. Validation of a Clinical Mate Pair Sequencing Assay to Characterize Apparently Balanced Structural Chromosome Variants Y. Cao, S. Smoley, B. Porath, R. Rowsey, J. Rustin, S. Johnson, G. Vasmatzis, S. Yerneni, J. Blommel, L. Peterson, K. Pearce, H. Kearney, U. Aypar, R. Jenkins, W. Sukov, N. Hoppman Mayo Clinic, Rochester, MN. Introduction: Mate Pair sequencing, an emerging powerful technology able to detect structural variation, is able to further characterize the breakpoints of apparently balanced rearrangements and better understand the genetic content of these regions. Here we validate the performance of a laboratory-developed whole genome Mate Pair sequencing assay to characterize apparently balanced rearrangements identified in 75 samples by validated Cytogenetic testing. Methods: DNA samples were library-prepped using the Illumina Nextera Mate Pair library preparation kit and sequenced on the Illumina HiSeq 2500 in rapid run mode. Pooled libraries are hybridized two samples per flow cell and sequenced using 101basepair reads and paired end sequencing. Data is aligned to the GRCh38 reference genome using BIMAv3 (in-house developed tools). Structural variants and their breakpoints are identified using SVAtools (in-house developed tools). Results: Twenty-five abnormal samples were collected from each tissue type, including peripheral blood (PB), bone marrow (BM) and fresh/frozen tissue (FT). To evaluate accuracy, we compared Mate Pair sequencing results to the structural variants previously identified by validated Cytogenetic testing. Mate Pair sequencing successfully detected 71/75 structural variants: 24/25 in PB, 23/25 in BM, and 24/25 in FT. Two abnormalities in BM were not detected probably due to the low level of abnormalities, 15% (3/20 metaphases) and 25% (5/20 metaphases) estimated by chromosome studies. Although, Mate Pair sequencing, with high reproducibility, is able to detect most abnormalities with variable DNA input (as low as 0.25ug) and variable level of mosaicism (as low as 5-10%), the detection level may vary among different variants due to the difference of sequence structures. Mate Pair sequencing also missed one translocation in PB and one uncharacterizable rearrangement in FT. Of note, previous Mate Pair sequencing with high coverage (88x) research setting did detect the constitutional translocation, but only with few data supports due to the repetitive sequence of this region, suggesting the failure of detection is probably due to the limited coverage (36x) in rapid run mode as well as the genomic repetitive nature of this region. Conclusions: Mate Pair sequencing successfully detected 71/75 (94.7%) structural variants: 24/25 (96.0%) in PB, 23/25 (92.0%) in BM, and 24/25 (96.0%) in FT. These results provide evidence supporting the clinical utility of mate pair sequencing to characterize apparently balanced rearrangements. Meanwhile, the failure of detection in 4/75 samples suggest the limitation of this assay: it may not be able to detect low level abnormalities or regions with complex genomic structures such as repetitive DNA sequence. During Cutting-needle Biopsy Guided by Computed Tomography S.I. Meireles, S.S. Koide, C.D. Godoy, R.A. Coudry Hospital Sírio Libanês, São Paulo, Brazil. Introduction: Cutting-needle biopsy guided by computed tomography (CT) is used largely for lung, pleural and mediastinal lesions and provides tissue samples that are suitable for histological evaluation. On-site assessment of tumor adequacy (rapid onsite evaluation) has been performed to maximize precise sampling for diagnosis. Since molecular analysis has a fundamental role to drive therapy in lung cancer, an incremental volume of tumor tissue has been demanded over time. In this study, we investigated the usefulness for molecular analysis of tumor material from rapid touch-prepared glass slides generated for assessment of adequacy during imaging guided procedure. Methods: Panoptic or Thionin stained cells from 26 cases harvested from the slides were submitted to DNA isolation and mutational analysis by next-generation sequencing (NGS) or capillary sequencing. Results: The number of cells in the touch-prepared slides ranged from 200 to 5000 cells and the neoplastic content was between 40% to 90%. There was only one case were the quantity of cells in the imprint was insufficient for further testing. 17 cases were successfully analyzed by NGS and 8 cases were applied in capillary sequencing using as low as 16 ng of DNA per reaction. Mutations were detected in 65.4% of the samples, including KRAS (5 cases), EGFR (6 cases), ERBB2 (1 case) and other genes (4 cases). Mutational status in the touch-prepared specimen was compared with the FFPE matched specimen in 8 cases and demonstrated 100% concordance for both presence and absence of mutation. FFPE material from the remaining 18 cases yielded insufficient number of cells or quantity of DNA for molecular testing. Conclusions: Material obtained from on-site assessment of sample adequacy from cutting-needle CT guided biopsy may represent a valuable specimen source for molecular testing. Additional cases are under evaluation to expand the comparison between touch-prepared glasses slides and matched FFPE samples. J. Eberlein, T. Harrison, I. McKittrick, M. Wemmer, L.M. Griffin, B.P. Culver, L.A. Johnson, B.A. Kudlow ArcherDX, Inc., Boulder, CO. Introduction: The adaptive immune system is involved in various disease conditions including cancer, chronic infection, autoimmune disease and transplant rejection. Adaptive immunity is mediated by B and T lymphocytes, which are activated upon antigen binding to antigen receptors expressed on their surface. Therefore, the spectrum of these antigen receptors, or immune repertoire (IR), provides a means to monitor adaptive immune responses to disease, vaccination and therapeutic interventions. Next-generation sequencing (NGS) of antigen receptor genes is a valuable tool in the study of disease states and responses to various interventions. Traditional amplicon-based NGS assays use opposing primers for targeted amplification of rearranged antigen receptor genes. Thus, large primer panels are required to capture the extensive combinatorial diversity exhibited by the IR. Quantification from such assays requires a complex system of synthetic controls to account for differential amplification efficiency across segment combinations. Here, we describe an Anchored Multiplex PCR (AMP)-based NGS assay to analyze the IR, employing a minimal set of gene-specific primers in conjunction with molecular barcodes (MBCs) to reduce amplification bias. Methods: AMP uses MBCs ligated to cDNA ends and gene-specific primers for amplification, enabling immune chain mRNA interrogation from a single side. This eliminates the need for opposing primers that bind within the highly variable V-segment, eliminating clone dropout due to somatic hypermutation. Furthermore, this facilitates CDR3 sequence capture from highly fragmented RNA inputs. We validated the quantitative reproducibility and sensitivity of AMP-based B-and T-cell IR assays using high-quality mRNA isolated from peripheral blood leukocytes and highly fragmented RNA isolated from formalinfixed paraffin-embedded (FFPE) samples. Results: Our data showed high reproducibility between replicates and quantitative clone tracking down to 0.01%. Furthermore, our data indicate that clonal diversity in sequencing data is driven by input quantity, total T-cell number, and, to a lesser degree, mRNA quality. Conclusions: AMP-based NGS with MBC quantification and error-correction is a powerful method to characterize the immune repertoire. This enables sensitive clone tracking and measurement of lymphocyte diversity from fragmented RNA samples. Introduction: Target enrichment of selected exonic regions for deep sequence analysis is a widely used practice for the discovery of novel variants, and identification and phenotypic association of known variants for a wide range of practical applications. Current available strategies for selective enrichment can be characterized as either hybridization-based enrichment, where long synthetic oligonucleotides are used to selectively capture regions of interest, or multiplexed amplicon-based, where pairs of short primer sequences leverage PCR to selectively amplify sequence targets. Although hybridization-based methods have proven to be a tractable approach for large panels scaling to whole exome, the approach presents challenges in a relatively high sample input requirement, longer workflows, and inability to scale to very focused panels. In contrast, multiplexed amplicon approaches have proven valuable for small, highly focused panels, yet suffer from inherent challenges including the inability to scale content, loss of specificity associated with PCR duplication, and difficulties annealing primer pairs to already degraded materials. Method: We have developed a technology that utilizes a novel approach to selectively enrich nucleic acid targets ranging from a single gene to several hundred genes, without sacrificing specificity. Fragmented DNA is rapidly hybridized to biotinylated oligonucleotide baits that define the 3´ end of each target of interest. The bait-target hybrids are bound to streptavidin beads and any 3´ off target sequence is removed enzymatically. The trimmed targets are then converted into Illumina-compatible libraries that include unique molecular identifiers (UMIs) and a sample barcode. A Cancer Hotspot panel has been developed to enable highly specific hybridization-based capture of 190 common cancer targets from 50 genes using this technology. Results: The combination of a short hybridization time with the enzymatic removal of 3´ off-target sequence enabled greater sequencing efficiency relative to conventional hybridization-based enrichment methods. With the Cancer HotSpot panel, 100% of bases were covered to 25% or higher of the mean target coverage (MTC), 98% of bases were covered to 33% or higher of the MTC, and 90% of bases were covered to 50% or higher of the MTC. A high percentage of sequence reads also mapped to targets when challenging sample types such as FFPE and circulating tumor DNA (ctDA) were used, and the panel provided minimized bias across sequence content. Conclusions: This one-day protocol enables the preparation of sequence-ready libraries with high specificity, uniformity, and sensitivity for the discovery and identification of nucleic acid variants, even with challenging sample types. B.J. Dokus, H.B. Steinmetz, G.J. Tsongalis, L.J. Tafe Dartmouth-Hitchcock Medical Center, Lebanon, NH. Introduction: Manual assessment of fluorescence in situ hybridization (FISH) slides is labor intensive, susceptible to interobserver variability and subjectivity and requires technologist and Pathologist review on a fluorescence microscope. We implemented an automated imaging analysis platform into our FISH workflow to alleviate variables associated with manual slide review. Methods: The BioView Allegro Plus (BioView Inc., Billerica, MA) is an automated imaging and analysis system with an 8 slide motorized stage which can scan multiple FISH probe types in a single batch. We validated this system using the PathVysion HER2 probe in FFPE breast and gastro-intestinal tissues (for which the instrument is FDA approved). This system allows for tissue matching of the FISH tissue slide with the H&E slide for the purposes of tumor localization and review of scored cells. The developed workflow includes the following steps performed by a technologist: 1) Scan H&E slide; 2) low resolution scan of FISH slide; 3) tissue matching of FISH and H&E slides to digitally align; 4) select 20 fields of view (FOV) from the regions of tumor previously marked by a Pathologist; 5) scan FOVs at the 60X resolution; 6) review and select 40 cells using the analysis software for classification and scoring using the Solo workstations; 7) second technologist review, formalize the case and electronically assign the case to the Pathologist for final review and verification. Results: Our HER2 FISH validation study included analysis of 50 samples on the Bioview system that have been previously analyzed manually (20 positive, 20 negative and 10 equivocal for HER2 amplification). A significant amount of time and effort was invested into optimizing the operating procedures and establishing a workable protocol before validation could begin. Although there was a significant learning curve, scanned results were comparable to the manual scoring. Harmonized analysis and improved turnaround times were also achieved. Conclusions: Adoption of the BioView Allegro Plus into our Laboratory, although requiring a significant initial investment of time and resources, ultimately allowed for more standardization and automation in the scoring of FISH slides, time savings for the reviewers and provides more accurate correlation with H&E slides. The pathologist can use the Solo web to review the case on a local or remote computer which can also potentially be linked to the laboratory's LIS. With tissue matching, the pathologist can review exactly which cells were selected and scored by the technologist and view the corresponding region on the H&E slide. Both methods require prior knowledge of the fusion partners, have no or limited multiplex capacity, and lack the ability to detect novel fusions. We evaluated the performance of a next-generation sequencing (NGS)-based assay utilizing Anchored Multiplex PCR technology (Archer FusionPlex, Boulder, CO) for detection of gene fusions in RNA isolated from formalin-fixed, paraffin-embedded (FFPE) sarcomas in comparison with FISH and cytogenetic findings. Methods: Total RNA was extracted from 17 cases of microdissected FFPE sarcomas including 1 alveolar rhabdomyosarcoma (ARM), 3 alveolar soft-part sarcoma (ASPS), 5 Ewing sarcoma (EWS), and 8 synovial sarcoma (SS). Sequencing analysis for gene fusions was performed using the Universal RNA Fusion Detection Kit (ArcherDX), the Archer FusionPlex Sarcoma Panel (ArcherDX) and the Ion Torrent personal genomic machine (Life Technologies, Carlsbad, CA). RNA from a EWSR1-FLI1 positive cell line (SK-N-MC) and two negative (RNAs from normal blood and FFPE) controls were included. Data were analyzed using the Archer Analysis Pipeline 4.0. All cases had been tested by FISH (14) and/or conventional cytogenetics (3). Results: Gene fusions with specific breakpoints were detected in 15 tumors: ASPSCR1-TFE3 in 3 cases of ASPS; 3 fusion variants in 5 cases of EWS: EWSR1-FLI (3), EWRS1-ERG (1), and FUS-ERG (1); 2 fusion variants in 7 cases of SS: SS18-SSX1 (3) and SS18-SSX2 (4). No fusion was detected in ARM (1) and 1 case of SS. Fifteen of 17 cases had concordant FISH or cytogenetics findings (88%). Two Ewing sarcomas were FISH-negative using the EWRS1 breakapart probe but were positive for EWRS1-ERG (low level) and FUS-ERG fusion, respectively, by this NGS-based assay. Conclusions: This NGS-based gene fusion detection assay is a reliable platform for simultaneous detection of multiple gene fusions from various sarcomas. The assay works well with RNA isolated from FFPE tissue. It is sensitive and can separate different fusions at a single base level. Using this assay, we were able to identify all the translocations identified by FISH or cytogenetics and found fusions that were missed by FISH. Most importantly, this assay allows discovery of new fusions without prior knowledge of the fusion partners, a unique characteristics that neither FISH nor RT-PCR has. Next-Generation Sequencing-Based Technologies L. Wang, J. Wang, M. Rao, R. Cimera, R. Aryeequaye, L. Cao, R. Benayed, M. Ladanyi, T. Hollmann, K. Busam, M. Hameed Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: Gene/locus-specific fluorescence in situ hybridization (FISH) probes are typically derived from cloned genomic regions presented in BAC (bacterial artificial chromosome) and PAC (P1-derived artificial chromosome) clones. The limitations of these conventional probes are the availability of specific clones and the large size of their genomic inserts (150 to ~300 kb). The use of array-and nextgeneration sequencing-based technologies is expanding our understanding of structural abnormalities at a previously unappreciated resolution, and many of the abnormalities detected are difficult to visualize using conventional FISH. Thus, there is a need to develop high-resolution (<100Kb) FISH probes. We present here our experience with oligonucleotide-based FISH (oFISH) probes which are derived from synthesized oligonucleotides targeting specific sequences in genome. Methods: Two cases with BRAFstructural aberrations revealed by high-resolution SNP-array analysis were studied by FISH using a custom BRAF break-apart probe set consisting of a 300kb probe targeting 5' BRAF (exons 1 to 7 plus 5' upstream sequence) and a 60kb probe precisely targeting 3' BRAF (exons 8 to 18). Probes were synthesized using Agilent's oligo-based SureFISH technology and labeled with FITC (5') and Cy3 (3') fluorophores. FISH was performed on FFPE tissue sections. For comparison, conventional FISH for BRAF gene rearrangement was performed using BAC-clone probes (Empire Genomics). Results: High resolution SNP-array analysis revealed intragenic copy number alterations in BRAF in both cases, showing gain of 3' portion of BRAF and loss or no copy number change of 5' BRAF with the change point mapped to intron 8. The boundaries of the gain segments were clearly defined by BRAF intron 8 and exon 18, whereas the 3' flanking sequence of BRAF was not involved in the gain in either case. FFPE FISH analysis using our custom oFISH probes successfully detected BRAF gene rearrangement with signals consistent with the pattern predicted from array analysis. In contrast, FISH using conventional probes failed to illustrate the BRAF alteration. Furthermore, a targeted RNA-seq assay (ArcherDX) identified in-frame BRAF fusion transcripts in both cases. Conclusions: oFISH probes can provide precise control over the sequence to be targeted, allowing for accurate analysis of regions as small as tens of kilobases. The resolution of oFISH is comparable to that of array-based technology in copy number analysis, therefore, can be used to develop complementary FISH assays accordingly. oFISH works well in FFPE tissue, and the implementation of oFISH probes in clinical diagnosis can significantly improve the precision and accuracy of FISH analysis in selected cases. Introduction: Success rate, accuracy and turnaround time of laboratory results being generated using increasingly complex technologies are of paramount importance in patient care and safety. Occupations that deal with risky technologies in high complexity environments (e.g. aviation, nuclear) apply proactive measures to avoid disastrous consequences of potential errors. We describe our proactive adoption of this model and development of strategies to achieve a High Reliability Organization (HRO) status in a clinical molecular diagnostics laboratory. Methods: We aligned our ongoing patient safety, quality and process improvement strategies with five HRO traits (see results) and Joint Commission Center for Transforming Healthcare's 3-domain (Leadership, Safety Culture, Robust Process Improvement) maturity model. New initiatives, engagement of staff, and proactive use of tools and monitors were added to achieve the goal of high reliability. Results: HRO trait 1-Sensitivity to operations: An operations command center allowed real-time monitoring of critical operational components (IT, staffing, reagents, instruments). An electronic dashboard allowed monitoring of production line, resources, quality, patient safety and training indicators. Industrial and quality engineering expertise were added to the laboratory operations. HRO trait 2-Reluctance to accept "simple" explanations for problems: Detailed root cause analyses and documentation systems allowed identification and remediation of multi-factorial issues. HRO trait 3-Preoccupation with failure: Systematic evaluation of laboratory processes allowed proactive implementation of preventive measures and detailed "Stop the Line" protocols. HRO trait 4-Deference to expertise: A patient safety team (frontline technologist on rotating membership) performed proactive compliance and patient safety risk assessments. A culture of safety was created through open discussions and active staff engagement. Also, a high level of informatics expertise was embedded throughout the lab. HRO trait 5-Resilience: Highly coordinated team efforts allowed the laboratory to continue offering high quality patient care during significant changes such as the laboratory move to a new location, national clinical trial adoptions, automation, implementations of a new LIS and a new institutional EMR, all within a total span of <1 year. Conclusions: Systematic evaluation of laboratory processes, engagement of staff and proactive implementation of laboratory-wide initiatives facilitate strengthening of operations, creates a culture of patient safety and helps to eliminate the risk of errors thereby providing high quality patient care and moving closer to the goal of becoming a High Reliability Organization. Introduction: Deregulation of the MET receptor tyrosine kinase is associated with aggressive phenotypes in a variety of human cancers, promoting proliferation, invasive growth and angiogenesis. Several types of genetic aberrations can drive MET deregulation, including gene amplification, overexpression, single nucleotide variants (SNVs), exon 14 skipping and fusions. MET is a target of intensive drug development efforts, although the various mutated forms of MET exhibit unique drug sensitivities. Therefore, detection of these mutations has the potential to guide treatments for cancers driven by MET deregulation. Next-generation sequencing (NGS) enables comprehensive detection of all mutation types from whole genomes and transcriptomes. However, low detection sensitivity, high input requirement and high costs render these approaches impractical for routine detection of mutations from low-input clinical sample types. Anchored Multiplex PCR (AMP) is a target enrichment strategy for NGS that, by its scalable and quantitative nature, is well suited to detect all modes of MET deregulation. Methods: AMP-based Archer VariantPlex and FusionPlex library preparation assays detect mutations from DNA and RNA, respectively. We designed AMP probes covering the MET gene to detect copy numbers and SNVs from DNA, and fusions, exon skipping and expression levels from RNA. Results: MET amplifications and resulting overexpression were detected in FFPE samples using the VariantPlex and FusionPlex kits and confirmed by FISH. Exon 14 skipping with concomitant splice site mutations was also detected using the Archer kits and confirmed by RT-PCR in both FFPE and cells. Interestingly, we detected a novel GTF2I:MET gene fusion and a Y1253D activating point mutation in FFPE samples. Conclusions: These results show that AMP-based VariantPlex and FusionPlex Assays enable comprehensive detection of multiple mutation types from low-input clinical sample types, such as FFPE specimens. Suspected B-Cell Clonality Y. Huang, A. Jacobsen, J. Panganiban, K. Hutt, D. Duong, J. Thornes, J. Miller, T. Stenzel Invivoscribe, San Diego, CA. Introduction: PCR-based methods targeting immunoglobulin heavy chain (IGH) frameworks 1, 2, and 3 (FR1, FR2 and FR3) are the current gold standard for the first line clonality testing in suspected B-cell proliferations. Recently, next-generation sequencing (NGS) based approaches for immune receptor geneshave been suggested to improve sensitivity and identify the specific V-J DNA sequence required to track clones in follow up testing. Here we report the development of a comprehensive IGH NGS assay which allows detection of all three FRs, simultaneously, in a single Illumina MiSeq run. Methods: The MiSeq IGH Assay consists of three master mixes, targeting FR1/J, FR2/J and FR3/J regions. The proprietary V and J consensus primers were designed and adapted to enable the PCR products to be sequenced on the MiSeq platform. Each MiSeq IGH FR master mix uses 24 indices, allowing analysis of 22 patient samples plus positive and negative controls. Multiplexed PCR was followed by amplicon purification using the AMPureXP PCR system. Purified equimolar amounts of amplicons from different samples and FRs were pooled to form a library. The harmonized and quantified library was sequenced using the MiSeq v2 Reagent kit (500 cycles). MiSeq output data was analyzed using proprietary bioinformatics software, which can sort the sequences by both index and FR, generate frequency distributions of V-J rearrangements, and determine the somatic hypermutation (SHM) rate of the IGHVregion based on the sequence of FR1/J amplicons. Results: The ability to detect FR1, FR2 and FR3 regions was verified by testing different known V-J rearrangement B cell line DNA. Cell line DNA, serially diluted into tonsil DNA, demonstrated the limit of detection (LoD) at 5% dilution with 50 ng DNA input. The assay achieved good linearity (R 2 > 0.95), and good reproducibility (<25% CV) was demonstrated over 2 operators, 2 MiSeqs and 2 lots of PCR master mixes. Preliminary testing of genomic DNA from peripheral blood (PB), bone marrow (BM) aspirates and formalin-fixed paraffin-embedded tissue (FFPE) demonstrated that the clonality detection missed by one FR could be detected by another FR, thus increasing overall detection rate. Conclusions: A comprehensive IGH NGS assay has been developed for the Illumina MiSeq platform that identifies clonal IGH V-J rearrangements and DNA sequences. The assay has demonstrated that combining FR1, FR2 and FR3 helps to decrease the false-negative rate due to somatic hypermutation in primer binding sites of the involved VH gene segments. Introduction: Accurate detection of low occurrence somatic mutations is paramount when using next-generation sequencing (NGS) based results to direct personalized management of cancer. Between March 2013 and March 2016, the Center for Personalized Diagnostics sequenced 3010 clinical cases on its amplicon based solid tumor panel (Sv1) which covers mutational hot spots across 47 genes. Our potential to include more genes, cover all exons and report variant allele frequencies (AF) of < 5% using low DNA input were limited due to the assay methodology and duplication errors of PCR and sequencing. To address these limitations, we designed our next solid tumor panel (Sv2) using Agilent Haloplex HS methodology that combines molecular barcoding and deep sequencing. We present our data from 48 samples and show that unique molecular identifiers can be used for accurate detection of variants below 5% AF with low DNA input. Methods: The final version of Sv2 captured 99.81% of the 583kbp target region designed to include all exons of 155 genes. We tested 48 DNA samples isolated from FFPE tissues (n=38), blood (n=6) and assay controls (n=4). DNA input range was 25ng to 100ng and all DNA were <30% degraded. Expected AF range was 0.05% to 100% for single-nucleotide variants (SNV) and insertion-deletions (indels). Data were analyzed using SureCall analysis software (v3.5.1.46) from Agilent Technologies. Variants were called only with unique reads, while at least three such reads were required to deem a variant to be true. Results: We successfully generated libraries at all DNA input ranges we tested. We observed >75% duplicate molecules when samples were >15% degraded and/or when DNA input was < 35ng. More than 80% of all target regions had mean coverage >200x when sample mean coverage was >500x and duplicate molecules were <60%. On-target specificity was consistently >98%. SNVs with expected AF <5% (n=28), had 100% sensitivity and specificity across all DNA inputs and sequencing runs. Below 1% AF, we had 100% concordance and accurately detected SNVs at 0.5% (n=3); 0.1% (n=7) and 0.05% (n=2). SNVs with AF >5% (n=38), had 100% concordance and on average AF's varied by 4.5% between runs. Sensitivity and specificity of indels and whole gene/exon alterations for all AF ranges are currently being determined. Conclusions: Accurate identification of somatic mutations in tumors by NGS is required for enrollment into mutation-specific treatment modalities and clinical trials. We show that our Sv2 can reliably detect lowlevel SNVs by tagging DNA fragments with more than a million unique molecular identifiers before PCR. This panel will greatly benefit patients whose tumors carry low-level actionable mutations and those who require monitoring for minimal residual disease. TT67. An Automated Cell-free DNA Extraction System for Clinical Diagnostics T. Ivanova, A. Fan, A. Yeo, P. Ariyaratne, G. Michel Vela Research Singapore Pte. Ltd., Singapore. Introduction: Circulating cell-free DNA (cfDNA) has emerged as an important biomarker in cancer diagnostics and non-invasive progression monitoring of various clinical conditions. This has resulted in the development of new in vitro diagnostics (IVD) cfDNA assays. Challenges encountered in these developments related to the efficient extraction of cfDNA from liquid biopsies, often yielding low quantities of highly fragmented DNA. As assays for cfDNA are typically intended to identify genetic variants present at very low allelic frequencies, many of the established detection technologies are driven to the edge of their performance. Methods: We developed a magnetic bead-based cfDNA extraction kit Sentosa SX cfDNA Kit (4x8) and optimized it for use on the Vela Sentosa SX101 platform. Sentosa SX101 is a CE-IVD certified robotic liquid handling system for nucleic acid extraction, PCR set-up and Next-Generation Sequencing (NGS) library preparation. We compared performance of the Sentosa SX cfDNA Kit (4x8) with a column-based cfDNA extraction kit. Integrity of cfDNA extracted by both methods was assessed using ALU repeats qPCR assay. Quality of the extracted cfDNA was tested using an NGSbased Sentosa SQ CRC Panel (4x8). Results: The Sentosa SX cfDNA Kit (4x8) utilizes 4 mL of human plasma as sample input and can process up to eight samples per run. In this pilot study DNA was extracted from plasma samples with 3.0, 1.5 and 0.75 ng spiked-in fragmented HCT116 gDNA (KRASG13D positive) using Sentosa SX cfDNA and column-based cfDNA extraction kits, respectively. Fragment size of HCT116 gDNA was ~170 bp (confirmed by Bioanalyzer). The ALU247/115 ratio for DNA extracted by the Sentosa SX cfDNA kit was 0.19-0.28 and for the columnbased extraction method >0.7 (expected ratio for cfDNA is less than 0.5 and for gDNA is 1.0). Amount and quality of DNA for all samples extracted by both methods was sufficient to prepare NGS libraries using Sentosa SQ CRC Panel (4x8). KRAS G13D mutation was detected in all samples extracted by the Sentosa SX cfDNA kit and no KRAS G13D mutation was detected in any of column-based extracted samples. Conclusions: The Sentosa SX cfDNA Kit (4x8) selectively extracts cfDNA over high molecular weight gDNA. The Sentosa SX cfDNA Kit (4x8) appears as an efficient and reliable solution for cfDNA extraction from human plasma samples. Integration into the Sentosa qPCR-and NGS-based workflows makes the Sentosa SX cfDNA Kit (4x8) a universal in vitro diagnostics tool, which can be used in combination with various IVD assays. Introduction: Plasma cell-free DNA is challenging to analyze due to a high degree of fragmentation and low concentration in circulation. Detection of low level mutations in cfDNA requires highly sensitive molecular detection methods. In this study, we compare the utility of the Bio Rad droplet digital PCR (ddPCR) and MassARRAY System, powered by UltraSEEK chemistry for detecting low level mutations in plasma cfDNA. Methods: cfDNA was extracted from plasma of 25 patients with metastatic solid tumors: 14 adenocarcinomas including 7 colorectal, 3 breast, 2 pancreas, 2 intestinal; 1 oral squamous cell carcinoma, 1 mucoepidermoid carcinoma, 6 melanoma, and 3 astrocytoma/glioblastoma using the QIAamp circulating nucleic acid kit within 2-16 hrs of specimen collection. Mutations in primary (n=14) or metastatic (n=11) tumors were tested using the Ampliseq Cancer Hotspot Panel v2 (Thermo Fisher). 10ng cfDNA was used for each assay. cfDNA genotyping was performed for 11 SNPs in IDH1 (R132H), PIK3CA (H1047R, E542K, E545K), KRAS (G12D, G12V, G13D), BRAF (V600E, V600M), NRAS(Q61R, Q61K) genes using ddPCR (Bio Rad) and 12 gene hotspot Ultraseek MassARRAY 1% MAF, all 37 expected calls were detected using the Accel TP53. Further, Accel TP53 assay detected 1 additional nonsense mutation (TP53. c.372C>A p.C124*) at 1.1% MAF (region not covered by TruSeq 53-gene hotspot amplicons). There was a good correlation of MAFs detected by the two assays (R 2 =0.98). The actual hands-on time was ~4.5 hrs (versus ~ 8 hrs on TruSeq 53-gene hotspot assay). Further optimization of the pipeline for improving the limit of detection is in progress. Conclusions: Accel TP53 is a sensitive, high-throughput "single-gene" NGS-based assay for detection of low-level TP53 mutations, suitable for MRD monitoring and early detection of sub-clonal mutations. Compared with the routinely used NGS-based multi-gene panels targeting hot-spot regions, the advantages of this assay include efficient resource utilization and comprehensive coverage of the entire coding region at reduced cost. The need to process small solid tumor specimens is increasing as early detection strategies become more effective and less invasive biopsy strategies are adopted. Moreover, the rapidly changing landscape of molecular testing argues that current strategies need to minimize sample input and preserve tissue for future testing. Processing small regions of tumor can be challenging as traditional manual macro-dissection and purification methods include multiple steps during which material can be lost. We evaluated an approach that combines automated tissue dissection with NGS library preparation directly from fragments of dissected tissue. Methods: FFPE tissue dissection was performed using the Roche Automated Tissue Dissection platform (Roche Diagnostics) which is currently in development. For DNA analysis, tissue fragments were processed using FFPE Direct Reagent (Thermo Fisher) and Ion AmpliSeq Cancer Hotspot Panel v2 (CHPv2) libraries were prepared and run on an Ion PGM system. For RNA analysis, tissue fragments were processed using the HTG Oncology Biomarker Panel (OBP) on the HTG EdgeSeq (HTG Molecular) and libraries were run on an Illumina MiSeq. Results: Automated tissue dissection was performed on slides of unstained sections from archived FFPE cases of melanoma, colon adenocarcinoma, pancreatic ductal adenocarcinoma and pancreatic neuroendocrine tumor. For DNA analysis, we dissected areas of tumor tissue from 50 mm2 to 1 mm2. We show that all CHPv2 libraries yielded excellent sequence data and that there was 100% concordance with previously reported mutations. We compared the performance metrics of libraries prepared from 1 mm2 to larger areas of tissue for the same case and found highly correlated amplicon coverages. For RNA analysis, we dissected areas of tumor tissue from 25 mm2 to 1 mm2. The HTG OPB libraries all passed QC metrics and yielded high quality gene expression data. In particular, samples processed from equivalent areas of tumor on adjacent sections yielded highly correlated data. Conclusions: The precision isolation of FFPE tumor provided by automated tissue dissection combined with extraction free DNA/RNA library preparation can be used to minimize material losses and reduce the amount of tissue input. The approach allows the processing of extremely small samples that are otherwise too small for standard NGS analysis. V. Kumar 1 , R. Webb 2 , N. Palomino 3 , E. Fang 1 , C. Perez 2 , J. Punia 1 , K. Fisher 1 , W. Parsons 1 , D. López-Terrada 1 , L. Suarez Ferguson 2 , A. Roy 1 1 Baylor College of Medicine, Houston TX; 2 Texas Children's Hospital, Houston, TX; 3 Texas Tech University, Lubbock, TX. Introduction: Internal tandem duplications (ITD) are a rare class of oncogenic genetic aberrations in cancers. The recent discovery of recurrent C-terminal ITDs in the X-linked BCL-6 co-repressor (BCOR) gene in diverse childhood solid tumors, including clear cell sarcoma of the kidney (CCSK), primitive neuroectodermal tumors of the brain, and soft tissue sarcomas, highlights the emergence of a useful marker for these diagnostically-challenging tumors. Herein, we describe the development and clinical implementation of a DNA-based fluorescent fragment analysis test to detect BCOR ITDs from formalin-fixed and paraffin-embedded (FFPE) and frozen tumor samples. Methods: Genomic DNA extracted from frozen and FFPE specimens of ITD-positive CCSK tumors (true positives, n=9), an ITD-negative CCSK and tumor-adjacent normal kidney specimens (true negatives, n=7), Wilms tumors (true negatives, n=23), and congenital mesoblastic nephromas (true negatives, n=9) were used. DNA was amplified with specific (6-FAM-labeled forward and unlabeled reverse) primers designed to amplify a fragment of the last coding exon of the BCOR gene that harbors ITDs. Fluorescent Fragments were detected using the ABI 3130 Capillary Electrophoresis system, sized with ABI GeneScan-500 LIZ ISS and analyzed with GeneMapper Software 5 (Life Technologies). Limit of detection studies were performed using serial dilutions of an ITD-positive case with known variant allelic fraction. Results were compared with Sanger sequencing and next-generation sequencing. Results: Fragment analysis-based ITD detection was successfully achieved with a total DNA input of >10ng (FFPE) and >1ng (frozen specimens), at an analysis threshold of 100 relative fluorescence units. Three different ITDs (90 to 96 bp) in eight of nine true-positive samples were detected by the fragment analysis assay (8/9: 89% clinical sensitivity), with the single discordant result being in a sample with degraded DNA. Clinical specificity of 100% was observed with all (39/39) true negative samples confirmed to be ITD negative. A high degree of concordance (0.97, Cohen's Kappa) was obtained between the two methods. Allelic ratio estimation using peak heights (ITD/wild-type) revealed a lower limit of detection of 5% tumor/normal DNA ratio. Both repeatability and reproducibility studies demonstrated sizing variation <0.15bp for the wild-type (expected size 287bp) and ITD fragments. Conclusions: The newly-developed fragment analysis test is a highly sensitive and specific clinical diagnostic tool that can be used on minimal archival FFPE specimens for the detection of BCOR ITDs in tumor specimens. Prospective application of this assay in a wide variety of BCOR ITDpositive tumor types would enable the evaluation of its clinical utility. Introduction: A majority of molecular diagnostic tests using formalin-fixed paraffinembedded (FFPE) tissue start from DNA extraction, a labor-intensive and timeconsuming step. To accommodate increasing demands of DNA-based testing on FFPE samples, we sought to decrease the labor and time of the deparaffinization step of extraction, and combine Zymo Research (ZR) Pinpoint isolation with Qiagen spin columns to facilitate automation of the DNA purification step on the QIAcube. Methods: FFPE slides were deparaffinized using a xylene alternative on the BiogenX Xmatrx NANO, followed by isolation of tissue from areas containing tumor cells using the ZR Pinpoint Solution and DNA extraction using ZR Pinpoint >40% Extraction Buffer and Proteinase K. The DNA purification was compared using the Zymo-Spin I columns (Zymo method) to Qiagen MinElute columns (hybrid method), which allowed us to use QIAcube. DNA quantity was determined with Qubit dsDNA BR Assay Kit and Agilent 2200 TapeStation. Results: Both methodologies require fewer unstained slides compared to our previously validated extraction and purification method, which involves scraping tumor from unstained FFPE slides, xylene deparaffinization, lysis, and purification on QIAamp Mini spin column. The ZR Pinpoint Solution allows for precise tissue selection and less tissue loss compared to scraping method. DNA yields were equivalent in both Zymo and hybrid methods, determined by Qubit BR assay and TapeStation (350-900 ng total), and are higher than our traditional scraping method. OD 260/OD280 ratio ranged from 1.85 to 2.00. Combining ZR Pinpoint DNA isolation with Qiagen DNA purification has enabled us to decrease hands-on time to 30 min from 60 min, when the QIAcube is incorporated; and the total extraction time to 6 hours from 18 to 20 hours. This method also allows for a decrease in error and recovery of more DNA. Conclusions: Automated deparaffinization of FFPE slides followed by ZR Pinpoint isolation of tumor tissue and Qiagen's spin column technology results in DNA yields similar to those of using ZR Pinpoint Isolation System alone. Pinpoint Solution allows for more precise tissue sampling and less loss of tissue than the traditional scraping method. Using Qiagen spin columns enables automation of DNA purification on the QIAcube, and results in significantly decreased hands-on time. Alternative deparaffinization eliminates xylene, thereby decreasing chemical exposure and disposal. This semiautomated hybrid method will increase efficiency as well as the quality of the workflow in our clinical laboratory. Rejection of Specimens Previously Judged Inadequate for Analysis S. Rapp, T.C. Greiner, A.M. Cushman-Vokoun University of Nebraska Medical Center, Omaha, NE. Introduction: Manual processing of specimens for Next-generation sequencing (NGS) is complex and time consuming. Automation can allow for better standardization and improved workflow. The Ion Torrent Chef requires 8 specimens be pooled on an Ion 318 chip, thus 16 specimens per 2 chip sequencing run. However, our lab volume is not sufficient to run two Ion 318 chips per week. We describe an alternative NGS automated workflow with combined data from two Ion 316 chips resulting in improved wet-bench turnaround time, reduced technician hands-on time, equivalent data and the ability to adequately analyze previously inadequate specimens. Methods: Twenty clinical specimens were previously sequenced using a manual workflow and the 50 gene Ion AmpliSeq Cancer Hotspot Panel, the Ion One Touch 2 and the Ion Torrent Personal Genome Machine (PGM). Eight additional specimens that previously failed or were excluded from clinical testing due to low quality or quantity DNA were also included in the study. Laboratory validation was then performed using the Ion AmpliSeq Kit for Chef DL8 and the Ion AmpliSeq Cancer Hotspot Panel v2 on the Ion Chef. Combined libraries were prepared in duplicate (8 specimens per Ion 316 chip including control) using the Ion PGM Hi-Q Chef Kit and Ion 316 Chip Kit v2 for clonal amplification, chip loading and sequencing. Aligned data from the same 8 specimens on two 316 chips were combined, using the Torrent Suite Software v5.0.4 with the variantCaller plugin v5.0.4.0, to achieve adequate coverage for detecting somatic mutations, rather than employing an Ion 318 chip. This strategy has not been previously reported in the literature or by the vendor. Results: For specimens with previously known clinical results (n=20), all expected mutations were identified. Additionally the 8 specimens, deemed inadequate for the manual workflow by lab experience, were successfully