|
GENETICS G01. C9orf72 GGGGCC Expansion Frequency amongst Patients with Negative HTT CAG Expansion Testing G03. Validation of an Orthogonal Next-Generation Sequencing Assay for the Mitochondrial Genome G04. Evaluation of CADD Scores for Identification of Pathogenic Variants in Non-Coding Regions in a Germline Cancer Risk Panel -Thalassemia Genotype Combinations Involving the PolyA Signal Mutation in -Globin Uncovered in the Population of Bahrain
Abstract
Introduction: C9orf72 intronic GGGGCC hexanucleotide expansion has been recognized as the most common genetic alteration associated with familial and sporadic amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD). Approximately 1% to 7% of patients with Huntington disease (HD) phenotype do not carry the CAG expansion within the HTT gene. Additionally, there might be phenotypic overlap between ALS, FTLD and HD, since these are neurodegenerative diseases that may clinically present with motor symptoms associated with variable levels of cognitive impairment. Recent studies have demonstrated C9orf72 expansion as the most common identified genetic alteration in patients with HD-like phenotype and negative HD genetic testing, with a prevalence of approximately 2% to 5% based on European studies. In the present study, we investigated the prevalence of the C9orf72 expansion in individuals with negative testing for HTT gene CAG expansion in a North American tertiary referral laboratory. Methods: De-identified DNA samples (n=236) that previously tested negative for CAG expansion of the HTT gene were evaluated by genotyping PCR. C9orf72 hexanucleotide repeat region was amplified using fluorescentlylabeled primers flanking the hexanucleotide repeat region and fragments were visualized by capillary electrophoresis. Allele size was determined using GeneScan 500 ROX dye size standard and GeneMarker software. Follow-up Southern blot analysis using XBaI and HindIII restriction enzymes separately and a 32 P-labeled oligonucleotide probe to specifically detect a region immediately upstream of the hexanucleotide repeat region was performed for apparently homozygous samples and those with intermediate (20 to 29 repeats) alleles. Results: Of the 236 cases, 3 (1.3%) tested positive for a C9orf72 expansion. Two cases had >2000 repeats, a repeat expansion range typically seen in our positive ALS/FTD samples. A single case had approximately 80 to 100 repeats. Of the remaining 233 cases, 1 (0.4%) had an intermediate allele (22 repeats) and 232 (98.3%) had <20 repeats, with 2 and 5 repeats as the most common and the second most common alleles, respectively. Conclusions: C9orf72 expansion was identified in a small subset of patients with negative HD genetic testing in a North American tertiary referral laboratory. This finding suggests that C9orf72 expansion is a genetic alteration that may be identified in patients in which HD is a clinical differential diagnosis prompting genetic testing, and supports consideration of C9orf72 expansion evaluation for individuals with HD-like phenotype and negative HTT gene expansion testing. Introduction: Clinical next generation sequencing (NGS) is rapidly becoming an established diagnostic tool. Because positive test results can influence significant medical decisions, many labs rule out false positive calls by confirming variants with an orthogonal method such as Sanger sequencing. The collective experience of many laboratories doing so is that the great majority of NGS positives indeed confirm, and therefore the need to broadly apply orthogonal confirmation has been questioned. However, identifying the highest confidence NGS calls and quantifying our degree of confidence in these calls can be challenging. A cross-laboratory framework to rigorously address this challenge has not yet been established but has clear value to the medical community. Methods: We report a framework that can apply across different assay targets, NGS protocols, bioinformatics algorithms, and QC criteria. In brief, our framework involves: 1) assembling confirmation data from each laboratory across clinical and reference samples, 2) determining the QC characteristics of subsets of variants that each data set adequately describes, and 3) using proper statistical metrics to both establish reliable filtering thresholds and to quantify the performance of variants that meet these thresholds. We propose the use of multiple threshold tiers: a strict tier, over which the likelihood of a false positive is exceedingly low, and a lower tier established to ensure sensitivity. Reported variants from the lower tier would need orthogonal confirmation whereas confirmation of the higher tier variants may depend on clinical factors (e.g. actionability or population prevalence) and/or operational criteria (e.g. positive sample tracking). Results: To date, we have applied this framework to data from 4 different laboratories. For example in one data set for a 184 gene NGS panel, 3082 variants were assembled as a complete and representative set of NGS calls that a) met that lab's top tier QC criteria (eg, read depth and other factors), and b) had high-quality orthogonal data available. All 3082 were confirmed as true positives. Further bioinformatic analysis identified a subset of 2394 with homogeneous properties in terms of genomic context, variant types and local sequence properties. The observed analytic false discovery rate (FDR) for the 2394 remains 0.0% and our statistical estimate is that these data demonstrate an FDR of at most ~0.1% for such variant calls (at p=0.05). Conclusions: Our framework is the first effort to combine data across clinical NGS labs to help evaluate the value of orthogonal confirmation and determine the appropriate burden of proof to potentially change practice in distinct cases. We believe these results and this framework can contribute to the ongoing community dialog on this subject. Introduction: Identification of mutations in the mitochondrial genome is critical to confirm multiple mitochondrial disorders. A next-generation sequencing (NGS) assay, using an orthogonal confirmation approach, was validated to sequence the entire mitochondrial genome. This assay detects single nucleotide variants (SNVs), up to 38 bp insertions, up to 44 bp deletions, and large deletions greater than approximately 500 bp. Methods: The mitochondrial genome was amplified in two overlapping products by long-range PCR. PCR products were quantified and mixed equimolar. NGS was then performed using two orthogonal methods: 1) TruSeq Nano library preparation sequenced on the Illumina MiSeq (primary) and 2) Ion Plus Fragment library preparation sequenced on the Ion Torrent PGM (Life Technologies) (confirmatory). Large deletions detected by the MiSeq were confirmed by agarose gel electrophoresis. Haplogroup classification was performed using Phylotree and HaploGrep software. To demonstrate accuracy, we tested 127 DNA samples (extracted from whole blood, blood spots, cultured cells, and muscle biopsy) previously genotyped using established methods. Precision was assessed using eight samples run in triplicate both on the same run and across three separate runs. Limit of detection (LOD) studies were performed to assess: 1) minimum input of DNA for long-range PCR, 2) minimum long-range PCR product input for library preparation, 3) minimum detectable variant frequency, and 4) 95% confidence of detection of variant frequencies. Results: Previously identified variants were detected for 126 out of 127 samples during validation. After excluding the discordant sample due to a sample mix-up, results were 100% concordant. 100% of variants detected by the MiSeq and within the LOD were orthogonally confirmed by the Ion Torrent PGM. Of the large deletions detected by the MiSeq, 100% were confirmed by agarose gel electrophoresis. The samples run for the precision study were in fact mixtures of samples, so as to create a range of variant frequencies (100% down to <1%). All results were 100% concordant among replicates down to the LOD. The LOD was determined to be 1.875 ng input of DNA for long-range PCR (standard protocol: 30 ng) and 25 ng input PCR product for library preparation (standard protocol: 100 ng). The minimum detectable variant frequency was determined to be 6% for SNVs and insertions/deletions, and 20% for large deletions. The lower 95% confidence levels were calculated for three different variant frequencies: 16.8% to 24.8% for 20% variants, 8.1% to 12.1% for 10% variants, and 4.2% to 6.6% for 5% variants. Conclusions: We have demonstrated validity of an NGS assay for sequencing and haplogroup classification of the complete mitochondrial genome. Introduction: Several in silico tools have been shown to have reasonable predictive value in classifying sequence variants in coding regions, but no method has been shown to perform this function in non-coding regions. Combined annotation dependent depletion (CADD) generates predictive scores for single nucleotide variants (SNVs) all areas of the genome, including non-coding regions. It has been shown that CADD compares well with other tools for nonsynonymous variants, but no one has explored whether this method can be used to effectively stratify variants in non-coding regions. Methods: We evaluated all unique SNVs with CADD scores for 624 patient samples submitted for germline mutation testing in a cancer risk gene panel.
entries in the 1000 genomes database. We used the Wilcoxin Rank Sum test to compare distribution of CADD scores of rare SNVs to that of common SNVs in our patient population and that of all possible SNVs defined by Kircher et al, stratifying by genomic region (eg, downstream, intergenic, etc.). We visually compared distributions that were found to be significantly different to determine if rare SNVs were appropriately overrepresented at high CADD scores. Results: We identified 783,387 total variants with CADD scores in the 624 samples. 12,391 variants were unique, and 8929 of these were rare. The median CADD scores of intronic and nonsynonymous variants were significantly different between rare and common SNVs (p<0.0001). On visual inspection, rare variants were overrepresented at higher CADD scores for these two regions. Rare downstream, intergenic, intronic, upstream and 5' UTR rare variants were underrepresented at the lowest CADD scores and overrepresented at the highest CADD scores compared to all possible variants (p<0.0001 for difference in median). A similar pattern was seen between common and all possible SNVs for intergenic and upstream regions (p<0.0001 for difference in median). When the rare SNVs with the highest CADD scores were evaluated, we found only one definitively pathogenic variant, which was a nonsynonymous SNV in a coding region. On further evaluation, other variants with disproportionately high CADD scores were deemed unlikely to be disease causing in the context of patients' cancer phenotypes. Conclusions: We found that the median CADD scores of rare SNVs are significantly 754 AMP Abstracts jmd.amjpathol.org ■ The Journal of Molecular Diagnostics higher for intronic and nonsynonymous regions compared to common variants in our patient population. However, the rare intronic SNVs with the highest CADD scores were unlikely to be disease causing in our patients. Thus, we determine that CADD scores are not useful for identifying pathogenic mutations in non-coding regions in clinical cancer risk samples.
of SYBR Green. Genomic DNAs with known SMN1 and SMN2 copy numbers were obtained from Coriell/NIGMS to standardize the assay, after which two other locally obtained DNAs were selected and defined as calibrator and as positive control for all subsequent routine quantifications, all of which were performed in triplicate having been setup robotically. The raw qPCR data were analyzed using LinReg which calculates the Cq and amplification efficiencies for each reaction; these two variables were then used to deduce to the SMN1 and SMN2 copy numbers using the data obtained from the calibrator sample and the reference gene targets. Results: The amplification efficiencies averaged 96% for the four targets. Genomic DNAs from 210 unrelated normal individuals were typed anonymously and the distribution of SMN1 and SMN2 copy numbers was in agreement with previously published data; three SMA carriers were detected in this sample. SMN2 copy numbers varied between 0 and 3 per person. Two unrelated infants with ages ~6 months with a clinical diagnosis of likely SMA were confirmed to be homozygote affecteds; their parents were confirmed to be heterozygote carriers. Conclusions: We describe a reproducible and inexpensive qPCR assay for quantifying SMN1 and SMN2 copy numbers. This test does not detect SMA '2+0' carriers nor carriers having deleterious SMN1 point mutations which represent ~5% of SMA carriers. In a normal working day using 96-well PCR plates, 25 experimental samples in addition to the calibrator and control samples can readily be processed.
jmd.amjpathol.org ■ The Journal of Molecular Diagnostics 78°C to 84°C and HLA-ABC melted from 86°C to 92°C. Eleven random unrelated individuals produced unique melting patterns and most individuals of a family of eleven siblings were distinguishable. The chance of 2 siblings matching at the 3 loci is 1 in 64. The chance of unrelated individuals matching is less than 5E-6. Conclusions: This system demonstrates the ability to rapidly match DNA samples in less than 30 minutes by rapid cycle PCR and high resolution melting in a closed system.
and VKORC1: Implication of Clinical Utility for Pharmacogenomics Testing H. Yao, T. Hsieh, J.A. Byers, M.A. Lettunich, K.T. Fobes, L.H. Hooper, W. Mo, J.M. Harrington, C.J. Sailey Molecular Testing Labs, Vancouver, WA. Introduction: Pharmacogenomics (PGx) is increasingly gaining acceptance as a valid method to utilize an individual's genotype to prescribe medications or medication doses. The goal is to identify important genetic variants that may affect patients' responses to drugs in order to optimize the efficacy of pharmacologic treatment and to minimize adverse drug reactions. The Cytochrome P450 (CYP) enzyme family, especially CYP2D6, CYP2C9 and CYP2C19, metabolize the majority of all prescribed drugs; multiple guidelines have been established for dosing medications based on the allele variants of these genes. One additional enzyme, VKORC1, is of interest because, together with CYP2C9, it is useful in determining the optimal initial dose of warfarin, a widely prescribed anticoagulant for preventing heart attacks, strokes and blood clots. Methods: This study involved 34,216 patients tested at Molecular Testing Labs (MTL) within a one-year period. All individuals have a complete data set of ethnicity, age, and valid genetic testing results. Alleles for CYP2C9, CYP2C19 and VKORC1 were determined using Taqman genotyping assays and CYP2D6 was tested using the Luminex xTAG CYP2D6 v3 assay. Analysis of variance and t-tests were used for statistical analysis. Results: Patients who ordered PGx testing ranged from one month to 110 years old, with 75.85% of them above 50 years old. We assessed how many patients have altered enzymatic activity. Genotyping analysis of all patients showed: poor metabolizer (PM) and ultrarapid-metabolizer (UM) of CYP2D6 was 4.98% and 4.87% respectively; PM and UM of CYP2C19 was 2.7% and 28.13%; PM of CYP2C9 was 3.12%, and 16.23% of patients were VKORC1 homozygous mutants. We analyzed ten medications that patients reportedly were taking and can be affected by the PGx testing results according to the Clinical Pharmacogenetics Implementation Consortium. To give an example, 42.59% (227/533) of patients who were taking Amitriptyline also showed altered enzymatic activities that would require an altered dose or an alternative drug. For Clopidogrel it was 58.95% (448/760) and for warfarin it was 100% (518/518). Finally, we used all these patient data to analyze allele distributions across different ethnicity groups. For example, although 6.47% of Caucasians were PM of CYP2D6, only 0.47% of Asians fell into this group. Conclusions: These data imply that PGx testing is a potential benefit to a large number of people who are taking multiple medications concurrently and may require altered dosing by their physicians. Ethnicity should also be taken into account as it often correlates with specific genetic variations.
Duplications M.P. Borgman 1 , J. Forcellini 1 , D. Pierce 1 , T. Hegerich 2 , T. Hartshorne 2 , K.K. Reynolds 1 , M.W. Linder 1 1 PGXL Laboratories, Louisville, KY; 2 Thermo Fisher Scientific, South San Francisco, CA. Introduction: CYP2D6 (2D6) gene duplications are observed by our laboratory at approximately 6% frequency. Phenotypic interpretation of 2D6 duplications is inconclusive when the genotype includes alleles with unequal activity and both alleles are candidates for duplication. A total copy number (CN) greater than three also confounds the interpretation. We observe these inconclusive phenotypes in 2.5% of patients tested. The inability to accurately define active CN impacts clinical guidance for many medications in pain management, psychiatry, and cardiovascular care. Digital PCR (dPCR) allows the CN of specific alleles to be directly determined. The purpose of our study was to assess this technological advancement and its feasibility for clinical testing. Methods: Initial 2D6 genotyping was performed on genomic DNA isolated from buccal swabs using TaqMan DME and CNV assays. Twenty-three representative 2D6 heterozygous samples with duplication were tested. Allele-specific digital PCR (ASdPCR) was performed in duplicate by restriction digesting genomic DNA followed by loading onto QuantStudio 3D Digital PCR 20K chips with 2D6 TaqMan DME assays targeting -1584C>G, 100C>T, 1023C>T, 1846G>A, 2850C>T, or 4180G>C variants depending on haplotype in question. After 40 cycles and end-point read, data was uploaded to QuantStudio 3D AnalysisSuite cloud software and analyzed with the Relative Quantification application. Analysis of % FAM positive reactions over Total FAM plus VIC reactions (%FAM/Total) for the allele-specific SNP was used to determine the duplicated allele. Results: ASdPCR detected the presence of each allele-specific SNP as expected and discerned the duplicated allele for 18 of 23 samples with a success rate of 78%. Successful samples with CN=3 were evident by a %FAM/Total ratio of approximately 33% or 66% depending on which location of FAM label and which allele was present twice. The remaining 5 samples with CN=3 generated 50% FAM/Total signal and thus were indeterminate. Interestingly, a *2/*4 sample with duplication and possible Extensive (EM) or Intermediate Metabolizer (IM) phenotype returned a total copy number of four and *2/*2/*2/*4 genotype predictive of Ultrarapid metabolizer status. Conclusions: ASdPCR is a feasible technology with simple workflow and reasonable DNA input requirements to decipher 2D6 duplication events in heterozygous samples. In samples with CN>3, possible phenotypic consequences may range from IM to UM depending on total active copy number. This highlights the need for accurate active CN detection and strong correlation between qPCR and dPCR CN analysis. Challenges remaining include optimizing methods for additional 2D6 variants, developing assay validation guidelines, and quality monitoring. A. Stanley, M. Beoris, A. Austria, A.A. Baca, J. Amos Wilson, J.A. Garces, A. Lukowiak AltheaDx, San Diego, CA. Introduction: Clopidogrel is a platelet inhibitor indicated for acute coronary syndrome, recent myocardial infarction, recent stroke or established peripheral arterial disease. In the package insert, clopidogrel is contraindicated for use (Warnings and Precautions) with omeprazole, esomeprazole, and other CYP2C19 inhibitors which are known to significantly reduce the antiplatelet response of clopidogrel. In addition, a boxed warning was issued by the FDA in 2010 highlighting that alternative treatments should be considered for CYP2C19 intermediate metabolizers (IMs) and poor metabolizers (PMs), which was further supported by the Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines in 2011. In this study we analyzed over 2000 patients on clopidogrel after 2013 for their CYP2C19 metabolizer status as well as their concomitant medications. Methods: Over 2000 patient samples were collected using buccal swabs and DNA extracted using a Qiagen based extraction platform. Highthroughput end-point and real-time fluorescent PCR methodologies were used to genotype patients for 80+ single nucleotide variants (SNVs) relevant to the pharmacokinetic and pharmacodynamic pathways of drug metabolism and action. Genotyping results, along with concomitant medications, were analyzed in aggregate to provide clinical recommendations for the use of multiple antithrombotic drugs, including clopidogrel. Results: Retrospective pharmacogenetic analysis demonstrated that ~30% of the population currently on clopidogrel were classified as either having an IM or PM status and would have been recommended for alternative antiplatelet therapy per CPIC guidelines. Additionally, ~14% of the same population was listed as concomitantly taking omeprazole which is contraindicated in the clopidogrel package insert. Interestingly, in many cases, non-clopidogrel alternatives for antiplatelet therapies were available which lacked gene-drug or drug-drug avoidances. Conclusions: To truly maximize the benefits of pharmacogenetic testing, medical practice will need to evolve to see pharmacogenetic testing as a proactive, rather than reactive, tool and service providers will need to complement gene-drug interactions with adverse drug-drug interactions to provide physicians with the most comprehensive opportunity for improving clinical outcomes.
Immunodeficiencies D.W. Close 1,2 , R. Adams 2 , E.M. Coonrod 2 , K.V. Voelkerding 1 , A. Kumanovics 1,2 1 ARUP Institute for Clinical and Experimental Pathology, Salt Lake City, UT; 2 University of Utah School of Medicine, Salt Lake City, UT. Introduction: Primary immunodeficiencies (PIDs) are a heterogeneous group of disorders that affect immune system development and function. PID severity varies widely and manifests as an array of syndromes and phenotypes. PIDs affect ~1:1,200 people with disease-causing variants identified for hundreds of different genes. Identification of genetic cause is critical for definitive diagnosis, genetic counseling, prognostication, and also because treatment, such as bone marrow transplantation or replacement therapy, can depend on the particular gene. Genetic diagnosis of PIDs is hindered by the fact that there is a significant heterogeneity in the presentation of patients with mutations in the same gene, and that the mutations in multiple genes can cause same or similar PIDs. The challenge of testing dozens of potential genetic contributors efficiently can be overcome using highly parallel next-generation sequencing (NGS) and targeted custom gene arrays. Methods: More than 200 genes reported to cause, or be associated with PIDs were selected and captured using Agilent's SureSelect custom capture assay followed by sequencing on the Illumina HiSeq platform. Exon, exon junctions, and custom regions containing known disease causing variants were targeted resulting in a total capture size of 1.0 Mb. Multiple runs were performed to assess precision of the capture and sequencing. A variety of samples were used to assess accuracy including the well-characterized HapMap NA12878, unaffected individuals, and at least 25 samples from individuals with PID or PID-related disease. 16 samples were multiplexed per HiSeq run and analyzed using a BWA-GATK pipeline. Results: SureSelect libraries were prepared successfully for 30 of 32 samples and sequencing revealed consistent distribution of coverage across targeted regions with high levels of precision between runs. Greater than 98% of the exonic regions are covered with 15 reads or more. Gaps in coverage are addressed using Sanger sequencing. The primary cause for gaps in coverage relates to pseudogenes which are addressed using long-range PCR and supplemental sequencing. On average, over 400 variants were called per sample and are part of ongoing validation and J.A. Lefferts, J.L. Gonzalez, A.R. Schned, S.A. Turner, L.J. Tafe Dartmouth-Hitchcock Medical Center, Lebanon, NH. Introduction: Products of conception (POCs) are routinely processed for histologic examination and aneuploidy analysis may subsequently be needed to rule out a molar pregnancy or other chromosomal abnormality based on histologic findings and/or clinical history. Aneuploidy analysis by fluorescence in-situ hybridization (FISH) is available for FFPE but this typically is limited to a small set of chromosome probes. Here we present a proof of principal study using molecular inversion probe technology (MIP) to perform genome-wide copy number analysis for aneuploidy detection in FFPE POC tissues. Methods: FFPE POC tissues were macrodissected from 8 unstained, 4 micron slides and DNA extraction was performed with the QIAamp DNA FFPE Tissue Kit (Qiagen). The DNA samples were processed using the Affymetrix OncoScan FFPE Array according to the manufacturer's protocol. Nexus Express and Affymetrix software packages were used for analysis. The majority of cases had previously been tested by FISH for chromosomes X, Y, 13, 15, 16, 18, 21 and, 22 or by cytogenetic analysis. Results: Seven FFPE POC tissues were selected for OncoScan analysis. Prior analysis had revealed trisomies in 16, 17, 18, 21 and, 22 (1 case each), a XXY partial mole and a diploid non-molar abortus. All previously identified trisomies were accurately detected with the OncoScan Array. In addition, testing by OncoScan revealed the following,a previously undetected results: trisomy 20 in the POC with trisomy 18; long continuous stretches of homozygosity (LCSH) totaling 70 Mb between chromosomes 3 and 13 in the case with trisomy 21; and a copy number gain of chromosome 3 (likely trisomy) in the case with no abnormalities detected with FISH. Although triploidy can often be identified with a SNP-based array, it was difficult to recognize the triploidy due to significant maternal cell contamination (MCC) from the decidua tissue also present in the same tissue block. With review of the histology slides, an accurate interpretation could be derived. Interestingly, a relative gain of chromosome 18 (copy number 4) was also detected in this triploid sample. Conclusions: OncoScan FFPE Array can successfully be used to detect single chromosome trisomies in FFPE POCs. Accurate determination normal chromosome complement and of triploidy (partial moles) can be challenging due to MCC and careful macrodissection is needed to enrich for fetal tissues. Additional samples will be tested to further expand our experience with this assay in FFPE POCs. Compared with FISH, OncoScan is able to detect genome-wide copy number changes and can also detect LCSHs which could be associated with uniparental disomy or consanguinity. M. Telatar, H. Hong, R.K. Pillai, C. Louie, V. Mehta, D. Gu, C. Fong, J. Weitzel, M. Afkhami, P. Aoun City of Hope National Medical Center, Duarte, CA. Introduction: We report development of a clinical validation protocol for NGS-based targeted resequencing of a panel of 28 genes, including BRCA1 and BRCA2, which predispose to increased risk of breast, ovarian and/or endometrial (BOE) cancers. Methods: Libraries were constructed and enriched for genes of interests using a hybridization capture-based enrichment system (Agilent SureSelectQXT). We designed custom capture "baits" targeting all exons in 28 genes associated with hereditary BOE cancers. Libraries were sequenced on MiSeq using 300-cycle Reagent Kit V2. NextGENe (SoftGenetics, LLC) was used for alignment, variant calling, and filtering. A total of 66 archival DNA samples were used in the validation to assess the overall performance metrics. Analytical accuracy was assessed by comparing with Sanger sequencing of all coding exons of 28 representative genes for each sample as reference gold standard. Reproducibility was determined by comparison of concordance within run and across runs. Accuracy was determined by screening 19 archival DNA samples with previously identified BOE cancer associated pathogenic mutations. Results: We achieved about 40% to 50% ontarget enrichment efficiency, and >99% of the target exons were covered at >100X at all bases, and > -homopolymer stretch regions, we observed 100% concordance between NGS and Sanger sequencing. Our reproducibility was > 99% for both intra-run and inter-run. We demonstrated 100% accuracy for all known pathogenic mutations in 19 blinded positive samples (6 deletions of up to 56bp, 4 insertions of up to 19bp, 3 indels, 6 substitutions). Our protocol was also able to identify large deletions, such as a 56bp deletion in BRCA1. Conclusions: We developed a clinical protocol for capturebased enrichment, NGS targeted sequencing and detection of inherited mutations in breast, ovarian and endometrial cancers with high accuracy and precision. Our protocol allows accurate identification of not only point mutations but also a wide spectrum of challenging mutation types including large deletions, insertions, and indels of varying sizes.
E. Cho, J. Jang, J. Lee, T. Lee, J. Park Green Cross Genome, Yongin, Kyunngi-do, Korea. Introduction: Non-invasive prenatal testing (NIPT) has been showing more than 99% of sensitivity and specificity in detecting trisomy 21, 18 and 13. However, for sex chromosomal aneuploidy (SCA), the specificity has been lower than that of those three autosomal trisomies. Recent reports explains that some NIPT false positive results were caused by maternal copy number variant (CNV) and fetal/placental mosaicism. Here, we present a false positive fetal SCA (47, XXY) result caused by maternal X chromosome copy number variation. Methods: A pregnant woman who is suffered from placental insufficiency at 17 weeks donated her blood to conduct NIPT research done by Green Cross Genome. Trisomy of 21, 18 and 13 was not detected; however, it was suspicious of having SCA (47,XXY). Results: G scores for X and Y chromosomes were -0.86 (normal female fetus: Gx:-2.58~3.04, normal male fetus:Gx<-2.58) and 6.31 (normal male fetus:Gy>5, normal female fetus:Gy<5). Real-time PCR for SRY gene confirmed male fetus and estimated fetal fraction was about 10%. To rule out segmental maternal CNV, which could skew the G score, the bin median value is calculated from the median of zscores measured per 50 Kb bin. Together, visualization was performed by plotting the per-bin z-score across the X chromosome. As a result, the bin median of chromosome X was not increased compared to that of normal male pregnancy and plotting of the z-score of the X chromosome was indicative for the presence of a maternal CNV across about 550 Kb region. Indeed, subsequent high resolution microarray analysis of maternal white blood cell DNA confirmed a duplication of approximately 521 Kb at chromosome Xq28 including MECP2, the common duplication region for male patients with severe intellectual disability, hypotonia at birth, progressive spasticity, epilepsies, recurrent infections, and early death.Conclusions: Zooming in to subchromosomal bins allows us to detect false positive results of NIPT. In case of NIPT results suggesting fetal trisomy including SCA, the confirmation of the presence of maternal CNV would be recommended to avoid unnecessary referral for invasive testing and also to evaluate the risk to the fetus.
H. Hong, M. Telatar, R.K. Pillai, C. Louie, V. Mehta, D. Gu, C. Fong, M. Afkhami, P. Aoun City of Hope National Medical Center, Duarte, CA. Introduction: We report development of a clinical validation protocol for NGS-based targeted sequencing of 11 genes which predispose to increased risk of hereditary colon cancers including Lynch Syndrome. Methods: Libraries were constructed and enriched for genes of interests using a hybridization capture-based enrichment system (Agilent SureSelectQXT). We designed custom capture "baits" targeting all exons of 11 genes associated with hereditary colon cancers. Libraries were sequenced on MiSeq using 300-cycle Reagent Kit V2. NextGENe (SoftGenetics, LLC) was used for alignment, variant calling, and filtering. A total of 66 archival DNA samples were used in the validation to assess the overall performance metrics. Analytical accuracy was assessed by comparing with Sanger sequencing of all coding exons of 25 representative genes for each sample as reference gold standard. Reproducibility was determined by comparison of concordance within and across runs. Results: We achieved 40% to 50% on-target enrichment efficiency, and >99% of the target exons in the 73 genes panel were covered at >100X at all bases, and > -homopolymer stretch regions, we observed 100% concordance between NGS and Sanger sequencing. Our reproducibility was >99% for both intra-run (total 1023 variants) and inter-run (total 1476 variants). We demonstrated 100% accuracy for all known pathogenic mutations in 11 blinded positive samples (5 deletions of up to 16bp, 2 insertions of up to 19bp, 2 indels, 2 substitutions). Although large deletions and insertions represent some of the most significant challenges in target enrichment capture and NGS alignment bioinformatics process, our NGS protocol successfully detected a 16bp deletion in APC and a 19bp insertion in MSH6. Conclusions: We have developed a clinical protocol for genomic capture-based enrichment, NGS-based targeted sequencing and detection of inherited mutations involved in hereditary colon cancers with high accuracy and precision. Our protocol allows accurate identification not only of point mutations but also a wide spectrum of challenging mutation types, including large deletions, insertions, and indels of varying sizes.
jmd.amjpathol.org ■ The Journal of Molecular Diagnostics CNVs across all disease causing genes is currently unknown, due in part to technical limitations. NGS-based CNV detection is now being used to identify deletions and duplications, allowing laboratories to target a wider variant spectrum. We developed and validated VisCap, an open tool for calling germline CNVs from targeted NGS data. Application to diagnostic gene panels for two inherited disorders illustrates a high overall utility but also critical limitations. Methods: VisCap uses the fraction of overall sample coverage for targeted genomic intervals to call CNVs with log2 ratios exceeding user-defined thresholds. Other features include automated data QC, iterative normalization to exclude low quality samples, and data visualization for easy scoring. We tested VisCap using 14 known CNVs and evaluated its performance in a cohort of 1,104 diagnostic cases using a visual scoring system and droplet digital PCR. The tool's clinical impact was assessed in >600 individuals with inherited hearing loss and in >1,400 individuals with inherited cardiomyopathies. Results: VisCap correctly identified all 14 previously identified CNVs. For the 1,104 diagnostic cases, 27 CNV calls were tested using ddPCR. The tool's accuracy showed a strong correlation with CNV length with none of 13 CNVs 3 exons or less in size and 71% (10/14) of larger CNVs confirmed. All 17 false positives could be removed by visual scoring, highlighting the need for manual review and orthogonal confirmation. Additionally, 72 copy number invariant regions of 1-2 exons were confirmed copy number neutral. 12.8% of validation samples did not pass automated QC procedures and could not be scored. Diagnostic detection rates varied greatly, with a clear added benefit for hearing loss associated genes (19% of all positive cases had intragenic CNVs leading to 4.4% added sensitivity) but limited utility for inherited cardiomyopathies where clinically significant CNVs were identified in only 9 of 1425 (0.6%) cases. Conclusions: VisCap is a sensitive method for inferring CNVs from targeted NGS data, although diagnostic implementation is challenging due to high failure rates and low specificity for small alterations. Currently, both visual scoring and orthogonal confirmation are necessary to ensure clinically acceptable accuracy, creating a high operational and financial burden. However, the ability to call CNVs from targeted NGS data has high utility for some disorders and reveals a major contribution of intragenic CNVs, which have historically been in the "blind spot" of diagnostic gene panels in molecular and cytogenetic laboratories.
H. Hong, M. Telatar, R.K. Pillai, C. Louie, V. Mehta, D. Gu, C. Fong, M. Afkhami, P. Aoun City of Hope National Medical Center, Duarte, CA. Introduction: We report development of a clinical validation protocol of NGS-based targeted resequencing of a panel of 22 genes associated with hereditary connective tissue disorders including Marfan and Marfan-like Syndromes. Methods: Libraries were constructed and enriched for genes of interests using hybridization capturebased enrichment system (Agilent SureSelectQXT). We designed custom capture "baits" targeting all exons of 22 genes associated with connective tissue disorders. Libraries were sequenced on MiSeq using 300-cycle Reagent Kit V2. NextGENe (SoftGenetics, LLC) was used for alignment, variant calling, and filtering. A total of 66 archival DNA samples were used in the validation to assess the overall performance metrics. Analytical accuracy was assessed by comparing with Sanger sequencing of all coding exons of 22 representative genes for each sample as reference gold standard. Reproducibility was determined by comparison of concordance within run and across runs. Accuracy was determined by screening 4 samples with previously identified connective tissue disorders associated pathogenic mutations. Results: We achieved 40% to 50% on-target enrichment efficiency, and >99% of the target exons in the 22 genes panel were covered at >100X at all bases, -homopolymer stretch regions, we observed 100% concordance between NGS and Sanger sequencing. Our reproducibility was greater than 99% for both intra-and inter-run. We demonstrated 100% analytical accuracy for all known pathogenic mutations in 4 blinded positive samples (1 deletion of 2bp in FBN1, 1 insertion of 1bp in COL5A1, 2 substitutions in COL3A1 and FBN2, respectively). Conclusions: We developed a clinical protocol for genomic capture-based enrichment, NGS-based targeted sequencing and detection of mutations involved in inherited diseases including hereditary connective tissue disorders with high accuracy and precision. Our protocol allows accurate identification of a wide spectrum of mutation types, including point mutations, deletions, insertions, and indels of varying sizes. Setting J.M. Devaney, B. Meltzer, K.P. Cusmano-Ozog, J.M. Campos, S.E. Hofherr Children's National Health System, Washington, DC. Introduction: The reliable and reproducible detection of single nucleotide variants and complex insertions/deletions is an integral part for the development of a clinical massively parallel sequencing (MPS) method. We sought to bring MPS into our lab to use as a technique for the discovery of variants in rare, inherited childhood disorders. To validate the test in our laboratory, we utilized the New York State Department of Health recommendations for "Next Generation Sequencing guidelines for somatic genetic variant detection" and modified the recommendations for the detection of germ line single nucleotide variants (SNVs) and complex variants (indels). Methods: The DNA for our validation work was from samples previously sent to reference labs or purchased from Coriell. The MPS was completed using the TruSight One panel (TSO; 4,813 genes) on a NextSeq 500 (Illumina). We tested 125 samples for the validation of the accuracy of our assay. In addition, we used three HapMap samples (NA12877, NA12878, and NA12880) and four patient samples sent to reference labs to test inter-run and intra-run sensitivity and specificity. The intra-run consisted of seven samples within the same run. The inter-run data was based on seven samples run four different times over four weeks. Alignment of our samples to the hg19 reference genome (Homo sapiens, build 37.2) and was performed with the Burrows-Wheeler Aligner (BWA). BWA (version 0.7.7) is based on a backward search with Burrows-Wheeler Transform to efficiently align short sequencing reads against a large reference sequence such as the human genome, allowing mismatches and gaps. BWA uses the Genome Analysis Toolkit (GATK; version 1.6) to call SNVs and Indels. GATK calls variants for each sample, analyzes variants against known variants, and then calculates a false discovery rate for each variant. Results: We were able to validate the 125 samples with previous SNV and Indel results from reference labs or from Coriell. The previously examined samples contained 175 different SNVs and 26 Indels. For the inter-and intra-sensitivity and specificity, we achieved 100% for the four samples previously sent to reference labs. For the HapMap samples, we achieved an average of 99.6% sensitivity and 97.3% specificity for SNVs in all 4,813 genes in the TSO panel. For Indels in the HapMap samples, our average sensitivity was 95.5% and specificity dropped to 83.5%. Conclusions: We validated our use of the TruSight One panel for the testing of clinical samples using the NextSeq 500 massively parallel sequencing system combined with BWA alignment. We are investigating the use of additional alignment procedures to improve to our sensitivity and specificity for SNV calls and Indels.
Ordering Behavior S. Yang, E.D. Esplin, D. Kamara, E.S. Gordon, S. Aradhya, S. Lincoln Invitae, San Francisco, CA. Introduction: Inherent in the concept of precision medicine is the customization of therapies and associated diagnostic testing. Multi-gene genetic tests have become increasingly available, but the pre-curated gene panels selected by laboratories may or may not always provide the exact information desired by a clinician or patient. Our laboratory offers both customized and a variety of pre-curated panels at a single price regardless of the number of genes ordered. In this study we sought to understand clinician preferences for these different genetic tests in the setting of hereditary cancer risk assessment. Methods: To initially assess the behavior of healthcare providers in a scenario where neither cost nor turn around time vary by the number of genes ordered, we selected 183 clinicians who had placed at least 2 orders for patients with a hereditary-cancer indication between January and November 2014. Orders from each clinician were categorized to reflect trends including breadth of orders (number of genes), degree of customization of orders, and ordering consistency across patients. To investigate the rationale underlying the ordering patterns observed, we identified providers who ordered multiple panel tests and invited them to participate in a brief interview. To date we have received detailed responses from 13 providers and we expect more before the AMP meeting. Results: 1743 tests with an indication of personal or family history of cancer were ordered by these 183 clinicians (on average, 9.5 tests per clinician). Most test orders were for pre-curated panels (53%), averaging 24 genes per test, although many ordering clinicians customized panels, at least sometimes. Over a third of all orders (37%) were for the full hereditary-cancer panel (29 genes). Customized tests had a smaller average of 10 genes per test. Roughly half of clinicians had consistent ordering patterns, whereas the others varied their orders more frequently. Tests for hereditary breast cancer were more likely to be customized by contrast with those for ovarian or colon cancer. Among providers that provided a rationale underlying their ordering patterns, 75% ordered primarily pre-curated panels. Also, 75% (but not the same list of respondents) reported to us that their guiding principle was to order the maximum number of potentially relevant genes. Half of providers indicated that personal and family history was the number-one factor guiding panel selection. About 70% of providers stated patient preference played a role in this decision. Conclusions: We observed a significant diversity of ordering behaviors for hereditary-cancer panels both between clinicians and by individual clinicians. This included a tendency toward larger pre-curated gene panels, with a sizable minority (13%) of the ordering providers customizing using the available panels. Factors the clinicians indicated influence their selection included clinical indications, patient preference, and characteristics of the offered panels, such as clinical utility of genes on the panel and insurance coverage. There is a desire within the genetics community to understand practices in the rapidly changing area of genetic testing. To help answer these questions, more extensive studies of provider ordering behavior and rationale will be highly valuable.
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org correlation between the levels of platelet miRNA, other laboratory test results, and the platelet activity after dual antiplatelet therapy. Methods: We enrolled a total of 60 participants, comprising 40 healthy volunteers and 20 ST-elevation MI (STEMI) patients to investigate the expression levels of circulating platelet miR-1, miR-21, miR-126, miR-150 and miR-223. Platelet-rich plasma was obtained by dual centrifugation, and platelet miRNA levels of participants were measured by real-time quantitative PCR. The results were analyzed using RNU 43 as an endogenous control. Results: In our study, we could find the decreased platelet miR-21 (80% lower) and miR-126 (75% lower) expressions as well as increased platelet miR-150 (2.64 fold) and miR-223 (4.71 fold) expressions which were statistically significant in patients with AMI (all p<0.01) than in control. However, we only observed that platelet miR-126 showed a good correlation with the plasma concentration of troponin I in AMI (r=-0.556, p=0.011). We did not find any significant correlation between the platelet miRNA (miR-21, miR-126, miR-150 and miR-223) and P2Y12 reaction units (PRU) or vasodilator-stimulated phosphoprotein (VASP)-platelet reactivity index (PRI) in AMI patients at pre-percutaneous coronary intervention (PCI) and 48 hours after PCI (all p>0.05). In the present study, the miR-1 expression detection was not successful. Conclusions: Platelet miRNA detection attracts more attention than plasma miRNA for cardiology research because platelets play major roles in the maintenance of hemostasis as well as thrombosis formation. The results of the current study suggest that platelet-enriched miRNA including miR-21, miR-126, miR-150 and miR-223 could be useful as sensitive diagnostic biomarkers for AMI. The potential of these miRNA to improve treatment stratification related to the prognosis warrants further study. Introduction: Myeloid malignancies are complex neoplasms with a wide range of gene alterations. Mutation profiling is essential in initial diagnosis and risk stratification of these patients, but is also valuable in the follow up to assess response to therapy. In this study we describe our experience using next-generation sequencing (NGS) of microdroplet PCR generated amplicons in the monitoring of patients with myeloid neoplasms. Methods: Blood and bone marrow samples received for routine molecular testing at multiple time periods of disease and treatment were selected. DNA was extracted, sheared and subjected to microdroplet emulsion PCR and sequenced on an Illumina MiSeq to an average depth of at least 500X. The testing panel interrogated mutational hotspots in 28 genes associated with myeloid malignancies (total of 847 amplicons designed with the RainDance DeepSeq system) (J Mol Diagn. 2014; 16:504-18) . Baseline mutation clones were recorded for each case and tracked across time points. Sensitivity for known mutations with this assay is about 1% minor allele frequency (MAF). For transplant patients, polymorphic SNP loci were monitored across samples to track disease and engraftment status. Results: A total of 1589 samples from 490 patients were analyzed during a 2 year period (range 2 to 16 samples per patient), of which 25% patients were in post transplant setting. Diagnostic samples harbored at least 1 mutation (Median: 2, range: 1 to 5); with CEBPA, DNMT3A, FLT3, NPM1 and TET2 as the most commonly mutated genes. Monitoring of clones stratified patients into 4 groups: 1) disappearance of founder clones, 2) disease with stable mutation profile, 3) disease with stable founder clones and development of minor clones, and 4) evolution and expansion of minor clones over the initial founder clone. Common mutations identified in subclones in the relapse/refractory setting impacted ASXL1, KRAS, NRAS, TP53 and TET2 genes. For post transplant patients, engraftment status was assessed by comparing genotypes of polymorphic SNP loci between donor, baseline host and post-engraftment samples. NGS data for engraftment status showed good correlation with concurrent STR analysis. Conclusions: Targeted NGS is a valuable method for monitoring of patients for low level residual disease, enabling the tracking of mutant clones and subclones in response to therapy and providing further insight into clonal heterogeneity and evolution. For post transplant patients it allows concurrent assessment of mutations and engraftment status. Monitoring of host and donor baseline SNPs allows higher sensitivity for detection of disease relapse than mutation profile. The multi-gene approach allows easy incorporation into clinical laboratory workflow to assess patients with multiple diseases.
Leukemia Using a 1200-Gene Panel (Onco1K) S. Kadri, N. Galanina, S. Sharma, A. Guo, P. Lu, G. Raca, G. Venkataraman, B. Long, M. Ming, W. Stock, S. Smith, L.V. Furtado, J.P. Segal, Y. Wang University of Chicago, Chicago, IL. Introduction: Bruton tyrosine kinase (BTK) plays a key role in the pathogenesis of chronic lymphocytic leukemia (CLL). Ibrutinib (Ibr) is a novel and highly specific irreversible BTK inhibitor that was recently approved for treatment of high-risk and refractory CLL. However, despite initial high overall response rates, 21% of patients develop ibr resistance at a median follow up of 3 years. Ibr-refractory patients have dismal outcomes with overall survival of 3.5 months. Thus, understanding molecular mechanisms underlying ibr resistance is important. Since, cancers evolve by reiterative process of clonal selection and expansion, we hypothesize that clonal evolution is a key mechanism responsible for CLL progression in ibr-relapsed patients. Onco1K, a 1168 gene panel developed at the University of Chicago, was used to study patients' longitudinal mutational profiles. We followed the clonal evolution of two CLL and two Richter Syndrome (RS)-transformed patients who relapsed on ibr throughout their clinical course to the time of disease progression. Methods: Onco1K is a hybrid-capture panel covering 300 genes of high clinical significance and 900 other cancer-associated genes, which is sequenced on the HiSeq 2500 (Illumina). Serial samples were collected from the patients before and after ibr treatment to the time of relapse (2-4 time points per patient) and sequenced with ~22 million reads per library. Internal custom-designed pipelines were used to process the large panel data to detect somatic mutations in the 1200 genes. Results: High (85% to 90%) overlap was found between variants at various time points for a given patient. K-means clustering of the profiles showed that the profiles for the 2 CLL patients clustered similarly, but not much overlap between the specific mutations. We detected several potential driver and passenger mutations that might be in the founder as well as relapsed clones for each patient. Besides the reported BTKC481S mutation that is detected in 3 of 4 patients, we identified a novel BTK mutation in a RS-transformed patient after ibr therapy. The new mutation was also confirmed by Sanger sequencing. Conclusions: Using Onco1K, we were able to study the mutational profiles of the four patients showing distinct patterns of clonal evolution in CLL relapse and progression. We found both previously reported and novel mutations in relapsed and RS transformed patients, respectively. Understanding genomic alterations during the course of the disease may help identify gene mutations that evolve through the process of clonal selection and may impart resistant phenotype. These mutations may be useful as predictive biomarkers of drug resistance and disease progression and may guide therapy selection for CLL patients. K.E. Fisher, M.M. Gramatges, H. Sayeed, S. Somvanshi, R.E. Rau, B.Y. Merritt, N.R. Patel, A. Roy, A. Marcogliese, J.N. Punia, A.A. Bertuch, M.L. Redell, K.R. Rabin, D.H. López-Terrada Baylor College of Medicine and Texas Children's Hospital, Houston, TX. Introduction: Pediatric hematologic malignancies harbor known diagnostic and prognostic molecular alterations many of which are detectable by next-generation sequencing (NGS). At our tertiary care pediatric cancer center, we designed a custom NGS panel to assess the presence of mutations in peripheral blood (PB) and bone marrow (BM) diagnostic specimens from pediatric patients with acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), or myelodysplastic syndrome (MDS). Methods: Representatives from pathology and pediatric hematology/oncology convened to select clinically actionable mutations and/or mutations frequently associated with pediatric leukemias and MDS in exonic, promoter, or intronic regions; chromosomal structural alterations and epigenetic modifications were excluded. Probe selection for the consensus regions was performed using Agilent's web-based application SureDesign. DNA from 8 patient samples (6 BM: 4 AML and 2 MDS, 2 PB: 1 AML and 1 B-ALL) with known cytogenetic or molecular alterations were selected, and Agilent SureSelect QXT custom kit libraries were prepared. All 8 samples were sequenced on a MiSeq flow cell in paired-end mode (2x75 bp). Read alignment to the reference human genome (GRCh37), variant calling, and annotation were performed using Agilent SureCall v3.0 software. Variants passing defined total depth of coverage (>50x) and variant allele fraction (>0.10) filters were reviewed for pathogenic significance. Results: The panel design targeted 278 regions spanning 87.5 kb in 46 genes: ANKRD26, ASXL1, BRAF, CALR, CEBPA, CNOT3, CREBBP, CRLF2, CSF1R, DNMT3A, ETV6, EZH2, FBXW7, FLT3, GATA1, GATA2, GATA3, IDH1, IDH2, IKZF1, IL7R, JAK1, JAK2, JAK3, KIT, KRAS, MPL, MYD88, NOTCH1, NPM1, NRAS, NT5C2, PAX5, PHF6, PTEN, PTPN11, RPL5, RPL10, RUNX1, SETBP1, TERC, TERT, TET2, TP53, VPREB1, and WT1. On average, 98.17% (range 98.14% to 98.19% ) of Q30 bases mapped to targeted regions with 415x mean depth of coverage (range 355x to 517x) per targeted region. Potentially significant variants were detected in 6 of 8 samples (75%: 4 AML and 2 MDS), including previously detected GATA2, RUNX1, and TERT pathogenic variants. Two activating variants, NRAS p.G60E and FLT3 p.D835Y, and alterations in JAK3, PAX5, SETBP1, TET2, and WT1 coding exons were also detected. However, routine analysis failed to detect a known FLT3 internal tandem duplication in an AML sample. Conclusions: The custom NGS panel shows promise for the detection of clinically-relevant mutations in pediatric hematologic malignancies. Additional samples are currently being tested. Further confirmatory, bioinformatics, and optimization studies are required for clinical validation, particularly for insertions, deletions, and duplications.
jmd.amjpathol.org ■ The Journal of Molecular Diagnostics H04. Targeted Mutation Analysis Provides Evidence of Clonality in a Subset of Patients with Chronic ITP and Subsequent Monocytosis Y. Liu, M.L. Gomez, J. Racchumi, J.T. Geyer, W. Tam Weill Cornell Medical College, New York, NY. Introduction: Several recent studies suggested an association between idiopathic thrombocytopenic purpura (ITP) and chronic myelomonocytic leukemia (CMML). The diagnosis of CMML in these studies was made based on 2008 WHO classification by which the presence of a clonality marker is not absolutely required if other criteria are met. However, the diagnosis of CMML in patients with ITP and persistent monocytosis can be challenging. A definitive exclusion of other etiologies for monocytosis is usually difficult due to concurrent or prior ITP treatment composed of multiple reagents. Moreover, CMML-like or MPN-like morphologic findings including hypercellularity and increased fibrosis have been described in patients receiving thrombopoietin receptor (TPO-R) agonists widely used in ITP treatment. Such morphologic features can be diagnostically confounding and demonstration of a clonality marker may be essential for establishing a diagnosis of CMML. We show targeted mutation analysis may provide critical evidence of clonality that aids in the differential diagnoses. Methods: Nine cases were retrieved from the Archives. Chart review and pathology evaluation were performed. All cases had either prior bone marrow (BM) sample morphologically compatible with ITP (7/9) or long history years) of clinically-diagnosed ITP showing initial response to ITP treatment (2/9). 8 cases developed persistent absolute monocytosis for 2 months). One case showed persistent relative monocytosis with an absolute monocyte count between 900/ml to 1000/ml). In this case, the monocytes accounted analysis including gene associated with myeloid neoplasm was performed on a subset of samples to identify potential clonality markers. Results: These cases showed wide age distribution (8 years old to 77 years old, med: 65) with no gender predominance (M:F=1:1.3). The intervals between the ITP diagnosis and the subsequent monocytosis ranged 14 months to 348 months (med: 132). All but 1 case received TPO-R agonists before the development of monocytosis (8/9). Targeted NGS analysis was performed on the subgroup of 5 cases showing intractable thrombocytopenia despite ITP treatments (Plt <50 K/ml) (5/9). Within this subgroup, a diagnosis of CMML was suggested in 3 samples based on increased blasts/dysplasia or the presence of mutations (1 showed SRSF2 in-frame insertion, TET2 missense and TET2 indel mutations; 1 showed an ETV6 nonsense mutation and 2 missense mutations in TP53). One of the 2 remaining cases with no dysplasia and no clonality marker within this subgroup showed subsequently resolved monocytosis. The other 1 was found to have significant lymphadenopathy that may explain the monocytosis. The other subgroup of cases showed improved platelet counts after ITP treatment (4/9) with no significant dysplasia or increased blasts. No case showed acquired cytogenetic abnormalities (0/9). Targeted NGS in the second subgroup is currently in progress. Conclusions: The diagnosis of CMML in patients with chronic ITP and subsequent monocytosis is challenging, especially in light of the morphologic overlap between myeloid neoplasm and morphologic changes associated with TPO-R agonists. Though the presence of a clonality marker is not absolutely required for a diagnosis of CMML by current 2008 WHO classification if other criteria are met, the demonstration of clonality marker by targeted mutation analysis is extremely useful in the differential diagnoses in this specific scenario.
W.M. Elbjeirami, A. Abdulwahab, H. Elsayed, N. Elnagdi, N. Abd Allatif, H. Al-Jedani, A. Al Shaikh, F.A. Al-Allaf King Abdullah Medical City, Makkah, Saudi Arabia. Introduction: Chronic myelogenous leukemia (CML) is one of the predominant hematological malignancies in Saudi Arabia (SA). It originates from reciprocal translocation of chromosome 9 and 22 t(9;22), and may give rise to different BCR/ABL fusion mRNAs due to different genomic breakpoints and alternative splicing. Most breaks occur immediately downstream of exon 2 or 3 of the M-bcr region and result in b2a2 (e13a2) or b3a2 (e14a2) fusion transcripts and the P210 BCR-ABL1 protein. The frequencies of one or other rearrangement in CML patients can vary in different ethnicities and geographic regions. Furthermore, significant correlation between sex and transcript type has been previously demonstrated. To our knowledge, there is no published data addressing the frequency of BCR-ABL1 fusion transcripts among Saudi CML patients. The aims of our study were: first, to determine the frequency of BCR-ABL1 transcript variants in Saudi CML patients presented at King Abdullah Medical City (KAMC) in Mekkah, Western region of SA, and compare it with the occurrence reported in other neighboring populations. Second, to determine whether sex and type of BCR-ABL fusion transcripts have any correlation. Methods: Peripheral blood and bone marrow samples were analyzed by nested and multiplex RT-PCR to detect BCR-ABL transcripts from 68 Saudi CML patients seen at KAMC from January 2011 to present. Clinical and laboratory data were obtained from the medical charts of the patients. Results: At diagnosis, the median age was 46 years, and there were nearly an equal number of males (55%) versus females (45%). All patients had their initial treatment with the tyrosine kinase inhibitor Imatinib. Out of 68 Saudi CML patients in the study, 37 (54.4%) positive patients showed b2a2 fusion transcript, whereas 31 patients (45.6%) showed b3a2 transcript. We did not observe any co-expression of both isoforms of BCR-ABL. There was no tendency of differential transcript expression associated with sex as b2a2 expression was found in 40.5% of females and 59.5% males. Similarly, expression of b3a2 was found in 51.6% females and 48.4% males. Thirty patients achieved major molecular response in one year of treatment with no disease related deaths during the follow up period which ranged from 2 months to 37 months with a mean/median of 1.47/1.42 year, respectively. Notably, five patients experienced disease progression and had tyrosine kinase domain mutation. Conclusions: These results confirm lack of any predominance of the BCR-ABL isoforms b2a2 or b3a2 in Saudi CML patients under investigation. This is discordant with similar studies conducted in other groups of neighboring countries such as Sudan, Pakistan, and Iran. There was no significant correlation between sex and type of BCR-ABL1 transcript.
H06. Flow Cytometric Cell Sorting of Plasma and Lymphoid Cell Clones prior to MYD88 L265P and CXCR4 Mutation Detection Improves Sensitivity and Specificity for WM/LPL Diagnosis B. Burnworth, Z. Wang, W. Fritschle, R. Bennington, A. Bennington, C. Wentzel, S. Verkamp, P. Nguyen, K. Ghirardelli, L.E. Brodersen, D.A. Wells, M.R. Loken, B.K. Zehentner Hematologics, Inc., Seattle, WA. Introduction: MYD88 L265P has been reported in high prevalence in patients with Waldenström's macroglobulinemia (WM) and lymphoplasmacytic lymphoma (LPL) and mutation analysis has been implemented in clinical practice to support the diagnosis of LPL/WM. However, MYD88 L265P is not exclusive to WM/LPL. Mutations are detected in patients with IgM monoclonal gammopathy of undetermined significance (IgM-MGUS) and rarely in diffuse large B-cell lymphoma (DLBCL), splenic marginal zone lymphoma (SMZL) and B-cell chronic lymphoproliferative disorders (B-CLPD). Recently, mutations in CXCR4 have been identified in a subset of patients with MYD88 L265P WM/LPL and the presence of mutated CXCR4 may negatively affect treatment response. In this study we demonstrate a novel approach to assess MYD88 L265P and CXCR4 mutation status in order to assist WM/ LPL diagnosis by combining flow cytometric cell sorting (FACS) with molecular analysis. Methods: FACS isolated clonal plasma and B lymphoid cell fractions of 69 WM /LPL, lymphoma, myeloma and CLL specimen were analyzed for MYD88 and CXCR4 mutations using Sanger sequencing. B cell clonality profiles in the lymphoid versus plasma cell compartments were determined using the Biomed-2 primer sets. Results: MYD88 L265P was detected in all specimens with confirmed WM (17/17). Notably, 8/17 cases demonstrated the mutation only in the sorted cell fractions, but not in the unsorted bone marrow. Identical gene rearrangement clonality profiles in both the plasma and the B cell collections were observed in 16 of the 17 cases. Of 21 specimens with LPL only 9 (43%) showed identical B cell clonality profiles in plasma and B lymphoid cells and 7/9 tested positive for L265P in both cell collections. None of the 12 LPL with unrelated clonality profiles revealed L265P in both cell fractions and 5/12 tested negative. 10 lymphoma, 12 myeloma and 5 CLL specimens tested negative for L265P, with the exception of one SMZL. In addition CXCR4 mutations were found in both plasma and B lymphoid cell fractions in a subset of MYD88 L265P WM cases. Conclusions: We conclude that confirming the presence of MYD88 L265P in both B-lymphoid and plasma cell fraction is an important prerequisite to distinguish LPL/WM from related disorders with high confidence. The L265P mutation was detected exclusively in plasma and B-lymphoid fractions but not in unsorted bone marrow of 8/17 WM and 4/9 LPL. In addition, the majority of WM and LPL with L265P in both cell fractions revealed identical clonality profiles. Furthermore, our approach can also be used to establish CXCR4 mutation status to potentially aid WM/LPL treatment decisions.
L.Y. Ballester 1 , S. Loghavi 2 , R. Kanagal-Shamanna 2 , B.A. Barkoh 2 , P. Lin 2 , L. Medeiros 2 , R. Luthra 2 , K.P. Patel 2 1 Baylor College of Medicine, Houston, TX; 2 MD Anderson Cancer Center, Houston, TX. Introduction: Waldenstrom macroglobulinemia (WM) is a B-cell lymphoma characterized by the accumulation of lymphoplasmacytic cells in the bone marrow and excess production of immunoglobulin-M. Approximately 90% of WM patients harbor the MYD88L265P mutation, which confers survival advantage to tumor cells through activated Toll-like receptor signaling and NF-KB activation. Whole genome sequencing identified somatic mutations in CXCR4 in ~29% of WM cases with MYD88L265P. The reported alterations are nonsense or frameshift mutations localized to the C-terminus of the CXCR4 receptor with the most common mutation being p.S338*. Recent studies have suggested that CXCR4 mutations may interfere with treatment response to ibrutinib. Our goal was to design and validate a clinical assay to detect CXCR4 mutations to facilitate therapy selection. Methods: Our study included Fifty-one low-grade B-cell lymphomas with plasmacytic differentiation (42 MYD88L265P and 9 MYD88WT) involving various samples types (bone marrow, peripheral blood and formalin-fixed tissue). We designed and validated Sanger sequencing to detect mutations in exon 2 (codons 292 to 353) and pyrosequencing assays to detect mutations codon 338 in CXCR4 in a clinical laboratory. Sanger sequencing assay required optimization of primer design using linker sequence. Results: We identified 8 cases with CXCR4 mutations, including 5 single nucleotide substitutions (4 resulting in p.S338*), and 3 insertion/deletions. All CXCR4 mutated
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org cases also had the MYD88L265P mutation. The analytical sensitivity for clinical use was determined to be 20% for the Sanger sequencing assay and 5% for the pyrosequencing assay. Among the single nucleotide substitutions we detected a novel missense variant (p.L326P) of uncertain clinical significance. Conclusions: We successfully validated a set of clinical assays to detect mutations in CXCR4 that identify WM patients with potential resistance to treatment with ibrutinib.
B.L. Betz 1 , A. Behdad 2 , N.A. Brown 1 , H.C. Weigelin 1 , K.S. Elenitoba-Johnson 1 1 University of Michigan, Ann Arbor, MI; 2 Northwestern University, Chicago, IL. Introduction: Calreticulin (CALR) mutation testing can aid in the diagnosis of primary myelofibrosis (PMF) and essential thrombocythemia (ET) and is recommended in patients who lack the JAK2 V617F mutation. Detection of CALR mutations is commonly achieved using PCR with capillary fragment-length analysis to detect the variably sized deletion and insertion mutations that lead to a +1 bp frameshift of the coding sequence. During preclinical validation of a fragment analysis assay we unexpectedly detected an in-frame 9 bp deletion in a negative validation specimen. We set out to characterize the frequency of CALR in-frame deletions, and to develop and implement a testing strategy to account for these variants. Methods: A PCR capillary fragment analysis assay capable of detecting all reported CALR length mutations was optimized. Genomic DNA from 590 healthy blood specimens and 694 cases of suspected myeloproliferative neoplasms (MPNs) were tested. Sanger sequencing was performed on cases with newly-encountered PCR fragment sizes to clarify the result and to generate a database of fragment sizes to guide future interpretation. Results: In-frame deletion variants (3 or 9 bp deletions) were detected in 6/590 (1.0%) healthy blood specimens and in 6/694 (0.9%) cases of suspected MPN. All in-frame deletions were present at an allele frequency of 50%. Frameshift deletions or insertions were detected in 111/694 (16.0%) of suspected MPNs. One case harbored both a 9 bp in-frame deletion and a pathogenic 52 bp frameshift deletion on the same allele. Conclusions: CALR exon 9 in-frame deletions are present at a frequency of approximately 1% in both the healthy population and in patients with suspected MPN, suggesting these are benign germline polymorphisms. Accurately distinguishing these benign variants from true frameshift mutations is critical to avoid reporting erroneous CALR test results. With careful implementation, we found that PCR capillary fragment analysis was a robust and sensitive method capable of detecting and distinguishing pathogenic frameshift mutations from germline in-frame deletions.
Is There a Role for Immunohistochemistry? L.Y. Ballester 1 , S.F. Sarabia 1,2 , C.E. Allen 1,2 , C. Webb 1,2 , A. Roy 1,2 , D.H. Lopez-Terrada 1,2 , K.L. McClain 1,2 , J. Hicks 1,2 , K.E. Fisher 1,2 1 Baylor College of Medicine, Houston, TX; 2 Texas Children's Hospital, Houston, TX. Introduction: Langerhans cell histiocytosis (LCH) is a proliferative disorder of a specific type of antigen-presenting cell (Langerhans cell histiocyte), and is more common in children. The BRAF p.V600E mutation is detected in 65% of pediatric LCH cases, its detection can assist with treatment options as well as monitoring treatment response, and recurrence. At our tertiary-care children's hospital, we test LCH patient samples for the BRAF p.V600E mutation using allele specific real-time PCR (AS-PCR). On occasion, samples yield inconclusive molecular results, and additional testing, such as BRAF V600E immunohistochemistry (IHC), is requested. However, evidence for the utility of BRAF V600E IHC in evaluation of pediatric LCH is limited. Here, we evaluated BRAF V600E IHC in pediatric LCH using molecular analysis as the correlative gold-standard. Methods: We identified 13 pediatric LCH cases (7 males, 6 females; age range 9 months-17 years) in which BRAF p.V600E molecular testing had been performed on formalin-fixed, paraffin-embedded (FFPE) tissue using the BRAF RGQ real-time PCR assay (Qiagen) within the last 2 years. Archived FFPE blocks were retrieved, 1 H&E and 2 unstained slides were cut for each, and percentage of lesional tissue cells was determined by H&E approximation (range 5% to 95%). IHC was performed on the Leica Bond automated stainer using a BRAF V600E monoclonal antibody (clone VE1, Spring Bioscience) at a 1:40 dilution and diaminobenzidine (DAB) as a chromogen. A BRAF-negative malignant melanoma and 2 BRAF-positive papillary thyroid carcinomas were used as negative and positive controls, respectively. Three pathologists evaluated and scored the IHC results for positivity, and consensus scoring was achieved for all cases by concomitant review. Results: The overall concordance rate between the two methods was 10/13 (77%). BRAF p.V600E mutations were detected in 8/13 (61.5%) of cases, and BRAF V600E IHC positive staining was seen in 7/13 (54.8%) cases. False-negative (n=2) and false-positive (n=1) results yielded a sensitivity and specificity of 75% and 80%, respectively. IHC interpretive analysis was hampered by suboptimal IHC parameters (eg, antibody staining intensity and edge effect), defining positivity in the aberrant Langerhans histiocytes, and non-specific uptake of the DAB chromogen by accompanying inflammatory cells. Conclusions: Molecular analysis remains the gold standard for BRAF p.V600E mutation testing in pediatric LCH. Additional studies are required to optimize IHC staining protocols, define scoring algorithms, and assess potential pitfalls associated with interpreting BRAF V600E IHC in this unique pathologic entity.
H10. Diagnostic Evaluation of a Targeted RNA Sequencing Panel for the Detection of Gene Fusions Associated with Ph-Like Acute Lymphoblastic Leukemia (ALL) K. Yap, L. Furtado, S. Kadri, J. Segal, S. Gurbuxani, G. Raca Unversity of Chicago, Chicago, IL. Introduction: Philadelphia (Ph)-like Acute Lymphoblastic Leukemia (ALL) is an increasingly recognized molecular subtype of high-risk B-cell ALL that responds poorly to standard chemotherapy. Since the molecular pathogenesis of Ph-like ALL mainly involves activation of tyrosine kinase (TK) signaling by expression of abnormal gene fusions involving a range of TK and cytokine receptor genes, patients with Ph-like ALL have been successfully treated with appropriately selected TK inhibitors. Due to the diversity of the implicated gene fusions, Ph-like ALL cannot be efficiently detected with clinically available cytogenetic and molecular assays. Methods: The goal of our retrospective study was to determine the diagnostic feasibility of using a targeted versus total RNA sequencing (RNAseq) approach to detect fusion transcripts in Ph-like ALL patients. RNA was extracted from bone marrow biopsies of 4 Ph-like ALL patients and used for the construction of total RNAseq libraries using the Illumina Truseq stranded total RNA preparation kit (2-day work flow). A panel of 13 Ph-like ALL associated kinase gene transcripts () was enriched ABL1, ABL2, CRLF2, CSF1R, PDGFRB, JAK2, EPOR, DGKH, IL2RB, PTK2B, TSLP, TYK2, NTRK3using a hybridization-based DNA oligonucleotide enrichment panel (IDT xGEN probes) for targeted RNAseq (1-day workflow). Using 150bp paired end sequencing and molecular barcoding, an average of 500,000-700,000 reads per sample were generated for targeted RNAseq on the Illumina Miseq system, and 50 million to 70 million reads per sample for total RNAseq on the Nextseq500 system. Fusion transcript analyses were carried out using the Fusioncatcher algorithm on a high performance computing environment with a focus to minimize false positive detection of fusion transcripts. Results: Using the Fusioncatcher algorithm, we were able to detect fusion transcripts (EBF1-PDGFRB, RCSD1-ABL1, RCSD1-ABL2, and CRLF2-P2RY8) unequivocally in a blinded fashion with minimal false positivity, by both targeted and total RNAseq approaches. On a high performance computing environment, the targeted RNAseq analyses were speedily completed in less than 20 minutes, whereas the total RNAseq data took approximately 20 hours of analyses. Junction reads generated by the next generation sequencing approach allowed accurate delineation of the fusion breakpoints, which were confirmed to be in-frame for the production of oncogenic fusion proteins. Conclusions: Targeted RNAseq is a time and cost economical approach to accurately detect Ph-like ALL associated gene fusions in a clinical diagnostic setting. We propose that the workflow developed in this study offers a powerful approach for the genetic characterization of Ph-like ALL cases, as well as other leukemias with uncommon fusion transcripts. A.A. Stence, A.D. Bossler, D. Ma University of Iowa Hospitals and Clinics, Iowa City, IA. Introduction: Somatic mutations of calreticulin (CALR), Janus kinase 2 (JAK2), and myeloproliferative leukemia virus oncogene (MPL) are important for diagnosis and prognosis in myeloproliferative neoplasia (MPN) and for determining treatment response to JAK inhibitors in primary myelofibrosis (PMF) patients. Traditionally, sequences of interest of these 3 genes were amplified separately and mutation analysis was performed sequentially depending on the clinical diagnosis. We developed a novel single-tube multiplex PCR assay for simultaneous amplification of CALR exon 9, MPL exon 10, and JAK2 exons 12 and 14, followed by single nucleotide primer extension (SNPE) assay and Sanger sequencing for mutation analysis of all three genes at the same time. Methods: Twenty-one samples from patients with a clinical diagnosis or suspicion of essential thrombocytosis (ET), PMF or polycythemia vera (PV), but negative for JAK2 p.V617F were included. Singletube, multiplex amplification of CALR exon 9, MPL exon 10, and JAK2 exons 12 and 14 was performed using archived DNAs. PCR product was purified and the product was split in two parts: one portion was used in SNPE assay for detection of MPL codon 515 mutations (the JAK2 p.V617F mutation can be detected in the same reaction), while the other portion was Sanger sequenced for detection of CALR exon 9, MPL exon 10, and JAK2 exon 12 mutations. Both the Sanger and SNPE products were analyzed by capillary electrophoresis (ABI 3130 XL). SNPE findings of MPL were confirmed by Sanger sequencing and CALR Sanger sequencing results were confirmed by fragment analysis (FA). PCR amplicons for CALR type 2 mutation, JAK2 exon 12 p.G543_D54del, and MPL p.W515L were cloned into the PCR2.1 TOPO cloning kit (Life Technologies, Carlsbad, CA) and the limit of detection (LOD) was determined by serial dilutions of the mutated plasmid DNA into wild-type DNA. Results: Mutations were detected in 11 of 21 patients (52%). Five (24%) patients (2 PMF and 3 ET) were positive for CALR mutations. Five patients (4 PMF and 1 ET) (24%) harbored mutations in MPL, and 1 PV patient (5%) had a JAK2 exon 12 mutation. There was 100% concordance between the MPL SNPE and Sanger results, and CALR Sanger and FA findings. The presence of the JAK2 exon 12 mutation in the PV patient was confirmed by an outside laboratory. The LODs of all these mutations by both SNPE and Sanger sequencing was 6%. Conclusions: We developed a novel single-tube multiplex PCR assay that can simultaneously amplify 4 exons of 3 different genes. The same PCR product was used for mutation analysis of CALR exon 9, MPL exon 10, and JAK2 exons 12 and 14. The assay is reliable
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org selection of genes involved in the etiology of a heterogeneous collection of bone marrow failure disorders providing the diagnostician critical information leading to diagnosis or the ability to rule out rare inherited disorders in the cases of idiopathic aplastic anemia.
S. Jiwani, J. Laudadio, W. Bellamy, G. Insuasti-Beltran University of Arkansas for Medical Sciences, Little Rock, AR. Introduction: Plasma cell myeloma (PCM) is a clonal neoplastic proliferation of plasma cells. Chromosomal abnormalities are among the most important prognostic parameters in risk stratification of patients with PCM. Fluorescent in situ hybridization (FISH) is now routinely performed to detect chromosomal abnormalities in PCM. The International Myeloma Working Group recommends a minimal FISH panel to include t(4;14)(p16;q32), t(14;16)(q32;q23), and 17p13 (TP53) deletion. Additional targets including t(11;14)(q13;q32), chromosome 13 deletion, hyperdiploidy, and chromosome 1 abnormalities are also suggested. The efficiency of FISH testing depends upon the presence of a sufficient quantity of plasma cells in the sample. Plasma cell enrichment of bone marrow (BM) aspirate with CD138 antibody coated magnetic beads is one method used to enhance efficiency of FISH analysis. Our observations indicate that despite plasma cell enrichment of BM aspirate samples there is a high rate of insufficient specimens. Flow cytometric analysis (FCA) is a rapid and quantitative method to determine cellular components in BM aspirates. It is routinely performed for qualitative and quantitative analysis of neoplastic plasma cells on all BM aspirates of PCM patients. We studied the value of FCA in selecting samples for further FISH analysis. Method: We performed a retrospective analysis of 488 CD138 enriched, multiple myeloma FISH assays routinely ordered between January-April 2015. Our panel included probes for t(4;14)(p16;q32), t(14;16)(q32;q23), t(14;20)(q32;q12), 17p13 deletions, t(11;14)(q13;q32), and chromosome 13 deletion/monosomy. We determined the number of tests that provided optimum results (all probes informative), partial results (1 to 10 probes informative) and uninformative results (0 probes informative). We also analyzed the data from FCA to quantitate the plasma cells in BM aspirates sent for FISH analysis. We then applied sample selection criteria based on percent plasma cells in BM aspirate and examined the effects of such sample selection on outcome of FISH analysis. Results: The efficiency of PCM FISH panel ordered on all bone marrow aspirates without sample selection was 24% optimal, 34.6% partial results and 41.4% uninformative. The percent of informative PCM FISH panels increased from 58.6% to 79.1%, 86.9% and 92.2% using 0.25%, 0.5% and 1% plasma cells by FCA quantitation as selection criteria for FISH analysis. Using 0.25%, 0.5% and 1% plasma cells as selection criteria would have excluded 6.7%, 12.6% and 18.8% samples, respectively, yielding informative results and excluded 2.2%, 4.0% and 6.5% samples, respectively, with abnormal FISH results. A selection cut-off of 0.25% plasma cells appears to provide the optimal combination of improving efficiency without compromising informative specimens. Conclusions: Quantitation of BM aspirate plasma cells by FCA provides a useful criterion for FISH sample selection. Introduction: JAK2 mutations are the most frequent genetic alteration in myeloproliferative neoplasms (MPN) including polycythemia vera, primary myelofibrosis and essential thrombocythemia. Molecular testing is therefore an important component of the workup of suspected MPNs. With the advent of broad next-generation sequencing panels, patients with other hematologic malignancies also get tested for JAK2 alterations as part of standard diagnostic screening. Here we describe the spectrum of myeloid neoplasms harboring JAK2 mutations detected through broad screening using our 28 gene myeloid NGS panel. Methods: Cases received for routine molecular screening using a broad myeloid panel over a 12 month period were reviewed. The testing panel encompassed 847 amplicons designed with the RainDance DeepSeq system, covering clinically relevant regions in 28 genes, relevant to AML/MDS and MPN. Samples were sequenced on an Illumina MiSeq. Results: Seventy five unique patients harboring JAK2 mutations were identified among 610 total cases, encompassing 55 MPN, 5 MDS, and 15 AML (8 de novo, 7 secondary). The V617F variant was detected in most cases (92%, 69/75). Less common variants (R564L, F85_G1132delinsLFLSLLLF, L933L) were seen in MPN, whereas L611S, D613G and G935R were seen in AML. Multiple coexisting alterations (range 2 to 7) were present in 71% of all patients, with AML cases having the most number of additional mutations. TET2 and DNMT3A were the most common mutations, accounting for 22% and 15%, respectively in MPNs and 27% for both mutations in AML. Other common co-existing alterations included AXL1 in 10% of MPNs and 5q-in AML (33%, 5/15). Among 7 AML patients sampled several times across different time periods, the JAK2 was maintained in all samples. Variant frequencies of the JAK2 mutation in most cases was among the highest compared to other coexisting mutations suggesting they represent a founder clone. Conclusions: Mutations in JAK2 are often considered a defining molecular event in myeloproliferative neoplasms. Based on broad NGS based mutation analysis, mutations in JAK2 may also be identified in up to 11% of AML and MDS.Further studies are needed to better define other coexisting mutations as acquired events which promote the evolution to AML or MDS.
M. Platt 1 , A.T. Fathi 1 , D.R. Borger 1 , A.M. Brunner 1 , R.P. Hasserjian 1 , L. Balaj 1 , A. Lum 2 , S. Yip 2 , D. Dias-Santagata 1 , Z. Zheng 1 , L.P. Le 1 , T.A. Graubert 1 , A.J. Iafrate 1 , V. Nardi 1 1 Massachusetts General Hospital, Boston, MA; 2 University of British Columbia, Vancouver, British Columbia, Canada. Introduction: Recurrent mutations in IDH1 and IDH2 have been described in a variety of malignancies, including acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS). Both IDH1 and IDH2 encode isocitrate --KG).
-KG to the oncometabolite 2--KG-dependent enzymes (eg, TET2), resulting in aberrant epigenetic modifications. Mutant IDH1 and IDH2 enzymes serve as promising therapeutic targets and specific inhibitors are currently in development. Recurrent IDH1 and IDH2 mutations have been described as primarily mutually exclusive. Here we describe the identification of dual mutations in AML and MDS, a finding that may carry significant treatment implications. Methods: Over a six-month period of clinical testing with a targeted next-generation sequencing assay (SNaPshot NGS), we evaluated a total of 92 patients with AML, MDS and chronic myelomonocytic leukemia (CMML) -a subset of whom had IDH1 or IDH2 mutations. SNaPshot NGS utilizes a multiplex PCR technology [Anchored Multiplex PCR tform for SNV and indel detection. The panel targets exons and hotspots in 39 genes. Droplet digital PCR (ddPCR) and ultradeep sequencing using the Ion Torrent PGM were used as confirmatory methods. Results: Five of 21 patients with IDH-mutated myeloid neoplasms had dual mutations. The mutations were present at substantially different allelic frequencies (AFs), ranging from 2% to 44% AF. Two of the five patients also had cooccurring non-IDH mutations (e.g. mutations in NPM1 or NRAS). IDH1 p.Arg132His variants were confirmed by ddPCR and all low AFs were confirmed by ultradeep sequencing. Conclusions: Concomitant mutations in IDH1 and IDH2 are not infrequent events. Using panel testing with sufficient depth of coverage, along with reporting AFs of identified mutations, may have considerable impact on diagnostics and clinical decisions regarding therapy. The identification of mutations at low AFs, often in the setting of extensive myeloblast marrow infiltration, suggests that IDH1/2 mutations do not always occur in the founding neoplastic clone. Although it remains to be definitively demonstrated whether IDH1/2 mutations co-occur in the same tumor cells or in discrete tumor cell subsets, our data support the co-existence of multiple tumor subclones. This finding may have practical consequences: though the initial clinical data for IDH-specific inhibitors may be encouraging, strategies such as simultaneous inhibition of both IDH1 and IDH2, in addition to other anti-leukemic therapies, may be needed for patients with IDH1-and IDH2-mutation-positive subclones.
H19. pSTAT3/pSTAT5 Signaling Patterns in JAK2 V617F versus CALR Mutated Myeloproliferative Neoplasms K.C. Schneider, J.R. Cook Cleveland Clinic, Cleveland, OH. Introduction: The JAK2 V617F mutation, found in >95% of polycythemia vera (PV) and approximately 50% of essential thrombocythemia (ET) and primary myelofibrosis (PMF), leads to constitutive activation of the JAK/STAT signaling pathway. CALR mutations, recently identified in many myeloproliferative neoplasms (MPN) lacking JAK2 V617F, have also been reported to increase pSTAT5 expression, suggesting a similar pathogenetic mechanism. We have previously demonstrated that the presence of JAK2 V617F or MPL mutations is associated with aberrant expression of pSTAT5 in megakaryocyte nuclei detectable by immunohistochemistry. In this study, we examine pSTAT3/pSTAT5 expression in CALR mutated MPN versus MPN with JAK2 V617F or MPL mutations. Methods: We identified 30 cases of non-PV, BCR/ABL1 negative MPN. Fifteen (15) cases had known JAK2 V617F mutations, and 15 were JAK2 wild type. DNA was extracted from fresh or formalin-fixed paraffin-embedded tissue. CALR exon 9 mutations were identified using PCR and fragment length analysis (ABI 3730, Life Technologies, Grand Island, NY) and MPL exon 10 was analyzed by Sanger sequencing. Bone marrow core biopsies were stained for pSTAT3 and pSTAT5 and were independently assessed for megakaryocyte nuclear staining by two pathologists. Discordant cases were resolved by a third pathologist. Results: CALR mutations were identified in 8 cases (27%), including 1 with and 7 without JAK2 V617F mutations (5 type 1, 3 type 2). MPL mutations were found in 2 cases, each negative for CALR and JAK2 mutations. pSTAT5 staining in megakaryocyte nuclei was found in 16 cases: 11/15 (73%) JAK2 V617F positive, 1/2 (50%) MPL positive, 1/8 (13%) CALR positive, 1/1 JAK2/CALR double mutant, and 2/6 (33%) unmutated. pSTAT5 staining correlated with JAK2 V617F mutation (p=0.009) but not CALR mutation (p=0.20). pSTAT3 was positive in megakaryocyte nuclei in 4 cases: 1/15 (7%) JAK2 V617F positive, 1/2 (50%) MPL positive, 1/8 (13%) CALR positive, and 1/1 JAK2/CALR double mutant. pSTAT3 staining did not correlate with JAK2 V617F or CALR mutation status (p=1.0 and 0.23 respectively). Conclusions: pSTAT5 staining in megakaryocyte nuclei correlates with JAK2 V617F in non-PV MPN, but does not correlate with CALR mutation. pSTAT3 megakaryocyte nuclei staining did not correlate with JAK2, MPL, or CALR mutation status. The
P. Szankasi 1 , W. Shen 1 , K. Frizzell 1 , J.A. Schumacher 1 , T.W. Kelley 2 1 ARUP Laboratories, Salt Lake City, UT; 2 University of Utah, Salt Lake City, UT. Introduction: A large number of recurrently mutated genes with diagnostic, prognostic or therapeutic significance have been identified in myeloid malignancies. Because many genes may be evaluated at once, NGS mutation panels are becoming an essential component of the laboratory evaluation of myeloid diseases. NGS has the advantage of generating variant allele frequencies (VAFs) which correlate with subclone size, allowing for a method to map tumor heterogeneity. Here we sought to use NGS data derived from a 53-gene NGS myeloid malignancies panel to reconstruct the clonal hierarchy in a large number of cases. Methods: NGS libraries were prepared from sheared DNA, enriched for regions of interest by solution capture (SureSelect, Agilent) and sequenced on the Illumina platform. Variants were identified by the FreeBayes and Pindel programs. We analyzed 162 cases that had >1 somatic variant. The general diagnostic breakdown was: 63 AML, 34 MDS, 35 MPN, 30 MDS/MPN (including 12 CMML). SNV and short insertion/deletion (indel) VAFs were validated by alternative quantitative tests and showed excellent correlation. Indels >4 bp were excluded from the analysis (including FLT3 ITD and CALR mutations). Previous analyses showed their VAFs were not accurate enough to predict sub-clone size. VAFs were considered different -linked VAFs were adjusted for gender, and possible LOH at known tumor suppressor genes was taken into consideration. Results: Each case had 2 to 9 (av. 3.6) clinically significant variants. The majority of cases had variants with significantly (up to 10-fold) different VAFs, indicating the presence of sub-clones. The number of clones ranged from 1 to 6. Those cases with only a single dominant clone had 2 to 5 variants of equal VAFs. The number of estimated clones per case in MPN (1.75±0.74) differed significantly from that in AML (2.43±1.03, P=0.0004), MDS (2.22±0.89, P=0.0143) , and MDS/ MPN (2.37±1.16, P=0.0086) . Mutation(s) with the highest VAF likely represent the founder mutation acquired early in tumorigenesis. Genes mutated in the founder clone include: DNMT3A (P=0.0001), IDH1,2 (P=0.028), TET2 (P=0.0159), splicing genes (P=0.0001), and JAK2 in MPN cases (P=0.0455). Other genes, such as ASXL1, RUNX1, TP53, and cohesins did not show a skewed distribution. Interestingly, RAS mutations trended towards the founder clone in MPN and MDS/MPN cases, contrary to a trend towards small and multiple (up to 4) subclones in AML and MDS cases. Conclusions: We provide a proof-of-principle analysis of VAFs as a tool to determine tumor heterogeneity. We show examples of recurrent gene and disease-specific trends that may indicate the usefulness of VAFs in cases with multiple significant mutations.
M.S. Ahooja, O.A. Shetty, M.Y. Gurav, A. Uttarkar, M. Gupta, M. Sengar, S. Epari, T. Shet Tata Memorial Centre, Mumbai, Maharashtra, India. Introduction: MYD88 serves as an adaptor for TLR signaling and the L265P mutation is reported to be a driving force in overactivation of NFkB which results in cell survival. This has been widely studied in Activated B-cell-like Diffuse large B-cell Lymphoma (ABC-DLBL) and Waldenstorms Macroglobulinemia and seen in about 40% and 80% to 85% respectively. The study was aimed at screening for this mutation in HIV associated lymphomas to explore whether or not this mutation is responsible for the aggressive nature of the disease. Methods: The study was performed on archived formalin-fixed, paraffin-embedded (FFPE) specimen of HIV associated lymphomas (n=25).These lymphomas were subclassified into plasmablastic lymphoma, post-germinal center (GC) MUM1 positive DLBL, grey zone between Burkitt and plasmablastic and grey zone between plasmablastic and DLBL.DNA was extracted from FFPE tissues and MYD88 gene sequencing was performed using Sanger sequencer. Sequencing data was analyzed on Chromas Lite software. Molecular data was correlated with the clinicopathological parameters. Results: In the preliminary study of 25 cases, 13 were plasmablastic lymphomas, 8 were post-GC MUM1 positive DLBL. Three were tumors with grey zone between Burkitt and plasmablastic lymphoma and one tumor was grey zone between plasmablastic and DLBL. The age of the study population was 9 years to 70 years of age, mean age was 38 years and male to female ratio was 3:1. All plasmablastic occurred at extranodal sites, whereas three post-GC DLBL had extra-nodal presentation. MYD88 L265P mutation was detected in control cases (2/5) lymphoplasmacytoid lymphoma. However, all the 25 HIV associated lymphoma cases were found to be wild type for MYD88 gene. Conclusions: Though most HIV associated B cell lymphomas are aggressive tumors and show NFkB activation, MYD88 mutation is not seen in these tumors and MYD88 independent pathway maybe responsible for NFkB activation. Since CARD11 links the B-cell receptor (BCR) to the canonical NFkB activation pathway, we are exploring the link between MYD88, CARD11 and NFkB pathway in this ongoing study.
H23. Next-Generation Sequencing of a 54 Gene Panel to Assess Myeloid Disease L. N. Toth, J.D. Peterson, P. Kaur, D.L. Ornstein, G.J. Tsongalis, F.B. de Abreu Geisel School of Medicine at Dartmouth, Dartmouth Hitchcock Medical Center and Norris Cotton Cancer Center, Lebanon, NH. Introduction: Myeloid diseases include a spectrum of diagnoses ranging from disturbances in proliferation, differentiation, and self-renewal of hematopoietic clonal stem cells. Identification of specific genetic alterations and combinations thereof plays an integral role in diagnosis, prognosis, and treatment guidance. This study describes the identification of multiple genetic mutations in patients with myeloid disease using the TruSight Myeloid Sequencing Panel. Methods: Eleven patient samples were sequenced using at least 50 ng of genomic DNA. Oligo primes were hybridized to target regions, followed by extension and ligation for library preparation. Indices and sequence adapters were added by PCR amplification. Libraries were then purified, normalized, pooled, and sequenced on the Illumina MiSeq System. Base-calling and sequence alignment were performed using the built in MiSeq Reporter Software. VCF files were generated using the Somatic Variant Caller, and then uploaded and analyzed using VariantStudio v2.1. Results: Sequencing identified mutations in 10 patient samples whereas one was negative for mutation. Among the positive samples, a total of 32 genetic alterations were detected. The average number of mutations detected per sample was 2.9 (range: 1 to 8). The most common mutations identified was TET2 (4 samples) followed by TP53 (3 samples), ASXL1 (2 samples), CDKN2A (2 samples), SRF2 (2 samples), and STAG2 (2 samples). A TET2 mutation independently acts as a favorable prognostic indicator, as seen in one sample, but is frequently associated with SRF2 mutations, which was seen in another sample. An SRF2 mutation, like a TP53 mutation, is an unfavorable prognostic indicator with decreased overall survival and increased risk for disease progression to acute myeloid leukemia (AML). CDKN2A mutations also show increased risk for disease progression to AML but also have increased risk of treatment failure when using tyrosine kinase inhibitors. Conclusions: The detection of clonal mutations in myeloid diseases is integral to predicting disease progression and directing treatment. Identification of single mutations can direct clinical practice, though single gene detection methods can be more time consuming and less cost effective. The TruSight myeloid sequencing
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org panel allows for the identification of multiple mutations with a single test providing clinicians with more information allowing physicians and patients to make more appropriate treatment decisions and improve clinical outcomes.
Digital PCR Technology M. Mai, D.S. Viswanatha, R. He Mayo Clinic, Rochester, MN. Introduction: Quantitative test of JAK2V617F (V617F) is important given the variable mutated allele frequencies in myeloproliferative neoplasms (MPNs) and its potential role in therapy response monitoring. Currently there is no reference standard for JAK2V617F allele specific quantitative polymerase chain reaction (ASqPCR) assays. Digital PCR (DPCR) is a new paradigm of nucleic acid quantitation independent of a reference standard. This study evaluated the utility of a chip-based DPCR system for JAK2V617F quantification and its role in test standardization. Methods: PCR primers and probes for V617F and wild-type (WT) JAK2 were designed using Primer Express V2 (Life Technologies, NY). Oligo controls were from Integrated DNA Technologies, Iowa. Patient samples previously tested by ASqPCR were tested on the Quantstudio 3D DPCR system (Life Technologies, NY) with a 20,000 reaction partition. Quantstudio software was used for data analysis. Results: In DPCR, serial dilutions of V617F oligo in WT oligo exhibited excellent accuracy and linearity (R 2 =0.99) over a 4 log dynamic range (100% to 0.01%). The reproducible limit of detection (LOD) was determined at 0.1% given noise signals seen in WT oligo and normal samples up to 0.01%. In accuracy study, 27/29 ASqPCR+ patient samples were positive by DPCR. The two DPCRsamples did show low level allele fractions (AF, 0.03% and 0.02%) but were below the 0.1% cutoff. The respective ASqPCR AF were also low at 0.01% and 0.05% (ASqPCR LOD=0.01%). Compared to ASqPCR, higher AF was consistently seen in DPCR. In the 27 positive patients, the DPCR/ASqPCR ratio was 7.7± 4.0 (mean±SD); the difference was more pronounced at low AF (10.4±2.7, n=12, AF<3% ASqPCR), and minimal at higher AF (1.6±0.5, n=3, AF>35% ASqPCR). 8 normal samples were negative by both ASqPCR and DPCR. Reproducibility in 2 patient samples with high and low mutant AFs were 84.5 ± 0.8% and 2.04 ± 1.7%; these results were not inferior to our ASqPCR assay. The HEL cell line-based ASqPCR calibrator (1:1 mix of HEL and HL-60) showed a mean V617F/WT ratio of 5.4 (5.4±0.3, n=4). Conclusions: DPCR has a strong potential for standardization of V617F quantitative test and provides independent absolute quantification. The analytic characteristics are similar to ASqPCR; however, LOD performance can be significantly improved by increasing the reaction partitions. Detected V617F AF was consistently higher by DPCR compared to ASqPCR, especially at lower AF, reflecting analytical platform differences and underestimation of the V617F/WT in a HEL-based ASqPCR calibration system. The latter suggests caution using HEL as a reference material for V617F quantification and underscores the value of DPCR to critically evaluate reference materials in quantitative clinical assays. Introduction: Chronic lymphocytic leukemia (CLL) is the most common type of leukemia in the western hemisphere, representing 30% to 40% of all leukemias worldwide. CLL is often an indolent disease, yet clinical course is very variable. Notch1 is a transmembrane receptor, its signaling pathway regulates cell-cell interactions. In CLL the vast majority (97%) of mutations are clustered in exon 34 and approximately 80% correspond to a dinucleotide frameshift deletion at the P2514 amino acid position. Mutations in NOTCH1 (frequency 4% to 15%) are associated negatively to diverse clinical scenarios such as therapy refractoriness particularly to rituximab, transformation to diffuse large B cell lymphoma, Richter's syndrome and short overall survival even after complete remission. Methods: Peripheral blood samples were collected from 50 patients diagnosed wirh CLL. Genomic DNA was extracted using a column based purification kit. High Resolution Melting Analysis (HRM) was performed with HPLC purified primers designed and aligned using NOTCH1 genomic reference sequence (NG_007458.1 RefSeqGene). The size of the HRM amplicon was 141 base pairs containing 65% of guanidine and cytosine content. The reaction contained the following reagents: high resolution melting master 2X (LightCycler 480 Roche Diagnostics GmbH Mannheim, Germany), 95°C for 10 minutes, followed by 50 cycles of 95°C for 15 seconds, 53°C for 10 seconds and 72°C for 10 seconds. Melting temperature ranged from 40°C to 70°C with a ramp of 0.02°C/s and 25 acquisitions per °C. LightCycler 480 (Roche Diagnostics GmbH Mannheim, Germany) real time instrument and LightCycler 480 1.5.0 SP4 software (Roche Diagnostics GmbH Mannheim, Germany) were used to perform the reaction and data analysis. All samples were studied by direct sequencing analysis to compare and confirm HRM analysis results using the Genetic Analyzer 3130 instrument (Applied Biosystems Inc. Foster City, CA) with sequence specific primers. Results: Discrimination of mutated samples from wild type (WT) was possible by analyzing the difference in the relative fluorescent signal during the melting step. The gene scanning program revealed 3 different melting patterns or groups: two positive groups and one negative group. Sequencing results showed the dinucleotide frameshift deletion P2514fs mutation, and a nonsense point mutation E2515X. Analytical sensitivity analysis was performed in duplicate using a mutated sample subjected to serial dilutions: 1:1, 1:4, 1:10 and 1:20 with a WT sample. The test showed detection sensitivity up to 1:10 in both replicates (10%). Conclusions: The test was able to detect the dinucleotide deletion p.P2514fs mutation which corresponds to 80% of the PEST domain mutations and also a premature stop codon p.E2515X nonsense mutation. The assay had a good performance and showed high analytical sensitivity. We present a reliable screening HRM method capable to detect the most frequent NOTCH1 mutation located in the PEST domain.
H26. Molecular Characterization and Clinical Impact of Acquired KMT2A Partial Tandem Duplication in Acute Myeloid Leukemia C. Gronseth, S. McElhone, J. Wilson, K. Kroeger, E. Estey, M. Fang Seattle Cancer Care Alliance, Seattle, WA. Introduction: Partial Tandem Duplication (PTD) of the KMT2A gene (aka MLL-PTD) can be identified at the molecular level by chromosome genomic array testing (CGAT) or RT-PCR; it cannot be detected by conventional cytogenetics or fluorescent in situ hybridization (FISH) techniques. KMT2A (MLL) gene rearrangements have been well described in patients with acute myeloid leukemia (AML) and are generally associated with a poor prognosis. However, clinical testing for MLL-PTD is not routinely performed for AML patients despite recommendation by the European LeukemiaNet. The recent evidence that MLL-PTD leukemic cells are sensitive to small molecule DOT1L inhibition and that liposomal bortezomib eradicates MLL-PTD/FLT3-ITD double-mutation AML underscore the importance of identification of AML patients with MLL-PTD for effective therapy. Methods: We performed chromosome genomic array testing (CGAT) on 129 AML patients (median age 53y) using CytoScanHD and analyzed the data with ChAS and Nexus software. MLL-PTD was identified along with chromosome copy number aberrations (CNAs) and copy-neutral loss-of-heterozygosity (cnLOH). Results were correlated with available cytogenetics, FISH, molecular mutation testing, and flow cytometry results, as well as clinical and outcome data. Results: Eight of 129 patients (6.2%) demonstrated MLL-PTD ranging from 16 Kb to 121 Kb (median size 27 Kb) and spanning exons 1 through 10 of the KMT2A gene, represented by 64 to 209 (median 106) probes. The minimum overlapping region (MOR) of the MLL-PTD in these patients is 16 Kb long and includes exons 2 to 8, with the genomic coordinates of 118,338,294 to these patients in addition to MLL-PTD although cytogenetics was only abnormal in three. Two of the patients were positive for FLT3-ITD, one of whom with 13q cnLOH. Another patient showed a RUNX1 mutation along with 17q cnLOH. Flow cytometry showed a median of 52% abnormal blasts. For the outcome study, 112 patients had follow-up data; five of them had MLL-PTD. There was no significant difference in overall survival, complete remission rate, and remission duration between patients with and without MLL-PTD. The four MLL-PTD patients who achieved CR had intermediate risk cytogenetics whereas the one not achieving CR had poor risk cytogenetics and complex CGAT results. One patient showed both MLL-PTD and FLT3-ITD and showed the shortest remission duration and survival. Conclusions: The 6% incidence of MLL-PTD in this AML study is consistent with the approximately 5% to 11% reported in literature. CGAT is an effective method to routinely identify this important prognostic and predictive marker in addition to chromosome aberrations.
Leukemia Using Next-Generation Sequencing S. Shin, I. Hwang, M. Choi, S. Lee, C. Jong Rak Yonsei University College of Medicine, Seodaemun-gu, Seoul, Korea. Introduction: Presence of minimal residual disease (MRD) after treatment of Blymphoblastic leukemia (BLL) has gained importance as prognostic factor. Immunoglobulin heavy chain (IGH) gene rearrangement test can be used as followup marker in B-lymphoproliferative disorders, including BLL. We conducted nextgeneration sequencing (NGS) using IonTorrent PGM (Life technologies) and LymphoTrack IGH Assay (InVivoScribe Technologies) to assess applicability and performance of NGS in MRD monitoring in BLL. Methods: We performed IGH gene clonality analysis using GeneScanning and NGS in initial diagnosis and follow up bone marrow samples of seven B lymphoblastic leukemia patients. GeneScanning and NGS were done using IdentiClone IGH Gene Clonality Assay (InVivoScribe Technologies) and LymphoTrack IGH Assay, according to manufacturer's instructions. The frequency of each identical sequence in NGS was derived by calculating the number of sequence reads divided by the total number of sequence reads in the sample. In NGS interpretation, we determined 5% frequency as cut-off for leukemic clone in initial diagnostic samples. The presence of MRD in follow-up samples was followed by identical sequence of leukemic clones in diagnostic samples. Results: All initial diagnostic samples were positive for IGH clonality with GeneScanning and NGS. With NGS, 1 to 3 leukemic clones were existed. Five patients had 2 leukemic clones, and the other 2 patients each had 1 clone and 3 clones. The frequencies of leukemic clones in diagnostic samples were 33% to 64.7% with NGS. Among 11 follow-up samples, 3 samples were both positive with jmd.amjpathol.org ■ The Journal of Molecular Diagnostics GeneScanning and NGS. The mean frequencies of leukemic clones were 13% (range, 13.7% to 15.4%) with NGS. Four follow-up samples were both negative with GeneScanning and NGS. The remaining 4 samples were only positive with NGS and negative with GeneScanning. The mean frequencies of leukemic clones were 1.49% (range, 0.14% to 3.1%). Conclusions: We performed IGH clonality analysis of initial and follow-up bone marrow samples of BLL patients using NGS. With NGS, we detected four more positive samples (4 out of 11, 36.4%) compared with GeneScanning. In other samples, NGS showed identical results with GeneScanning method. The prognostic significance of low-level leukemic clone detected by NGS can be established in further studies. NGS can be applicable in MRD monitoring in BLL.
H28. Improved Diagnosis for Familial Myelodysplastic Syndromes and Acute Leukemia Using Next-generation Sequencing and Splicing Analysis L. Guidugli, A. Knight Johnson, G. Alkorta-Aranburu, V. Nelakuditi, K. Arndt, F. Kobiernicki, D. Waggoner, J. Churpek, D. Del Gaudio, S. Das, Z. Li University of Chicago, Chicago, IL. Introduction: Myelodysplastic syndromes (MDS) and acute leukemia (AL) are hematological malignancies characterized by abnormal production of the bone marrow cells. Patients with familial MDS/AL are suspected when 2 or more first-or second-degree relatives within a family develop MDS or AL or have chronic unexplained low blood counts. Currently, clinical diagnostic testing is available in a few CLIA-licensed laboratories only for 5 genes associated with MDS/AL and is based on Sanger sequencing. To improve familial MDS/AL diagnosis, our CLIAlicensed laboratory has recently developed a next generation sequencing (NGS)gene panel for MDS/AL syndromes. In this study we explored the clinical utility of this panel and performed RNA studies on a selected variant. Methods: Ten patients with familial MDS/AL were analyzed on the NGS panel that includes 8 known predisposing genes (CEBPA, RUNX1, GATA2, ANKRD26, SRP72, TERC, TERT and TP53) . The DNA obtained from cultured skin fibroblasts or peripheral blood was used for preparing DNA libraries with the SureSelectXT Enrichment System. Next, DNA libraries were sequenced on an Illumina MiSeq instrument and the NGS data was analyzed with a custom bioinformatic pipeline. Reverse transcription PCR, Sanger sequencing and quantitative PCR were utilized for splicing analysis. Results: Pathogenic sequence changes were identified in 2 samples (20%) and were both novel missense changes in the GATA2 gene. One of the variants, c.857C>T, was predicted to cause aberrant splicing by in-silico prediction tools. The RNA analysis confirmed the generation of a cryptic donor splice site that resulted in a frameshift and a premature stop codon. Quantitative PCR showed a significant reduction in the RNA expression of the GATA2 gene in the patient harboring the variant compared to controls. These results indicate that the c.857C>T is a likely pathogenic sequence change and the predisposing mutation for this patient's MDS. Conclusions: This study demonstrates the effectiveness of NGS-gene panel analysis coupled with RNA splicing assay in uncovering molecular diagnoses in patients with MDS/AL. This is also the first study to report a GATA2 missense change that affects splicing. The identification of GATA2 likely pathogenic variants in 20% of patients highlights the critical contribution of this gene in MDS/AL predisposition. Such contribution may have been underestimated due to the clinical heterogeneity of the phenotype of patient's with GATA2 mutations. Overall, this study has provided additional evidence to better understand the genetic etiology and broaden the gene mutation spectrum of familial MDS/AL syndromes.
H29. TAF15-ZNF384 Rearrangement in Acute Lymphoblastic Leukemia S. Shin, K. Ham, K. Hong, H. Kim, S. Lee, J. Choi Yonsei University College of Medicine, Seodaemun-gu, Seoul, Korea. Introduction: The TAF15 RNA polymerase II, TATA box binding protein (TBP)associated factor (TAF15) gene at chromosome 17q12 encodes a RNA-binding protein as a component of multi-subunit transcription initiation factor complexes. The zinc finger protein 384 (ZNF384) gene at chromosome 12p12 encodes a C2H2-type zinc finger protein functioning as transcription factor. Translocation between TAF15 and ZNF384 has been reported rarely in acute lymphoblastic leukemia. The clinical characteristics and prognostic impact of this fusion is not well known. Here we report a B lymphoblastic leukemia with t(12;17)(q13;q11.2); TAF15 -TNF384 translocation. Methods: A 32-month-old girl was admitted for 4 days of fever and erythema of both legs. Initial complete blood count result demonstrated pancytopenia, with a hemoglobin level of 5.5g/dL, a white blood cell count of 2.63x109/L, and a platelet count of 123x109/L. Leukemic blasts were observed 5% in peripheral blood and 84.9% in bone marrow. The blasts were positive for CD34, CD19, CD13, CD33, cytoplasmic CD79a, and terminal deoxynucleotidyl transferase (TdT) and negative for CD2, CD7, CD10, CD14, myeloperoxidase, and cytoplasmic CD3 on immunophenotyping tests. For cytogenetic analysis, the bone marrow sample was cultured for 24 hours in mitogen free media. Chromosome study was performed using G(Giemsa)-T(Trypsin)-G-banding technique and karyotyped by observing 20 metaphases. For confirmation of fusion transcript, complementary DNA was synthesized by total RNA isolated from EDTA anti-coagulated bone marrow sample from the patient. Fusion gene product was amplified and analyzed by polymerase chain reaction and direct sequencing using primers specific for TAF15 and ZNF384 genes. Results: Cytogenetic study revealed a karyotype of specific primers was approximately 800 bp. Aligned sequence of the product revealed breakpoints between exon9 of TAF15 and exon3 of ZNF384. The patient was diagnosed with B lymphoblastic leukemia and achieved complete remission after induction chemotherapy at 11 month after diagnosis. Conclusions: We described a rare translocation of t(12;17)(q13;q11.2) in B lymphoblastic leukemia patient. TAF15 -ZNF384 fusion was confirmed by polymerase chain reaction and direct sequencing of complementary DNA. Our patient achieved complete remission after first induction chemotherapy. More data is needed to determine the prognostic impact of TAF15 -ZNF384 fusion in acute lymphoblastic leukemia.
Large B-cell Lymphomas D. Xia 1 , P. Michaels 1 , A. Guttapalli 2 , R. Joyce 1 , J. Houldsworth 2 , R. Dewar 1 1 Beth Israel Deaconess Medical Center, Boston, MA; 2 Cancer Genetics Inc., Rutherford, NJ. Introduction: Diffuse large B-cell lymphoma (DLBCL) is the most common high grade lymphoma in the Western Hemisphere. The disease exhibits a wide spectrum of clinical aggressiveness, and is characterized by heterogeneity at the histologic, cytogenetic, and molecular levels. One goal of this study is to identify novel molecular prognostic biomarkers in DLBCL patients, and to integrate these with existing prognostic biomarkers. In a previous study using publicly available DLBCL datasets, we defined a set of 50 common copy number variations (CNVs) involving 36 genomic loci, for which robust scoring criteria were established (Dias L, Thodima V, Friedman J, Guttapalli A, Mendiratta G, Syrbu SI, and Houldsworth J. December, 2014. Robust Assessment of Genomic Imbalance in Diffuse Large B-Cell Lymphoma Confirms Inferior Outcome Is Associated with Genomic Complexity and Identifies Potential Therapeutic Pathway Targets. Poster presented at: American Society of Hematology Annual Meeting 2014). Here, we present the initial results of an attempt to systematically validate these CNVs using formalin-fixed, paraffin-embedded tissue blocks from an additional group of DLBCL patients. Methods: Patients diagnosed with de novo DLBCL and treated with R-CHOP from 2003 to 2009 at the Beth Israel Deaconess Medical Center (BIDMC) were identified through retrospective review of electronic medical records (EMR). Sections of one formalin-fixed, paraffin-embedded block from each patient were submitted for custom array comparative genomic hybridization (aCGH) at Cancer Genetics Inc., and each sample was scored according to the established criteria for the presence/absence of each of the 50 CNVs. Patient survival data was obtained through the BIDMC EMR. Kaplan-Meier survival analysis was performed using the R statistical software.Bonferroni correction was employed to adjust for multiple testing (cutoff p-value < 0.001). Results: To date, the 50 CNVs were successfully assayed in samples from 45 DLBCL patients with survival data. In univariate analyses, the presence of four of these CNVs were associated with poor survival at a p-value of <0.05. One of the four CNVs (loss at 1p13.1) was significantly associated with poor outcome, after adjusting for multiple testing (p = 1.43 x 10^-5). Conclusions: Our ongoing systematic analysis of common copy number variations in DLBCLs identified several biomarkers potentially informative about patient survival. Further validation with additional cases and correlation with immunostains for cell of origin studies (ABC versus GCB subtypes) are in progress. The utility of each potentially useful novel biomarker will be assessed via multivariate analyses along with prognostic clinical variables. The somatic mutation JAK2-V617F is associated with BCR-ABL1negative myeloproliferative neoplasms (MPNs). Detection and quantification of the JAK2-V617F mutation is now part of clinical diagnostic algorithms and may provide a method to monitor response to therapy. It has been shown that the level of JAK2-V617F detectable in the peripheral blood (PB) of some patients with MPNs can be as low as 1% at the time of diagnosis, which is below the detection limit of some standard diagnostic methods. Previous studies have emphasized the usefulness of quantitative PCR (qPCR) assessment of residual JAK2-positive cells to predict outcome following allogeneic stem cell transplantation (SCT). A significantly higher risk of relapse and poor overall survival have been associated with the detection of JAK2 mutant levels >1% early after SCT. Thus, a reliable method to quantify low levels of JAK2-V617F mutations is critical. Recently, Qiagen introduced a new assay for the detection and quantification of the JAK2-V617F: the JAK2 RGQ PCR kit. This assay has been optimized to improve the sensitivity of the detection of JAK2-V617F mutation by using newly designed primers, standards, and probe mixes in line with recent international technical recommendations. We evaluated the analytical performance characteristics of this assay and compared results with two currently available Qiagen assays. Methods: The study involved distribution of 72 DNA samples among 3 laboratories. Qiagen JAK2 RGQ PCR reagents and reaction conditions to be used were provided by Qiagen. The qPCR reactions were performed on the RotorGene Q 5-Plex platform. Laboratories evaluated the distributed material in parallel with their established in-house method (ipsogen JAK2 MutaQuant or ipsogen JAK2 MutaScreen, Qiagen). Raw and analyzed data were submitted by participating laboratories for central independent analysis and statistical evaluation (Genetics Associates Inc, Nashville, TN). The accuracy, precision, analytical measurement range, analytical sensitivity, and analytical specificity of this assay were evaluated and compared with other Qiagen assays currently used by many laboratories. Assay workflow and cost analysis were also assessed. Results: Twenty-nine JAK2-V617F-positive samples were identified among 72 samples analyzed, with a mutation concordance rate >97%. Specificity of the assay was demonstrated by absence of mutant sequence amplification up to 38 cycles of PCR; sensitivity experiments indicated a limit of detection of 0.05% mutant sequence in a background of wild type sequence. The test measured linearly over a wide logarithmic range and exhibited excellent reproducibility. Compared with other Qiagen assays currently available for the detection of the JAK2-V617F mutation, the new JAK2 RGQ PCR assay demonstrated a lower limit of detection, a simplified workflow, and a decrease in cost per run. Conclusions: The new Qiagen JAK2 RGQ PCR assay is a robust and reliable assay with a sensitivity suitable for monitoring minimal residual disease in therapeutic trials designed to target JAK2-V617F-positive MPNs, as well as predicting outcome and guiding management of patients undergoing allogeneic transplantation. H32. A State of the Art Custom "Duplexed" Real-Time PCR Assay for Minimal Residual Disease Monitoring of Chronic Myeloid Leukemia Using Locked Nucleic Acid Probes. S. K. Joshi, P. Subramanian, S. Chaudhary, P. Tembhare, S. Gujral, H. Doshi, S.K. Joshi, N. Patkar Tata Memorial Centre, Kharghar, Maharashtra, India. Introduction: In India, treatment of Chronic Myeloid Leukemia (CML) is supported through an International Patient Assistance Program in which Imatinib mesylate is provided to economically disadvantaged patients at no cost. As a result, nearly all patients opt for treatment and need regular molecular monitoring. However, commercial kits for BCR-ABL monitoring are cost prohibitive for our patients. It is also currently recommended to test for BCR-ABL copy numbers in triplicates to ensure making confident calls at low levels placing further economic strain in resource constrained settings especially if all quality control measures are to be followed. In that context it is required to develop cost effective BCR-ABL monitoring technologies. Globally, a majority of centres use the EAC protocol for CML, an assay that suffers from background noise. We addressed these problems by modifying the legacy real time PCR (qPCR) process. We cloned a duplex qPCR compatible plasmid that enables us to multiplex BCR-ABL and ABL reaction in a single well using FAM-black hole quencher1 (BHQ1) and HEX-BHQ1 probes modified with locked nucleic acids (LNA). The aim of the study to develop a custom duplex PCR compatible plasmid and develop and standardize a LNA probe based assay for BCR-ABL and ABL. Methods: A 1600bp amplicon from a newly diagnosed CML (e13a2) patient into a pJET1.2/blunt cloning vector. This was linearized using Avogadro number, an estimate of copy numbers was arrived at and a standard curve was created. Accuracy of dilutions was confirmed by using Ipsogen standards. Using Modified EAC protocol, BCR-ABL probe was truncated to 16 nucleic acids labelled with FAM and BHQ1 with 3 LNA modifications whereas the ABL probe was 19 nucleic acids (HEX and BHQ1) with 4 LNA modifications. Calibration to the WHO calibrator was done using a secondary standard (Asuragen Inc) to derive an international scale conversion factor. Assay precision was monitored using 2 levels of controls in the duplex assay (median NCN 0.1% and 0.05%). We compared the normalized copy number (NCN) derived from the duplex assay with the simplex assay in 87 samples of CML (NCN ranging from 0.002 to 108.5, median 0.14%). Results: The plasmid was successfully cloned and confirmed by Sanger sequencing. The duplex assay did not lead to loss of MMR in any patient. In addition, we did not face any issues with false positivity. We found a good correlation between the simplex and duplex assay (r2=0.97). In all runs, the slope was between -3.2 to -3.6, R2>0.98. The international scale conversion factor was 1.13. Over 120 runs the BCR-ABL assay had a CV of 50.84% and 58.1% for high and low precision controls. Conclusions: This assay enabled us to maintain a high level of quality control (runs in triplicates, precision, no template as well as no reverse transcription controls), and to process 22 samples per run. As a result, the assay became cost-effective. We would be happy to share this plasmid to other users from resource constrained setting such as ours. To the best of our knowledge; this is the first study to have used LNA probes for CML minimal residual disease (MRD) monitoring. The use of a duplex PCR enables us to significantly decrease the cost of the assay ($5/patient), yet maintain a high standard of quality control. The core binding factor acute myeloid leukemia (CBF AML) defined by the presence of t(8;21)(q22;q22) or inv(16)(p13q22)/t(16;16)(p13;q22), constitutes about 17% of AML and is known to have a favorable prognosis because outcome is better than with other subtypes of AML. The detection method has not been assigned for risk assessment, although c-KIT exon17 mutation is a poor prognostic factor in CBF AML. It is necessary to evaluate the analytical performance and clinical significance for the new application of sensitive methods with higher detection rate. Methods: We developed an allele-specific, real-time quantitative PCR (AS-qPCR) assay and evaluated its performance for detection of common c-KIT mutations in CBF AML. AS-qPCR and melting curve analysis were performed and compared. In cases with discrepant results between AS-qPCR and melting curve ana -modified oligonucleotides PCR were used to confirm the mutations. Results: A total of 111 CBF AML patients (74 with t(8;21) and 37 with inv(16) ) were included in the study. Sixty nine patients (62.2%) had c-KIT exon 17 mutations (D816 and/or N822). Among 69 patients with c-KIT mutations, 30 patients (43.5%) had multiple mutations, which indicate the heterogeneous populations of leukemic cells. AS-qPCR was more sensitive than melting curve analysis and detected additional 28 cases with c-KIT mutation with low mutant allele level. We set the cut-off value of the relative mutant allele level as 10 10, n=26) showed significantly inferior OS (P =0.005; 3-year OS, 35.6%) and EFS (P=0.03; 3-year EFS, 28.9%), whereas low mutant level (<10) did not influence the prognosis. Monitoring of c-KIT mutant level after treatment revealed similar trend with fusion transcripts level (RUNX1-RUNX1T1 or MYH11-CBFB). Interestingly, 2 patients who harbored double c-KIT mutations at diagnosis showed selective proliferation of minor clones with low mutant level at relapse. Conclusions: We have successfully developed a highly sensitive detection method to identify common c-KIT exon 17 mutation and quantify the relative mutant levels. The small populations of leukemic cells with c-KIT mutations could be detected and the linearity was excellent. The incidence of c-KIT exon 17 mutations in CBF-AML is highest and the heterogeneous populations of leukemic cells exist as minor subclones along with the instead of c-KIT mutation positivity. The highly sensitive, quantitative measurement of the c-KIT exon 17 mutation is a valuable tool in evaluation of prognosis at diagnosis and subsequent patient follow up. A.A. Almohammedsalim, R. Luthra, C.Y. Ok, S. Loghavi, M. Mehrotra, R. Abraham, M. Harmon, X. Lu, L.J. Medeiros, K.P. Patel, R. Kanagal-Shamanna University of Texas MD Anderson Cancer Center, Houston, TX . Introduction: Traditionally, fluorescence in situ hybridization (FISH) and conventional karyotyping (CK) have been used for identification of genetic alterations of prognostic significance in chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL). Recently, array-based comparative genomic hybridization (aCGH) has gained acceptance as a high-resolution platform for genome-wide assessment of DNA copy number alterations (CNA) in CLL/SLL. In this study, we compared the genetic abnormalities detected by a high-density CGH+SNP microarray with FISH and CK in a large series of CLL/SLL cases in a CLIA-laboratory setting. Methods: One hundred and sixty-eight consecutive CLL/SLL cases involving the bone marrow (BM)/ peripheral blood (PB) were tested using Agilent CGH+SNP Microarray (SurePrint G3 4x180K). DNA from BM/ PB was extracted using automated ReliaPrep and Maxwell systems (Promega). Selection criteria included > 20% involvement by morphology/ flow. Results were analyzed using Agilent CytoGenomics software v2.9 and compared with FISH and CK. Results: Array-CGH identified recurrent CNAs in chromosomes 11, 12, 13, 17, 2p, 6q and 8p in 117 cases (67%). FISH and CK results were available in 101 and 82 cases respectively. Concordance between aCGH and FISH was 80% and CK was 84%. Sixteen FISH positive cases were negative by aCGH. In all these cases, the aberrations were present in <20% of cells (below the analytical sensitivity of aCGH). After excluding the balanced translocations, in 13 cases, abnormalities in CK were not detected by aCGH. Of these, 8 cases showed aberrations in 1/20 metaphases; 3 were of composite karyotype; and 2 had marker chromosomes. Array-CGH identified additional CNAs not detected by FISH or CK in 4 and 20 cases respectively. Novel recurrent 3p loss was identified in 9 cases. Copy-neutral loss of heterozygosity was noted in a subset of cases. TAT for aCGH from sample receipt to report was <7 days, with 57% of cases yielding aCGH results prior to CK/FISH. Conclusions: We have successfully implemented the CGH+SNP microarray for routine clinical testing of patients with CLL/SLL. Compared to FISH, CGH+SNP is a cost-effective and less laborious platform for genome-wide assessment of novel and recurrent CNAs and CN-LOH with a comparable TAT. D.L. Duncan, C.J. Civalier, K.E. Weck, N.M. Patel University of North Carolina at Chapel Hill, Chapel Hill, NC. Introduction: There is increasing clinical demand for DNA sequencing of myeloid neoplasms based on research detailing genes involved in prognosis and pathogenesis in acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), and myeloproliferative neoplasms (MPN). Therefore, we validated 3 bioinformatically defined panels based on sequencing produced using the Illumina TruSight Myeloid kit to address this demand. Methods: Twenty-eight archived DNA samples, extracted from either blood or bone marrow, and 1 cell line standard (Horizon Diagnostics) were used for validation. Of the archived samples, 14 had documented mutations that were expected to be identified by the assay. The remaining 14 samples were either non-neoplastic (6 of 28), or clinically neoplastic with no documented mutation (8 of 28). PCR-generated libraries were sequenced using V3 chemistry on an Illumina MiSeq instrument. Each sequencing run multiplexed 8 specimens on a single flow cell. Variant calling was performed by MiSeq Reporter v2.3 with subsequent analysis using Illumina VariantStudio v2.2 and IGV. Collaboration with oncologists and hematopathologists and literature review were used to generate lists of relevant genes for 3 indications: AML, MDS and MPN. Bioinformatic filtering using the VariantStudio software generated output restricted to variants encompassed by predefined gene lists. Results: All 29 samples generated successful libraries, demonstrating that DNA extraction and library preparation procedures are robust. Of the 14 samples with previously documented variants, all variants were confirmed. The commercial cell line possessed 5 variants covered by the assay, all of which were identified. Of note, 1 of the clinically neoplastic samples with no previously documented gene variants was found to have the L367fs*46 52 base pair deletion in CALR, verified by Sanger sequencing. Based on this validation set, overall analytical sensitivity and specificity for the assay were 100%. Limit of detection was established using dilutions of the commercially available cell line standard, with successful detection of all variants at 5% allele frequency. Conclusions: We validated 3 myeloid mutation panels supported by Illumina technology to detect clinically relevant genetic variants in AML, MDS, and MPN. These panels have replaced other testing methodologies for identifying commonly reported variants in NPM1, JAK2, and the FLT3 tyrosine kinase domain. The panels offer coverage of many genes with newly recognized clinical importance in myeloid disease while limiting detection of variants of unknown significance in genes of unknown clinical relevance. This broader coverage allows us to assist in diagnosis as well as fine tune prognosis in myeloid neoplasms.
Lymphocytic Leukemia N. Rabade, A. Bibi, P. Amare, S. Joshi, S. Gaware, S. Chaudhary, F. Mishra, H. Doshi, R. Bane, P. Tembhare, P. Subramanian, S. Gujral, N. Patkar Tata Memorial Centre, Kharghar, Maharashtra, India. Introduction: The B cell receptor (BCR) contains the immunoglobulin (IG) molecule as a key component. Alterations in the BCR form a foundation on which CLL immunogenetics is based upon. There is no data from India which has addressed the immunogenetics of CLL. The determination of the mutational status of rearranged immunoglobulin heavy chain variable (IGHV) genes in patients with CLL has shown strong and independent prognostic value. We analyzed clinically important variables such as identification of immunoglobulin heavy chain (IGVH) mutations & IGVH gene usage. Methods: Seventy-seven consecutive cases of CLL (based on WHO 2008 criteria.) were prospectively accrued from May 2013 to May 2015. Genomic DNA was isolated and amplified with "leader" primer sets for the V regions and the downstream joining (J) region of the IGVH gene. Monoclonal amplicons spanning the variable (V) region and the complementarity-determining regions (CDR) were sequenced bidirectionally. Comparative sequence analysis of the tumor and germline IGVH sequences were performed on IMGT reference database. The mutation status was correlated with clinical staging, immunophenotype and cytogenetics. Results: All patients showed clonal VH gene rearrangements. Of the 77 cases a majority of patients showed lack of somatic hypermutations (53.3%). A biased VH usage was seen towards VH3 & VH1 genes (33.7% and 28.5% respectively). Common VH genes used were IGHV2-5 (10.3%), IGVH1-2(9.2%), IGVH 1-69, IGVH 3-30, IGVH 3-33, IGVH 4-59 and IGVH 4-34 (6.5% each). On immunophenotypic analysis (96.7%) of the CD19, CD20 positive B cells co expressed CD5 and CD23 and 96.8% cases showed CD43 and CD200 coexpression. CD38 expression was more common in unmutated cases (32.5%) as compared to the mutated ones (12.9%). 13q deletion was the commonest clonal cytogenetic abnormality detected in 50% of the cases. p53 deletion which was more common in the unmutated cases (19.8%) as compared to the mutated ones (7.1%). Conclusions: The immunogenetics of Indian CLL is unique in showing a higher predominance of IGVH unmutated cases. This data has not been previously reported from India. This may account for clinical heterogeneity and varying outcomes in CLL in the Indian population. F. Sabato, C.I. Dumur, P. Morris, D. Surve, A. Ferreira-Gonzalez Virginia Commonwealth University Health System, Richmond, VA. Introduction: Quantification of BCR-ABL1 transcripts by reverse transcription realtime PCR (RT-qPCR) has proven to be the most sensitive method available and has prognostic impact with regard to disease progression for CML patients undergoing TKI therapy. Achieving defined levels of BCR-ABL1 on the international scale (IS) within specific time frames is an important measure for assessing patient response to treatment and probability for relapse and progression. The 100% BCR-ABL1 IS corresponds to the standardized baseline value and the 3 Log reduction to 0.1% BCR-ABL1 IS (defined as major molecular response, MMR or MR3). With the advent of second generation TKIs that produce faster and deeper levels of responses, a method that is able to detect MR4 or even MR4.5 with high precision and accuracy would be useful. Herein, we describe the performance evaluation of a set of reagents, Quantidex BCR-ABL IS CMR (Asuragen, Austin TX), for quantitating BCR-ABL1 transcripts e13a2 and/or e14a2 using RT-qPCR for patient monitoring.
Methods: RT-qPCR was performed using the Quantidex BCR-ABL IS CMR reagents. Precision and linearity studies were conducted using 4 reference samples constructed from cell lines that corresponded to 10% to 0.01% BCR-ABL1 IS (MR1, MR2, MR3 and MR4). These samples were tested over 4 separate runs on an ABI 7500 Real-Time PCR Standard System. Lower limit of detection (LOD) was determined by testing 4 contrived CML human RNA reference samples (MR 4.0, MR 4.5, MR 4.6, MR 4.7) in quadruplicate. A correlation study was performed using 95 archival, de-identified RNA samples from CML patients previously tested with our Laboratory Developed Procedure (VCU-LDP). Of these samples, 42 were paired samples from 16 CML patients at baseline and 12 to18 months after initiation of treatment. Results: The Quantidex reagents showed linearity for the interval tested (R2=1) with a maximum observed SD (%IS) of 2.05. LOD was defined at 0.002% BCR-ABL1 IS or MR 4.7. High level of total precision was achieved for all 4 levels of reference samples tested with a maximum %CV of 22%. All undetectable samples by VCU-LDP were undetectable with Quantidex. High correlation was obtained for the 80 positive samples analyzed by both methods (R = 0.9687) with VCU-LDP values being lower on an average of -0.18% Log BCR-ABL1 IS (limits of agreement of 0.59 and -0.95) and a positive bias for samples with > 1.55% Log BCR-ABL1 IS. MR results obtained with VCU-LDP for the 21 cases correlated well with the Quantiplex reagents (R = 0.9857). Conclusions: The Quantidex BCR-ABL1 IS CMR reagents showed high performance for measuring MMR and its clinical utility should be further evaluated. A. Bibi, K. Chopra, S. Mirgal, P. Amare, S. Joshi, S. Chaudhary, B. Arora, S. Banavali, G. Narula, M. Sengar, H. Jain, P. Tembhare, P.G. Subramanian, S. Gujral, N. Patkar Tata Memorial Centre, Kharghar, Maharashtra, India. Introduction: Copy number alterations have been described in precursor B lineage acute lymphoblastic leukemia (B-ALL) providing meaningful insight into its pathogenesis. These are non-random alterations involved in the critical B cell developmental pathways. The most common of these novel genetic alterations are PAX5 and IKZF1. Multiplex ligation probe assay (MLPA) has been able to detect these CNAs in a cost effective manner. The aim of our study was to identify the baseline frequency of IKZF1 deletions in all consecutive patients of B-ALL and to evaluate the genetic determinants of MRD positivity as evidenced by copy number alterations. Methods: Genomic DNA was obtained from healthy controls as well as 118 patients of B-ALL at diagnosis. For cases that were positive by immunophenotyping for minimal residual disease (MRD), DNA was examined using the SALSA MLPA kit P335 which includes 57 probes including IKZF1, CDKN2A/B, PAX5, EBF1, ETV6, BTG1, RB1 and the PAR1 region (CRLF2, CSF2RA, IL3RA). Approximately 150ng of DNA was hybridized to the probe mix and subjected to amplification as per manufacturer's recommendations. PCR amplicons were subjected to capillary electrophoresis on an ABI3500 genetic analyzer. Analysis was done on Coffalyzer software (v.140). In all patients including MRD negative cases, a breakpoint PCR was used to identify the IKZF1 deletion. Results: We analyzed 118 samples over a 6 month period. The age range was from 1 to 39 years, the median age being 7 years. The male to female ratio was 2:1.The patients were classified into good risk, intermediate risk and high risk cytogenetics. The distribution of the patients in various risk groups were as follows: good risk comprised of 41/118 (34.7%) cases whereas the intermediate and high risk groups constituted 23/118 cases (19.5%) and 11/118 cases (9.3%) respectively. The overall frequency of IKZF1 deletion in our study was found in 29/118(24.6%) cases. A high frequency of IKZF1 deletion upto 50% was found in BCR-ABL1 positive ALL.The most common genetic abnormality by MLPA in MRD positive cases was IKZF1 deletion 11/35 cases (31.4%). This was associated with deletion of PAX5 gene in 6 out of 35 cases and ETV6 and CDKN2A deletion in 8 out of 35 cases each. Conclusions: This study determines the frequency of copy number alterations in B-ALL which has not been previously documented from India. is an important member of the leucine zipper family of transcription factors that is required for granulocytic differentiation. On average, mutations are identified in 15% of acute myeloid leukemia (AML) patients with a normal karyotype. Most mutations occur at the C-terminal region of
The most common benign polymorphisms reported include c.690G>T and c.590_595dupACCCGC. We report a novel genetic alteration, c.584_589delACCCGC that was identified in a patient with AML. Methods: The cDNA reference sequence used was NM_004364.3. The single exon of the gene was amplified followed by Big Dye sequencing in both forward and reverse directions from the patient's genomic DNA. Sequencing was read on an automated 3130XL Genetic Analyzer. Mutation Surveyor software was used to analyze and interpret the results of sequencing. Results: sequencing analysis in both peripheral blood and saliva of this AML patient demonstrated a novel genetic alteration, c.584_589delACCCGC, which leads to loss of Proline 194 and Histidine 195 from a highly polymorphic region. This particular deletion has not previously been reported in normal individuals or AML patients. Conclusions: The deletion found in this individual in the same region as the wellknown duplication, has not been reported previously. The confirmation test performed on the saliva suggested a germline alteration. Considering its high polymorphic location, c.584_589delACCCGC is likely to be a new benign variant in the gene. Introduction: CLL is one of the most common leukemias in adults. It is recognized by a gradual accumulation of neoplastic B-cells in the bone marrow. Cytogenetic analyses are an important part for the diagnosis and also are used to aid clinicians in determining treatment protocol. Whole genome SNP microarray analysis using DNA isolated from bone marrow is increasingly becoming a test of choice, especially for those patients where chromosome studies are normal. Methods: Chromosome analysis and FISH were done using standard methods.The FISH panel currently in use for CLL at the Georgia Regents University Cytogenetics laboratory consists of the following probes: 1) DLEU1/13q34, interstitial deletions or whole arm deletions of 13q; 2) CEP 12, copy number of chromosome 12; 3) ATM, loss or gains of 11q22.3; 4) p53, loss or gains of 17p13; 5) IGH break apart probe, rearrangements and copy number of 14q32. High resolution SNP microarray using CytoScan HD Microarray (Affymetrix Inc, Santa Clara) was performed on DNA isolated from methanol-acetic acid fixed bone marrow pellets. Results: Of the 22 cases with microarray results, 9 were chromosomally normal and 2 additional cases had no karyotype. Two of these cases were normal by FISH and microarray. Three cases were normal by microarray but had findings by FISH analysis, including: t(11;14); bcl6 rearrangement; and hyperdiploid clone found at 10%. The remaining 6 cases had abnormalities identified by both FISH and microarray analysis. Three cases had FISH abnormalities observed at <15% missed by the array. These cases had additional array findings, including: a 30% loss of 3p21.31 and a loss of 22q11.22; a deletion at 14q11.2, an amplification of IGH, 3-4 copies of 16p, and mosaic LOH 20q; and a mosaic loss of 22q11.22. One case had a loss of DLEU on 13q14 observed in 25% by FISH which was not observed by array; however, the array observed a 15% loss of ATM and a 15% gain of chromosome 12. The remaining 2 cases had abnormalities visualized by FISH which were confirmed by array along with additional abnormalities including: a confirmation of a 13q loss and an additional loss on 22q; and the confirmation of 13q14 loss, and additional gain of chromosome 1, and mosaic loss of chromosomes 19 and 22. Conclusions: Microarray analysis is a powerful tool for use on hematological malignancies to identify copy number abnormalities and regions of LOH. This is especially true for CLL, where the abnormal clone may not be identified by chromosome analysis. However, abnormalities of less than 15% to 20% may still be missed by microarray which continues to necessitate the use of FISH for prognostic aberrations.
C.K. Belludi, C. Vnencak-Jones Vanderbilt University Medical Center, Nashville, TN. Introduction: Calreticulin gene (CALR) mutations in myeloproliferative neoplasms (MPN) often occur in exon 9 as frameshift mutations due to insertions or deletions, and are generally mutually exclusive with JAK2 and MPL gene mutations. We report an incidental finding of a distinct clone with a CALR in-frame deletion in a 55-yearold male with a JAK2 V617F+ myelodysplastic/myeloproliferative disorder (MDS/MPN) and KIT D816V+ associated systemic mastocytosis(SM) that persisted throughout treatment and in tissue manifestation of the disease. Methods: DNA was extracted from peripheral blood (PB), bone marrow (BM) and purified CD3+ and CD33+ cells submitted through the disease process, and which included both pre and post RIC PBSCT specimens. CALR mutation analysis was performed using fluorescent PCR amplification of exon 9 coupled with capillary electrophoresis (CE) and Sanger sequencing to characterize the observed novel mutation. JAK2 V617F and KIT D816V mutation analysis were performed using allele-specific fluorescent PCR and CE. Donor engraftment in post-PBSCT specimens was determined by fluorescent PCR analysis of STR markers and quantified by CE. Mutation profiling for common point mutations in IDH1, IDH2, FLT3, DNMT3A and KIT was performed using a laboratory developed single base extension assay. NPM1 and FLT3-ITD mutations were analyzed by fluorescent PCR and CE. Results: DNA extracted from the PB at the time of BM diagnosis of MDS/MPN with associated SM and no abnormal cytogenetic findings, showed JAK2 V617F and KIT D816V mutations. Retrospectively, this specimen revealed a novel CALR exon 9 mutation, characterized as c.1175_1184delTGAGGATGA. The relative percentage of JAK2, KIT and CALR mutations were 9%; 10% and 50%, respectively. Through disease progression, the JAK2 mutation disappeared early and remained undetectable thereafter, the KIT mutation was as low as 1%, yet the CALR mutation remained at high levels (47%). A myeloid sarcoma specimen collected 13 months after diagnosis contained exclusively the CALR 9-bp deletion. Neither the initial PB nor myeloid sarcoma specimens showed any additional common point mutations or a NPM1 insertion or FLT3-ITD. Conclusions: We report a unique case that possesses a clone with a unique CALR in-frame 9-bp mutation co-existing with a JAK2 V617F+ MDS/MPN clone and a KIT D816V+ SM clone. Clonal distinction is evidenced by differences in the relative percentage of each clone at various stages of disease progression. Generally, CALR-mutation positive MPN portends a better prognosis compared to JAK2-/MPL-/CALR-diseases; however, the novel CALR mutation clone in our case demonstrates resistance to decitabine therapy and PBSCT, and is associated with myeloid sarcoma. There was a significant skewing of SNVs to occur within exons as opposed to introns (3.5:1), at C or G nucleotides (2:1), and within AID motifs (3.5:1), all at p<0.001; no significant associations were observed with categories of FL grade or treatment status. In contrast to the 5'-UTR of BCL2, MLL2 mutations were more dispersed, preventing detailed analysis of tumor heterogeneity. However, 9 of 42 specimens had SNVs at high (>5%) but sufficiently different frequencies, suggestive of FL subclones with different MLL2 genotypes or ongoing mutations in those clones with inactivated MLL2. Conclusions: Ultradeep sequencing allowed discernment of FL subclones based on MLL2 genotype. These mutation patterns are different from 5'-UTR of BCL2 in mutation density, but retain the bias towards SNVs arising at AID motifs. These SNVs are more frequently observed in exons, consistent with a clonal selection of the driver/facilitor role of MLL2 loss of function mutations in FL.
Routine IGH and TCRG Clonality Assessment M.E. Arcila, T. Baldi, M. Syed, I. Rijo, J. Bakas, Z. Momin, M. Ladanyi, K. Nafa Memoral Sloan Kettering Cancer Center, New York, NY. Introduction: IGH and TCR gamma clonality testing is an important component in the diagnosis of lymphoproliferative disorders. Capillary electrophoresis (CE) of multiplexed PCR products is a common method of analysis. Although robust, simple and highly reproducible, it does not provide a full characterization of clonal sequences that would allow tracking in subsequent samples. We describe our clinical validation and implementation of an NGS based assay for initial clonal characterization and minimal residual disease (MRD) assessment through patient specific clone tracking. Methods: Blood, bone marrow and FFPE tissue samples submitted for routine clonality assessment were selected. DNA was extracted and tested using both standard CE and LymphoTrack IGH +TRG -MiSeq assays (Invivoscribe). Positive, negative and no template controls were run with all assays. Sensitivity and limit of detection (LOD) were assessed based on serial dilution studies for detection of the initial clonal population and for tracking of a precharacterized clone at MRD levels, respectively. Results: A total of 160 samples were analyzed including 126 clonal (defined by initial CE testing, 92 IGH, 34 TCRG) and 34 non-clonal samples with 94% concordance between the 2 methods. Discordant results (clonal by CE, non clonal by NGS) were attributable to pseudoclonality in the post treatment setting. Based on a minimum input of 50ng of high quality DNA, analytical sensitivity was 5% for diagnostic samples (uncharacterized clone) with good Inter and intra-assay reproducibility. Further dilution studies to establish LOD for tracking a previously characterized clone showed accurate detection at 1x10-5 with 1-2ug DNA input. The mean number of reads per sample was approximately 500,000. Conclusions: Assessment of clonality by NGS methods provides significant improvement over existing clonality assays using fragment analysis by CE. Sensitivity for initial detection of a diagnostic clone is similar to the CE assays but provides full characterization of the clone to enable tracking in subsequent samples at MRD level. NGS testing readily resolved pseudoclonality calls in post-treatment samples by differentiating clonal products of jmd.amjpathol.org ■ The Journal of Molecular Diagnostics same size but different sequences interpreted as clonal by the CE method. However, the LymphoTrack assays remain more expensive and with higher TAT compared to CE. Introduction: Transcript levels of p210 BCR-ABL1 are used to monitor response to tyrosine kinase inhibitor (TKI) therapy in patients with Philadelphia chromosome positive chronic myelogenous leukemia (CML). The standardized baseline, established in the IRIS study, is defined as 100% on the International Scale. A reduction of 3 logs or more from the baseline is considered a major therapeutic response. A RT-qPCR assay with a sensitivity of more than 4.5 logs below the baseline levels is recommended as assays with this sensitivity level allow monitoring of complete molecular response, defined as undetectable p210 BCR-ABL1 transcripts in two consecutive samples by a test with a limit of detection of 0.01% on the International Scale. TKI therapy is also approved for patients with Philadelphia chromosome positive acute lymphoblastic leukemia (ALL), but monitoring of ALL BCR-ABL1 transcripts has not been standardized. We sought to examine the performance of digital droplet PCR on the RainDrop instrument (RainDance, Inc) for detecting p190 and p210 BCR-ABL1 transcript levels to establish a standard approach for minimal residual disease detection in ALL and to improve sensitivity of p210 detection in CML. Methods: RNA was extracted from K562 (BCR-ABL1 breakpoint p210 e14a2) and SUP-B15 (BCR-ABL1 breakpoint p190 e1a2) cell lines on the CSC Maxwell extractor (Promega). Ipsogen BCR-ABL1 standards were purchased from Qiagen. TaqMan primers and probes for BCR-ABL1 p210 e14a2 and p190 e1a2, validated by routine real-time PCR, were used. Emulsified reaction mixes were amplified on a PTC-200 cycler. Linearity of the Ipsogen standards was determined from absolute numbers of BCR-ABL1 positive droplets. A series of K562 RNA dilutions were spiked into a constant background of 1ug RNA to establish the sensitivity of BCR-ABL1 e14a2 transcript detection. Results: Linearity of our p210 e14a2 and p190 e1a2 digital droplet PCR based assays, established using the Ipsogen BCR-ABL1 standards, achieved a r2 value of 1 for both assays. A similar r2 was obtained with serial 10-fold dilutions of K562 RNA over a range of 1 μg to 0.1 ng, with 287598 to 28 droplets counted respectively, representing a sensitivity of 1:10,000. Concentrations measured based on absolute counts of positive droplets were half those of Ipsogen standards. Conclusions: Digital droplet PCR based assays provide a high degree of linearity across a wide dynamic range for the accurate detection BCR-ABL1 transcripts. The increased accuracy of digital droplet PCR should improve BCR-ABL1 transcript quantification for TKI response monitoring once the dynamic range for the detection of p190 BCR-ABL1 is established and the accuracy of both assays is validated using curated sets of clinical standards.
Sequencing-based Transcriptome Profiling D. Munafo, B. Langhorst, C. Sumner, C. Chater, L. Mazzola, J. Bybee, D. Rivizzigno, S. Russello, F. Stewart, E. Dimalanta, T. Davis New England Biolabs, Inc., Ipswich, MA. Introduction: Peripheral blood can reflect molecular profile changes occurring in other tissues during pathological events even before clinical symptoms have appeared. Blood is also more accessible and easily collected than other tissues making it ideally suited for biomarker discovery. Whole blood transcriptome analysis by next generation sequencing allows for accurate detection of known as well as unknown transcripts with low-and high-abundance and it is beginning to demonstrate utility in early clinical diagnostics. One limitation of the whole blood RNA sequencing is that, the high content of hemoglobin transcripts (constituting up to 75% of total transcripts population) can mask detection of lower abundant biological meaningful transcripts. In this work, we demonstrate that globin depletion significantly increases sensitivity of transcript detection without altering transcript expression profile of the non-globin transcripts. Methods: Here, we present a method to deplete for hemoglobin transcripts. We applied this method to deplete multiple blood-derived samples (peripheral blood, leukocytes, bone marrow and umbilical cord blood) from adult, fetal and embryonic hemoglobin transcripts. Using next generation RNA sequencing, we assess depletion efficiency and we compare library complexity and transcript expression correlation before and after globin depletion. Results: Whole blood RNA sequencing from single male donor reported to have 50% of total sequencing reads being hemoglobin transcripts. At 50 million paired-end read coverage, 23% hg19 annotated genes were detected. On the contrary, after globin depletion, only 0.1% of sequencing reads mapped to globin transcripts increasing sensitivity of transcript detection by 30% (allowing for reliable detection of 2,787 additional transcripts). We achieved high depletion efficiency (up to 99.9% globin depletion) with minimal off target effects (high FPKM correlation between depleted and non-depleted libraries; R2= 0.87486). Conclusions: Globin depletion by this method is not altering the transcript expression levels. Globin reduction increases the coverage of less abundant transcript. This method can potentially be expanded to deplete for other highly express transcripts with minimal biological relevance. Moreover, it is amenable to high-throughput sample preparation and robotic automation to easy implement in a clinical setting. The accurate classification and diagnosis of cutaneous T cell lesions is challenging, often requiring a combination of clinical history, multiple small biopsies, multiple immunohistochemical stains and molecular studies to arrive at a definitive diagnosis. Furthermore, clonality can be seen in multiple reactive conditions including drug reactions. We present an evaluation and optimization study of the LymphoTrack TRG next generation sequencing (NGS) assay for monoclonal T cell receptor gamma chain gene rearrangement as also utility in cutaneous T cell lesions. Methods: The pathology database was queried for all skin cases where a T cell receptor gene rearrangement study was ordered to identify cases between 2007 and 2013. Charts and slides were reviewed to determine outcome and definition of true positive and true negative cases. DNA was extracted from formalin fixed paraffin embedded (FFPE) tissue blocks, and the samples were analyzed with the LymphoTrack TRG assay on the IonTorrent PGM system. Next generation sequencing results were assessed using the LymphoTrack PGM Software. Statistical analysis was performed using Microsoft Excel and STATA 11. Clinically confirmed positive cases were used to ascertain sensitivity, specificity, read cut offs and percentage of total reads to assess clonality. Results: Charts were reviewed for 201 skin cases, representing 183 patients. A total of 40 cases were analyzed by NGS. Clinical follow up ranged from 6 months to 7 years. Sequence reads ranged from 3,776 to 531,621, with most results falling between 10,000 and 200,000 reads. Reads below 2000 were deemed unacceptable. Clonality cut-offs were set at 10% of total reads. Results were interpreted as polyclonal, oligoclonal, and monoclonal. The most common monoclonal gene rearrangements involved Vg4Jg1/2. When oligoclonal results were considered negative, NGS had a sensitivity of 92% and a specificity of 100% for the diagnosis of MF. By contrast, the fluorescence based sequencing assay showed a sensitivity of 85% and a specificity of 100% in this same subgroup. Several cases that were only suspicious for a monoclonal T cell gene rearrangement were found to have distinct reproducible monoclonal gene rearrangements. Known cell lines were ran in a dilution series and demonstrated a limit of detection of 10% clonal cells in a polyclonal population. Conclusions: Using clinical follow up as the gold standard for the diagnosis of MF, NGS was found to have a sensitivity superior to fluorescence based sequencing assays, with a similar specificity. Samples with <500x coverage in >10% of all amplicons were re-assayed. Amplicons with <500x coverage in >50% samples were either not analyzed, or considered with reduced sensitivity. The analytical sensitivity of the NGS assay was 5% AVF at >500x coverage, and 20% AVF at 100-500x coverage. The precision of the NGS assay was 98.4% and reproducibility was 95.7%, based on multiple intra/inter runs of clinical samples diluted to the limit of detection. For the validation set, 100% accuracy was obtained for variants verified by an independent methodology. As expected, large indels in 6 samples (including CALR and FLT3-ITD) were not be detected by NGS, confirming the need for testing by independent methodology: fragment analysis (5% sensitivity) and Sangers sequencing (20% sensitivity). Due to the overall poor coverage observed for the CEBPA gene, mutation analysis was also performed by Sangers sequencing. Conclusions: To facilitate disease identification and/or risk stratification in AML, MDS, and MPN, we have developed a multi-platform approach, permitting robust mutation analysis of 54 relevant genes with high accuracy, precision, reproducibility, analytical sensitivity, and specificity. Two samples with rearrangements detected by FISH, t(9;22)(BCR-ABL1) and t(8;21)(RUNX1-RUNX1T1), and 3 samples with translocations reported by G-band BCR-ABL1 control were tested. Our BCR-ABL1 sample was negative by quantitative polymerase chain reaction (PCR) assay. NGS libraries were created using anchored multiplex PCR. The libraries were then sequenced on an Illumina MiSeq sequencer. Generated FASTQ files were analyzed through the ArcherDx bioinformatic pipeline. Strong evidence fusions were defined as those with 10% or higher frequency and excluding fusions between introns and exons, and fusions with partners of significant similarity (pseudogenes). Results: All translocations were confirmed by the NGS assay except for the inv(3). The expected fusion sequences, MLLT3-MLL and KMT2A-AFF1 were confirmed in t(9;11), and t(4;11) samples respectively. The sample with t(4;11) also demonstrated a novel fusion, t(10;22)(BCR-TCTN3). Interestingly, the sample reported having inv(3) by karyotyping had 2 previously unreported fusions: t(3;22)(MECOM-GUCD1) and t(21;X)(RUNX1-XIST). No other strong evidence fusions were seen. Conclusions: The ArcherDx FusionPlex Heme panel is promising in detecting all the commonly seen recurrent translocations associated with myeloid malignancies. Alternate fusions may also be detected if one of the targeted genes is involved. In our case that was reported as inv(3), a fusion involving MECOM was discovered, which could have mimicked the appearance of inv(3) in karyotyping. Careful examination of the fusions reported by the informatics pipeline is critical, as many false positive results were present, including alignment to pseudogenes and biologically unlikely exon/intron fusions. The cutoff of reporting fusions should be defined by testing samples with low level but definitive fusion genes. One of the potential utilities of the NGS result is to design patient specific minimal residual disease testing based on the unique fusion sequences. and acute myeloid leukemia (AML) represent the two most common myeloid malignancies in both adults and children. Genomic mutation profiling has started playing important roles in risk stratification, therapeutic decision, and disease monitoring in MDS and AML. Methods: We performed massive parallel deep sequencing on 229 MDS/AML samples using a custom-design next generation sequencing (NGS) panel that covers 3,600 mutations in 48 genes that are known commonly mutated in hematological malignancies. Of the 229 samples (196 cases), 51 were MDS (26%) and 145 AML (74%). The specimen types tested included blood, bone marrow, bone core, and FFPE. Results: A total of 326 mutations were identified in 133 cases (67.9%). Ninety-one cases (91 of133 cases, 68.4%) harbored 2 or more mutations. The most frequently mutated genes were TET2 (47) and TP53 (46), followed by DNMT3A (21), RUNX1 (17), ASXL1 (16), SRSF2 (15). The mutations mainly affect 5 major biological functions including epigenetic regulators (TET2, DNMT3A, ASXL1, EP300, IDH1, IDH2 and EZH2), signal transducers (FLT3, NRAS, JAK2, PTPN11, CBL, JAK3, KRAS, ABL1, CSF1R, FBXW7 and KIT), tumor suppressors (TP53, WT1 and CDKN2A), transcription factors (CEBPA, NPM1, RUNX1 and NOTCH1) and spliceosome proteins (SRSF2, U2AF1 and SF3B1). The most commonly mutated genes in both MDS and AML were epigenetic regulators (37.08% and 30.21%). Other mutations in MDS were more frequently associated with transcription factors (17.98%), spliceosome proteins (16.85%), and signal transducers (15.73%); but in AML were signal transducers (26.81%), tumor suppressors (21.7%), and transcription factors (13.19%). In 25 cases with follow-up studies, the results clearly documented therapeutic responses, disease evolution and relapse. A patient was found to have 4 mutations at diagnosis, which were disappeared after BMT and reoccurred at pre-clinical relapse. Results of follow-up studies may also suggest germline mutations or the presence of multiple tumor clones. An AML case evolved from MDS showed 4 mutations in NRAS, RUNX1, SRSF2, and TET2 at diagnosis. After treatment, the mutant allele frequencies of SRSF2 and TET2 reduced, but the frequencies of NRAS and RUNX1 remained the same (45% to 65%), indicating that one or both of the NRAS and RUNX1 mutations could be germline or the presence of a clone with the NRAS and RUNX1 mutations that was resistant to the current treatment. Conclusions: Our results demonstrate the clinical utility of NGS in MDS/AML. In addition to patient risk stratification and therapeutic selection, further studies with larger cohorts of MDS patients could establish genomic mutation signatures that can predict stable disease or leukemia transformation. Introduction: Immune sequencing, which allows for the study of complex immunological diseases by sequencing millions of V(D)J combinations from B-cell antibody and T-cell receptors, has gained popularity 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. Method: 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 identifier (UID). This enables all PCR copies of each mRNA fragment to be collapsed into a single consensus sequence, making the assay extremely accurate by resolving PCR bias and sequencing errors, as well as allowing quantitative digital molecule counting. Results: Immune sequencing libraries were generated from total RNA extracted from PBMCs in duplicate from a single patient and sequenced. The use of the UID enabled absolute quantification of starting RNA molecules present in the original sample and therefore accurate ranking of the antibody clone abundance by avoiding PCR or sequencing bias. 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 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. Conclusions: 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.
Y. Liu, A.M. Cheng, J. Racchumi, M.J. Kluk, W. Tam Weill Cornell Medical College, New York, NY. Introduction: False-positive (FP) variants occur not infrequently in amplicon-based Next-Generation Sequencing (NGS). Thus, a higher allele frequency (AF) cut-off is required to minimize false positivity for correct mutation calling; however, higher cutoff decreases the sensitivity for residual disease and subclonal mutation detection. Generally speaking, high-sensitivity molecular assays require duplicated reactions to minimize false results due to cross-contaminations and potential technical artifacts. However, due to its relatively high costs, duplicated reactions are not commonly used in NGS. Conceivably, the use of duplicates will lower the possibility of false-positive variant calling even at a low AF cut-off. It may also increase the chance of detecting a low-AF mutation due to an increased coverage of targeted regions by virtue of an additional reaction. We aim to explore the impact of duplicated reactions on mutant calling in the targeted amplicon-based NGS assay. Methods: Seven DNA samples with known mutation profiles are retrieved from Pathology archives. These samples are separately diluted with HAPMAP DNA (NA12878) for a final AF close to 3% to 5% in the known mutations. Duplicated libraries occasions. The library preparation was performed with the Raindance ThunderStorm system using a customized multiplexamplicons. The complexity level was 5 in each droplet generated by the ThunderStorm system. The analysis pipeline removed the technical artifacts (sequencing or alignment-related) systemically based on the analysis of normal controls. Results: Total 42 duplicated reactions from 7 separate samples were evaluated. Examination of the effect of cut-off variation for the duplicates demonstrates that lowering the AF cut-off from 4% to 2% in these fairly diluted samples identified an additional 1.5 variant in each duplicate on average while only generating an additional 0.25 false-positive variant calling per duplicate. In a subset the duplicates identified all known mutations in the sample, with an AF as low as 1% to 2%, whereas required higher cut-off (4%). Conclusions: Using duplicated reactions in targeted amplicon-based NGS assay largely eliminated potential false-positive variant calling in a setting of low AF cut-off (eg, 2%). The strategy facilitates the accurate identification of low AF, subclonal mutations as well as detection of residual disease in post-treatment samples. It also improves the detection rate of mutations when compared with singlet reaction that requires high cut-off. Duplicated reactions may be a potential solution for monitoring residual disease, especially when information regarding patient's prior mutations is available. The Journal of Molecular Diagnostics ■ jmd.amjpathol.org confirmed diagnoses, 69% demonstrated one or more disease-associated mutations whereas 3% harbored an unclear variant and 27% showed a normal sequencing profile. In cases suspected of myeloid neoplasia, 38% showed one or more diseaseassociated mutation at greater than 4% allele frequency. A number of these samples showed disease-associated mutations at high allele frequencies. For example, a peripheral blood found to have a 0.5% CD34(+) myeloblasts of uncertain significance was found on NGS testing to contain an SRSF2 mutation (p.Pro95_Arg102del) at 49.49% allele frequency as well as TET2 mutation (p.Gln847*) at 39.83% allele frequency. In addition, within the samples with unconfirmed diagnoses, 6% carried an unclear variant and 57% resulted in a normal sequencing profile. Conclusions: Over one third of peripheral blood or bone marrow samples suspected of myeloid neoplasia but with an unconfirmed diagnosis were found to harbor a disease associated mutation. A subset of these samples were found to have mutations at high allele frequencies indicating a significant clonal burden. Mutational profiling of genes involved in myelodysplastic syndromes may aid in diagnosis in certain situations. LG life sciences, Korea) based on multiplex real-time PCR was developed for simultaneous detection of 14 respiratory viruses (RVs). We compared the performance of the AdvanSure assay with the Seeplex RV 12 ACE (Seeplex; Seegene, Seoul, Korea), a multiplex end-point PCR kit. Methods: A total of 454 consecutive respiratory specimens were tested with both the AdvanSure and Seeplex assays in parallel. A positive or negative result from both methods was considered to be a true positive or negative, respectively. The specimens with discrepant results between the 2 assays were further confirmed by monoplex PCR and sequencing in a blind manner. Results: The AdvanSure and Seeplex assays detected 153 (33.7%) and 145 (31.9%) positive cases, respectively. The positive percent agreement, negative percent agreement, and kappa value for 2 assays were 87.2% (95% CI, 80.3 -92.1), 91.1% (95% CI, 87.2 -93.9), and 0.77 (95% CI, 0.70 -0.83), respectively. Analytical sensitivity, specificity, positive predictive value, and negative predictive value of the AdvanSure assay were 91.6% (95% confidence intervals, 85.8% to 95.3%), 97.3% (94.6% to 98.8%), 94.7% (89.4% to 97.5%), and 95.7% (92.6% to 97.6%), respectively, whereas those of the Seeplex assay were 87.1% (80.5% to 91.8%), 98.0% (95.5% to 99.2%), 95.7% (90.6% to 98.3%), and 93.6% (90.1% to 96.0%), respectively. The AdvanSure assay had shorter turnaround time (TAT) and hands-on time (HOT) (3 hr and <1 hr) than the Seeplex assay (8 hr and 2 hr). Conclusions: In conclusion, the AdvanSure assay produced results comparable to the Seeplex assay for RV detection, exhibiting a tendency to have better sensitivity. Furthermore, the AdvanSure assay had shorter TAT and HOT, which could contribute to more rapid diagnosis and timely initiation of antiviral therapy for patients.
S. Khedr, J. Friderici, C. Bissaillon, K. Lebel, F. Moore Baystate Health, Springfield, MA. Introduction: The gold standard for diagnosing HSV encephalitis is PCR testing of cerebrospinal fluid (CSF). At our institution prior to June 2013, CSF samples from suspected cases were sent to a reference laboratory. Beginning June 2013, HSV PCR was performed in-house. Since the turnaround time of the result was significantly decreased after the switch to in-house testing, we hypothesized that patients' length of hospital stay (LOS), total hospitalization charges, and days on acyclovir would be reduced. Methods: Study Design: Pre-Post, quasi-experimental design, retrospective review of hospital records. The "pre" period was defined as 10/1/2011 through 3/31/2013; and the "post" period as 7/1/2013 to 9/30/2014. We allowed for a 3 month implementation period (4/1/2013 to 6/30/2013). Trends in LOS, hospitalization costs, and acyclovir duration were examined using piecewise linear regression. Stata 13.1 was used for all analyses. Patient demographics data and Infectious Disease (ID) consults were also collected. Results: A total of 526 samples were sent for PCR analysis over the study period (247 pre; 35 implementation; 244 post). The mean ± SD patient age was 38.0 ± 28.5 years and 21.8% were < 1 year old, 56.2% were male; 64.3% white; 24.9% Hispanic; 8.8% Black. About half (48.6%) of all patients had an ID consult. Patient demographics were similar before and after in-house PCR implementation. After implementation of in-house PCR, geometric mean interval from admission to test result decreased by 3.8d (95% CI -3.0d, -4.6d), and LOS decreased by 3.7d (95% CI -1.8, -5.6). Surprisingly, both rebounded gradually but significantly in the year after in-house implementation (admission to test result +0.3d/quarter, p=0.007; LOS +0.9d/quarter, p<0.001). A similar "reduction-rebound" pattern was noted for total hospitalization charge (initial change of -43%, 95% CI (-17%, -60%), followed by +11.5% (p=0.005) increase per study quarter. Interestingly, the time from result reporting to discontinuation of acyclovir increased from 0.6d pre to 1.4d post (p=0.03). Observed changes appear to be neither related to seasonal influence nor number of consults by ID specialists. However, testing rate increased from an average of 15 tests per month pre to 17 post (p=0.08), whereas the rate of positive tests decreased 3.3% to 0.8% (p=0.08), suggesting relaxed testing criteria after in-house PCR. Conclusions: Immediate benefits of in-house PCR were realized after implementation; however a perplexing rebound to previous values was seen during the first year after implementation. Future studies will focus on hospitalist behaviors as a potential explanation for the rebound.
ID03. Simplexa C. difficile Direct: A Solution for Moderate-Complexity Labs Combining Real-Time PCR Detection with Simplified Workflow M. Tabb, E.X. Huang, A. Bologa, Y. Parocua, R. Martin, S. Dempsey, C. Cheng, L. Huang Focus Diagnostics, Cypress, CA. Introduction: Clostridium difficile infection is a leading cause of hospital-associated gastrointestinal illness. The Simplexa C. difficile Direct assay is in development as a real-time PCR test for C. difficile; it directly tests unformed stool samples without requiring a separate nucleic acid extraction step. The sample-to-answer format and simplified workflow of the assay allow for testing up to 8 samples in about 1 hour. In this study, we evaluated the performance of the Simplexa C. difficile Direct assay and compared it to results from both the Cepheid Xpert C. difficile assay and enriched culture. Methods: The limit of detection (LoD) study was performed with toxigenic C. difficile strains ATCC 43255 and NAP1A. A panel of 126 bacteria and viruses in stool matrix was tested (106 CFU/mL for bacteria, 105 TCID50/mL for viruses) to evaluate cross reactivity and microbial inhibition. Twenty toxigenic C. difficile strains were evaluated for analytical reactivity. Potentially interfering substances were tested for inhibition. The reproducibility study was performed with medium-and low-positive panels. A set of 100 patient samples was evaluated with Simplexa C. difficile Direct, Xpert C. difficile, and enriched culture assays in the method comparison study. Real time stability studies for reaction mix and positive control were also conducted for 5 months at -20 celsius storage. A sample stability study was conducted by incubating stool in TE for up to 6 hours at room temperature prior to testing with Simplexa C. difficile Direct. Results: LoD of Simplexa C. difficile Direct was 0.5 CFU/mL for strain ATCC 43255 and 1.3 CFU/mL for strain NAP1A.
None of the 126 organisms tested for cross-reactivity were detected by the Simplexa C. difficile Direct assay, and none of them inhibited detection of C. difficile. All 20 toxigenic C. difficile strains were detected in analytical reactivity tests. None of the substances tested interfered with C. difficile detection. Inter-and intra-assay reproducibility studies yielded <5% coefficient of variation. Positive and negative agreement was 96.7% (29 of 30) and 95.7% (67 of 70), respectively, between Simplexa C. difficile Direct and Xpert C. difficile. Positive and negative agreement was 93.1% (27of 29) and 94.4% (67 of 71), respectively, between Simplexa C. difficile Direct and enriched culture. Real time stability studies showed the performance of both reaction mix and positive control remain the same after stored at -20 celsius for 5 months. Swabbed sample stability study showed that swabbed stools in TE were stable for 6 hours under room temperature without result change. Conclusions: 1) Simplexa C. difficile Direct can provide an option for simplified C. difficile testing. 2) Simplexa C. difficile Direct was comparable to both Xpert C. difficile and enriched culture for identifying C. difficile in patient samples. 3) The assay is in development; it is not currently available for sale, and it is not FDA cleared.
R.R. Gullapalli, K.G. Sanchez, J. Wu University of New Mexico, Albuquerque, NM. Introduction: Gallbladder Cancer (GBC) is a relatively rare cancer. It is the 6 th most common malignancy of the gastrointestinal tract. Despite its relative rarity, it has a distinctly poor prognosis with less than 8% 5-year survival in patients of Stage III and above. Epidemiologically, GBC has a distinct incidence pattern across the world with high incidences in Chile, India, Poland and the state of New Mexico within the USA. Specific ethnic populations seem to be more prone to the development of GBC. The Native American population of New Mexico has been shown to have a 6 to 8 times higher incidence of GBC compared to Caucasians for currently unknown reasons. Multiple other risk factors in the development of GBC include female gender, presence of gallstones and potential association with chronic Salmonella bacterial carrier status. To examine the potential infective associations with GBC, we conducted an unbiased 16s Metagenomic Microbiome analysis of 28 GBC patients at our institution. Methods: We collected GBC FFPE tissues after obtaining prior IRB approval and included 28 patient samples which were eligible for the study. The cohort comprised of 10 Caucasian, 10 Hispanic and 8 Native American ethnicities. Ion Torrent based 16s Metagenomic protocol to sequence the samples on a 318 chip to ensure high throughput. A custom-built, 24-step, bioinformatics protocol was used to analyze the 16s data based on the open-source Mothur software (v.1.3.4). Extensive validation of the sequencing as well as bioinformatics protocol was performed to ensure the accuracy of the results. A mock community comprising of 20 known bacterial species derived from the Human Microbiome Project (HMP) was used as positive control. A negative control was included in every run. Results: 15.43 million 16s reads were obtained from 4 different 318 chip runs. High quality reads containing read lengths of 150bp to 296bp were chosen for further downstream analysis. A phylotype based analysis was chosen for the classification of bacterial sequences instead of an OTU based approach in order to ensure greater accuracy of the data. The bacterial species were classified at the order level in order to obtain an accurate taxonomic classification. We observe "unclassified" species are observed in abundance indicating the presence of a "rare" biosphere within the gallbladder microbiome. The alpha and beta diversity indices showed adequate sampling of the tissues under study. Conclusions: We show, for the first time, that it is feasible to obtain an unbiased sampling of the local microbiome using the GBC FFPE tissues. In contrast to the commonly held observation of the role of Salmonella in GBC causation, we did not observe a definite role for Salmonella in the GBC tissues studied in our study.
Direct Assay C. Cheng, H. Gregson, B. Yu, L. Geller, M. Tabb Focus Diagnostics, Cypress, CA. Introduction: Group A streptococcus (S. pyogenes) infection can result in serious sequelae if not treated. Therefore, a quick and accurate diagnosis is important. Traditional antigen detection tests have limited sensitivity and require additional confirmatory culture testing to reduce the risk of false negative results. To provide an alternative solution for S pyogenes detection, we developed a real-time PCR assay that detects group A streptococci (GAS) directly from throat swabs in approximately 1 hour. The Simplexa Group A Strep Direct Assay (Simplexa Direct assay) targets the conserved exotoxin B gene of S. pyogenes. Here, we tested the of the Simplexa Direct assay for microbial inhibition, sample stability, and compatibility with other transport systems.Methods: Patient specimens were collected in E-swab transport medium. For each assay, 50μL of transport medium with patient specimen and 50μL of Direct Mix were added to their respective wells on the Direct Amplification Disc. For the microbial inhibition study, specimens containing S. pyogenes were spiked with 64 potentially interfering pathogens. For the sample stability study, a panel of 10 contrived E-Swab specimens spiked with GAS bacteria was stored at 2ºC to 8ºC for up to 7 days or at room temperature up to 2 days. For the compatibility study, a panel consisting of 24 specimens spiked with GAS bacteria using liquid Amies or liquid Stuart's transport systems. In each study, specimens were tested for detection of S. pyogenes using the Simplexa Direct assay. Results: None of the tested potentially interfering pathogens inhibited detection. All of the specimens stored at 2ºC to 8ºC for up to 7 days and the specimens stored at room temperature up to 2 days showed consistent detection. Preliminary results showed that specimens collected in liquid Amies or liquid Stuart's transport systems showed comparable GAS detection to the E-Swab collection system. Conclusions: The Simplexa Direct assay is specific for GAS detection in the presence of potentially interfering pathogens and can consistently detect GAS in samples stored at 2ºC to 8ºC for up to 7 days or at room temperature up to 2 days. Our data also suggest that the Simplexa Direct assay is compatible with other transport media in addition to the E-Swab system, though further studies are required to confirm this. The Simplexa Group A Strep Direct assay is only FDA cleared and moderate complexity for use with E-Swab throat specimens. It has not been FDA cleared for use with other transport media. It provides a rapid molecular solution that does not compromise detection sensitivity ID06. Assessing Laboratory Performance in the Molecular Detection of Human Papilloma Virus through International Proficiency Testing Programs C. Di Lorenzo 1 , E. Schuuring 2 , P. Wallace 1 , H. Niesters 2 , A. Van Loon 1 1 QCMD, Glasgow, Scotland; 2 University Medical Centre Groningen, The Netherlands. Introduction: Development of invasive cervical cancer is caused by persistent infection with high-risk types of human papilloma virus (HPV). Although cytological screening has reduced the incidence of cervical cancer in countries with organized screening programs, the disease is still a major cause of concern worldwide. The establishment of a universal cervical cancer screening strategy continues to be a subject of much debate. Inclusion of molecular testing for the detection of HPV DNA in cervical cancer screening programs is now becoming increasingly common. Participation in proficiency testing (PT) programs is of outmost importance in order to improve the overall quality and reliability of HPV testing in the respective laboratories. Here we report the results of four years of international PT programs focused on the molecular detection of different high-risk HPV types. Methods: HPV positive and negative samples were prepared using HPV-positive cervical cancer cell lines or cell lines that did not contain HPV DNA. Panel members were prepared in PreservCyt and distributed to participants annually worldwide. The results were collected through a dedicated online system, before being analyzed by the QCMD Neutral Office. Results: The number of laboratories taking part in the QCMD HPV PT program steadily increased from 2010 (n=155) to 2014 (n=228). The majority of the PT results were generated using commercial assays. In general, the results showed a trend towards an improvement in performance going from 59.1% of correct datasets received in 2011 to 87.2% in 2014. The percentage of false positive results has also decreased within the last 4 years going from 2.8% in 2011 to 1.1% in 2014. Although the performance on the samples containing HPV16 and HPV18 was similar between in-house and commercial assays, in-house developed assays performed better on the lower titre HPV16 samples. Within the last 4 years, 60% of datasets contained typing information with the majority of results consistently reporting the correct HPV type 16, 18 and 45. On average, 65% of datasets contained clinical interpretation across the years with the majority of datasets reporting a clinically negative result on the low titre HPV16 samples. Conclusions: The majority of laboratories demonstrated an acceptable level of proficiency. Although a decrease in false positive results was observed, assay contamination remains an issue of concern. Although not part of the formal laboratory assessment, collection, analysis and presentation of clinical and typing results is of valuable importance as it allows laboratories to compare and evaluate their clinical results in relation to current clinical practice guidelines in their region and with laboratories using similar methodology. Introduction: Virtually all cervical cancer, and a large proportion of other cancers, are caused by persistent HPV infection. Although HPV infections are extremely common, very few cause cancer. For unknown reasons, there is huge variability in risk conferred by different HPV types and, remarkably, strong differences even between closely related variant lineages within each type. HPV16 is a uniquely powerful carcinogenic type, causing approximately half of cervical cancer and most other HPV-related cancers. Methods: To permit the large-scale study of HPV genome variability we developed a high-throughput next-generation sequencing (NGS) whole-genome method utilizing a custom HPV16 Ion AmpliSeq research panel with 47 overlapping amplicons. Degenerate primers were employed to cover the genome for all known HPV16 variant lineages. This assay required as little as 2 microliters of crude exfoliated cervical cell extract (equivalent to 1/500 of a 1ml Virapap sample) with minimal nucleic acid isolation. We applied this approach to sequence HPV16 DNA from 796 HPV16-positive research samples from the Kaiser Permanente Northern California NCI HPV Persistence and Progression cohort. We evaluated several metrics to assess the quality and reliability of our HPV16 NGS data: overall and per sample genome completion rates, concordance among 18 replicated samples (replicated 1 to 5 times across each plate, total replicates=28), concordance with Sanger sequencing data, overall HPV16 genome coverage and the coverage distribution. Eight water controls were also included (1 per plate) to assess potential HPV16 contamination.Results: To validate our approach, we determined the concordance between Sanger sequencing and the Ion Proton Sequencer in a random subset of 89 research samples. The mean concordance across the 7906bp HPV16 genome by sample was 99.97%. We discovered many novel SNPs; large deletions associated with cancer (OR of 27.3, 95% CI 3.3-222, P=0.002); frequent HPV16 variant lineage coinfections; and strong lineage risk associations. Notably, the HPV16 non-European D variant lineage was associated with adenocarcinoma in situ (OR 11.3, 95% 4.1-33.3, P=3.7x10-07) and adenocarcinoma (OR 38.1, 95% 8.9-200.7, P=1.9x10-08) histologic subtypes. Conclusions: This research method represents an innovative high-throughput, ultra-deep coverage technique for HPV genomic sequencing, which, in turn, enables the investigation of the role of genetic variation in HPV epidemiology and tumorigenesis. B.L. Voss, D. Schora, K.A. Mangold, S. Das, R.B. Thomson, I. Dusich, M. Wright, B. Smith, L.R. Peterson, K.L. Kaul NorthShore University HealthSystem, Evanston, IL. Introduction: Preventing the transmission of multi-drug resistant organisms (MDRO) within health care settings and the community necessitates accurate and rapid screening. We compared traditional culture methods to a real-time PCR assay targeting genes CTX-M, KPC, and NDM-1 in Gram-negative Enterobacteriaceae. Methods: Three screening swabs were collected at nasal/other respiratory site, axilla, and perianal regions. For culture, swabs were inoculated onto a VACC (vancomycin, amphotericin B, ceftazidime, clindamycin) plate; bacteria found growing were identified and subjected to susceptibility testing. For PCR, swabs were inoculated into an enrichment broth with a 30μg ceftriaxone disc for overnight incubation at 37°C, allowing for maximum recovery of resistant bacteria. By adding 250uL of broth to 200uL of lysis buffer (1% Triton X-100, 0.5% Tween 20, 10mM Trisobtained, present in the supernatant after microcentrifugation. The test was run utilizing the LightCycler 480 or LC 96. Primers, probes, and PCR parameters were previously published by Mangold et al (J. Clin. Microbiol. 2013; 51:3423-25 and J. Clin. Microbiol. 2011; 49:3338-9) . The presence of resistance genes was confirmed by amplification and compared to culture results. Results: Over a period of 14 months, 1191 samples were tested. Relative to culture, PCR yielded: 96 true positives, 1066 true negatives, and 29 discrepants. Of the discrepant samples, 5 were positive by PCR but negative by culture, 1 was negative by PCR but positive by culture, following further culture after broth enrichment. There were 23 discrepants that were negative by PCR but positive by culture, and were determined to be a multitude of extended spectrum beta-lactamases, and positive for untested resistance genes (SHV, TEM, OXA) by alternate PCR testing. Thus, ultimate PCR assay performance exhibited 98.9% sensitivity and 99.5% specificity compared to culture when comparing genes targeted by the assay. Also, PCR was up to 48 hours faster than culture. Conclusions: The molecular approach excelled in detecting instances of resistant bacteria over traditional culture methods for known surveillance targets, but failed to identify MDROs containing genes that were not specifically targeted. Culture was able to detect a broader spectrum of MDROs, with sensitivity and specificity comparative to PCR, but can take longer for results versus PCR. Emergence of other resistance mechanisms will make phenotypic identification (by culture) necessary for epidemiologic detection, and flexibility in molecular assays will increase in importance.
ID09. Evaluation of the Solana GAS Assay for the Detection of Streptococcus pyogenes from Pediatric Throat Swabs K.B. Pierce 1 , S. Holt 2 , A.J. Blaschke 1 , K. Ampofo 1 , K. Korgenski 2 , A. Phillips 2 , M. Dickey 2 , J.A. 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. GAS remains the most common bacterial cause of pharyngitis, and timely diagnosis and treatment are known to decrease the duration and severity of infectious symptoms and lessen disease transmission. Microbiologic culture remains the gold standard for diagnosis of GAS pharyngitis, but takes 24 to 48 hours to reach a definitive conclusion. New molecular tests are challenging culture's efficacy in this role. The Solana GAS Assay (Quidel Corporation, San Diego, CA) is a rapid molecular test designed to identify the presence or absence of GAS in throat swabs from symptomatic patients via qualitative isothermal helicase-dependent amplification (HDA) endpoint detection by fluorescent probe within 1 hour of swab receipt. Methods: Throat swabs obtained for GAS testing at Primary Children's Hospital (PCH) in Salt Lake City, Utah per standard of care and at physician discretion over a 60-day period in January and February of 2015 were included in the study. Each throat swab was first plated using PCH standard procedures and then tested with the Solana GAS Assay following the package insert protocol. Pledgets from culture swab transport tubes were forwarded to Quidel for culturebased testing for GAS and for testing with an alternative nucleic acid test (PCR). Both PCH and Quidel culture outcomes were used to determine the consensus culture result. The PCR testing was used for discrepancy resolution between consensus culture and Solana. Results: A total of 213 swabs were tested by both Solana and consensus culture. Sixty-nine swabs (32.4%) were culture positive for GAS. Of these, 68 (98.6%) were positive on Solana testing. Of 144 culture-negative swabs, 136 (94.4%) were also negative by Solana. In all 8 cases where culture was negative and Solana positive, PCR showed the presence of GAS DNA within the sample. In the single case that was culture positive and Solana negative, PCR identified no GAS DNA in the sample. After discrepancy resolution, sensitivity and specificity of the Solana GAS Assay were 98.7% and 98.5%, respectively. Accuracy was 98.6%, PPV 97.4% and NPV 99.3%. Conclusions: The Solana GAS Assay surpassed culture for accuracy in our study of pediatric throat swabs. Use of this rapid method for identification of GAS in clinical samples has the potential to refine current methods for the diagnosis of GAS pharyngitis.
R. Luna 1,2 , A.R. Magee 2 , J.K. Runge 2 , A. Venkatachalam 2 , E.B. Hollister 2 , J.J. Dunn 2 , J. Versalovic 2 1 Baylor College of Medicine, Houston, TX; 2 Texas Children's Hospital, Houston, TX. Introduction: In the emerging field of medical metagenomics, microbiome characterization is rapidly evolving into a companion diagnostic with potential clinical utility. By evaluation of bacterial community composition, microbiome profiling can be aid in screening unknown infections as well as pre-and post-intervention assessment. Working in parallel with routine clinical diagnostics, this next-generation sequencing (NGS) based approach has been employed in a variety of patient populations. Methods: Upon request, an aliquot was obtained from specimens submitted to the clinical laboratory for routine diagnostics, including microbial culture in most cases. Bacterial DNA was extracted from a variety of gastrointestinal (GI) and respiratory specimen types. NGS targeting the 16S rRNA gene yielded ~5,000 sequences per sample. Utilizing a variety of bioinformatics approaches, microbiome profiles were evaluated individually, for the identification of potential pathogens, as well as evaluated in comparison to control microbiome profiles (matched based on sex, age, and disease). Results: Microbiome profiles were successfully obtained from all samples. GI-based evaluations of patients with prolonged GI symptoms (including patients with C. difficile infection, ulcerative colitis, autism spectrum disorder, and unknown GI disorders) all exhibited noticeably different profiles, with an absence or decrease in Bacteroides sp., when compared to healthy controls. Respiratory-based evaluations ranged from ventilated neonates to adolescents with cystic fibrosis (CF). In the ventilated neonate, retrospective microbiome evaluation of a tracheal aspirate provided a profile consistent with E. coli sepsis, with a microbiome dominated by Escherichia/Shigella associated sequences. Profiles of sputum and bronchoalveolar lavage specimens from patients with CF yielded beneficial findings when compared to patients with a similar clinical profile. Beyond comparisons to microbial culture results, identification of a sputum-based microbial profile with large amounts of sequences attributed to Helianthus annuus (sunflower) highlighted the importance of adequate specimen collection procedures. Conclusions: Initial exploration of microbiome characterization as a companion diagnostic has provided valuable insights. In some cases, obvious pathogenic microbiomes were identified. In other cases, shifts in the overall community were found to be consistent with symptomology or confirmatory of previous clinical laboratory findings. Beyond diagnosis and confirmation of pathogenic infections, identification of microbiome profiles relevant in specific patient populations may have potential utility for individual patients in regards to treatment selection and development of prognostic criteria. (HBV) DNA is best documented by the sensitive nested polymerase chain reaction (PCR) technique. Objective: to determine the presence of HBV DNA extracts from liver tissue in patients who are incidentally detected to be positive for HBV in liver section by immunohistochemistry (IHC). Methods: All enrolled patients were serology negative for HBsAg with no obvious clinical features of hepatitis. Wedge biopsies or liver resections were carried out for non hepatitis conditions. Paraffin sections of liver exhibited positive IHC staining for HBV surface or/and core antigens. DNA extracts of liver tissue were subjected to nested PCR for HBV DNA. Results of PCRs were correlated with liver histology and IHC findings. Results: Fifty patients who were positive for HBV in IHC were studied. Male:Female=25:25, median age in male 41.7 and in female 42.9 years, 48 biopsies showed strong IHC positivity for HBcAg, cytoplasmic in all 48 and nuclear in 10. HBsAg was positive in 22 biopsies. Nested PCR was positive in 47 biopsies. Negative controls were serology and IHC negative histologically normal liver tissue, and positive controls are serology PCR positive samples for HBV DNA. Histological features: normal in 10, chronic hepatitis with activity in 18, metastasis in 10, changes EHPVO in 6, Budd Chiari in 2 and cholangitis in 2, and 1 each of non cirrhotic portal fibrosis, granuloma and venoocclusive disease. Twenty five patients were followed up for 6 to 12 months and none of them developed any kind of hepatitis-like features. One patient was positive for anti HBc IgG on serology whereas all others had negative serology for HBV including PCR for HBV DNA. Conclusions: occult hepatitis B appears to be common among individuals who presents with long standing clinical problems and complications. Routine serology screening possibly is not sensitive enough to identify certain group of HBV infected individual, especially those who already have certain health problems. Liver appears to be the only ideal sample to document HBV infection. Introduction: Influenza related mortality was substantially higher in the 2013-2014 season than the preceding or following seasons; this correlated with the semiquantitative nanoamperes (nA) reported from our multiplex respiratory viral panel test (eSensor, GenMark Diagnostics, Carlsbad, CA). Averaging nA for a population would reduce variability in sample collection and source (e.g. nasopharyngeal, nasal, bronchoalveolar lavage), and might reflect severity of illness, on the assumption that sicker patients would seek care earlier in their illness, since viral titers in Influenza are highest within the first several days. Methods: The eSensor assay detects the matrix gene for influenza A, and subtype genes for H1, H3 and 2009 H1N1. Since the matrix gene sequences used for PCR and capture probe are conserved across subtypes, comparison of H1N1 and H3N2 strains is reasonable. For consistency, the H1N1 data from 2012-2013 were combined with 2013-2014 since there are sequence differences in the matrix gene between H3N2 and H1N1 subtypes. Corresponding average nA for the matrix gene were 121.1 ± 109.5 (N=91), 172.6 ± 70.2 (N=194) and 65.1 ± 37.5 (N=172) for the 3 seasons respectively, p<0.0001, ANOVA for the 3 groups. The denominators differ from the mortality data because the nA data were retrieved by hand from lab worksheets and inclusion of outpatients in some cases, whereas the mortality data were based strictly on inpatient data from the electronic health record (Epic). Although mortality data was not available from the BayCare Health System, Clearwater, FL, corresponding nA results were similar to those of UFHealth Shands for the matrix gene nA: 74.8 ± 77.8 (N=19), 165.4 ± 76.1 (N=232) and 50.2 ± 25 (N=152), respectively for the 3 seasons, p<0.0001 for 2012-2013 and 2014-2015 (H3N2) seasons versus the 2013-2014 (H1N1), but were not statistically different between the H3N2 seasons 2012-2013 and 2014-2015. nA for the subtype genes had the same general pattern at both UFHealth Shands and BayCare: the 2013-2014 averages were significantly higher than those of the preceding and following seasons, p<0.0001, ANOVA. Conclusions: Significant changes in population Influenza A matrix gene nA were observed over 3 influenza seasons. Further study is needed to determine the relative contributions of severity of patient illness and sequence drift to explain the findings. Population averages of semi-quantitative PCR data could be helpful as Influenza and other respiratory virus sequence drift occurs.
N. Guaring-Angulo 1 , D.S. Grosser 2 , A. Rezaei 1 , W. Watson 1 , M. Montgomery 1 , N. Dhiman 1 1 med fusion, Lewisville, TX; 2 Baylor University Medical Center, Dallas, TX. Introduction: Human Rhinovirus (HRV) and Coronavirus (HCoV) infections cause considerable pulmonary morbidity and mortality in high-risk immune-compromised patients with cancer/transplant and critically-ill hospitalized individuals. Recent developments in molecular diagnostic tools have led to the easy and rapid codetection of a large number of HRV and HCoV strains. The aim of this study was to determine the frequencies and clinical significance of HRVs and HCoV in background of other respiratory pathogens using a sensitive molecular detection technique in specimens from high risk patients with acute respiratory illness. Methods: Respiratory samples were obtained from 261 individuals between February 2015 and May 2015. Limited viral targets based on a laboratory developed respiratory viral panel (Influenza A/B, Respiratory Syncytial Virus (RSV), Para-Influenza (ParaFlu) 1, 2 and 3 and Human metapneumovirus (HMPV)) were reported as the part of routine diagnostic care. In addition, an expanded 20 pathogen respiratory panel containing 17 respiratory viruses including HRVs/HCoV and 3 bacterial targets by multiplex real-time PCR (BioFire Diagnostics, Inc.) was also performed. HRV/Enterovirus results were further characterized using Enterovirus specific assay (Cepheid). Retrospective chart review was performed for HRV and HCoV positive patients for underlying disease indicative of immunosuppression, transplantation, and immunosuppressive therapy. Patients were considered immunecompromised if they had an actively treated malignancy, HIV infection, or rheumatologic conditions on immunosuppressive therapy or were recipients of solid organ or hematopoietic stem cell transplants. Results: HRV and HCoV were detected in 14.56% (N=38) of the cohort with 11.88% and 2.68% positivity rate, respectively. Among the positive cases, multiple virus detection was observed in 4.98%. Other frequently detected viruses were HMPV (5.36%), RSV (3.83%), and ParaFlu 3 (3.07%). Of the 38 patients in HRV and HCoV positive subset, 8 (21.05%) were BMT, 7 (18.42%) were SOT/immunosuppressed and 23 (60.53%) were critically-ill and/or hospitalized patients. Due to lack of respiratory infectious etiology, 29 (76.31%) patients were placed or maintained on antibiotic therapy. Only 6 (15.79%) of these had laboratory confirmation of respiratory or non-respiratory bacterial infection. Conclusions: Detection of HRV and HCoV may help in explaining uncharacterized respiratory illness, particularly in patients with pulmonary co-morbidities. Identifying HRV and HCoV provides opportunities for discontinuation of antibiotic therapy and timely implementation of infection control. Introduction: The Human Microbiome Project (HMP) represents a major advance in our understanding of the human microbiome. Bacteria are, on an average, 10 times more abundant than the total number of human cells within the body. The microbiome comprises of approximately 1% to 3% of the total human body mass. Due to the advances in the Next Generation Sequencing (NGS) technology, it is now possible to conduct large scale, high throughput experiments to analyze the native microbiome. Microbiome dysbiosis has been studied and is associated with an increasing number of common pathological conditions such as Crohn's disease, colon cancer and even in lung cancers. Microbial sources of material for the study of the human microbiome include diverse resources such as sputum, feces, urine and site specific swabs. Often, this material is freshly obtained in a prospective manner. However, there exists a vast untapped resource in the form of archived formalinfixed paraffin embedded (FFPE) tissue in Pathology departments dating back to decades. We examined the feasibility of using such tissue for a project within the lab focusing on gallbladder cancer (GBC). Methods: We obtained archival FFPE tissues of patients suffering from GBC in the institution. The FFPE samples were initially evaluated for adequacy. Paraffin sections were obtained and tissue microdissection was performed to obtain the DNA material from the gallbladder mucosal interface. DNA was extracted using a Qiagen FFPE DNA extraction kit. A Ion 16s Metagenomics kit was then used to amplify the extracted DNA. The Ion kit is comprised of 6 variable region amplicon primers of the 16s RNA gene from bacteria (V2, V3, V4, V5-6, V8 and V9). After the amplification, we ran the DNA on the gel in order to obtain the size specific product. The DNA was extracted from the gel and subsequently sequenced on an Ion Torrent PGM instrument using a 318 chip. Results: Adequate amplification was obtained in all of the samples included in the current study (n=28). The quality of the DNA obtained from individual specimens was variable. We observe that the age of the specimen did not have a significant impact on the DNA quality extracted. Specimens which were nearly a decade provided good 16s DNA as well. Special precautions were taken to prevent environmental contamination. Conclusions: We demonstrate, for the first time, the successful use of archived FFPE tissue as a resource to obtain 16s RNA gene material to study the cancer microbiome. Due to the archived nature of these specimens, they represent a potentially valuable resource to examine and understand the role of microbiome in cancer causation in a tissue specific manner. The Journal of Molecular Diagnostics ■ jmd.amjpathol.org ID17. Performance Evaluation of orfX-ACME MRSA Limit of Detection in Cepheid Xpert MRSA Gen 2 and Gen 3 Assay K. Chan, L. Oon, A. Tan, T. Koh, E. Yau, P. Ong, M. Lau, K. Chan Singapore General Hospital, Singapore. Introduction: Molecular detection of Methicillin-resistant Stahylococcus aureus (MRSA) colonization, as compared to culture, has drastically reduced turnaround time from days to hours. Rapid detection aids in early infection control measures in healthcare settings but culture still remains as the gold standard for detection of MRSA carriage, as molecular assays are predisposed to false positive and false negative results. This is due in part to the high genetic diversity of SCCmec-orfX junction, which most commercial molecular assays target. An example would be MRSA strains with Arginine Catabolic Mobile Element (ACME) inserted between SCCmec and orfX junction. It was shown in a previous study that the Xpert MRSA Gen 2 assay was unable to reliably detect these orfX-ACME MRSA strains. As the Xpert MRSA Gen 3 assay has recently been made available, we seek to ascertain if it is able to detect MRSA strains with orfX-ACME insertion. The Xpert MRSA Gen 2 and Gen 3 have a stated limit of detection (LoD) for MRSA at 80 CFU/swab and 256 CFU/swab respectively. This study evaluated and compared the LoDs of both the Cepheid Xpert MRSA Gen 2 and Gen 3 assays for an orfX-ACME MRSA strain. Methods: A series of dilutions was performed from 0.5 McFarland bacterial suspension of an orfX-ACME MRSA strain. To express the LoD value as CFU/swab, 100μl aliquots of the dilutions were pipetted into micro tubes. Dry dacron swabs were then placed in the micro tubes and left to completely absorb the suspension. The moistened swabs were then processed in accordance to the manufacturer's guidelines. Five bacterial concentrations, with viable counts ranging from 4.52E4 to 7.08E2 CFU/swab, were tested in replicates on Xpert MRSA Gen 2 and Gen 3 assay concurrently. Parallel aliquots of 100μl suspensions were also plated in triplicates, and colony counts performed after 24 hours incubation. The LoD was calculated using Probit analysis on IBM SPSS statistics ver. 21. Results: The limit of detection of Xpert MRSA Gen 3 assay was determined to be 394 CFU/swab at 95% confidence level. Conversely, the Xpert MRSA Gen 2 assay was not able to detect the orfX-ACME MRSA strain in any of the replicates even at the highest concentration tested, ie, 4.52E4 CFU/swab. Conclusions: This study confirmed that the Xpert MRSA Gen 2 is unable to reliably detect the orfX-ACME MRSA. On the other hand, the newer Xpert MRSA Gen 3 assay readily detects orfX-ACME MRSA, at close to its stated LoD. The capability to detect orfX-ACME MRSA strains is particularly important in Singapore as these strains are not uncommon, accounting up to 5% to 9% of MRSA isolates in certain hospitals, and thus an inability to do so may greatly hinder infection control efforts . The microbial community of the lower genital tract is known to influence the reproductive health of women. Bacterial Vaginosis (BV), a familiar condition, has been characterized as a reduction of vaginal lactobacilli and overgrowth of facultative anaerobic bacteria. Understanding the relative composition of individual organisms, their interactions with each other and the resulting impact on vaginal health is difficult because of their diversity. Therefore, we sought to develop a high-throughput molecular platform to assess the vaginal microbiome of symptomatic and asymptomatic patients. Methods: DNA was isolated from 837 clinical specimens collected and stored in ThinPrep (Hologic) or BD Universal Viral Transport Media (swab). Using the QuantStudio 12k flex system (Life Technologies), 2664 real time PCR reactions can be run in parallel on an OpenArray in a span of 8 hours, providing a wealth of data from a single specimen. Custom OpenArray plateswere designed to contain assays for 29 organisms, encompassing a large spectrum of vaginosis/vaginitis-associated pathogens. For relevant BV-associated organisms, relative quantitative data was generated by comparing to total bacterial load measured using the 16s rRNA gene. By comparing this data in symptomatic and asymptomatic patients, statistically determined clinical cut-offs were established and a clustering algorithm was applied to characterize the composition of these communities and their relationship to symptomatic and asymptomatic status. Results: Our analysis of 9 BV associated organisms and 4 lactobacillus species discerned consistent clustering patterns that broadly subdivided patients into Normal, Borderline and Abnormal categories. All clusters contained mixtures of anaerobes with or without Lactobacillus spp. Clustering of microbial profiles, and the application of a statistical model for the prediction of symptomatic status, demonstrated high sensitivity and specificity (90.1% and 88.5%, respectively). As expected, lactobacillus species displayed a negative correlation with the symptomatic state. Of note, L. iners was prevalent in vaginal microflora from both symptomatic and asymptomatic patients with and without depletion of other lactobacillus species. All BV associated organisms were detected in low numbers in normal vaginal microflora and occasionally in the Borderline stage, and differed in their relative proportions between the clusters. The presence of Megasphaera Type 2 was found to be highly correlated with symptomatic status. The OpenArray platform also performed robustly in the detection of Candida spp. as well as aerobic vaginitis pathogens, with limits of detection ranging from 500 to 50,000 organisms per sample. Overall concordance with reference laboratory results was 94%. Conclusions: The vaginal panel developed on the OpenArray is a rapid, high-throughput, sensitive and specific assay to aid in the accurate diagnosis of vaginal infections by evaluating a large panel of pathogenic and commensal organisms, and a potential tool to monitor the efficacy of antimicrobial therapy.
J.S. Thomas, D. Chung, J. Kilby, S. Sefers, J.D. Chappell Vanderbilt University Medical Center, Nashville, TN. Introduction: Diagnostic and treatment guidelines for solid-organ and hematopoietic stem-cell transplant recipients include frequent monitoring of circulating cytomegalovirus (CMV) loads. Current methods rely heavily on quantitative real-time PCR (QRT-PCR) laboratory-developed tests (LDTs), which demonstrate marked differences in performance, reproducibility, and calibration. To understand the interrelated effects of specimen type, calibrants, and assay methodology on CMV viral load, we compared: 1) a plasmid-calibrated QRT-PCR LDT (blood and plasma), 2) the same QRT-PCR LDT calibrated to the 1st WHO International Standard for Human CMV for Nucleic Acid Amplification Techniques (blood and plasma), 3) Roche COBAS AmpliPrep/COBAS TaqMan CMV Test calibrated to the WHO standard (FDA-approved for plasma), and 4) Bio-Rad QX200 Droplet Digital PCR (ddPCR) System using QRT-PCR conditions without external calibration (blood and plasma). Methods: The study included 25 CMV-positive and 20 CMV-negative adult whole-blood specimens submitted to Vanderbilt University Medical Center for CMV load testing using the plasmid-calibrated QRT-PCR LDT (reference method). Positive specimens ranged from 261 to 244,820 copies/ml. Residual blood and postcentrifugation plasma fractions were stored and tested by additional quantitative methods. Percent mean bias in log10 viral load was determined for pairwise method comparisons using Deming regression (R-value range, 0.50 to 0.99). Results: All methods showed 100% qualitative agreement with the reference method. Plasma yielded higher CMV loads compared to blood by all methods. The WHO standard did not improve agreement between plasma and blood CMV loads measured by QRT-PCR LDT (4.2% and 6.9% mean bias using plasmid and WHO standards). In plasma, calibration of the QRT-PCR LDT with the WHO standard increased agreement with the COBAS assay (change in mean bias from 9.2% to 6.8%). The impact was more pronounced in a comparison of the QRT-PCR LDT and ddPCR (change in mean bias from 22.0% to 6.1% with WHO standard). The reference method (blood) calibrated to the WHO standard compared to the COBAS assay (plasma) showed 0.20% mean bias, representing the closest agreement between any two methods. Conclusions: PCR-based measurements of CMV loads are subject to the combinatorial effects of multiple variables. Calibration with the WHO standard harmonizes the results of a classically designed QRT-PCR LDT with a commercial CMV load assay. DdPCR shows good agreement with assays calibrated to the WHO standard and offers promising clinical capability for absolute quantification of CMV. Further comparative studies, to include the role of patientspecific variables on inter-assay agreement and individual assay precision, are needed to define optimal conditions for CMV viral load testing.
C. Hildenbrand 1,3 , L. Wedekind 1 , G. Li 2 , R.Y. Zhao 2,3 1 University of Maryland Medical Center, Baltimore, MD; 2 University of Maryland School of Medicine, Baltimore, MD; 3 Institute of Human Virology, Baltimore, MD. Introduction. Cytomegalovirus (CMV) infection is often associated with complications of organ-transplant and opportunistic co-infection of HIV-infected individuals. It is also a leading cause of hearing loss, vision loss, and mental retardation in congenitally infected children. The FDA-approved Roche COBAS AmpliPrep/ COBAS TaqMan CMV Test measures CMV DNA viral load in plasma, which is to be used in conjunction with clinical presentation and other laboratory markers in the diagnosis and management of patients at risk for CMV-associated diseases. Besides plasma, however, CMV is often found in other human body fluids such as urine, CSF and BAL. Thus, monitoring of CMV in non-plasma samples for the clinical care and follow-up of organ-transplant becomes necessary. The objective of this study is to carry out analytic and clinical feasibility study of the Roche COBAS AmpliPrep/ COBAS TaqMan CMV Test in non-plasma sample types including CSF, urine and BAL. Methods. Twenty of the CMV(-) BAL, CSF or urine samples, which were tested by the Roche LightCycler TaqMan quantitative PCR method, were pooled together for this study. Commercially available CMV viral DNA in the standardized concentration of 1x106 IU/mL was purchased from Acrometrix Inc. The CMV viral DNA was diluted in 1:10 increments to generate 4 levels of viral loads, i.e., from 2.30 to 5.27 log10 IU/mL. These diluted viral DNA were then divided up in triplicate for the determination of analytical measurement range (AMR). The low limit of detection (LOD) was determined by testing BAL, CSF or urine samples spiked with serially diluted and standardized levels of CMV viral DNA from 50, 100, 200 to 300 IU/mL, individually. Results. The AMR and the linearity between the expected and observed viral load were calculated and plotted based on the test results of the serially diluted samples with the CMV viral load from log10 2.30 to log10 5.27. The differences between the expected and observed viral loads were less than 0.1 log at all levels. The correlation of regression between the expected and observed viral 3 types of samples. Known quantities of CMV viral DNA at the levels of 50, 100, 200, or 300 IU/mL were tested in 2 or 3 triplicate runs for each level, and the average of these results was used to define LOD. Although 100% of to 100% of the samples were detected at the 50 or 100 IU/mL, respectively for BAL, CSF and urine. To evaluate the assay specificity, 18 to 20 CMV(-) BAL, CSF and urine specimens were run against the COBAS AmpliPrep/ COBAS TaqMan CMV Test. Only 1 out of 18 BAL and none of the 20 CSF or urine samples were tested positive, suggesting 94.4%, 100% and 100% of specificity for BAL, CSF and urine, respectively. The assay precision was demonstrated by calculating the standard deviation from the average of the 9 values for each level (log10 2.30-5.27 IU/mL). Precision for these serially diluted standardized samples were all in the acceptable ranges. Conclusions. Based on the results of the above described testing parameters, these data showed that the BAL, CSF and urine samples could be used in the COBAS AmpliPrep/COBAS TaqMan CMV Test to measure CMV viral loads in the range of 200 to 1.875 x 105 IU/mL. Introduction: In the US, screening for cervical cancer includes routine PAP smear and high risk HPV (hrHPV) testing from a liquid cytology specimen. This is not the case in many low-and middle-income countries (LMICs) where testing is traditionally performed using a smear-to-slide PAP, and HPV testing is not performed. Without a liquid specimen, pathologists have not had samples suited to use in the existing PCR-based screening devices. In this study, we evaluated the feasibility of using a dried cervical swab to collect specimens for hrHPV testing using the Cepheid GeneXpert hrHPV test. Methods: Thirty cervical swab specimens were collected from Honduran women in a one-day clinic with consent and IRB approval. Swabs were air dried before being transported to the laboratory for testing. All swabs were rehydrated in 2 mL of 1X PBS for 15 minutes. One mL of rehydrated sample was loaded into the Cepheid hrHPV cartridge. Dried control swabs were prepared from cell pellets of liquid cytology specimens that were previously tested for hrHPV using the Roche Cobas assay. Results: Of the 30 samples, 73% (22) were negative for HPV and 27% (8) were positive for hrHPV. Of the positive samples, 50% (4) were hrHPV 16; 12.5% (1) hrHPV 18/45; 25% (2) "other" hrHPV genotypes; and 12.5% (1) had co-infections of hrHPV genotypes 18/45 and "other" hrHPV types. Eight controls including ThinPrep patient samples (16, 18, "other," and Negative) and mock flocked swab (16, 18, "other, " and Negative) were in 100% concordance with the Roche Cobas assay. Conclusions: The Cepheid hrHPV assay can be used with cervical swab specimens and is not limited to liquid cytology samples. This demonstrates the Cepheid assay is suitable for use in settings where liquid cytology is not available, such as rural locations, one-day clinics that typically operate out of schools or churches, and throughout LMICs where health care resources are limited. Dried swab collection makes testing more accessible, less costly, and more feasible and robust when liquid cytology samples do not need to be transported. Introduction: During influenza season, rapid virus identification provides critical information for treatment, management and isolation of infected patients. Historically, rapid virus identification was performed using antigen tests; these tests are being replaced by better performing molecular methods. The goal of this study was to compare the FDA cleared Roche Cobas Liat influenza A/B assay to our laboratory developed, real-time PCR respiratory virus panel (RVP) for influenza detection. Additionally, we modeled implementation of the Liat in our clinical laboratory as an initial screening test, and examined the effects on turn-around time (TAT) and cost. Methods: During the 2014 to 2015 influenza season, 104 samples were evaluated on the Liat. Positive samples included: influenza A (n=27), influenza B (n=27), respiratory syncytial virus (n=4), parainfluenza (n=6), adenovirus (n=2) and human metapneumovirus (n=2). Six indeterminate (high Ct) and 30 negative RVP samples were also analyzed. Discordant samples underwent replicate testing on both platforms. To model the implementation of the Liat in our laboratory, we analyzed data on all RVP specimens from the 2014 to 2015 influenza season (n=3865). Results: Liat analysis of influenza A or B positive samples with presumably high (20/21), respectively. For samples with viral burdens near the RVP limit of detection (Ct > 36.0), the positive agreements decreased to 50% (4/8) and 83% (5/6) respectively. We did not detect cross reactivity with other respiratory viruses and the agreement with RVP negative samples was 100%. The model with the greatest effect on TAT (from receipt in lab to result) for our hospital-based patients (~40% of samples) involves initial testing performed using the Liat, with reflex RVP testing for negative samples. Approximately 15% of these samples (n=202) are expected to be Liat positive and their TAT would decrease from 8 hrs to 40 min. For Liat negative samples, the change in TAT would be negligible. In this workflow, laboratory costs would increase by 1.5 fold or 2.5 fold for Liat positive and Liat negative samples respectively. Additional data from other models of Liat implementation will be presented. Conclusions: Based on our experience, the Liat results were highly concordant with our RVP results for influenza samples with an apparently high viral burden (low Ct). Conversely, the Liat did not reliably detect influenza samples with presumably low viral burdens (high Ct), most notably influenza A positive samples. Although one model of Liat implementation would increase laboratory costs, there would be a drastic reduction in TAT for hospital-based patients with influenza which may reduce other health care costs. Introduction: Herpes simplex virus types 1 and 2 infect and cause disease in almost every system of the body. Rapid identification will allow early antiviral intervention and appropriate infection control. This may allow a reduction in both the duration of symptoms and period of infectivity. The Simplexa HSV 1 & 2 Direct (Focus) assay is a multiplex, real-time PCR test which allows typing of HSV. This study assesses the test performance of the Simplexa HSV 1 & 2 Direct (Focus) assay compared to the LightCycler HSV1/2 (Roche) real-time PCR assay for CSF, urine, skin/oral, genital, and respiratory specimens. Methods: The Simplexa HSV 1 & 2 Direct assay is currently FDA cleared on the 3M Integrated cycler for the qualitative detection and differentiation of HSV-1 and HSV-2 in CSF. The comparator method in this study was the LightCycler HSV1/2 (Roche) real-time PCR assay (ASR), which requires a nucleic acid extraction step using the Roche MagNA Pure Total Nucleic Acid kit. Forty delinked specimens consisting of CSF (n=4), urine (n=6), skin/oral (n=15), genital (n=10), and respiratory (n=5) were tested. Results: The Simplexa HSV 1 & 2 Direct assay revealed an average limit of detection (LOD) Ct of 37.1 (34.8 to 39.5) for HSV-1 and a single determination Ct of 37.0 for HSV-2. The Roche LightCycler LOD showed an average Ct of 31.6 (31.2 to 31.8) for HSV-1 and 33.1 for HSV-2. This represents a sensitivity increase of 123.5X (14.5 to 304.7) for HSV-1 and 18.1X for HSV-2 for Simplexa. Viral detection for HSV1/2 in CSF and urine was completely concordant for HSV-1 (2/2 positive, 3/3 negative) and HSV-2 (2/2 positive, 4/4 negative). Skin/oral, genital and respiratory specimens were partially concordant at 6/6 positive, 7/7 negative; 5/5 positive, 4/4 negative; 4/4 positive respectively. This reveals a higher sensitivity for the Simplexa assay compared to the Roche LightCycler assay. Additional specimens identified by Simplexa consisted of 2 skin/oral (Ct: 35.3, 37.8), 2 genital (Ct: 37.4, 38.4) and one respiratory (Ct: 35.9). Conclusions: The Simplexa HSV 1 & 2 assay demonstrated an increased sensitivity for both HSV-1 and HSV-2 in non-CSF samples over the current Roche assay. Simplexa identified five additional positive results, which were in low concentration in the specimens as evidenced by the high Ct varies that were not detected using the Roche LightCycler HSV-1/2 assay. Introduction: As liquid media culture has become a routine practice in mycobacterial identification, rapid and accurate identification of Mycobacterium tuberculosis (MTB) complex and nontuberculous mycobacteria (NTM) in positive liquid culture is critical in mycobacterial culture practice. However, there are few data on the performance of molecular assays for this purpose. In this study, we evaluated four molecular assays for differentiation between MTB complex and NTM in positive liquid culture. Methods: A total of 100 nonselective consecutive mycobacterial culture-positive liquid media were obtained. Each culture broth was analyzed by the conventional multiplex PCR assay (Myco-ID V3, YD Diagnostics, Yongin-si, Korea) and three real-time PCR assays including AdvanSure TB/NTM real-time PCR (AdvanSure; LG life sciences, Korea), Real-Q MTB & NTM kit (Real-Q; BioSewoom, Seoul, Korea), and Genedia MTB/NTM Detection Kit (Genedia; Green Cross Medical Science Corp., Chungbuk, Korea). Results: Eighty-eight samples (88%) showed concordant results by all four molecular assays, whereas 11 (11%) and one (1%) samples gave discordant results in one and two out of four assays, respectively. When a concordant result in three of the four assays was considered to be a consensus, the sensitivities of the AdvanSure, Real-Q, Genedia, and Myco-ID V3 assays were 98.0% (95% confidence intervals, 92.2-99.6%), 99.0% (93.7-99.9%), 99.0% (93.7-99.9%), and 92.9% (85.5-96.9%), respectively. Analytical specificities assessed by using 85 reference strains including 1 MTB, 59 NTM, and 25 closelyrelated non-mycobacterial species were 100%, 98.8%, 98.8%, and 89.4%, respectively. Conclusions: Three real-time PCR assays presented comparable performance with each other, whereas the conventional multiplex-PCR assay showed inferior results to the real-time PCR assays.
Udhayakumar 2 1 Meridian Bioscience Inc., Cincinnati, OH; 2 Centers for Disease Control and Prevention, Atlanta, GA. Introduction: Current malaria diagnostic tests rely on parasite detection by microscopy methods or antigen-based rapid diagnostic tests (RDT). Molecular methods like polymerase chain reaction (PCR) can be used to increase the sensitivity of detection. However, practical deployment of molecular testing should address certain 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. Loop mediated isothermal amplification (LAMP) is a highly sensitive, rapid molecular method which can be used to detect Plasmodium DNA. We report on the performance of a simplified malaria assay in an easy to use LAMP platform. Methods: The Meridian illumigene Malaria DNA Amplification Assay (research use only (RUO), not cleared for use in USA) uses LAMP to detect Plasmodium parasite at the genus level. During LAMP amplification, an increase in turbidity occurs due to the magnesium-pyrophosphate built up. The turbidity is measured by the Meridian illumipro-10 instrument and a result is determined. Two methods were designed to extract DNA from blood. Both procedures rely on chemical lysis, and produce amplifiable DNA within 1 to 2 minutes (simple filtration, Meridian SMP PREP) or 7 to 10 minutes (gravity-driven gel filtration column, Meridian M-prep) of sample preparation. Both methods were evaluated using five well characterized Plasmodium falciparum standards (strains US05FPHI, US05FFC27/A3, US05FBeninI, US05FSantaLucia, US08F NigeriaXII), 50 archived specimens and the five human-infecting Plasmodium species (P. falciparum, P. vivax, P. ovale, P. malariae and P. knowlesi). A blood sample was mixed with lysis buffer, and the lysate transferred to either SMP PREP or M-Prep. The collected eluates from each device were directly added to the illumigene Malaria test device containing lyophilized Malaria test and control reagents. Results: The Limit of Detection (LoD) of P. falciparum was determined by Probit modelling to be 1.95 parasites/μl of blood with the SMP PREP method and 0.44 parasites/μl with M-Prep method. The assay also detected all tested human-infecting Plasmodium species. Sensitivity for both sample preparation methods was determined to be 98.0% by 50 microscopy or RDT positive, retrospective samples, and 100% after resolving with inhouse PCR method. No cross-reactivity was observed with human genomic DNA. Conclusions: The RUO LAMP based illumigene Malaria assay is capable of detecting the Plasmodium species at genus level with analytical sensitivity recommended by WHO (2 parasites/μl), while using an extremely simple procedure in less than one hour. This provides a much needed alternative to the more complex molecular test for malaria diagnosis.
Introduction: A simple, efficient and precise method for blood collection via finger prick, which removes cells and stores plasma at ambient temperature, is needed globally. Current collection requires electricity driven processing, temperature control and packaging. Dried blood/plasma spots on filter paper offers an alternative to shipping worldwide but with limited success as reduced assay precision/accuracy are linked to filter paper. Herein we describe functionality/performance of a novel prototype plasma separation card composed of a separation membrane and absorbent wick (ViveBio LLC, Alpharetta, GA). Cards collect capillary blood separating cellular components by gravity and store plasma at ambient temp pending laboratory testing. Methods: Varying volumes whole blood (WB) were pipetted drop wise (as if finger prick) onto 3 different membranes of varying size, each using an oversized absorbent wick (Wicks were inspected and weighed pre/post WB addition to evaluate percent plasma recovered, hemolysis and time of separation. Five different prototype cards were tested and WB allowed to absorb through the separator membrane and onto the wick. Prototype cards with single combination separation membrane/absorbent wick were selected to evaluate functionality for HIV viral load testing. HXB2 spiked WB (100 l) was pipetted onto cards and stored overnight. The next day wicks were removed from cards, added to tubes containing 1,000 l of SPEX buffer (Roche) and transferred to an Eppendorf Thermomixer and incubated for 10 min at 56°C and 1,000 rpm. by Roche COBAS TQ HIV-1 v2.0 for use with High Pure Extraction System. Results: For 2 of 3 separation membranes, sample passed through membrane within 4 minutes to 15 minutes (min) and plasma was captured on absorbent wick (~12% to 63% of expected volume). One of 3 membranes allowed WB to pass through and was disregarded from further testing. Two of 5 cards exhibited plasma wicking through within 1 min. Two of 5 cards exhibited slow flow to wick (~10 min to 20 min). WB on 1 of 5 cards did not flow through the membrane. HIV-1 RNA was recovered from wicks and quantitated, with mean viral load of 4.34 log c/ml and mean SD of 0.12 log. Conclusions: This proof of concept study demonstrates the prototype plasma separation card can be utilized for collection, separation and storage of plasma from WB without the need for electricity or cold chain storage and can be used with downstream viral load testing. Use of this device can offer global solutions and increase accuracy/reproducibility of healthcare services.
S. Ashraf 1 , R.P. Schaudies 2 , D. Diorio 3 , J. Pulliam 1 , D. Mittar 1 1 American Type Culture Collection, Manassas, VA; 2 GenArraytion Inc, Rockville, MD; 3 Cincinnati Children's Hospital Medical Center, Cincinnati, OH. Introduction: Noroviruses (NoV) are the most common cause of epidemic gastroenteritis, accounting for 95% of viral gastroenteritis outbreaks worldwide. Given that NoV are unculturable, ATCC has developed quantitative synthetic molecular standards for NoV GI and GII that are compatible with lab developed and commercially available PCR assays for the detection and quantification of this virus. Here, we describe the application of these standards for the development of new assays and as positive controls for assay validation. Methods: To develop new qPCR assays, the NoV GI and GII synthetic molecular standards were serially diluted from 10 -1 to 10 -6 , then tested with various NoV GI and GII specific primer/probe sets. The limit of detection (LOD) for each assay was calculated to determine the optimal primer/probe set for the assay. In an independent validation study, NoV GI and GII standards were serially diluted and assayed using Norovirus analyte specific reagents (Focus Diagnostics, Inc.) in order to select the optimal dilution to use as a control for all the subsequent assays. Assays using the synthetic NoV GI and GII controls were run (n=32) over a period of 2 months by 3 different technologists to determine the assay reproducibility and to establish a crossing threshold range (Ct). Results: The ATCC synthetic molecular standards were compatible with all the tested primers and probe sets. Of the two selected sets, set 1 specifically detected NoV GI and GII with a LOD of 470 and 480 genome copies, respectively, whereas primer set II detected NoV GI and GII with an LOD of 4.7 and 4800 genome copies, respectively. In the assay validation study, the average Ct value obtained from 32 runs was 30.27 with a standard deviation (SD) of 1.09 and co-efficient of variation (CV) of 3.6% for the NoV GI standard and a SD of 2.15 and CV of 6.4% for the NoV GII standard. The controls yielded reproducible data that fell within the 2.5 SD range for each of the NoV GI and GII standards. Conclusions:
The LOD data during our assay development studies and the consistency of Ct values from daily run positive controls clearly demonstrates the utility of the ATCC NoV GI and GII synthetic molecular standards for these and other applications for the detection and quantification of NoVs.
Introduction: The Luminex NxTAG Respiratory Pathogen Panel (RPP) is a research use only (RUO), qualitative nucleic acid multiplex test that provides simultaneous detection and identification of 19 viruses and 3 atypical bacteria associated with respiratory tract infections. NxTAG RPP is a ready to use system requiring minimal hands-on time and is performed in a closed PCR vessel, reducing the chances of contamination. Nucleic acid is added directly to pre-plated, lyophilized reagents for RT-PCR and bead hybridization. Results are read on the MAGPIX instrument. The objective of this study was to evaluate the Hamilton Microlab STAR platform's ability to automate extraction of total nucleic acid from contrived samples with subsequent addition of the extracted material to the NxTAG RPP reaction vessels. Methods: Contrived samples were created by suspending ATCC and Zeptometrix control strains in Remel Micro Test M5 media. Fifty microliters of contrived specimen was then added to 1 mL of BD UVT in preparation for extraction. Extraction of samples was performed by the Microlab STAR using the Promega Maxwell HT Viral Total Nucleic Acid kit or by the Biomerieux NucliSENS easyMag robot, reagents and Generic 2.0.1 protocol. The easyMag-extracted products were manually delivered to the RPP reaction vessels. The Microlab STAR robotically delivered extracted products to the RPP reaction vessesl. RT-PCR was performed on an Eppendorf Mastercycler Pro S using the NxTAG RPP RUO package insert-defined protocol. Turnaround time was measured from initial sample aspiration through extraction, adding extracted material to NxTAG RPP reaction vessels, thermal cycling and finally data acquisition on the MAGPIX. Results: A comparison of the turnaround times for easyMag-extracted samples versus Microlab STAR-extracted samples produced the following mean results, respectively: 24 sample runs were completed in 213 and 193 minutes; 48 sample runs were completed in 296 and 226 minutes; 72 sample runs were completed in 379 and 259 minutes; and 96 sample runs were completed in 462 and 292 minutes. The data showed a reduction in turnaround time for all run sizes: 10%, 31%, 46% and 58%, respectively. The positive and negative results from each extraction method matched for all samples tested. Conclusions: The Microlab STAR instrument represents a novel automated extraction method that substantially reduces turnaround time for the NxTAG RPP assay while providing comparable assay performance to the traditional easyMag extraction method. jmd.amjpathol.org ■ The Journal of Molecular Diagnostics years of age. Infections are predominately asymptomatic; however T. vaginalis has been linked to inflammatory reproductive tract disease syndromes including vaginitis, cervicitis and pelvic inflammatory disease. Pregnancy-related complications include pre-term birth and infertility. The objective of this study was to comparatively assess three nucleic acid amplification tests for T. vaginalis detection from urogenital specimens. Methods: A laboratory-developed, quantitative TaqMan PCR (LDT) was adapted and optimized for use with the User Defined Workflow software (UDF) for the cobas 4800 system. Using the UDF, T. vaginalis screening was performed with cobas 4800 eluates or spin column-extracted DNA from several lower reproductive tract specimen types collected from 5 LSU-affiliated hospitals/clinics in Louisiana (2013 to 2014; n=2000). Assay performance was evaluated, and parallel comparisons were made with two TIB MOLBIOL T. vaginalis tests also run on the UDF system. Results: For residual cobas 4800 DNA eluates, once-weekly freeze/thaw cycles for up to 4 weeks had no significant impact on template quantitation indicating extended stability of these eluates at -20°C. Using the LDT, DNA templates derived from serial dilutions of T. vaginalis organisms showed a linear range of detection from 1x10 1 to 1x10 8 organisms with % coefficient of variations (CV) ranging from 0.1 to 2.5. Spiking T. vaginalis organisms into reproductive tract specimens showed similar inter-assay reproducibility of the extraction and detection system (<1.7 %CV). The analytical sensitivity was 1x10 1 organisms per reaction for the LDT, which was detected 100% of the time; analytical sensitivity values for the TIB MOLBIOL tests were less than 1 organism per reaction. Statistical comparison of results to those of the TIB MOLBIOL test showed a Cohen's kappa correlation coefficient of >0.67 (95% CI 0.50 to 0.85) for liquidbased cytology specimens; kappa coefficients for all conditions were >0.6 with qualitativie agreements of "good" and "very good" among the specimen types. Conclusions: These results highlight the utility of the UDF system for qualitative and quantitative T. vaginalis detection from female urogenital specimens. Introduction: Wound infections can be severe and result in surgery, sepsis, or death without rapid diagnosis and treatment. Skin punctures or internal concussions facilitate a unique microenvironment allowing bacteria or toxin migration to the bloodstream. Current diagnostics can take 48 hours or more to accurately identify bacterial organisms and require specific growth conditions. Multiplex screening for bacteria from necrotic wounds is either not available or has poor sensitivity due to suboptimal extraction or amplification. Rapid and sensitive multiplex detection of bacteria from necrotic wounds is necessary to improve outcomes for infected patients. Methods: Nested primers were designed to amplify gene targets specific for Bacteroides fragilis, Clostridium septicum/novyi, Clostridium perfringens, Staphylococcus aureus, Kingella kingae, Streptococcus pyogenes, Staphylococcus lugdunensis, methicillin resistance, and the Panton-Valentine Leukocidin gene. Specificity was verified with SYBR real-time PCR, and all target primers were combined into a Target Enriched Multiplex Polymerase Chain Reaction (TEM-PCR) Necrosis Panel primer mix. DNA extraction methods were evaluated using S. pyogenes (1e2 to 1e5 cfu/mL) and C. perfringens (1e4 and 1e1 cfu/mL). TEM-PCR was performed, and PCR products were hybridized to probes coupled to barcoded magnetic beads and analyzed on the Applied BioCode-2000. PCR enzymes were evaluated for multiplex PCR amplification efficiency using a selection of panel organisms. Results: Two of ten extraction methods indicated S. pyogenes and C. perfringens detection at 1e2 and 1e1 cfu/mL, respectively. One extraction protocol did not isolate detectable levels of organisms. Two of seven PCR enzymes resulted in comparable results with the detection of S. aureus, S. lugdunensis, and C. septicum K. kingae and S. pyogenes B. fragilis and C. perfringens -Valentine Leukocidin genes were detected in S. aureus BAAamplified internal controls but did not amplify panel targets. Lot-to-lot variability testing of the two comparable enzymes yielded similar performance between lots of each enzyme and higher detection signals (41% to 548%) observed for one enzyme.
Conclusions: Process optimization of reagents and protocols improves sensitivity and performance of diagnostic assays, particularly with more efficient extraction and amplification methods. The TEM-PCR Necrosis Panel provides sensitive, specific, and rapid results. Accurate and prompt identification of bacteria from necrotic wounds can improve patient outcomes by preventing severe complications, such as amputation, sepsis, or death.
S. Dempsey, E. Huang, Y. Parocua, A. Bologa, P. Naranatt, M. Tabb Focus Diagnostics, Cypress, CA. Introduction: Influenza vaccine effectiveness was low during the 2014 to 2015 respiratory season. In addition, outbreaks of avian influenza in wild and farmed birds have increased worldwide, leading to increased concern about human infection. To ensure effective detection, testing of the Simplexa Flu A/B & RSV Direct assay (Simplexa Flu Direct) was conducted using 56 influenza A, B and RSV strains, including current circulating strains causing human infection, flu vaccine strains and avian and swine influenza subtypes. Simplexa Flu Direct is validated for nasopharyngeal (NP) swabs in universal transport medium (UTM), but is in development for use with nasal aspirate/wash, a common sample type for pediatric patients. We evaluated Simplexa Flu Direct by testing limit of detection for influenza A and B, and RSV in nasal aspirate/wash matrix and clinical performance using nasal aspirate/wash patient samples. Methods: Analytical reactivity testing was conducted using viral stocks from BEI Resources, Influenza Reagent Resource & ATCC. Strains tested included 2015-16 influenza vaccine strains; influenza A/Switzerland/971523/2013 (H3N2) & A/California/02/2014 (H3N2); avian influenza subtypes H5N1, H5N2, and H9N2; and swine flu subtypes H1N2 and H3N2. Viral sequences were obtained from GISAID EpiFlu, Influenza Research Database and NCBI's Nucleotide. Nasal aspirate/wash clinical samples were collected by Dell Children's Hospital. Limit of detection was tested using viral strains spiked into nasal aspirate/wash matrix. Results: Viral sequences were evaluated using in silico analysis against the Simplexa Flu Direct Scorpions and primers. Each strain tested was predicted to be detected with the assay. With the exception of inactivated viruses without titer information, viruses were detected at 100 TCID50/mL or 100 CEID50/mL. LoD studies using nasal aspirates showed comparable limits of detection to those previously reported for NP swabs for influenza A/PR/8/34 (H1N1), influenza A/Hong Kong/8/68 (H3N2), influenza B/Malaysia/2506/2004, influenza B/Great Lakes/1739/54, RSV-A2 and RSV B CH93-18. Simplexa Flu Direct detected 8 Flu A-positives, 5 Flu B-positives, and 5 RSV-positives from 77 nasal aspirate/wash clinical samples. Cepheid's Xpert Flu/RSV XC was used to verify Simplexa clinical performance. Conclusions: Simplexa Flu A/B and RSV Direct can detect 2015 to 2016 influenza vaccine strains, current circulating strains, avian influenza subtypes H5N1, H5N2, and H9N2 and swine influenza subtypes H1N2 and H3N2. Simplexa Flu Direct was originally validated for nasopharyngeal swabs in UTM; preliminary analytical and clinical results suggest the assay can also be used for nasal aspirates/washes. Simplexa Flu A/B and RSV Direct is not FDA cleared for nasal aspirate/wash specimens.
The Aga Khan University, Karachi, Sindh, Pakistan. Introduction: The extensively drug-resistant tuberculosis (XDR-TB) has emerged worldwide as one of the biggest threats to public health and TB control programs. The XDR-TB is defined as tuberculosis (TB) caused by Mycobacterium tuberculosis (MTB) strains that are resistant to at least rifampin (RIF) and isoniazid (INH) among the first-line anti-TB drugs and resistant to a fluoroquinolones and to at least one of the three injectable second-line drugs. The resistance to first-and second-line antituberculous drugs has been associated with single nucleotide polymorphisms (SNPs) in particular genes. It is has been shown that efflux pumps play an important role in mechanism of resistance in bacteria including MTB. Up regulation of these efflux pumps can decrease the intracellular concentration of drugs and reduce their efficacy. Methods: Whole Genome Sequencing (WGS) was performed on (n=37) XDR MTB strains using Illumina paired end HiSeq2000 technology, the WGS raw sequence data was mapped distinctively to H37Rv reference genome to identify SNPs using the SAMtools/BCFtools. We also analysed the non-synonymous (nsSNPs) in the efflux pump genes of (n=37) XDR MTB using the KVarq software and compared with H37Rv reference genome. Results: Of the (n=37) XDR MTB strains analysed by kvarq software, 8 (19.5%) XDR MTB strains were wild type for rpsL, rrs (500 region) and gidB genes with nsSNPs (aspartic acid to histidine) in the drrA efflux pump gene at position 3273138. Three XDR MTB strains, wild type for rpsL, rrs, gidB and gyrB genes were phenotypically streptomycin sensitive and 5 XDR MTB strains were streptomycin resistant.All XDR MTB strains, wild type for rpsL, rrs, gidB and gyrB genes were fluoroquinolone (ofloxocin) and ethambutol (EMB) resistant. 32% (n=13) XDR MTB strains had mutations in the rpsL, rrs, gidB and drrA efflux pump genes. 3 XDR MTB strains wild type for rpsL, rrs and gidB genes showed nsSNPs (Isoleucine to Valine) in the Major facilitator superfamily (MFS), Rv1634 efflux pump gene at position 1839306. Efflux pump protein genes Rv0194, Rv2688c, pstA2, Rv0143c, Rv3193c and arsC were also examined and 36/37 XDR strains had SNPs in at least one of these genes. Conclusions: Resistance to anti-tuberculosis drugs may be attributed to polymorphisms in efflux pump genes. Therefore, diagnostics for MTB drug resistance should include new targets to determine the full extent of resistance in isolates. Introduction: Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) represent significant global public health problems. Point of care (POC) testing can drastically reduce disease burden of these organisms by providing rapid results which reduces time to treatment. Additionally, POC testing can be deployed in resource poor settings where complex diagnostic technology is not available. A novel molecular system, io, that generates results in 30 minutes using a fully-integrated cartridge comprising of sample prep, PCR amplification and proprietary electrochemical detection was previously developed for detection of CT. In this study, we sought to use this technology for detection of two distinct NG targets as
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org well as an internal control (IC) with the promise of combing the NG and CT in a single cartridge using one PCR chamber, and two detection chambers respectively for each organism. Methods: Inclusivity was tested using 8 WHO NG strains and 25 NG clinical isolates. Sixteen species, closely related to NG were tested to ensure no cross-reactivity. Sixty clinical samples were extracted, purified and the eluted DNA used to reconstitute dried amplification reagents followed by ultra-rapid amplification. In the detection chamber the amplification products are recognised by target-specific probes labelled with electrochemical labels. The resulting double strands are cleaved by an exonuclease releasing the electrochemical labels, which are then detected on electrodes by pulse voltammetry. Results: All WHO strains and clinical NG isolates were amplified and detected by the assay. The closely-related Neisseria species were not amplified or detected. For the clinical samples, a cut-off threshold for negative and positives was established. Five negative and 2 positive samples for NG were removed from the analysis due to mechanical failures. The assay detected 28/28 positives and 24/25 negative samples resulting in a sensitivity of 100% and a specificity of 96%. The test was developed to be combined in a cartridge with a previously developed test for CT which was shown to have 98.1% sensitivity and 98% specificity. Conclusions: The inclusivity and exclusivity results indicate the assay can detect a wide range of N. gonorrhoeae strains and that there is no crossreactivity with closely related species. The clinical sample data results show sensitive and specific detection of NG using multiplex PCR and electrochemical detection using elements of a fully-integrated POC system. This novel molecular system can accurately detect CT and NG in a clinical sample at Point Of Care in 30 minutes. This technology should allow for simultaneous detection of CT and NG in a sensitive, low-cost, rapid PCR POC device that could improve management and treatment of CT and NG infections.
Introduction: Manufacturers of molecular controls rely on cold chain logistics to deliver products to laboratories. Controls packed on dry ice require insulated shipping containers increasing overall cost. A method to store and transport control material in dried and thermostable format has advantages for manufacturers and laboratories. To illustrate Alere recently received FDA clearance/CLIA waiver for i Influenza A/B assay which utilizes dried thermostable controls. In this study two sets of HCV controls and one HIV control set (2 commercial, 1 lab developed) were analyzed in their native frozen format and compared to ViveST dried, thermostable format (ViveBio LLC). Methods: A HIV-1 Genotype Performance Panel (SeraCare, PRD201) was loaded (0.5mL) onto ViveST, dried overnight and stored RT. Samples were recovered using 1mL molecular grade water (mgw) and analyzed using the ViroSeq HIV-1 Genotyping System v2.0 (Abbott Molecular). Accurun 305 Series 200 and Series 400 HCV RNA Control material (SeraCare) were loaded onto ViveST (n=9 each series, 1mL aliquots), dried overnight and stored RT. Samples were recovered using 1mL mgw and analyzed with frozen Accurun samples using Abbott RealTime HCV Assay (ART). bioMONTR HCV RNA Control Panel samples were loaded onto ViveST (1.2mL aliquots, 5 levels in duplicate, n=10), dried overnight and stored RT. Samples recovered using 1.2mL of mgw and analyzed with paired frozen replicates on Roche COBAS AmpliPrep/Cobas TaqMan HCV Assay (TQ). Results: Sequence analysis for HIV-1 Genotype panel demonstrated 100% concordance with expected results from SeraCare Panel PRD201. Accurun 305 Series 400 panel (frozen) yielded average HCV RNA VL of 4.61 LOG IU/mL (SD=0.06) whereas samples on ViveST yielded average VL of 3.66 LOG IU/mL (SD=0.05). When compared to frozen, the average HCV RNA reduction was 0.95 LOG (ART). Accurun 305 Series 200 panel samples (frozen) yielded average VL of 2.83 LOG IU/mL (SD=0.04) whereas ViveST samples averaged 2.20 LOG IU/mL (SD=0.07). When compared to frozen the average reduction HCV RNA was 0.63 LOG (ART) whereas the SD's were similar. For the bioMONTR HCV control panel, the overall Pearson coefficient among paired frozen/ViveST samples was 0.9971 (TQ). When compared to frozen, average difference was -0.30 LOG and linear regression analysis on paired samples yielded R 2 =0.9942. Conclusions: These data support ViveST as an alternate storage and shipping method for molecular controls in dried, stable format. ViveST allows the amount of control material loaded to be adjusted depending on the application. ViveST eliminates need for cold chain storage and provides access to new and emerging markets with a method that keeps control material stable at room temperature.
M. Espy, C. Irish, M. Binnicker Mayo Clinic, Rochester, MN. Introduction: Infection with herpes simplex virus (HSV) can be found in a wide range of clinical specimens, including genital, dermal, respiratory, ocular, whole blood and cerebrospinal fluid. Although detection of HSV-1/2 by real-time PCR is sensitive and specific, most assays require preanalytic extraction of nucleic acid. Recently, Focus Diagnostics (Cypress, CA) developed a real-time PCR system (Simplexa) that incorporates both extraction and detection of HSV-1/2 in a closed system. The assay provides rapid results (~65 min) and is FDA-cleared for cerebrospinal fluid (CSF). In this study, we sought to evaluate the performance of the Focus HSV-1/2 Direct assay using a variety of non-FDA-cleared sample types, including blood, respiratory and ocular specimens. Methods: Sixty-three clinical determined to be positive for HSV-1 (n=48), HSV-2 (n=11) or HSV non-typeable (n=4) 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). Following routine testing, an aliquot of each sample (50 μL) was tested by the Focus HSV-1/2 Direct assay on the 3M Integrated Cycler (Focus) according to the manufacturer's instructions for CSF. Results were analyzed by comparing the Focus results to the results obtained by routine testing, which was considered the reference standard for this study. Results: Following testing of the sixty-three specimens, the Focus HSV-1/2 Direct assay demonstrated a sensitivity of 100% (48/48) for HSV-1 and 91.0% (10/11) for HSV-2 when compared to the Roche HSV-1/2 ASR. The four samples that were resulted as "HSV non-typeable" by routine testing were determined to be positive for HSV-1 by the Focus assay. The specificity of the Focus HSV-1 and HSV-2 assays was found to be 100%. Conclusions: Direct testing of blood, respiratory and ocular specimens by the Focus HSV-1/2 assay showed comparable performance to routine testing that included nucleic acid extraction. The Focus HSV-1/2 Direct assay allows for the rapid and sensitive detection of HSV-1/2 from a variety of sample types, including those (i.e., ocular) where specimen volume is typically limited. Introduction: Adenoviruses (AdV) cause a wide range of illnesses, from mild respiratory infections in young children to life-threatening multi-organ disease in people with a weakened immune system including transplant recipients. The implementation of a quantitative real-time PCR assay to monitor AdV viral load is expected to improve prognosis and management of AdV infection in transplant patients. Methods: Plasma specimens were separated from whole blood collected in EDTA containing tubes drawn from transplant patients and stored at -70ºC until analysis. Additional viral isolates from nasal specimens were obtained from North Shore-Long Island Jewish Health System Laboratories and spiked into plasma. AdV Reference Material: AcroMetrix Adenovirus Plasma Panel containing specific concentrations of Adenovirus (Type 3) in a human plasma matrix (1x10e3 to 1x10e7 copies/mL) was obtained from Life Technologies and utilized in determining the limit of quantification. Specimens were extracted on the QIAcube and quantified using the Altona Diagnostics RealStar Adenovirus PCR Research Use Only Kit on the Rotor-Gene Q instrument. Results: Viral loads for 44 clinical specimens, 22 of which were reported positive, were determined using the Altona real-time PCR kit and compared to values obtained by a reference laboratory. A good correlation between observed and expected values (± approximately 0.5log) was obtained over at least 6 logs of detection with a coefficient of determination (R2 value) of 0.98 when compared to the reference laboratory. For the AdV-positive reported samples viral load ranged in concentration from 273 to 1,437,688,203 copies/mL (mean= 77,645,149 and median= 11,982), and was on average within 0.17 log with the Altona real-time PCR assay. The assay was linear (R2=1) over 4 logs of detection (2.7-7.0 log10 copies/ml) using a commercially available AdV quantification panel and assay precision ranged from approximately 1% to 3% CV. The specificity was 100% with all 22 samples having viral loads below the limit of quantification of 200 copies/mL. Conclusions: The Altona Diagnostics RealStar Adenovirus real-time PCR assay permitted rapid, sensitive and specific detection and quantification of Adenovirus and may be useful in monitoring the course of infection in transplant patients. 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 19% of total child deaths. High-throughput multiplex assays can aid in rapid identification of pathogens that can cause outbreaks of diarrhea and for infection control in healthcare settings. Using the proprietary barcoded magnetic bead (BMB) technology, Applied BioCode has developed a molecular diagnostic assay for detection of gastrointestinal pathogens. In parallel, we have developed an automated high-throughput system with a 96-well format. Methods: The BioCode MDx3000 platform integrates and automates PCR, post-PCR sample handling and detection steps in a 96-well format. Following extraction of nucleic acids with an automated system, DNA and RNA targets were amplified by RT-PCR. PCR products were captured by target-specific probes coupled to BMBs, and the presence of target sequence(s) was detected by a fluorescent conjugate. Qualitative results were determined by a median fluorescent index (MFI) value relative to assay cutoff. The BioCode GI Pathogen Panel is a 17-plex molecular assay for detection of gastrointestinal pathogens which include bacteria (Campylobacter, C. difficile toxin A/B, Salmonella, Shigella/enteroinvasive E. coli, enteroaggregative E. coli, enteropathogenic E. coli, enterotoxigenic E. coli, shiga toxin-producing E. coli, E. coli O157, Vibrio, Yersinia enterocolitica), viruses (norovirus group I/II, adenovirus F, rotavirus A), and parasites (Cryptosporidium, Entamoeba histolytica, Giardia lamblia). Results: The BioCode GI Pathogen Panel did not cross-react with organisms tested in this study. Preliminary limit of detection was 1.2x10 3 CFU/mL for C. jejuni, 1x10 1 cysts/mL for E. histolytica, and 5.0x10 2 TCID50/mL for adenovirus 41. No well-to-well contamination was observed. Liquid transfer precision by the MDx3000 was < 5% CV for 5 μl to 50 μL. Equivalent performance was observed for unpreserved stool and stool in Cary-Blair medium. Assay performance for 287 stool specimens showed 91% positive agreement and 94% negative agreement with a FDA-cleared multiplex assay. Out of 282 specimens, 52 (18.4%) gave positive results for more than one pathogen. Conclusions: The BioCode MDx3000, automated for integration of PCR, post-PCR sample handling and detection steps, enables multiplex molecular detection in 96-well format whereas BioCode GI Pathogen Panel specifically detects several 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 the contamination risk.
S. Das, R. Perez, A.G. Ward, S. Dunbar Luminex Corporation, Austin, TX. Introduction: US hospital systems and diagnostic laboratories are continuously looking for more effective management strategies to maximize productivity, improve workflow, optimize staff time, and reduce the time to deliver results back to healthcare providers. In addition, there is a desire to shift from traditional testing methods to faster, more sensitive, and more cost-effective molecular methods. In this report, we describe a time and motion study conducted with the Luminex ARIES system. ARIES is a fully integrated, automated, sample-to-answer platform that performs extraction of nucleic acid from clinical samples, followed by real-time PCR detection, data analysis, and results reporting. Methods: This study compared the hands-on time for ARIES and Cepheid GeneXpert systems. The study was designed to compare the hands-on time required to set up various numbers of samples, including data entry, patient/sample information entry, and then starting the run for both systems. The Xpert GBS LB and ARIES HSV 1&2 assays were used for this study as the Xpert GBS LB assay does not require any pre-processing steps and is most similar to ARIES workflow. Two different scenarios were tested for ARIES-i) entering test order, sample, and assay information at the instrument using the barcode reader provided, and ii) test orders for samples are previously created by a Laboratory Information System (LIS). The times measured include the time required for removing cassettes/cartridges from the kit box and individual pouches, adding the sample, placing cassettes in a magazine (ARIES), and loading the tests onto the systems. Results: The average time needed to load one sample was 71 secs for ARIES standard workflow, 34 secs for ARIES in LIS mode, and 52 secs for GeneXpert. As the number of samples increased to 16, ARIES standard workflow required 9 mins 52 secs, ARIES in LIS mode was 4 mins 58 secs, and GeneXpert was 10 mins 10 secs. A considerable reduction in hands-on time was realized when ARIES was configured in the LIS-enabled mode and the orders were created by the LIS. It was observed that loading 12 samples into ARIES in LIS mode reduced the hands-on time by 45% as compared to GeneXpert. Conclusions: In this time and motion study, we found that as the number of samples approached six, the set up time favored ARIES as less hands-on time and fewer user interactions were required. This pattern repeated at each multiple of six samples, since specific steps are constant on ARIES for groups of one to six samples. However, setting up a run for a single sample was 19 secs faster on GeneXpert. Overall, ARIES demonstrated enhanced simplicity, reduced hands-on time, and less chance for user error as compared to the GeneXpert system.
The Ohio State University Wexner Medical Center, Columbus, OH. Introduction: Herpes simplex virus (HSV-1 and HSV-2) causes a wide range of clinical manifestations with morbidity and severity in all age groups including neonates, children and adults. Nucleic acid amplification tests are more sensitive than culture for the detection of HSV. In this study, we report our experience with 3 different amplification assays and protocols for the detection of HSV-1/2 in various clinical specimen sources. Methods: One hundred four specimens genital (n=42), oral (n= 18), skin (n=18), CSF (n= 26) were collected and extracted with NucliSENS easyMag (bioMerieux). The extracted nucleic acids were amplified and detected with Focus Simplexa HSV1/2 ASR primer pairs (Focus indirect), Qiagen Artus (QA) HSV-1/2 PCR and Eragen MultiCode RTx (EM) HSV 1 & 2 assays. Additionally, a prospectively collected panel of de-identified genital swab specimens (n=89) were tested as part of a clinical study with Focus Simplexa HSV 1 & 2 Direct versus the culture-based ELVIS HSV ID and D3 Typing Test System (Diagnostic Hybrids, Inc). Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of these assays was calculated by comparing test results with culture and/or molecular tests including bi-directional sequencing. Results: The overall sensitivity, specificity, PPV and NPV for the 104 specimens tested with the Focus Simplexa indirect assay was 100%, 98.6%, 97.1% and 100% compared to the Eragen MultiCode assay and was 100%, 95.8%, 91.9%, 100% compared to the Qiagen Artus assay for HSV-1. For HSV-2, the sensitivity, specificity, PPV and NPV was 100% across the board for the above 3 assays. For the 89 genital swab specimens tested with the Simplexa HSV 1 & 2 Direct and ELVIS assays, the sensitivity, specificity, PPV and NPV were 100%, 98%, 90.9% and 100% for HSV-1 and were 69.2%, 91.8%, 78.3% and 87.5% for HSV-2 respectively. Comparison of Simplexa HSV 1 & 2 Direct with bi-sequencing, revealed the sensitivity, specificity, PPV and NPV to be 98.5%, 94.7%, and 98.5% for HSV-1 and 100% across the board for HSV-2 respectively. Conclusions: The Focus Simplexa HSV 1 & 2 Direct and indirect assays detected HSV-1 and HSV-2 reliably from several clinical specimen sources. Simplexa assays depicted high sensitivity and specificity in comparison to other molecular assays. Simplexa HSV 1&2 Direct is easy to use and can identify additional HSV positive samples not detected by the culture assay. Simplexa HSV 1 & 2 Direct also has the added ability to detect specimens without up-front nucleic acid extraction allowing for a rapid (~1h) turn-around time. This assay is currently awaiting FDA approval using genital swab specimens. Introduction: FDA clearance for diagnostic devices requires sensitivity and reproducibility studies prior to approval. Isolates of pathogens capable of being used as biological weapons are uncommon, but development and validation of diagnostic devices for these pathogens is needed. To mitigate this problem, standardized methods to generate mock clinical samples by spiking cultured pathogens into blood or plasma have been developed. The goal of this study was to assess the sensitivity and reproducibility of mock samples produced by standardized methods that can be used for validation of diagnostic devices for detection of potential biological weapons. Methods: Mock samples were prepared by the FDA by spiking Bacillus anthracis, Staphylococcus aureus, Leishmania donovani, Yersinia pseudotuberculosis, and Dengue virus into blood or plasma of healthy or symptomatic donors at high, medium, and low concentrations. The isolated nucleic acids (NAs) were quantified in PCR Detectable Units/mL (PDU/mL) using TaqMan assays. The same NAs extracted from the mock samples were blinded and tested with Target Enriched Multiplex Polymerase Chain Reaction (TEM-PCR). Amplicons were hybridized to probes coupled to barcoded magnetic beads and analyzed on an Applied BioCode-2000 reader. Results: There was 100% correlation of TEM-PCR and TaqMan Assays at the highest concentrations, with one exception. The TEM-PCR assay detected 90% of B. anthracis samples at 1e5 PDU/mL, compared to the FDA TaqMan Assay, which detected 100% at the same concentration. Results of the two assays did not vary more than 15% at medium and low concentrations, with one exception. L. donovani was detected at 40% with the TEM-PCR assay and 70% with the TaqMan Assay at the lowest concentration (1e3 PDU/mL). There was one false positive in the unspiked blood sample with TEM-PCR. There were no false positives in un-spiked samples with the TaqMan Assay. There was no difference in results using spiked blood or plasma from symptomatic or healthy donors at high concentrations with a correlation of 100% between two datasets. There was variation from 4% to 43% between the two datasets at medium and low concentrations of spiked blood, 0% of spiked plasma. Conclusions: Testing of blinded mock samples by end-point TEM-PCR correlated with results generated with TaqMan Assays. This correlation strongly supports the validity of the methods used to prepare mock samples. Standardized methods for creating mock samples may allow a means for meeting FDA sensitivity and reproducibility studies for clearance of diagnostic devices for detecting rare pathogens. Introduction: Mycoplasma genitalium is emerging as a prevalent sexually transmitted infection that has been linked to several inflammatory syndromes in women with HIV, including pelvic inflammatory disease and cervicitis. Such inflammation may enrich the population of HIV-infected cells in lower reproductive tract tissues, and in turn enhance the sexual transmission of HIV. The objective of this study was to utilize a novel lab-developed nucleic acid amplification test (LDT) to determine the prevalence of M. genitalium in a longitudinal cohort of HIV+ of Louisiana women, and then characterize the inflammatory response both cytologically and histologically in those positive for M. genitalium. Methods: To detect M. genitalium, we utilized a novel laboratory-developed, quantitative TaqMan PCR assay to screen a longitudinal cohort of 108 HIV(+) New Orleans women enrolled into the HIV Outpatient Program (HOP). Study participants visited the clinic approximately once every three months from 2009 to 2014. To characterize the inflammatory response in the cervix, we measured cytokines and chemokines from cervicovaginal lavages (CVLs), quantified leukocytes from liquid-based cytology specimens, and characterized leukocytic infiltrates in cervical tissues using immunohistochemistry. Results: The prevalence of M. genitalium and T. vaginalis
A.S. Ang, K. Poon, M. Khoo, L. Chiu, E. Koay National University Health System, National University Hospital, Singapore. Introduction: Cervical cancer is the first cancer to be causally linked to a virus, the human papillomavirus (HPV). For many years, the cervical cytology-based PAP smear test had been the primary screening assay in detecting precancerous cervical lesions and cervical cancers. However, there is an increasingly strong recommendation for the utilization of HPV DNA testing either as a co-testing strategy or as a primary screening test for cervical cancer. The Roche Cobas 4800 HPV Test is currently the only FDA-approved molecular assay which has an established clinical cutoff in addition to an analytical lower detection limit, and offers individual detection for HPV16 and HPV18, as well as pooled results for 12 high-risk (31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68) HPV types. We implemented its replacement of our previous method, the Hybribio HPV16/18 Real-Time PCR, for clinical use in our hospital in January 2015, after thorough evaluation of its performance. We present our evaluation data and also the results of a two-year assessment of the prevalence of different HPVs on Singaporean women at the National University Hospital. Methods: During method evaluation, 159 paired samples collected separately in the liquid-based SurePath preservative fluid and Digene DNAPap Cervical Samplers were prospectively procured for the concurrent testing by Cobas 4800 HPV Test and Hybribio HPV16/18 Real-Time PCR, respectively. The HPV prevalence study was based on cervical swab specimens from 1642 women, two-thirds of which were collected using Digene DNAPaP Cervical Samplers and tested using the Hybribio test kit (from January 2013 to December 2014) and one-third were collected from patients (between January and May 2015) using SurePath preservative fluid and ran on the Roche Cobas 4800 HPV system. Results: Overall agreement of 96.8% (154/159) was achieved in concurrent testing by both Cobas 4800 HPV Test and Hybribio HPV16/18 Real-Time PCR assays. Among the 1642 samples tested during 2013-2015, 240 (14.61%) tested HPV positive. Of those women tested positive, including the ones with co-infections, 12.5% (N=30) were found to be HPV 16 positive, 5.42% (N=13) for HPV18, and 85% (N=204) were positive for other high risk HPV. Of those that were other high-risk HPV positives, 45% were found to be HPV 52 or HPV56. Conclusions: Contrary to the findings in women with HPV infections in Western countries, in which HPV16 was the most common HPV type, followed by HPV18, our initial data (N=1642) did not show the same trend in the Singapore women tested. Substantiation of these preliminary results with a larger cohort would be beneficial in evaluating the costeffectiveness in carrying out national HPV vaccination programs for future cervical cancer prevention strategies. Introduction: Luminex next-generation NxTAG Respiratory Pathogen Panel (NxTAG RPP) pilot assay for research use is a high-throughput assay designed to simultaneously detect and discriminate nucleic acids from 19 viruses and 3 atypical bacteria extracted from respiratory samples. NxTAG RPP is a ready to use system with a throughput capability of up to 96 reactions per run. An aliquot of extracted nucleic acid is directly added to pre-plated lyophilized reagents. Multiplexed RT-PCR and bead hybridization is carried out as one single cycling program in a closed PCR vessel. Data is acquired on the Luminex MAGPIX in <4 hours for a batch of 96 samples and the results are analyzed with SYNCT software. No post-PCR sample handling is required. This study evaluates the pilot assay performance with nasopharyngeal swabs collected from symptomatic subjects. Methods: The NxTAG RPP pilot assay was evaluated with remnant de-identified clinical samples collected during the 2014/2015 flu season from 303 subjects. All samples were tested with BioFire FilmArray RP assay which was the primary comparator method for the common targets between the two assays. For human Bocavirus and Legionella pneumophila (not probed by the FilmArray assay), in house real-time PCR assay and culture were used as primary comparator methods. Bi-directional sequencing was used for discrepancy analysis. Results: Overall positive agreement between the NxTAG RPP pilot assay and FilmArray RP for the 16 common targets was 98.3% (289/294). Positive agreement was 100% for 11 targets: H3 (35/35), RSV (40/40), Adenovirus (45/45), Human Metapneumovirus (hMPV) (10/10), Parainfluenza (PIV) 1 (7/7), PIV2 (4/4), PIV3 (8/8), PIV4 (6/6), Coronavirus 229E (4/4) and NL63 (8/ Introduction: It is common to encounter granulomas in lymph nodes draining carcinoma and in a population where tuberculosis is rampant; this is often treated as tuberculosis. As nodal tuberculosis is paucibacillary, it generates clinical and diagnostic dilemma, whether the granuloma is due to tuberculosis, cancer itself, or chemotherapy induced reaction. The present study aims to address the issue of determination of etiology of granulomas in draining lymph nodes in breast cancer patients using molecular techniques. Methods: Excised lymph nodes (FFPE) from breast cancer patients (n=63) with granulomas in draining nodes were analyzed. Nested PCR for IS6110 was performed using DNA extracted from QIAamp DNA FFPE tissue kit (Qiagen, GmbH) for detecting Mycobacterium tuberculosis (MTB) with appropriate controls and CK19 PCR was performed using RNA extracted from RecoverAll Total Nucleic Acid Isolation Kit (Ambion, Life Technologies, USA) for occult metastatic tumor cells. Positive results were confirmed by sequencing on 3500 GA ABI. Molecular analysis was correlated with the clinical and histological data (Table 1) . Results: Granulomas were detected in axillary nodes (n = 48), supraclavicular node (n = 13), neck node and mediastinal node in one each. Of the jmd.amjpathol.org ■ The Journal of Molecular Diagnostics 63 patients, 26 had received chemotherapy; 29 did not and in 8 patients details were unavailable. Out of 63 patients, the draining nodes showed metastasis in 42. A total of 11 out of 63 cases showed CK19 positivity and 6 were positive for MTB. One case was positive for both CK19 and MTB. In MTB positive cases, nodal granulomas and metastatic tumor were seen in different nodes. Three patients had nodal metastasis; whereas 3 patients were node negative. In CK19 positive cases, 4 had granulomas and metastatic tumor in different nodes, 3 had nodal metastasis with granulomas in the same node whereas 4 patients had no nodal metastasis. As seen in Table 1 , the presence of subcapsular or perisinusoidal granulomas was significantly associated with CK19 PCR positivity. Both mycobacterial granulomas and granulomas as part of chemotherapy related tumor damage are seen in nodes draining cancer. Presence of peri-/sinusoidal granulomas is a feature that favors metastasis over tuberculosis.
J.H. Chen 1 , 5 , J. Xu 2 , 5 , Y. Jiang 3 , E.S. Ma 4 , B.S. Tang 4 , H.Y. Lam 4 , C.C. Yip 1 , V.C. Cheng 1 , H. Zhang 2 , K.Y.
A. Rosado, I. Rivera, J.R. Paz, M.A. Noy Laboratorio de Patología Dr. Noy, San Juan, Puerto Rico. Introduction: A correlation subsists between premalignant and malignant lesions and oncogenic human papilloma virus (HPV) types. Segregation of HPV DNA has been achieved from 46% to 100% of in situ and invasive squamous cell carcinomas (SCCs) of the anus. Epidemiologic studies demonstrate that up to 93% of anal SCCs are related to HPV infection. Presently, routine screening tests for HPV on the anus do not exist. Therefore, the objective of this study is to determine an evidence-based method to screen for High Risk Anal HPV, particularly the genotypes 16, 18 and/or 45, since medical professionals in Puerto Rico (PR) have shown a tendency to request analysis for these specific types. Methods: Two hundred nineteen anal Pap smear samples were collected from several medical clinics in PR from years 2013 and 2014. Samples were submitted through the following techniques: PCR analysis with Cobas HPV Test and Aptima HPV Assay, to identify High Risk HPV genotypes; and through flow cytometry (FC) with HPV OncoProbe, to detect the presence of virus activity. Results: Seven samples were positive for High Risk HPV genotypes types 16, 18 & 45 on PCR analysis and FC, representing the true positives. One hundred twenty six samples were negative on both PCR & FC analysis, representing the true negatives. Fifteen samples were positive on PCR analysis and negative on FC analysis, representing the false positives. Seventy one samples were negative on PCR analysis and positive on FC analysis, representing the false negatives, which were then labeled as unknown genotypes for high risk HPV, due to the fact that samples were negative to HPV genotypes 16, 18 and/or 45. Conclusions: Both technologies together (i.e., PCR and FC) showed that there is a low prevalence for anal high-risk HPV genotypes 16, 18 and/or 45. Nonetheless, the presence of other anal high-risk HPV genotypes have been identified and labeled as unknown, until further research can be completed. Both PCR and FC should be used together to screen for anal high-risk HPV, but not limited to HPV genotypes 16, 18 and/or 45, as it follows: PCR analysis for virus detection; and FC analysis as a viable adjuvant test to identify virus activity. Cytologic analysis is suggested as another means to confirm the presence or absence of viral infection, and hence, give the proper diagnosis.
C. Parikh 1 , R. Qi 1 , P. Brzoska 1 , N. Mulakken 1 , B. Buchan 2 , N. Ledeboer 2 , C. Lee 1 1 Thermo Fisher Scientific, South San Francisco, CA; 2 Medical College of Wisconsin, Milwaukee, WI. Introduction: Conventional microbiology methods for identification and characterization of microbial pathogens in human research samples are limited by time consuming bacterial culturing, required sterility of source tissue, and the need for significant amounts of target microbial organisms. Here, we describe a fast and sensitive targeted approach for pathogen identification using semiconductor-based sequencing. Methods: A single tube, multiplex PCR assay consisting of 196 primer pairs was developed, targeting known signature genomic regions in common bacterial species like E.coli, S.aureus, S.epidermidis and S.pneumonia. Libraries were constructed from amplicons generated using this assay, and sequenced on a semi-conductor based Next Generation Sequencing platform. Results: Testing the custom assay on relevant clinical research samples from external collaborators has demonstrated up to ~99% sensitivity, allowing for species identification, strain discrimination and detection of antibiotic resistance genes. Conclusions: We demonstrate a robust and highly sensitive approach for identifying microbial species in research samples. Importantly, the entire workflow takes just 1.5 days and holds great promise for the future in rapid pathogen identification in diverse applications as well as metagenomics. The global spread of fluoroquinolone (FQ)-resistant E. coli sequence type (ST) 131 over the past decade has been reported using multilocus sequence typing (MLST). This study investigated the sequence type (ST) distribution of ciprofloxacin (CIP) non-susceptible E. coli blood culture isolates collected from 2013 to 2014 in a tertiary care hospital, and the epidemiological characteristics were compared with our previous study (2015, 25th ECCMID, poster session V, P1024). Methods: Ciprofloxacin non-susceptible E. coli non-duplicate isolated from blood culture were collected from January 2013 to September 2014. The susceptibility test of CIP and extended spectrum beta-lactamase (ESBL) production was performed using Microscan Walkway (Siemens Healthcare Diagnostics, Deerfield, IL) and the nonincluding adk, fumC, gyrB, icd, mdh, purA and recA for MLST were amplified and sequenced. Medical records of patients were reviewed and compared with previus study. The proportion of each ST was compared with previous study conducted at same hospital in 2005 to 2010. Results: A total 160 E.coli blood culture isolates (including 33 ESBL producers) were collected, and the 59 (36.9%) isolates of them were non-susceptible to CIP. The clinical characteristics including gender, community or hospital acquired, history of operations, procedures, antibiotic use within 60 days and cancer history of patients were not different from current study and previous study. Among the 59 CIP non-susceptible E.coli, the most common ST type (n, %) was ST131 (22, 37.3%) and the proportion was not significantly different from previous study (18/80, 22.5%) (p=0.086). The proportion of ESBL producers in ST131 in current study (12/22, 59.1%) was not significantly different from previous study (9/18, 50%). The ST393 showed significantly decreased proportion (2/59, 3.5%) in current study compared to previous study (34/80, 42.5%)(p<0.001). Other discovered STs (n) were ST10 (1) (14), ST2505 (1) and ST4952 (1).
, ST23 (1), ST38 (1), ST69 (2), ST70 (1), ST73 (1), ST95 (1), ST405 (2), ST448 (1), ST457 (1), ST602 (1), ST648 (2), ST744 (1), ST1158 (1), ST1177 (2), ST1193
The CIP non-susceptible E.coli ST131 was the most prevalent in current study whereas the ST393 was decreased in our hospital. The clinical characteristics of patients in 2013 to 2014 with ST131 were not different from previous study of 2005 to 2010. Introduction: Klebsiella pneumoniae is one of the most important pathogens causing hospital and community-acquired infections. Klebsiella pneumoniae sequence type 11 (ST11) has been detected worldwide as the main, international high-risk clones with clonal complex 258/11 (CC258/11) and the sequence type 258 (ST258). This study investigated the ST distribution of ciprofloxacin (CIP)-resistant K. pneumoniae isolates and compared them with the distribution of quinolone resistance mechanisms and ESBL production. Methods: Twenty CIP-resistant K. pneumoniae isolates were collected from blood cultures from 2008 to 2010. Seven housekeeping genes for MLST were amplified and sequenced. Results: ST11 was the most prevalent (50.0%) followed by ST15 and ST23 (15.0% respectively). exclusively found in ST11 (80.0%) and ST15 (66.7%). All ST11 isolates had Ser83Ile in the quinolone resistance-determining regions (QRDRs) of gyrA and Ser80Ile in parC, and one of them (10.0%) had additional gyrB mutation (Ser359Ala and Ser367Thr). All ST15 isolates had Ser83Phe and Asp87Ala in gyrA and Ser80Ile in parC, showing more number of the QRDR mutation than that of ST11. Simultaneous possession of plasmid-mediated quinolone resistance (PMQR) genes, -Ib-cr, qnr, and oqxAB, was exclusively observed in 70.0% of ST11. The most prevalent PMQR gene in ST11 was oqxAB (100.0%), which was not observed in other STs. Conclusions: ST11 was the most prevalent ST in blood culture isolates from 2008 to 2010. ST11 was higher in the distribution and the number of PMQR gene than ST15, however, lower in the number of QRDR mutations than ST15. The Journal of Molecular Diagnostics ■ jmd.amjpathol.org encephalitis in the United States and presents as fever, behavioral changes, and altered consciousness, resulting from localized temporal lobe involvement. Without treatment, mortality exceeds 70%, and few survivors recover normal neurologic function. Early treatment can reduce morbidity and mortality; however, residual neurologic impairment is common. HSV PCR of CSF is recommended in guidelines as the test of choice, being more sensitive than culture. The objective of this study is to characterize the performance of the cobas HSV 1 and 2 Test with CSF specimens collected from patients in Belfast, Ireland, compared to a published real-time TaqMan assay for these targets [van Doornum et al, 2003 . JCM 41(2):576-Methods: Retrospective residual CSF specimens previously evaluated by routine procedures with a validated lab developed PCR assay for the detection of HSV 1 and HSV 2, were obtained from the Infectious Diseases BioArchive (Regional Virus Laboratory, Belfast) and were tested with the cobas HSV 1 and 2 Test. Residual CSF, 200ul or less depending on volume availability for each specimen, was introduced into 1.3 mL of MSwab media (Copan, Brescia, Italy) and tested directly on the cobas 4800 system. A series of 111 CSF specimens previously reported as negative were also evaluated. These specimens were included as being representative of the small volume CSF specimens that are often received for testing from neonates. Testing of prospectively collected CSF specimens is underway for this study. Results: The cobas HSV 1 and 2 Test detected 17 of 17 (100%) previously positive HSV CSF specimens diluted in MSwab media, and none of the negative CSF specimens (0/111) generating a positive and negative percent agreement of 100%. There were 15 HSV 1 and 2 HSV 2 CSF specimens detected with the cobas HSV 1 and 2 Test. Conclusion: The cobas HSV-1/2 test, run on the fully automated cobas 4800 system, exhibited excellent performance characteristics and was found to be suitable for detecting HSV-1 and HSV-2 from CSF specimens.
L. Glaser, K. Alby University of Pennsylvania and the Hospital of the University of Pennsylvania, Philadelphia, PA. Introduction: Detection of herpes simplex virus (HSV) 1 and 2 and varicella zoster virus (VZV) from lesions helps guide patient treatment and infection control practices. Molecular assays for viral nucleic acid detection provide a rapid and sensitive approach to testing. We compare the Lyra Direct 1+2/VZV (Quidel, San Diego, CA) assay with a laboratory-developed assay on the BD MAX open platform (BD, Franklin Lakes, NJ). Methods: Samples (n=160) from skin and mucosal sites collected for routine clinical testing in viral transport media between March and June 2015 were prospectively tested on both systems. The Lyra assay was performed per manufacturer's instructions using an ABI 7500 Real-Time (RT) PCR system. The BD MAX assay was developed and validated in our laboratory and consists of a duplexed HSV1 and 2 RT-PCR assay with a separate VZV assay. Both assays contain a manufacturer provided process control. Samples with discordant results were tested at second laboratory by a different laboratory developed molecular assay. Statistics were performed using GraphPad software. Results: The overall agreement between the 2 assays is 94% (Kappa 0.89, 95% CI 0.81 to 0.96). The positive agreement for HSV-1, HSV-2 and VZV is 78%, 100% and 89%, respectively. Based on a third arbiter assay the Lyra assay had 1 false positive and 4 false negatives, whereas the BD MAX assay had 4 false positives. In all discordant cases the cycle threshold (Ct) value of the positive test was above 30. Clinicians ordered both HSV and VZV approximately 5% of the time at our center. The multiplexed HSV and VZV Lyra assay detected an additional 2 cases of VZV in patients who were only tested for HSV. Conclusions: Overall these 2 assays have excellent agreement. The BD MAX assay offers increased sensitivity for detection of HSV-1 in our population. The Lyra assay offers the advantage of testing HSV and VZV in a single reaction.
University of Texas Health Science Center at San Antonio, San Antonio, TX; 2 Hospital Universitario La Paz, Madrid, Spain. Introduction: Azole antifungals (itraconazole, voriconazole, posaconazole, and isavuconazole) are effective treatments for invasive aspergillosis. However, drug resistance in A. fumigatus has been reported. The main mechanism of azole resistance in A. fumigatus is due to cyp51A missense mutations. Mutations in different codons of cyp51A are related to different azole susceptibilities or resistance patterns. To our knowledge, the status of A. fumigatus cyp51A mutations in the US has not been fully characterized. We determined the cyp51A mutation status in azole resistant A. fumigatus isolates collected in the US. Methods: Thirty-seven A. fumigatus clinical isolates confirmed to be azole resistant (voriconazole & clinical samples were sent to our mycology reference laboratory from institutions across the US. DNA was extracted using a combination of cetyl trimethyl-ammonium bromide (CTAB) buffer and physical disruption by bead beating, followed by manual chloroform/ ethanol extraction. Primers targeting the A. fumigatus cyp51A promotor and coding region were used for PCR. PCR products were then bi-directionally sequenced using internal primers targeting reported mutation codons (N22, S52, G54, TR34/L98, G138, Q141, H147, P216, F219, M220, S297, P394, Y431, G434, T440, G448, Y491, F495, and TR46/Y121F/T289A). The G54, TR34/L98, and M220 mutations have been reported to be related to azole pan-resistance. Sequences were assembled and analyzed using DNASTAR software (DNASTAR, Inc., Madison, WI). Mutations were identified by comparing obtained isolate sequence to the cyp51A wild-type sequence. Results: Cyp51A mutations were found in 19 of 37 (51%) clinical isolates, which include 6 isolates with G54 (G54E, G54R, G54W), 1 isolate with G138C, 1 isolate with F219S, 4 isolates with M220 (M220I, M220K, and M220V), 4 isolates with G448S, 1 isolate with TR34/L98H, and 2 isolates with TR46/Y121F/T289A mutations. The 2 isolates with TR46/Y121F/T289A mutations were phenotypically highly resistant to voriconazole and isavuconazole. Neither TR34/L98H nor TR46/Y121F/T289A mutations have previously been reported in A. fumigatus isolates from the US. Cyp51A mutations were not detected in the remaining 18 of 37 (49%) isolates. Conclusions: Cyp51A missense mutations and tandem repeat insertions associated with azole resistance were detected in this US collection. Other mechanisms of resistance might also be responsible for azole resistance in A. fumigatus. Introduction: Rapid and accurate identification (ID) of clinically significant fungi is essential. DNA sequencing of specific gene targets has become an important tool for fungal molecular ID. Although the internal transcribed spacer (ITS) and domains 1 and 2 of the large ribosomal subunit (D1/D2) are recommended for most fungi by the CLSI guideline MM18-A, these targets alone may not be sufficient to discriminate between certain species.Methods: In addition to the ITS and D1/D2 primers pairs, 5 more sets of primers targeting different gene regions were clinically validated: translation elongation factor (EF-nd largest subunit -tubulin (TUB), calmodulin (CAL), and glyceraldehyde-3-phosphate dehydrogenase (GPD). Eleven other sets of primers targeting the actin gene (3 sets), RNA polymerase II largest subunit (RPB1, 1 set), alternate regions of RPB2 (1 set), EFclinical fungal isolates, DNA was extracted by physical disruption (bead beating), followed by either manual chloroform/ethanol or automatic (EZ1 DNA Tissue Kit, Qiagen) extraction. Sanger DNA sequencing was performed using primers specific to the gene regions stated above. Targeted regions were amplified, and PCR products were bi-directionally sequenced. Sequence data were aligned using Lasergene software (DNASTAR, Inc). Sequences were queried in the CBS-KNAW database, and final fungal IDs were attained by combining the molecular sequencing data with morphologic and physiologic characteristics. Results: Five primer sets were validated against 132 fungal isolates as follows: TUB/45 spp. (including Aspergillus, Penicillium, Paecilomyces, Phaeoacremonium); EF-1 and RPB2/39 spp. (Fusarium); CAL/23 spp. (including Scedosporium, Aspergillus); GPD and RPB2/25 spp. (including Curvularia, Alternaria). An additional 11 primer sets are currently under validation and are being used on a research basis with various fungal spp. (75 isolates). From August 2012 to May 2015, over 2400 clinical isolates have been identified using combined molecular data and phenotypic features. A lack of ID or genus-only ID in approximately 5% of the isolates suggests more informative target sites may be needed and/or a lack of credible deposits in public databases for comparative sequence-based ID. Conclusions: Molecular ID using multiple targets in combination with phenotypic data increased the accuracy of fungal ID, and allowed for the accurate and fast ID of over 2400 clinical isolates. Introduction: Clostridium difficile is a significant cause of pseudomembranous colitis, and is the most commonly recognized cause of infectious diarrhea in healthcare settings. The lack of point of care diagnostics allows disease spread through healthcare settings suggesting the need for a rapid, accurate and CLIAwaivable C. difficile test for patient triage at point of care. Here we report on the development of a CLIA-waivable Loop-mediated isothermal AMPlification (LAMP) based diagnostic assay and describe its analytical performance, tolerance to potential interferents and preliminary clinical data. Methods: LAMP primers were designed against the C. difficile pathogenicity locus. Reactions contained a thermostable DNA polymerase with strand displacement activity (Lucigen), primers, and fluorescent dye for amplification detection. C. difficile and other bacteria were spiked into negative stool matrix. Sample prep involved dilution in lysis buffer, filtration to remove particulates, and heat treatment to lyse cells. Assays were carried out at 68ºC. Reactions that crossed a threshold within 25 minutes were considered positive. Results: Serial dilutions of 3 toxigenic C. difficile strains in negative stool matrix demonstrated analytical sensitivity from 6E+03 to 1.4E+04 CFU/ml. To assess assay inclusivity, a total of 23 different C. difficile clinical isolates that varied by year, location and toxinotype were tested. All the isolates were correctly identified as C. difficile. Test specificity was measured by assaying 44 non-C. difficile organisms that jmd.amjpathol.org ■ The Journal of Molecular Diagnostics may be found in diarrheal stool specimens. The set included 16 Gram positive organisms, 5 non-difficile Clostridium species, 26 Gram negative organisms, and 1 yeast strain. No cross reactivity was detected with any of the strains tested. The ability of 15 potential interferents to cause false negative or false positive test results was investigated using whole blood, fecal fat, antacids, laxatives, anti-diarrheal medications, spermicide, and antibiotics. None of the agents exhibited positive or negative interference to the test. Test performance using clinical specimens was evaluated using 54 C. difficile positive and 49 C. difficile negative, deidentified residual stool samples. The assay had a 94% PPA and 100% NPA to the comparator method, AmpliVue test. Conclusions: The LAMP based diagnostic test is a rapid, accurate and unaffected by common inhibitors. In preliminary testing, using residual clinical specimens, the assay performed with high concordance to a commercially available molecular diagnostic test. The test has the potential for CLIAwaiver to serve as an aid to timely patient management and infection control
Comparison to the Esensor Respiratory Viral Panel S.T. Jacob, A. Seth University Hospital, Newark,NJ. Introduction: Rapid and accurate identification of various viral and bacterial infectious agents responsible for respiratory tract infections is important for effective patient management. The Luminex NxTAG Respiratory Pathogen Panel (RPP) is an extended multiplex panel that identifies 19 viruses and 3 atypical bacteria associated with respiratory tract infections. Objectives: To evaluate the RPP panel for its analytical performance characteristics and utility for patients at the University Hospital. The assay was compared with the eSensor RVP assay and its performance characteristics were evaluated. Methods: A total of 114 archived respiratory specimens including nasopharyngeal swabs, nasal washes and broncho alveolar lavages were analyzed using the RPP assay. Specimens were previously reported by the eSensor RVP assay as positive or negative or determined positive using either real time PCR assays or the RPP assay performed in external laboratories. Extraction of nucleic acids was performed using the bioMerieux easyMag automated extractor. The NxTAG RPP assay was performed following manufacturer's protocols. Results: Previously determined positive specimens (n=114) re tested by the RPP assay included respiratory syncytial virus (RSV) (n=19), rhinovirus/enterovirus (n=19), influenza (n=16), metapneumovirus (hMPV)(n=5), parainfluenza virus (n=18), adenovirus (n=23), coronavirus (n=10), bocavirus (n=10), Chlamydophila pneumonia (n=10), Mycoplasma pneumonia (n=10), Legionella pneumophila (n=10). These pathogens were present individually or as coinfections previously determined by the Genmark RVP assay or alternative assays. The overall concordance between the RPP assay and other assays used for determination of positives was 98%. The RPP assay showed 100% concordance with the eSensor RVP assay for detection of influenza A and B, hMPV, RSV and rhinovirus. Concordance between the RPP and RVP assays varied between 83% to 87% for coronavirus, adenovirus and parainfluenzavirus. All discordants displayed low current values on the eSensor assay and will be sequenced for resolution. A 7% increase in detection of coinfections was observed using the RPP panel. The most commonly detected additional coinfection was Adenovirus and Bocavirus. These specimens are being sequenced for confirmation. The 3 bacterial targets tested with the RPP assay demonstrated 100% concordance with real time PCR assays and sequencing. Subtyping results for influenza A, RSV and parainfluenza were comparable in both assays. Conclusions: The assay offers an advantage with respect to detection of additional pathogens from a single specimen with a faster turnaround time and can prove beneficial for appropriate treatment of patients at University Hospital.
M. Stonebraker 1 , L.L. Malone 1 , S. Brzezinski 1 , D. Stalons 1 , P. Brzoska 2 , N. Fantin 2 , K. Varma 2 , J. Trotta 2 , E. Grigorenko 1 1 Diatherix Laboratories, Inc., Huntsville, AL; 2 Thermo Fisher Scientific, South San Francisco, CA. Introduction: The progression of antimicrobial resistance is a threat to public health worldwide. The assessment of antimicrobial susceptibility patterns of bacterial isolates is one of the primary responsibilities of the clinical microbiology laboratory. Molecular testing can offer a rapid and sensitive approach compared to bacteriological testing, and such tests can have a significant impact on patient care. In this study, we developed and validated a TaqMan Assay panel to screen for 26 multidrug-resistance genes encoding the most clinically prevalent mechanisms of resistance to 4 -lactams, fluoroquinolones, and macrolides. Methods: TaqMan Assays were designed using gene specific sequences available through the NCBI database. Assays were printed for use on a high-throughput real-time platform (QuantStudio 12K Flex). DNA was isolated from samples using a KingFisher Flex instrument and subsequently amplified using a panel-specific PreAmp primer mix. Pre-amplification products were then processed on custom OpenArray Real-Time PCR Plates. Ligation-mediated real-time PCR (Check-Points, Netherlands) and Target-Enriched Multiplex PCR (TEM-PCR) were used for confirmation of gene resistance profiles for selected samples. Results: Target specificity of TaqMan Assays was validated using clinical isolates (n=109) with known, sequence-validated gene resistance profiles and patient samples (n=517) for which phenotypic data were available. No crossreactivity was detected for following gene targets: TEM, SHV, CTX-M Groups 1, 2, 8/25, and 9, ACT/MIR, KPC, NDM, VIM, VEB, IMP, OXA-1, OXA-48, ACC, GES, FOX, ermA-C, qnrA, qnrS, DHA, CMY, and PER. Clinical isolates yielded 100% correlation when compared to resistance profiles. One isolate, characterized as blaNDM-1 positive, was not detected by this panel, but repeated sequencing indicated loss of the NDM-1 plasmid. Screening of patient samples revealed a complex pattern of up to 14 genes encoding resistance for 4 major classes of antibiotics, with TEM and ermA-C being the most prevalent (48% and 91% in respiratory and gastrointestinal samples, respectively). CTX-M Groups 1, 2, and 9; SHV; ACT/MIR; qnrA and qnrS were also detected. Results generated with TaqMan Assays were confirmed with alternative PCR-based methods. Conclusions: TaqMan chemistry in combination with high-throughput real-time PCR provides a powerful technological platform for molecular testing for a broad spectrum of multidrug resistance genes.The implementation of rapid tests for identification of antimicrobial resistance profiles can lead to targeted antibiotic therapy, reduction of unnecessary antibiotic use, and overall improvement of patient care.
Fluids of Patients with Fungal Infection D. Hooper, S. Muralidhar, V. Bolton, S. Sutton RealTime Laboratories Inc, Carrollton, TX. Introduction: Rapid and early diagnosis for invasive fungal infection by Aspergillus and Candida species is needed especially for transplant recipients and patients with hematological malignancies with high mortality and morbidity ratio. Conventional methods such as blood culture and ELISAs for detection of antigens (e.g. galactomannan and glucan) are suboptimal in sensitivity and specificity. Consequently, nucleic acid detection techniques have come into use for the differential diagnosis and follow-up of fungal infections. The goal of these studies was to determine if target specific DNA from various species fungi could be identified in human tissue and body fluids from patients exhibiting signs of a fungal infection. Methods: Assays utilize primers and proprietary specific hydrolysis probes designed in-house for various human fungal pathogens including 4 Aspergillus (18S Ribosomal RNA Genes) and 4 Candida (26S Ribosomal RNA Genes) speciesspecific targets. Assays were modeled, and tested in the laboratory for specificity, efficiency, and precision. All assays were developed and validated on qualitative PCR platforms most commonly used in clinical laboratories such as the Cepheid SmartCycler and Applied Biosystems 7500Fast. Upon validation, 18 human upper respiratory (UR) specimens, and 34 tissues from biopsies, and/or autopsy specimens were tested from patients with histories of exposure to toxic molds. 12 negative (UR) specimens and 15 autopsy/surgical specimens from patients with no known history to fungal infection were used as negative controls. Results: Nucleic acid from 10 different fungal spores was detected in 35% of UR specimens and in 42% of human tissue from exposed patients tested. Control specimens included noninfected tissue and were negative. Conventional cultures and special stains on all specimens were negative. Aspergillus niger was detected in UR, lung, liver, brain, and skin specimens. Aspergillus species (flavus, sydowii, ustus) , Penicillium simplisticum and Eurotium amstelodami were also detected in various tissues and UR specimens. The internal control and negative control were correlated as expected in all samples. Data for Candida specimens are also presented. Conclusions: Assay designs were selected and validated for each fungal DNA targets with specificity, efficiency (>90%), linearity (>.99) and precision. Clinical sample analysis results indicate and provide evidence for a more rapid and reliable assay by species-specific PCR for the diagnosis of various fungal pathogens when compared to current strategies such as blood cultures and galactomannan ELISA. These tests could be applied to monitor the effectiveness of therapy after diagnosis and potentially assist in predicting outcomes.
Cleveland Clinic Foundation, Cleveland, OH. Introduction: Herpes simplex virus (HSV) causes chronic cutaneous and mucocutaneous infections and viral meningitis. Herpes Simplex Type 1 is associated with infections of the oropharynx and eyes although, there is an increasing prevalence in genital infection. Herpes Simplex Type 2 is associated with genital and neonatal infections. We compared the illumigene HSV 1&2 DNA amplification assay (Meridian Bioscience, Inc., Cincinnati, OH) to HSV cell culture in our clinical lab for sensitivity, specificity, ease of use and turn-around time. Methods: Genital and oral swabs in viral transport media submitted to the lab for routine HSV cultures (N=248) on cutaneous and mucosal specimens were used as the comparator. Positive HSV cultures were typed using the ELVIS HSV ID and D3 Typing Test System (Enzyme-Linked Virus Inducible System HSV ID and D3 Typing Test System, Quidel Corp., San Diego, CA). A 50μl aliquot of the viral transport media was used to set up the illumigene assay per protocol. The illumigene assay is a qualitative molecular test for the direct detection and differentiation of HSV 1 + 2 performed on the illumipro-10instrument. The illumigene assay utilizes loop-mediated isothermal DNA amplification (LAMP) technology to detect HSV-1 + HSV-2. Discrepant analysis was performed according to protocol by another FDA cleared molecular method (AmpliVue HSV 1+2 Assay, Quidel Corp, San Diego, CA). Results: Of 248
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org specimens tested, 145 were negative by both ELVIS and the illumigene, 91 were positive by both methods and 12 were positive by the illumigene and negative by ELVIS. The sensitivity and specificity of the illumigene compared to ELVIS is 100% and 92.4% respectively. The 12 discrepant samples were adjudicated using the AmpliVue test with the following results: 7 ELVIS neg/illumigene pos /Amplivue pos, 4 ELVIS neg/illumigene pos /Amplivue neg, 1 sample unable to test by Amplivue. Of 91 pos ELVIS cultures, 51 were HSV-1, 39 were HSV-2 and 1 did not type. Of 103 positive illumigene samples, 57 were HSV-1, 45 were HSV-2 and 1 sample was pos for both HSV-1 and 2. Conclusions: The sensitivity and specificity of the illumigene compares favorably to ELVIS culture. The illumigene assay detects co-infection with HSV-1 and 2. The illumigene demonstrated ease of use and a faster turn-aroundtime for results (1.5 hours). ELVIS culture requires 16 hour incubation, tissue culture facility and expertise at staining and reading ELVIS cultures. The illumigene assay has a simple specimen prep, and amplification and detection occur on the illumipro-10 instrument which has a small footprint. No extractor or thermocycler is needed. This is an excellent test for labs that wish to offer a molecular testing option for identification and typing of Herpes virus. Rosa, T. Oliveira, N. Muto, R. Sitnik, C. Mangueira, J. Pinho Hospital Israelita Albert Einstein, São Paulo, Brazil. Introduction: Chikungunya virus (CHKV) is an RNA virus from the genus Alphavirus which causes a tropical disease that is transmitted to humans by infected Aedes mosquitoes, the same vector of dengue virus (DENV). Chikungunya infection was initially described in Africa, Asia and the Indian subcontinent, causing fever, rash and severe joint pain. In December 2013, an outbreak of CHIKV was notified in the Caribbean and for the first time a mosquito-human cycle in the Americas was described. On September 2014, the first autochthonous CHIKV infection was confirmed in North Brazil, indicating a potential risk for virus establishment in Brazil, where a DENV still represent a public health problem. Clinical diagnosis is difficult due to similar symptoms with DENV. Accurate diagnosis is crucial for patient management and disease control given the economic and social impact in endemic areas. Laboratory diagnosis can be accomplished by testing samples to detect viral RNA and immunoglobulin M. However, only molecular test can precisely identify CHIKV infection in acute first days of illness. To offer a high quality molecular CHKV diagnostic test in our service, we validated the Light Mix Kit Chikungunya -Virus (TIB MOLBIOL) according to CAP guidelines. Methods: We analyzed 33 blood samples obtained from the routine laboratory from a private hospital in São Paulo, Brazil, from December 2014 to May 2015. Nucleic acids were obtained with NucliSens Easy Mag equipment (bioMérieux, Geneva, Switzerland) and the amplification was performed using the Light Mix Kit Chikungunya -Virus (TIB MOLBIOL) with the (COBAS Z 480) instrument, according to manufacturer's instructions. Data were analyzed using manufacture's provided software. Results: Frequency of the CHIKV in our hospital was 6%. Analytical sensitivity of the method is 800 copies/mL calculated using a standard curve constructed with commercial available RNA control. These 2 cases were probably infected in other countries (Porto Rico and Colombia) from where the patients have recently passed. Conclusions: The assay was successfully used for CHKV detection in patients with suspected cases that might be clinically confused with dengue and Zika virus infections that are also present in our country. Molecular methods are rapid and sensitive tools for the diagnosis of this viral infection in the early viremic stages of illness. This technique should be used to provide a better management of patients, as well as for the detection and control of outbreaks. Since CHIKV and DENV virus produce similar clinical signs and DENV is endemic in Brazil, it is important for physicians and epidemiologists to differentiate them rapidly.
R.F. Santana, R.C. Petroni, G.F. Dastoli, V. Fusco, J. Souza, V.M. Oliveira, S.
Assay for the Identification of Clostridium difficile North American Pulse-Field Type 1 (NAP-1) T. McsMillen, M. Kamboj, N. Babady Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: Clostridium difficile (C. difficile) strain 027/NAP1/B1 is the most common strain type in the United States and is associated with increased morbidity and mortality. Molecular typing methods including multi-locus sequence typing (MLST) or PCR ribotyping are traditionally used for strain typing of C. difficile. Identification of C. difficile NAP-1 type may also be accomplished using the Xpert C. difficile/Epi (Cepheid Inc., Sunnyvale, CA) assay, a rapid, real-time PCR with minimal hands on time and manipulation. In this study we compared the Xpert C.difficile /Epi assay with MLST for identification of the C. difficile NAP-1. Methods: From September 2013 to May 2014, all stool specimens positive for C. difficile by the Xpert C.difficile/Epi assay were saved and stored at -80°C prior to typing by MLST. Discordant C. difficile NAP-1 results were further evaluated using PCR-ribotyping, tcdc gene sequencing and Matrix Assisted Light Desorption-Time of Flight mass spectrometry (MALDI-TOF MS). Percent agreement between MLST and Xpert C.difficile/Epi were calculated. Results: Five hundred thirty stool samples were positive for C. difficile by Xpert PCR with 64 samples identified as presumptive C. difficile NAP-1. A total of 401 C.difficile isolates were recovered and typed by MLST including 52 C. difficile NAP-1 identified by Xpert PCR. 42/401 (10%) were identified as MLST type 1 (NAP-1). The total percent agreement for identification of C. difficile NAP-1 between MLST and Xpert PCR was 97.51% (k=0.8; 95% CI 0.81-0.95). Sequencing of the tcdC gene and PCR ribotyping confirmed 9 of 10 and 6 of 10 strains as C. difficile NAP-1 respectively. MALDI-TOF MS confirmed 9 of 10 discordant strains as C. difficile with similarity between isolates ranging from 50% to 80%. Additionally, 10 strains with concordant types between Xpert and MLST were also tested by the same methods. All 10 strains were identified as C. difficile NAP-1 by sequencing and PCR ribotyping. Nine of the 10 concordant samples were identified as C.difficile by MALDI-TOF MS with similarity between strain profiles >85%. Overall the sensitivity of the Xpert PCR assay for C. difficile 027/NAP1/B1 identification was 100%, with a specificity of 98%. Conclusions: The limited amount of hands on time and rapid turnaround time make the Xpert C.difficile/Epi PCR assay an advantageous method for the rapid identification C.difficile 027/NAP1/B1 when compared to MLST.
A. Gray, T. Moses, K.M. Bennett Texas Tech University Health Sciences Center, Lubbock, TX. Introduction: Chlamydia trachomatis is the most commonly reported sexually transmitted infection in the United States. There are many differences between patients in the persistence and progression of infection that are believed to be strongly influenced by the genetics of the host. Previous studies in our lab have shown a correlation between host genotype and C. trachomatis infection status. The objective of this study was to quantify percent CpG methylation in targeted promoter sequences in the abl-interacter 1 (ABI1) and the protein disulfide isomerase 2 (PDIA2) human genes, which have been shown to play a role in the intracellular life cycle of C. trachomatis. Methods: CpG methylation analysis was performed on a total of 22 subjects, half of which were positive for C. trachomatis infection, as determined by the Roche Amplicor CT/NG assay. DNA was isolated from deidentified urethral or endocervical swab specimens, using the Qiagen QIAmp DNA isolation kit. Genomic DNA was bisulfite treated using the EpiTect Fast DNA Bisulfite Kit (Qiagen), followed by PCR amplification using the Qiagen PyroMark PCR kit. DNA methylation was quantified at specific CpG sites using the Qiagen PyroMark Q24 pyrosequencing instrument. Percent methylation was compared between C. trachomatis positive and negative samples for each CpG site using a z-test for 2 proportions. Results: For the ABI1 gene, the target site for analysis was the binding site for transcription factor E2F2, and the PDIA2 target site was a GC-rich sequence within the promoter region. The examined CpG sites in ABI1 were not highly methylated (Mean Site 1: 4.1%, Site 2: 3.2%), Although in contrast, the sites in PDIA2 were highly methylated (Mean Site 1: 96.2%, Site 2: 94.7%, Site 3: 92.1%). There was no significant difference in percent methylation between C. trachomatis positive and negative specimens for ABI1 (Site 1: p=0.65, Site 2: p=0.31) or PDIA2 (Site 1: p=0.68, Site 2: p=0.34). Conclusions: Although no significant difference was found in percent methylation for the examined CpG sites between C. trachomatis positive and negative specimens, there are other CpG islands throughout both genes that are candidates for additional studies. Very little research has been performed previously on the correlation of epigenetic markers with C. trachomatis infection status, so further studies will continue to reveal novel data that can further our understanding of the host-pathogen interactions.
A. Seth, C. Patel, A. Baisre University Hospital, Newark, NJ. Introduction: The Center for Liver Diseases and Transplantation at University Hospital requires reliable and accurate CMV viral load determination for the management of liver transplant patients. Quantitative viral load assays can show variations in results obtained due to differences in pre analytic procedures employed by laboratories. Since extraction procedures introduce variability, we evaluated the performance of the Qiagen artus CMV RGQ MDx assay using traceable standards employing extraction methods identical to plasma specimens. Objectives: To evaluate the analytical performance of the artus CMV RGQ MDx assay using extracted traceable standards (Acrometrix CMVtc panel). To compare assay performance with DNA standards supplied with the CMV RGQ kit in order to analyze variability introduced due to specimen extraction procedures. Methods: CMV viral loads were determined using the CMV RGQ assay on the Rotor-Gene Q according to manufacturer's recommendations.The AcroMetrix CMVtc panel standards were used to verify performance specifications of the RGQ assay. Clinical specimens and standards were extracted on the bioMerieux easyMag automated extractor. Results: Limit of quantification assays using the extracted acrometrix standards detected all CMVtc panel members with expected values of 100 IU/ml and 75 IU/ml. Using this standardized panel, the assay (r2 =0.982) was linear in the range of 2.0 to 6.4 log10 IU/ml. The CMVtc panel members showed colinearity with the supplied standards that are not extracted (QS1-4) across the linear range of 2.0 to 5.4 log10 IU/ml (mean variability 0.13 log10 IU/ml). A higher variability in quantification was observed at viral loads of 6.4 log10 IU/ml between the extracted CMVtc and nonextracted (QS1-4) panels with the latter showing an increase in quantified viral load values by 0.3 log10 IU/ml. Clinical specimens (n=58) previously tested using the artus CMV LC jmd.amjpathol.org ■ The Journal of Molecular Diagnostics PCR kit were quantified using the CMV RGQ MDx kit. An overall concordance of 86% was observed between both assays and all discrepant results corresponded to low viral loads (less than 2.6 log10 IU/ml). The CMV RGQ kit assay was more sensitive and reproducible in detection of low viral loads. Conclusions: The artus CMV RGQ MDx assay offers a more sensitive and reproducible method of CMV viral load quantification in comparison to our current laboratory method. A standardized panel (CMVtc panel) was useful in assessing the variability introduced across the entire dynamic range of the assay due to pre analytical processing procedures in our laboratory.This assay will be beneficial for CMV viral load monitoring in transplant patients at University Hospital. Introduction: There are recent changes in interim guidance for the use of HPV testing in primary screening, which include genotyping. The recommended alternative algorithm allows for efficient triage of women to colposcopy. The FDA approved Roche COBAS 4800 is one of the molecular-based platforms routinely used for HPV testing and is approved for use in ThinPrep (TP). At present there is limited data comparing results from TP versus SurePath (SP) samples on the COBAS 4800 in high risk women and an ethnically diverse screening population. This study is part of ongoing New York State post-validation molecular testing for the use of COBAS 4800 in SP. Methods: Women aged 18 to 65 who are attending colposcopy or well-women clinics are actively being prospectively recruited at Montefiore Medical Center, Bronx, NY. Two specimens (TP and SP) are collected from each subject. Cytology is performed on the SP specimen, after which the residual SP cell concentrate and the TP samples are tested for high risk HPV (HR-HPV) on the COBAS 4800. SP and TP samples are tested on the same day to ensure limited bias. The SP and TP performance and correlation with cytology results were compared using standard 2 by 2 and univariate analyses. Results: In the first 79 cases tested in this ongoing prospective study, results were concordant for overall HR-HPV status in 74 of 79 cases (94%). Four HR-HPV discordant cases were HR-HPV positive in SP and HR-HPV negative in TP samples. The cytology diagnoses in these 4 cases were: 1) Negative for SIL with reactive changes, 2) ASCUS, 3) LSIL and 4) SIL rule out HSIL, diagnoses associated with a higher probability of HR-HPV positive results. The fifth discrepant sample failed on SP and was negative on TP. Among the 42 concordant HR-HPV positive cases, HPV genotyping was concordant in all but one, which was positive for HPV 16 on TP and positive for both HPV 16 and other HR-HPV types on SP. A trend of lower cycle threshold (CT) values was noted in SP samples. As expected, there was a high correlation of HR-HPV results with the cytology diagnosis: cytology Negative, 15% HR-HPV+; Negative for SIL with reactive changes, 28% HR-HPV+; ASCUS, 72% HR-HPV+; LSIL, 86% HR-HPV+; >LSIL, 100% HR-HPV+. Conclusions: In the planned interim analysis of this study, there was increased detection of HR-HPV and a trend toward lower CT values with COBAS testing of SP sample concentrates, suggesting higher sensitivity compared to TP samples. Although not FDA approved, testing of SP sample concentrates may be a better source for HR-HPV testing in our diverse urban patient population.
Genotype Assay Using ThinPrep Liquid Cytology Media S.S. Beqaj, D. Simmons Pathology, Inc., Torrance, CA. Introduction: Human papillomavirus (HPV) is one of the most commonly sexually transmitted infectious pathogens causing cervical cancer in women. Testing for HPV infection is a standard of care as a reflex for ASCUS cytology results in women 21 and older and as a primary adjunctive screen with a Pap test in women 30 to 65 years of age. The purpose of this study was to evaluate the performance of the APTIMA HPV and HPV 16 and 18/45 Genotype (HPV-GT) Assays in comparison to the HPV Assay on the Roche cobas 4800 system. The cobas HPV Test detects simultaneously a pool of 12 high-risk genotypes and individually HPV 16 and HPV18. The Aptima HPV assay non-specifically detects the presence of 14 HPV high risk genotypes, whereas Aptima HPV-GT identifies the presence of HPV16 and/or HPV18/45 genotypes. Methods: A total of 201 Thinprep patient specimens were tested with the HPV assay and 139 with the HPV-GT assay. All specimens were tested on the Gen-Probe Tigris using 1 mL of PreservCyt media aliquotted into an Aptima tube. These results were compared to the Roche cobas 4800. All discrepant results were sent for genotyping by Roche Linear Array (LA). Precision, reproducibility and repeatability were performed using validation panels containing 10 positive specimens for 16 and 18/45 genotypes and 5 specimens negative for HPV. All specimens were tested over 3 days/runs and results compared. Results:
Of the 201 ThinPrep specimens tested, 120 of 141 were positive and 58 of 60 were negative by Aptima giving a sensitivity of 85.1% and specificity of 96.7% when compared to the cobas HPV assay. Of the 23 discrepant results analyzed by LA, 3 matched Aptima, 18 matched Roche, and 2 were invalid. After testing discrepant results by LA, the sensitivity of the Aptima HPV assay increased to 87.23% with a specificity of 98.33% and an overall agreement of 90.55% when compared to Roche cobas. For Aptima HPV-GT there was a sensitivity of 81.58% and specificity of 98.02% with an overall agreement of 93.53% when compared to Roche cobas. The lower sensitivity of the Aptima HPV assay was not surprising since the assay uses transcription mediated amplification technology based on messenger RNA or viral expression, which would indicate active infection as opposed to direct detection of the virus by DNA amplification as performed on the Roche cobas. Interestingly, there were 4 specimens run on Aptima that were positive for 16 or 18/45 but negative on the HPV assay.An internal validation on BD SurePath media was attempted but not successful. We believe this to be due to the type of extraction technology employed by the Aptima system which does not allow adequate isolation of nucleic acids bound within viral capsids. Conclusions: In summary, the Hologic Aptima HPV assay and HPV 16 and 18/45 genotyping performed on the Tigris platform should be carefully evaluated when considered for HPV screening.
D. Lucic 1 , C. Dunn 1 , C. Herman 1 , L. Mazur 2 , G. Cloherty 1 1 Abbott Molecular Inc., Des Plaines, IL; 2 ACL, Rosemont, IL. Introduction: In resource limited settings, the use of dried blood spots (DBS) for HIV-1 has become an important tool for treatment monitoring. The objective of this study was to compare the performance of the HemaSpot and Whatman 903 devices using the Abbott RealTime HBV assay. Methods: Normal whole blood was obtained from ProMedDx (Norton, Ma). HBV positive plasma was used to generate whole blood panels targeting 2-6 log IU/mL. Seventy five microliters of whole blood was spotted on Whatman 903 and HemaSpot devices and air dried overnight at room temperature. Panels were placed at 8C overnight for long term storage. One full spot from each device was incubated in 1.3mL of m2000 proprietary elution buffer for 45 minutes at room temperature. Each panel member was tested in replicates of 5. Samples were gently vortexed and transferred to m2000sp for nucleic acid extraction. DNA whole blood nucleic acid extraction was performed using a 1ml extraction protocol and the performance of the RT assay was assessed. Results: Both devices had 100% detectability at target concentrations of 3-6 log IU/mL. HemaSpot device demonstrated 100% detectability at 2 log IU/mL target concentration whereas Whatman 903 device demonstrated 80% detectability at the same target concentration. Linearity for all conditions tested exceeded R² > 0.95. In comparison to spiked whole blood, both devices demonstrated a lower overall negative bias of 0.24 log IU/mL and 0.38 log IU/mL for HemaSpot and Whatman 903 respectively. Standard deviation for the two devices did not exceed 0.35 log IU/mL. Both devices were left in the sample tube during the extraction and no extraction errors were generated during the process. Conclusions: Both devices were suitable for use with the automated m2000 platform. Additional testing is ongoing to establish limit of detection and inter run reproducibility.
A. Rector 1 , C.R. Webb 1 , L. Pritchard 1 , J. Dunn 2 1 Texas Children's Hospital, Houston, TX; 2 Baylor College of Medicine, Houston, TX. Introduction: The Roche MagNA Pure 96 System (MP96) is an automated extraction platform utilizing magnetic particles to extract and purify nucleic acids from a variety of specimen types. The MP96 simultaneously extracts 96 samples in approximately 1 hour, providing quality nucleic acid for various downstream, high through-put applications. In this study, the performance characteristics of the MP96 and the QIAGEN QiaSymphony (QS) were compared for the recovery and purification of viral nucleic acid from a variety of respiratory samples. Methods: A total of 205 respiratory specimens, including nasal/nasopharyngeal swabs (NS, n=61), nasal washes (NW, n=89), bronchoalveolar lavages (BAL, n=27), tracheal aspirates (TA, n=25), bronchial washes (BW, n=2), and sputums (SPT, n=1) were extracted on the MP96. Real-time PCR amplification and detection of respiratory viruses was performed using the Prodesse (Hologic) ProFlu+, ProPara+, ProhMPV+, and ProAdeno+ Detection Kits on the ABI7500, a lab-developed influenza A subtyping assay on the Lightcycler 480 and a lab-developed rhinovirus assay on the Lightcycler 2.0. Results obtained using the MP96 were compared to those obtained from the same specimens extracted on the QS. Results: Initial studies on the MP96 resulted in an overall agreement of 92.2% (115 of 131 positive, 74 of 74 negative). The 16 discrepant samples included 10 NW, 4 NS, 1 BAL and 1 TA and detection of viruses other than influenza A and rhinovirus. In addition, direct comparison of target cycle thresholds (Cts) revealed values 2-3 cycles higher for the majority of positive samples extracted on the MP96. Fourteen of 16 discrepant samples were available for re-extraction on the MP96 with the addition of ~4 ng/ul per sample of carrier RNA. After retesting, 12 of 14 samples yielded concordant results thus improving the overall agreement to 98.1% (201 of 205). Samples extracted on the MP96 with carrier RNA demonstrated target Cts similar to those obtained using the QS. Analytical sensitivity studies utilizing serial dilutions of spiked specimens are in progress. Conclusions: The clinical sensitivity of respiratory virus detection using samples extracted on the MP96 was significantly enhanced with the addition of carrier RNA resulting in 98.1% overall agreement with the QS-extracted specimens. The ability of the MP96 to process 96 samples in less than 1 hour allows for a more efficient workflow and higher through-put compared to the QS which requires more than 6 hours.
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org ID67. Comparison of xTAG RVP Fast Assay (Luminex) to the eSensor RVP (GenMark Dx) and FilmArray RP (BioFire) C. Ratner, S.S. Saxena, R. Henry, R.C. Naeem, A.S. Fox Montefiore Medical Center, Bronx, NY. Introduction: Respiratory viruses cause infection in the lungs and respiratory tract and are an important cause of hospitalization, most significantly in immunocompromised patients. Symptoms of these viruses can range from a mild cough to a more severe illness which may lead to fatalities; therefore, an accurate and timely diagnosis is imperative. Objective: To determine which respiratory virus panel is most accurate and comprehensive in diagnosing patients. Methods: This study was conducted utilizing specimens collected from patients with upper and lower respiratory tract infections from the 2013 to 2014 influenza season. The samples were obtained from nasopharyngeal swabs that were stored at -70°C. Three hundred and twenty-eight samples were tested on eSensor and xTAG RVP. Sixty of the 328 samples were randomly selected and tested on the FilmArray. Results: Three hundred twenty-eight samples were run on the xTAG RVP and eSensor. Of these samples, 14 were excluded from analysis because of the absence of coronavirus on the xTAG RVP panel. Two hundred and seventy-five samples were detected by both the xTAG and the eSensor. Thirty-nine samples were discordant. Out of the 328 samples, 60 samples were pulled at random and were ran on the FilmArray. Out of the 60 samples, 14 samples were excluded from analysis because those 14 samples tested positive for coronavirus. Thirty-five samples were detected by both the xTAG RVP and FilmArray. Thirteen samples were discordant.
Twenty-three samples were tested for influenza B by any methods. Twelve of those samples were tested on all 3 instruments. xTAG RVP tested positive for 5 of 12, eSensor tested positive for 12 of 12 and the FilmArray tested positive for 4 of 12. Due to the discordance, the 7 samples were also tested in tissue culture. These 7 samples were the ones that the eSensor tested positive but the xTAG RVP and the FilmArray both tested negative. Samples placed into tissue culture were negative for influenza B. Conclusions: Although both the eSensor and FilmArray detected more targets than the xTAG RVP, the benefit to clinical outcomes is currently being explored. The inability for the xTAG RVP to diagnosis certain targets, specifically for coronavirus, may be critical when evaluating the platforms and further analysis is needed. Based on this preliminary study, it appeared as if the eSensor assay was more sensitive than the FilmArray, which was substantiated by the eSensor's lower limit of detection for the analytes. The eSensor includes a full extraction method whereas the extraction process of the FilmArray is shortened, which may explain the increased sensitivity of the eSensor. However, this will be validated in a larger sample size. Both the eSensor and FilmArray have a larger target panel, shorter run time, and yielded results more quickly than the xTAG RVP.
Cerebrospinal Fluid and Whole Blood Specimens P. Lauryn 1 , D. Cuevas 1 , C. Webb 1 , J. Dunn 2 1 Texas Children's Hospital, Houston, TX ; 2 Baylor College of Medicine, Houston, TX. Introduction: Herpes simplex virus (HSV) types 1 & 2 are common viruses that can cause life-threatening infections in the central nervous system and bloodstream. Due to the high morbidity and mortality rate associated with these infections, rapid diagnosis and initiation of therapy is critical. Real-time PCR testing has greatly increased the ability to rapidly detect the presence of HSV. This study compares the performance of the Simplexa HSV 1&2 Direct Kit and HSV 1&2 analyte specific reagents (ASR) (both from Focus Diagnostics) to the MultiCode-RTx HSV 1&2 Kit (Luminex) currently in use at Texas Children's Hospital. The performance characteristics of all 3 assays were compared for both CSF and whole blood specimens. Methods: A total of 24 CSF and 17 EDTA whole blood specimens originally tested using the MultiCode kit were retested by the Simplexa Direct and ASR assays. All samples for the ASR and MultiCode were extracted using the NucliSENS easyMag (bioMerieux) with off-board lysis. Whole blood specimens were tested with the Simplexa Direct kit using nucleic acid extracted on the easyMag and diluted 1:2 in PBS whereas CSF samples were run directly without prior nucleic acid extraction. Serial dilutions of quantified viral material for each target were tested by all 3 assays to compare analytical sensitivity. Results: Compared to the MultiCode, the overall clinical sensitivities for the Simplexa Direct and ASR assays were 100% (30 of 30) and 32% (8 of 25), respectively. Discrepancies observed with the ASR could not be resolved due to specimen availability. One CSF sample originally testing positive by the MultiCode tested negative on the Simplexa Direct. However, repeat testing of the original specimen on the MultiCode also gave a negative result. Analytical sensitivities for the Simplexa Direct and the MultiCode were 100cp/mL for both HSV 1&2. The analytical sensitivity of the ASR was equivalent for HSV-2 at 100cp/mL, but less sensitive for HSV-1 at 250cp/mL. Conclusions: The Simplexa HSV 1&2 Direct assay demonstrated equal or greater clinical sensitivity when compared to the ASR or MultiCode assays. It also involved the fewest hands-on steps for CSF and the shortest time to result at ~60 minutes compared to 2.5 to 3 hours for the ASR and MultiCode-RTx. The direct assay requires only 50μL of CSF, which may be an important factor in cases where CSF sample volume is limited. The Simplexa Direct assay performed comparably to the MultiCode using whole blood samples; however, the assay does require extraction prior to testing and off-label validation. Of the 3 assays, the Simplexa Direct was the easiest to perform and provided the shortest time to result, a factor that is critical for prompt initiation of effective treatment of HSV meningitis and viremia. This study was performed during performance verification (TPV) and validation of the cobas Cdiff Test in a large, multicenter trial using the cobas 4800 system. We prospectively collected and stored unformed stool from patients suspected of C. difficile infection (CDI). Archived samples were stored at -20°C or colder. Direct and enrichment toxigenic culture (DETC) was performed at a reference laboratory before shipping samples frozen (dry ice) to Roche Molecular Systems, Inc, (Pleasanton, CA) for the cobas Cdiff Test. A random subset was returned frozen for repeat DETC. A panel of 100 contrived specimens also was prepared using fresh negative stool specimens spiked with toxigenic C. difficile at varying densities (2.5, 5.0, 10, and 20 x LoD). These were tested fresh and following storage for 32 days at -20±5°C. The cobas Cdiff Test also was run on 683 samples comparing fresh and frozen sample results. Lastly, repeat DETC was performed on a subset of clinical samples (15 culture-positive and 72 culture-negative) stored at --sided Fisher's exact test and the likelihood ratio of the chi-square test. Results: 124 of 234 (53%) random TPV samples collected between July 25, 2012 and September 21, 2012 showed a positive percent agreement (PPA), negative percent agreement (NPA) and the overall percent agreement (OPA) between DETC using fresh or frozen stool specimens of 97.1% (33 of 34), 97.8% (88 of 90), and 97.6% (121 of 124), respectively ( statistic=0.0201; p value=0.8873). The PPA between the cobas Cdiff Test results using the contrived specimens tested before and after one freeze-thaw cycle was 100.0% (100/100; 95% CI = 96.4-100.0%). There was also no difference in the distribution of positive and negative cobas Cdiff Test results when comparing 90 fresh (18 positive, 72 negative) and 593 frozen (121 positive, 472 negative) specimens from the multicenter trial (Fisher's exact test p=1; chi-square test likelihood ratio p=0.9291). Lastly, there was 100% agreement between direct and repeat DETC results from 87 samples cult days at -20°C (Kappa statistic: 1.0). Conclusions: There was no significant difference in the recovery of toxigenic C. difficile by DETC or the direct detection of the toxin B (tcdB) gene in the cobas Cdiff Test when using fresh and archived stool samples in CDI test development and clinical validation. (CT) and Neisseria gonorrhoeae (NG) are common sexually transmitted infections that can cause serious effects on a woman's reproductive track. The US Centers for Disease Control and Prevention (CDC) guidelines recommend women less than 25 years of age need screening for CT and NG. The CDC guidelines also recommend that all pregnant women in the United States be screened for CT and pregnant women at risk for NG or live in an area of high NG prevalence should also be screened. There are commercially available assays to detect CT and NG which utilize different automated instruments. Methods: The product insert for the Aptima Combo 2 Assay, Panther System (Hologic/GenProbe) includes endocervical, vaginal and urethral swabs, male urine, and liquid based cytology specimens (PreservCyt). This study evaluated female urine specimens and specimens from pregnant women because these populations were not evaluated by the manufacturer. Accuracy, reproducibility, and range of detection were evaluated with patient samples that were shared with a different laboratory or by seeding a known negative specimen with CT or NG. Results: There was 97.5% overall agreement between the different laboratories. Conclusions: Our study determined that the analytical sensitivity and specificity for both female urine and specimens from pregnant women were at similar levels as the other sources evaluated by the manufacturer. Introduction: Toxoplasmosis, a zoonosis caused by the parasite Toxoplasma gondii, infects both birds and mammals, particularly cats, and has a relatively high incidence in the US, where 15% to 29% of the population is seropositive.
Transmission occurs by foodborne route, rarely via transfusions or organ transplantation and, importantly, by vertical transmission transplacentally. Although symptoms are usually mild in immunocompetent hosts, infection during pregnancy can produce catastrophic effects on the fetus; including visual impairment, encephalomyelitis, developmental delay, chorioretinitis, hydrocephalus, and brain calcification. Treatment is available, once diagnosis is established -by serology, immunofluorescence or by molecular methods. Methods: Toxoplasma gondii laboratory-developed test (LDT) is based on real-time PCR utilizing in-housedesigned primers and probes and was run on Roche LC480 instrument. Analytical sensitivity and LOD were determined by run dilutions of the Toxoplasma gondii RH type I at concentration 18,000 cp/uL x 250 copy (equivalent 4.5M cp/uL -Advanced Bio Technologies) and LDT plasmid from MuMeTel ITT concentration of 250 cp/uL).
In addition, 73 previously tested samples (representing different matrixes) were tested, following extraction using NucliSENSE EasyMag protocol. Results: Analytical specificity (exclusivity) was 100% when run against 49 different microbial and viral targets. Additionally, GenBank reference strain AF146527 for Toxoplasma gondii sequence was used for "in-silico" analysis of coverage primers and probes specific to 529bp "repeated fragment". Analysis confirmed 100% specificity (inclusivity study) to highly conserved nucleotide sequence among T. gondii genotypes I, II and III. Clinical sensitivity of the Toxoplasma gondii assay was evaluated by analyzing 23 Toxoplasma gondii positive samples (positivity range 27 to 230,000 cp/mL). Twenty-three of 23 samples were positive producing a clinical sensitivity of 100%. Fifty of 50 clinically negative samples were not detected by the assay producing a clinical specificity of 100%. The Toxoplasma gondii LDT assay has the following characteristics: LOD = 200 cp/mL; slope, -3.300; R 2 value=0.999; PCR efficiency, E=2.004. Assay linear range extends over an interval of at 7 orders of magnitude. Conclusions: This study demonstrates that the LDT assay for T. gondii developed by ACL laboratory has very good performance characteristics and is suitable for clinical testing from different sample matrixes. Introduction: Human cytomegalovirus (CMV) viremia should be monitored in immunocompromised patients including solid organ or stem cell transplant recipients due to the potential for various morbidities including graft damage and loss. CMV monitoring is generally accomplished by measuring CMV DNA in the plasma using qPCR. We verified performance characteristics of an FDA-approved CMV qPCR assay using a combination of clinical patient specimens and commercial control material. Methods: CMV viremia was assessed and quantified in 98 patient plasma specimens and plasma-based control material using the COBAS AmpliPrep/COBAS TaqMan CMV Test (Roche Molecular Systems) for automated DNA isolation and qPCR of CMV DNA according to the product insert with results reported in IU/mL as "target not detected" (TND), "<137" or a value between 137 and 9,100,000 IU/mL. Results from all patient plasma specimens were compared to results obtained from an outside laboratory using an aliquot of the same plasma sample with the same Roche CMV assay. The AcroMetrix CMVtc Panel was used to assess assay performance at CMV DNA concentrations covering a 5 log range (300 to 3,000,000 IU/mL). CMV panel members were also diluted with AcroMetrix EDTA Plasma Dilution Matrix to assess performance at other concentrations. Results were considered concordant qualitatively if results from the 2 tests were: both TND, <137 or >137; TND and <137; or <137 and within 0.2 log of 137. Results: Results were qualitatively concordant with the outside lab for 98 of 98 (100%) of specimens. Quantitative comparisons using a Bland-Altman plot were possible with 37 of these specimens with results by the outside laboratories were on average 0.11 log IU/mL above the value from our laboratory with the difference between the outside lab value and ours ranging from 0.42 to -0.31 log IU/mL. The differences between the 2 values was more pronounced with CMV DNA below 3.2 IU/mL. The CMV panel members were tested in triplicate and the assay was found to be linear across the 5 log range with an R2=0.9945. Precision was assessed using replicates of 3 dilutions: 2.14 (LLOQ), 3.64 (Mid) and 5.64 (High) log IU/mL. Results >137 were obtained for 16 of 20 (80%) of the LLOQ replicates (4 of 20 were <137). The values for the mean values for 15 replicates of the mid and high level samples were 3.65 and 5.56 log IU/mL with SDs of 0.082 and 0.097, respectively. Analytical sensitivity studies confirmed detection of CMV in 20 replicates diluted to the published LOD (91 IU/mL). Conclusions: The COBAS CMV Test performed well in our laboratory with respect to analytical sensitivity and precision. The assay accurately quantified CMV DNA in plasma specimens as compared to outside laboratory assessment using the same test.
J. Jones, J. Painter, C. Derecho, T. Astacio, P. Shah, J. Baden Janssen Diagnostics, Raritan, NJ. Introduction: Influenza (IFV) and respiratory syncytial birus (RSV) are some of the most common human respiratory viral infections and are associated with significant morbidity and mortality. Clinicians want diagnostic tests that provide a cost-effective, reliable result rapid time frame to aid and support treatment decisions. We have developed the investigational Janssen Diagnostics (JDx) Idylla RESPIRATORY (IFV-RSV) Panel, a rapid, real-time RT-PCR assay designed to qualitatively detect and differentiate Influenza A (and subtypes H1, H3, and pdm H1), Influenza B, RSV-A and RSV-B, and the mutation that confers resistance to Oseltamivir (Tamiflu) in Influenza A pdm H1positive samples all within 50 minutes on the Idylla platform with only one minute of hands-on time. Methods: Performance of the Investigational IFV-RSV panel was conducted with archived banked nasopharyngeal (NP) swab samples and virus cultures. An FDA cleared Molecular Diagnostic assay was used as the comparator. Concordance (Negative and Positive Percent Agreement), Limits of Detection, Exclusivity, and Inclusivity were evaluated. Results: The initial Limit of Detection (LoD) for influenza virus A (including subtyping assays), B and was RSV A Analytical specificity (Exclusivity) of 100% was observed from testing 24 cultures including 13 viruses, and 11 bacteria. Analytical reactivity (Inclusivity) was assessed with 9 influenza A H1N1 (pre-2009), 7 influenza A pdm H1N1, 10 influenza A H3N2, 7 avian influenza A (H5 and H7), 8 influenza B, 3 RSV A and 4 RSV B strains, and all were reported correctly. Pre-clinical data were derived from a prospective study. With NP swabs (n=100), the investigational IFV-RSV panel demonstrated 100% positive and 100% negative agreements for detection of Influenza A, Influenza A H1, Influenza A H3, Influenza A H1 (pandemic 2009) and Influenza B; 100% positive and 98.8% negative agreements for RSVA and 100% positive and 98.8% negative agreements for RSVA and 100% positive and 97.8% negative agreements for RSVB relative to the Nanosphere Verigene Respiratory Virus Nucleic Acid Test. RSV discrepants were analyzed via bidirectional sequencing and the presence of RSV was confirmed. Additionally, 3 of 10 H3N2 samples tested were H3N2(v). All 3 of these samples were detected by the Investigational IFV-RSV Panel. Four of the 7 influenza A pdm H1N1 samples tested contained the H275Y mutation that confers resistance to Oseltamivir. All 4 of these mutations were correctly identified by Investigational IFV-RSV Panel.
The Idylla platform is designed to couple the sensitive performance of molecular testing with extreme ease-of-use. The Investigational JDx Idylla RESPIRATORY (IFV-RSV) Panel correctly identified and subtyped samples containing influenza virus A and correctly identified samples containing influenza virus B, RSV-A and RSV-B, with demonstrated detection of recent variants commonly escaping detection by conventional rapid diagnostic tests.
University of Alabama at Birmingham, Birmingham, AL. Introduction: Next-generation sequencing (NGS) is an area of rapid growth in modern molecular genetic pathology (MGP). One obstacle of implementation of clinical NGS is the complexity of NGS data analysis, which is beyond the scope of practice of most practicing molecular pathologists. NGS data analysis currently is not strongly emphasized in most MGP fellowship training curricula; where such training does exist, it is usually cursory and does not touch upon the quantitative foundations of the discipline. We sought to solve this problem by designing, implementing, and testing a customized bioinformatics curriculum and practicum in molecular diagnostics which culminates in the construction of a complete, clinically-validated GATK-based NGS data analytics pipeline. Methods: Our curriculum begins with introduction to computer science, including systems architecture, procedural and object-oriented programming, and software engineering principles, which lasts two weeks. Trainees are tasked to build and host a standard LEMP (Ubuntu Linux Server 12.04 LTS; nginx 1.6.3; MariaDB 10.1.2; PHP 5.6.3) stack, followed by an intensive practicum in programming and shell scripting on that stack. Following multiple small programming and analytics projects that serve to build the trainee's abilities in and understanding of data science, the final project begins: the synthesis of an exome analysis pipeline utilizing the Broad Institute's Genome Analysis Toolkit (GATK) Best Practices (current software versions: BWA 7.12, Picard 1.130, SamTools 1.2, and GATK 3.3). A curated set of previously-sequenced genomic data from a reference laboratory was used to formally validate the pipeline. Results: Several trainees have completed this training. The majority of the trainees neither have preexisting knowledge and skills of computer science beyond the levels of medical students, nor have experience in programming. It took an average of one week for the trainees to install all the programs required by the GATK best practice, and 1 to 2 months to validate the pipeline. Every trainee was able to achieve the training goal and validated their pipelines successfully. Conclusions: This experience shows the utility and feasibility of informatics training in MGP. This approach allows trainees to gain fundamental knowledge of NGS data analysis, which prime them for better understanding of NGS result and interpretation in their future career. Introduction: Clinical NGS is subject to the same regulatory requirements for proficiency testing (PT) as other clinical molecular tests, and methods-based proficiency testing (MBPT) is ideally suited to NGS. MBPT surveys based on cell lines engineered to harbor specific mutations challenge both the "wet lab" and "bioinformatic" aspects of NGS tests but can be limited by the difficulty and sustainability of creating a full spectrum of mutations (ie, SNVs, indels, copy number variants, and translocations) and range of variant allele frequencies (VAFs) needed to comprehensively evaluate assay performance over a broad range of genetic loci. Engineered cell lines also often contain genetic artifacts arising from the introduction of the sequence variants that are not found in patient specimens and which may confound analysis. To address these limitations, we are developing in silico proficiency testing (ISPT), a type of MBPT focused only on the bioinformatic component of NGS. For ISPT, sequence data from a well-characterized specimen are manipulated by computerized algorithms to introduce a spectrum of sequence variants. The resulting modified data files challenge an NGS test's bioinformatic pipeline from alignment through variant detection, annotation, and interpretation. This study was undertaken to evaluate the feasibility of ISPT for amplification-based NGS of an oncology gene set. Methods: The College of American Pathologists' reference genome was sequenced using the Illumina TruSeq Amplicon cancer panel and the Ion Torrent AmpliSeq cancer Hotspot v2 panel. Using a custom GATK locus walker (MutationMaker v0.3), 24 SNVs and 2 small indels (all modeled from somatic mutations reported in the COSMIC database) were introduced into the sequence files in a subset of genes at various VAFs; flow space data from the Ion Torrent were modified to match the variants introduced. Participating clinical labs (blinded as to the in silico mutations) downloaded the files and applied their bioinformatic pipeline to identify variants and associated VAFs, which were reported using a standardized form. Results: Three labs using the AmpliSeq panel completed the model ISPT survey. These labs correctly identified an average of 24 of 26 variants overall (SNVs+indels), and 23 of 23 SNVs with VAFs>10%; no false positive variants were reported. For these labs, the mean difference of the reported versus true (introduced) VAF across all variants was 2.9% (range 0.16% to 14%). The single lab using the TruSeq panel identified 25 of 26 variants overall, and 23 of 24 SNVs with VAFs>10%; no false positive variants were reported. For this lab, the mean difference of the reported versus true VAF across all variants was 2.3% (range 0.0008% to 15%). Conclusions: The results indicate that ISPT is a feasible approach for MBPT to challenge bioinformatic pipelines for amplification-based NGS of oncology specimens. The results suggest that ISPT is likely to be useful in a broader range of settings given that it can be employed to create mixtures of variants that mimic the complexity of clinical samples, for testing of both germline and somatically-acquired variants. (TAT) is an important quality measurement of a molecular diagnostic laboratory. In our laboratory, the target TAT of the BK virus quantitation test is 3 to 7 days. Upon requests from clinicians for a shorter TAT, we set our quality improvement goal to complete 95% of the specimens within 5 days in October 2011. The purpose of this study is to evaluate the utility of informatics in driving real-time TAT and allowing staff to identify issues with workflow. Methods: A Dell Precision T3600 (Intel Xeon E5-1603 @ 2.8 GHz; 16GB DDR3 SDRAM; 256GB SSD; 1TB HDD; Microsoft Windows 7 x64) was used to host a standard LEMP (Ubuntu Linux Server 12.04 LTS; nginx 1.5.6; MariaDB 5.5.32; PHP 5.4) stack virtual machine on Oracle VM Virtualbox virtualization software. Using this LEMP stack, we implemented a new module atop a previously reported Web Real-Time Quality Assurance Platform code-named "Alchemy" to calculate TAT for BK virus quantitation. The input data was autogenerated monthly as comma-separated value (CSV) files from the clinical pathology laboratory information system. Results: Alchemy allowed us to analyze QA statistics in real time and accurately, with minimum time required (decreased from 45 to 60 minutes per month to 15 minutes per month). Alchemy also allowed us to monitor TAT more closely. 80.1% (range 47.8% to 99.4%) of BK virus quantitation tests were resulted within 5 days during a 36-month period between October 2011 and September 2014 prior to Alchemy implementation. This metric changed to 87.4% (range 83.6% to 92.0%), as measured during a 6-month period between October 2014 to March 2015 after Alchemy implementation (p=0.19, by Student's twin-tailed t test). Factors contributing to longer TAT include staff shortage, instrument and reagent issues. Conclusions: Although Alchemy does not improve TAT per se and our goal of TAT is not met at the time of this report, Alchemy enabled the laboratory to monitor TAT in real time, identify the causes of delay, troubleshoot existing workflow issues, and optimize resources in the lab to improve the TAT. Going forward, we will continue to utilize our custom real-time QA platform in driving Lean production methodologies in our laboratory. This project clearly demonstrates the utility of informatics in molecular diagnostics workflow and in resident and fellow education.
Gene Hybrid-Capture, NGS-Based Platform S. Middha, J. Hechtman, J. Shia, Y. Zarina, M. Berger, K. Nafa, L. Zhang Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: Microsatellite instability-high (MSI-H) status is the result of deficiency within the mismatch repair (MMR) system, resulting in increased frameshift mutations at microsatellite loci across the genome. This phenotype has clinically important prognostic, germline, and predictive implications. Conventional testing for MSI-H or MMR deficient status included PCR and immunohistochemistry, respectively. Here, we describe the clinical utility of integrating MSI testing into a broad clinical NGS-based tumor genotyping assay. Methods: MSK-IMPACT (Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets) is a hybrid capture-based NGS assay that detects copy number alterations, substitution and small indels, and translocations in all exons and select introns of 410 genes (JMD 2015 May; 17:251-64) . Sequencing results were analyzed via MSIsensor, a C++ program for NGS data that assesses the number and length of homo-polymers/ microsatellites within the targeted regions and assigns a continuous (rather than categorical) MSI score (Bioinformatics 2014 Apr 1;30:1015-6). Loci were considered unstable (somatic) if k-mer distributions were significantly different between the tumor and matched normal using a standard multiple testing correction of x2 p-values. The percentage fraction of unstable sites was reported as the score for MSIsensor. MSIsensor scores were correlated with previously clinically validated MSI PCR (Promega) and MMR IHC results. Results: Using an MSIsensor cut-off MMR IHC for 9 MSI-stable (PCR),11 MSI-H (PCR), 122 MMR proficient (IHC) and 14 MMR deficient (IHC) cases. Conclusions: MSIsensor scoring on MSK-IMPACT had 100% sensitivity and specificity for MSI-H/ MMR deficiency. Integrating MSIsensor into the analytical pipeline of a broad clinical NGS assay has advantages in terms of workload, turn-around-time (TAT), cost, and tissue sample utilization for the detection of MSI-H status / MMR deficiency while simultaneously detecting copy number changes, mutations, and gene fusions.
Introduction: A translational research pathology lab utilizes a complex workflow strategy to process a variety of samples and specimens (i.e. assets). The need for specimen tracking in a research lab is as relevant and important as in the clinical lab. Tracking samples in a research lab introduces unique challenges. Custom designed database software solutions offer a low cost, easy to use, powerful solution to asset management. Our objective is to develop and implement a database software solution for asset tracking in a translational research core lab to ensure compliance with institutional, local, and federal requirements. Methods: The project was developed using consultations between CTSI Biomedical Informatics Software Engineer and lab personnel. The following datasets were captured: specimen, accession, requisition, investigator profile, workflow, and pricing structures. The following data was analyzed and reported: invoice, asset tracking, lab productivity, institutional utilization, and internal review board (IRB) compliance. Results: REDCap provides a low cost tool to rapidly conduct data searches. Reports are used for monitoring lab productivity, anticipating future operational demands, and communicating with institutional and non-institutional clients. Data can be archived and reevaluated for additional purposes giving users the ability to share information among lab personnel and researchers Conclusions: Specimen tracking and data collection tools are essential for statistical analysis of lab productivity and asset management. Our work demonstrates that data is easily captured with REDCap allowing significant improvements with specimen tracking, productivity reporting, and compliance with institutional requirements. The hexanucleotide repeat expansion of CCGGGG in the C9orf72 gene has been recognized as a major cause of ALS and Frontotemporal Dementia (FTD) and been implicated in other neurodegenerative and psychiatric disorders. Approximately 30% of familial ALS cases, 20% of familial FTD cases, 6% of sporadic ALS cases and 5% of sporadic FTD cases have been attributed to the C9orf72 repeat expansion. Although the exact pathological cut-off has not yet been established, the C9orf72 repeat is considered expanded when it counts more than 30 hexamers (>180bp). Since it is so prevalent, all genetic studies in ALS and FTD must be stratified or adjusted for C9orf72 repeat status, requiring a continuous and separate RP-PCR analysis. Moreover, the exact length of the repeat expansion is jmd.amjpathol.org ■ The Journal of Molecular Diagnostics difficult, if not impossible, to determine and currently involves a combination of RP-PCR, Southern blotting and targeted sequencing. Here, we present a tool to accurately determine the length of any 3 to 6 nucleotide repeat, even when they are extremely expanded, in paired-end whole-genome sequence data. Methods: Sequencing was performed on 349 individuals (including 77 with the C9orf72 expansion) on an Illumina HiSeq 2000 using 2x100bp paired reads to ~40x depth. The tool first extracts all reads that belong to the repeat of interest by searching for 1) reads that span the repeat and contain flanking sequence on both sides, 2) reads with flanking sequence on 1 side, 3) read pairs of which 1 read consists entirely of the repeat, and 4) read pairs of which both reads consist entirely of the repeat. The number of repeat units are counted and normalized using the average genomic read depth to calculate the length of the repeat. Results: Of the 77 C9orf72 samples that RP-PCR identified as expanded, only 1 sample had a repeat count of only 2 according to the tool, whereas 6 samples had a repeat count between 20 and 30 and the rest of the samples had a repeat count higher than 30. Of the 272 C9orf72 samples that were not expanded, only 1 sample had a repeat count of 260 according to the tool, whereas 2 samples had a repeat count between 20 to 30 and the rest of the samples had a repeat count lower than 20. If a count of 30 is the pathological cutoff, sensitivity of the tool is 90.9%, and specificity 99.6%. If the true cutoff is 20, these numbers are 98.7% and 98.9% respectively. Conclusions: Further validation of this tool is ongoing in independent samples and for other repeat expansions, which will be presented. Since microsatellite expansions are a common genetic risk factor in neurodegenerative diseases, accurately determining the length of a repeat at known genomic locations using NGS data and also discovering unknown repeat expansions will allow studying the impact of repeat expansion on neurodegenerative disorders. Unfortunately, NGS tests are more complicated to evaluate and validate than single analyte tests. All clinical assays offered in CLIA-certified labs must undergo a validation process where the platform used for the assay, the assay itself, and all subsequent informatics analysis steps are shown to accurately and reproducibly identify individuals positive for the marker of interest in the assay. For NGS assays, this means the ability of the assay to detect specific variants (SNVs, INDELs, CNVs, etc.) that might be important for treatment. Clinical labs validate these assays by sequencing samples containing known variants (gold standard data set) that have been verified using some orthogonal technique. Run and sample level sequencing metrics must be monitored to ensure high confidence data is being generated. We have created an easy-to-use tool for analyzing complex NGS data for validation called GenomOncology (GO) Validation Reporting Framework. Methods: The GO Validation Reporting Framework, a component of GenomAnalytics, leverages a robust NGS data parsing and annotation workflow combined with data analysis using the R statistical computing language. Variant data sets (.bam and .vcf files) are imported into GenomAnalytics and combined with the set of known, gold standard reference variant calls for the sequenced validation samples. The system determines if each variant is a True Positive or False Positive or if there is a missing variant in the sequenced data set (False Negative). This data is then passed to custom, configurable R scripts that carry out the subsequent analysis of the data and generate a PDF or Excel based report that the user can view and download to their computer. Results: We compared the validation results of 5 independent labs introducing the test (validation) and over time (proficiency testing). Valiation results varied from lab to lab and over time. Each of the labs improved over time as they became more acquainted with the hardware and software. Reports provide the lab the essential information needed to prove the accuracy of their assay and provide a permanent record of all data analysis carried out for the validation process and for subsequent proficiency testing. Conclusions: Analyzing and presenting validation data for NGS is a difficult and time-consuming task faced by clinical laboratories with limited bioinformatics resources. Commercially available gold standard DNAs and reference materials along with dedicated software have make validation and proficiency testing much easier. Introduction: In somatic testing for oncology, next-generation sequencing (NGS) is commonly used to evaluate for mutations and insertions/deletions. Detection of copy number variations (CNVs) can be achieved in NGS; however the library generation method can significantly impact evaluation of CNVs. Amplicon-based methods present challenges to evaluation of CNVs due to potential PCR duplicates skewing analysis; however informatics methods evaluating depth of coverage (DOC) compared to an expected distribution can be used. Here we present our initial evaluation of an algorithm in which variance in DOC can be utilized to screen for samples in which further evaluation by fluorescence in situ hybridization (FISH) may be of value. Methods: Sequencing data from tumor specimens evaluated by the Colorado Molecular Correlates Laboratory at the University of Colorado using the Illumina TruSight Tumor Panel were analyzed using an in-house CNV algorithm equivalents (CNE) based on NRD for one of 6 genes (EGFR, ERBB2, KIT, KRAS, MET and PIK3CA) were classified as positive (+) for NGS-based copy number gain (NGS-CNG +). Specimens with CNE <4 were classified as negative (-) for NGS-CNG (NGS-CNG -). 44 samples, 20 (NGS-CNG +) and 24 (NGS-CNG -), were further evaluated for CNV by FISH. FISH results were classified as (+) for gene -) when these criteria were not reached. Specimens (-) for GA were classified as (+) for FISH-Results: Forty four cases investigated by both NGS and FISH included 10 EGFR, 2 ERBB2, 1 KIT, 4 KRAS, 20 MET and 7 PIK3CA. Of 20 NGS-CNG+ cases, 16 were (+) for GA, 2 were (+) for FISH-CNG, and 2 were (-) for GA and FISH-CNG. All 16 cases classified as (+) for GA by FISH were identified by CNE. Of the 24 NGS-CNGcases, 13 were (-) for both GA and CNG by FISH; 11 cases were (-) for GA but showed CNG by FISH. Nine of 11 cases with NGS-CNG-but FISH-CNG+ were in MET. Nine 4 gene copies. Conclusions: Normalized read depth from a targeted NGS amplicon-based tumor panel was applied as a tool to identify patient specimens for further testing by FISH for gene amplification. In these initial studies, NGS was highly concordant with FISH for detection of GA. However, this normalized CNE approach had reduced sensitivity for detecting instances of CNG without GA, potentially related to tumor heterogeneity. As this approach shows promise in helping to identify cases where gene amplification may be present, further algorithmic refinement and validation is warranted. Although clinical utility data for these panels is currently being generated, the theoretical utility and actionability of these expanded panels can be determined. Methods: Four expanded gene panels (DF, FMI, JAX, and MSK) were ranked for actionability using the 72-gene list and actionability tier ranking by Hayes et al, 2015: J Clin Invest. The tiers were limited to TIER 1 (standard of care), TIER 2A (FDA targeted drug), and TIER 2B (targeted therapy in a clinical trial). The Ion Torrent AmpliSeq hotspot panel was also included in the analysis. Furthermore, eight genes on both JAX-CTP and AmpliSeq panel were compared for the number of additional actionable mutations that could be theoretically identified in the expanded panel. Results: For TIER 1 genes, all four expanded panels had 91% (10/11) of the genes, whereas the AmpliSeq panel covered 82% (9/11). In TIER 2A, gene coverage differed slightly among the expanded panels: DF 77% (20/26), FMI 92% (24/26), JAX-CTP 85% (22/26), and MSK-IMPACT 96% (25/26). The AmpliSeq hotspot panel had 54% (14/26) coverage of TIER 2A genes. In TIER 2B gene coverage was 74% (26/35) for DF, 80% (28/35) for FMI, 77% (27/35) for JAX-CTP, 83% (29/35) for MSK-IMPACT and 31% (11/35) for AmpliSeq. It was further determined that JAX-CTP could theoretically detect 17, 5, 5, 0, 9, 3, 4, and 6 additional actionable mutations in AKT1, ALK, BRAF, EGFR, ERBB2, GNA11, JAK3, and MET, respectively, over the AmpliSeq hotspot panel. In addition, JAX-CTP has identified an additional 55 variants in 126 samples that the AmpliSeq panel would not have been able to detect. Conclusions: Expanded targeted panels have the potential to call more actionable mutations than smaller hotspot panels. In general, the larger panels have better coverage of actionable genes, even when limited to a stringent evidence ranking system. Due to the full exon coverage of all genes on the panel, the larger panels also have the theoretical ability to identify more actionable mutations within genes that are also found on smaller hotspot panels. Introduction: Whole-exome sequencing (WES) has been increasingly adopted for the diagnosis of rare disorders. The "diagnostic odyssey" to identify the causal variants typically involves time-consuming review processes performed by highly trained personnel. Therefore, the sensitivity of the test in identifying the pathogenic variant and the turn around time of the test remain the biggest challenge for a clinical exome lab. In addition to computational methods designed to predict a variant's The Journal of Molecular Diagnostics ■ jmd.amjpathol.org deleteriousness, efforts have been made to leverage patient's phenotypes to prioritize candidate genes and variants. We have developed Phenoxome, an innovative network-based approach to generate a personalized gene panel (PGPs) derived from patients' phenotypes and prioritize the genes carrying variants. Methods: Phenoxome generates a PGP and then intersects it with genes carrying variants, followed by a random walk model to prioritize relative phenotypes and genes accordingly. To compile a PGP and rank phenotypes, we utilized the Human Phenotype Ontology (HPO) to extract gene-phenotype associations. For each patient's phenotype, Phenoxome first traverses the ontological tree of HPO, and then generates a PGP retrieving genes associated with all of its direct and indirect children and immediate parent terms. Following this, a PageRank with priors model is implemented on the sub-network of navigated terms to assign a significance score to each phenotype. This score is then used to prioritize genes by ranking those associated with more significant terms on top. Results: For validation, we used a combination of genomic data from 36 clinical WES cases and simulated data of 32 Mendelian diseases with known causal genes from OMIM. We used patient phenotypes to generate the PGPs for clinical cases. Similarly, 64,000 PGPs were compiled taking into account of penetrance data of the 32 Mendelian diseases. Phenoxome captured all causative genes in the test PGPs (100% sensitivity). To evaluate the ranking of disease causing genes, for each of the above-mentioned Mendelian diseases, 1,000 patients per disease were simulated. Since the application relies on user input, we modeled human factors by adding noise to the synthetic profiles, ie, irrelevant phenotypes. The median causative gene rank was 1 for both optimal and noise scenarios with areas under the curve of 0.99 and 0.98, respectively. For the positive clinical exome cases, all target genes were within the top 20 of the ranked list. Conclusions: In summary, Phenoxome is a computational application that improves the efficiency of the interpretation of WES/WGS data by providing a prioritized list of genes coupled with variant analyses from presenting patient phenotypes. It is the first implementation of a PageRank model on HPO to our best knowledge to rank genes. We expect more intelligent computational models will be developed to increase accuracy in variant classification and further reduce the processing time.
The Jackson Laboratory for Genomic Medicine, Farmington, CT; 2 The Jackson Laboratory, Bar Harbor, ME; 3 The Geisel School of Medicine at Dartmouth and Dartmouth Hitchcock Medical Center, Lebanon, NH. Introduction: Somatic tumor profiling can generate a multitude of associated targeted therapeutic treatment options. However, the strength of evidence in the literature for therapeutic efficacy varies depending on clinical data, preclinical data, reported variant specificity, and tumor type. Often, it is known that a therapy targets a specific activating mutation in a gene, but it is unclear if this data can be extrapolated to all activating mutations of that gene. Therefore, we generated at 25 level tier system, which ranks efficacy evidence data based on clinical data, preclinical data, variant specificity, and tumor type. Methods: DNA was isolated from 56 solid tumors and subjected to Illumina next-generation targeted sequencing using a 358-gene panel. FASTQ files were submitted to a Clinical Genome Analytics (CGA) bioinformatic pipeline to identify SNP's, indels, and CNV's. All variant calls were mapped to an in-house Clinical Knowledgebase (CKB) to generate a clinical report that matched somatic variants to targeted therapies. Efficacy evidence in each report was scored on a 25 tier ranking system, with tier 1 having the highest efficacy ranking. Quality and consistency of reports were statistically assessed before and after implementation of the ranking system. Results: An average of 3.16 ± 1.15 actionable variants were found per patient sample. Each report had an average of 1.36 ± 1.63 FDA approved therapies, 8.52 ± 5.42 FDA therapies approved in other indications, and 12.07 ± 9.01 investigational therapies. The number of clinical trials averaged 72.68 ± 43.01 per clinical report. Ranking of efficacy evidence per variant/therapy combination produced an average ranking of 2.73 (high) for FDA approved, 16.97 for FDA approved in other indications, and 19.50 (low) for investigational therapies. Prior to implementation of the ranking system, reports had a mean efficacy score of 19.60 with a variance of 34.90. After implementation, the mean was 16.70 with a variance of 59.30 and comparison between before and after was statistically significant (p<.00005). Conclusions: The increase in variance of the mean efficacy evidence score indicates increased specificity of reporting after implementation of the tier ranking system. The system further controls quality and consistency among multiple report writers in objectively selecting data to include and the clinical report. In addition, this evidence-based reporting system will aid the oncologist and selecting therapies and/or clinical trials for patients who have exhausted standard-of care treatment options.
D.N. Baker, N. Welker Associated Regional and University Pathologists Institute for Clinical and Experimental Pathology, Salt Lake City, UT. Introduction: Barcoded Molecular Families Tools (BMFTools) is a suite of computational tools for manipulating barcoded reads providing both error correction and accurate PCR duplicate identification. Massively-Parallel Sequencing (MPS) offers greater breadth of information than first-generation sequencing and digital droplet PCR, but at the expense of significant error rates. Even with the higher quality of instruments such as the HiSeq, with typical error rates <1%, these errors accumulate in unbarcoded sequencing to produce spurious calls. Methods: The HD734 standard was acquired from HorizonDx. This standard was sequenced to a unique observation depth of 30,000 on the Illumina NextSeq 500. An FFPE lung cancer biopsy was selected from ARUP's available validation materials after fragmenting to a mean insert size of 200Kb. This sample was sequenced to a unique observation depth of 1600 on the Illumina MiSeq. An 18-gene custom hybrid capture panel was acquired from Integrated DNA Technologies, consisting of full exonic coverage of these genes and the entirety of ALK intron 19. After DNA extraction, the random molecular barcodes were ligated to the template molecules, amplified, captured, and sequenced. Each set of reads sharing a molecular barcode (a "Barcoded Molecular Family") is collapsed into a single observation by performing a probabilistic base-by-base meta-analysis. This method provides both error correction and PCR duplicate removal. Additionally, we have developed a barcode-aware SNV caller which takes advantage of the additional information gained from the read collapsing process. Results: BMFTools also identify PCR duplicates rather than attempting to infer them as in standard NGS analysis, as recommended by GATK's best practices. Standard PCR duplicate removal inference tools, such as samtools rmdup and Picard Tools' "MarkDuplicates," infer PCR duplicates by examining the genomic coordinates and selecting the "best" read of the set. With deep coverage or amplicon sequencing, many reads will have the same mapping, invalidating the method. After removing identifying and collapsing true PCR duplicates with BMFTools, we ran these tools to see how much of our already de-duplicated data they would incorrectly remove. For this first sample, default samtools rmdup removed 70.0% of the 87,963,178 read pairs, samtools rmdup -S removed whereas Picard's MarkDuplicates removed 74.5%, and samtools rmdup -S removed 85.6%. Conclusions: Through the appropriate use of molecular barcodes, BMFTools facilitates error correction and precise PCR duplicate identification and removal. The use of molecular barcodes increases the effective yield of the sequencer. Accurate quantitation from this PCR duplicate removal has applications in a number of NGS analysis challenges, including copy number alteration calling and detecting differential expression. Introduction: Determining the eligibility of patients for molecular-based clinical trials and the interpretation of data emerging from clinical trials is significantly hampered by the two primary factors: 1) the lack of specific reporting standards for biomarkers in clinical trials and 2) the lack of adherence to official gene and variant naming standards. Clinical trial registries need specifics on the mutation required for enrollment as opposed to allowing a generic mutation entry such as, "EGFR mutation." The use of clinical trials data in bioinformatics analysis and reporting is also gated by the lack of robust, state of the art programmatic access support. Methods: The need for biomarker reporting standards was determined by searching ClinicalTrials.gov (CT.gov) on two biomarkers. First, 300 trial records in ClinicalTrials.gov were analyzed for lack of consistency in gene naming using VEGFR2 and KDR search terms. Second, searching CT.gov on "EGFR and mutation" and then examining the variability in the biomarker eligibility inclusion/exclusion criteria assessed the lack of specificity in gene variant nomenclature. An in-house syntax method for gene naming and referential integrity using HUGO and NCBI gene Id, respectively, was implemented for manual curation of biomarkers in clinical trials. This also required the development of regex system for gene variant nomenclature based on HGVS and provided proof of concept for a method to map patient variant calls to curated knowledge. Results: Searching CT.gov on KDR versus the synonym, VEGFR2, returned a differential of 93 trials. In addition, 8.6% (5/58) irrelevant trials were returned for the search term, KDR. Manual curation of clinical trials identified, 41% (43/106) of trials recruiting on nonspecific EGFR mutations (ie, EGFR mutation, EGFR sensitizing mutation) and using an EGFR inhibitor. Over 200 patient samples were mapped to an in-house clinical knowledgebase demonstrating proof of concept for a gene variant nomenclature standards paired to a regex system. This same system enabled the curation of over 1,000 clinical trials with biomarker data. Conclusions: Clinical trial registries as well as reporting of outcome data should have an obligatory and standardized format, which would facilitate accruals and support clinical utility for predictive, prognostic, and diagnostic biomarkers. We have demonstrated the lack of consistency of biomarker data in clinical trial registries, proposed standards for reporting biomarker data, and demonstrated a proof of concept for a standardized system. This approach will facilitate improved identification of relevant clinical trials, aggregation and comparison of information across independent trials, and programmatic access to clinical trials databases. Introduction: Genes subject to focal copy number (CN) gain are important in the progression of cancer. Current methods to assess CN gain (microarray, polymerase chain reaction, and fluorescent in situ and/or scalability to multiple targets. Next-generation sequencing (NGS) has the potential to provide a rapid solution for multiple targets using limited sample input. Methods: We describe a method of assessing CN alterations using an Ion AmpliSeq targeted sequencing cancer research panel that includes amplicons targeting 19 genes subject to recurrent somatic CN gain as well as other genes that are subject to recurrent hotspot mutations and gene fusions. Read count normalization approaches were designed to address intrinsic (amplicon length, GC content) as well as extrinsic sources of variability. The latter used a Principal Components approach to characterize variability associated with assay effects within technical replicates from a large library of sample runs. A metric "Median Absolute Pairwise Difference" applied to assessment of the per-sample reliability of CN estimation was useful in identifying samples at increased risk of generating false positives.Results: Twentyone retrospective FFPE solid tumor samples of multiple cancer types were characterized by NGS and FISH to estimate CN for 8 gene targets. Relative to FISH, the NGS method demonstrated good overall concordance for estimating CN values ranging from 2 to ~50 and 100% sensitivity and specificity for calling cases of CN gain >4 compared to cases with no copy number gain. Titration experiments using mixtures showed a linear relation of expected CN in the range 2 to 10. Conclusions:
We demonstrate an effective method of CN estimation using an Ion Ampliseq targeted sequencing panel. When combined with detection of mutations and gene fusions, this NGS-based approach may be useful in characterizing relevant somatic alterations in cancer research samples.
M.J. McGinniss, S.A. Nahas, j. Kines, S. Gadient, D. Taylor, G. Sims, M. Panjikaran, R. Kanchiravi, C. Scott, A. Graber, B. Dabbas Genoptix Medical Laboratory, Carlsbad, CA. Introduction: In the delivery of actionable results from NGS data, an unmet need exists for data management, clinical interpretation, and bioinformatics tools to analyze, interpret, and database these variants. Methods: We have developed a pan-cancer assay using NGS methods to detect actionable genomic alterations across 173 genes. This panel includes probes to detect single nucleotide variants, copy number variations (CNV) in 45 genes, and 16 of the most common translocations/gene rearrangements for lung and certain hematologic malignancies. We collaborated with a third party clinical genomics software provider (Cartagenia) to automate our bioinformatics pipeline. Cartagenia Bench Lab NGS allows for data management and provides clinical interpretation support tools to assess the significance of genomic alterations. In addition, we collaborated with another third party (CollabRx Inc.) to configure a customized report that identifies therapies and clinical trials associated with genomic alterations. We used a modification of the ACMG classification scheme (Richards S et al 2015. Genet Med 17:405-423) for the interpretation of sequence variants. Results: Validation results support a minimum average sequencing depth of 500x coverage, a lower limit of detection (sensitivity) of 5% mutant alleles for single nucleotide variations and translocation fusions, 10% for insertions/deletions, and CNV detection (>6 copies for amplification and <0.3 for gene deletions). We established filters to determine clinically actionable pathogenic alterations, CNVs, and translocations. For example, we validated a classification scheme using >10 discrete filters such as: is this in our variant list or in publicly available databases (eg, COSMIC, dbSNP) for assessing the population frequency. Our listing comprises 32,212 variants including 1,455 pathogenic or likely pathogenic variants, 538 variants of uncertain clinical significance, and 30,219 likely benign or benign polymorphisms. This system allows technologists, clinical laboratory geneticists, and clinicians to visualize, assess, and report clinical information. This automated platform may result in significant labor savings (~50%) for geneticists doing the assessment and annotation of genomic alterations. Both systems allow for version control, defined user roles, security access control, and audit traceability. Conclusions: We validated an automated bioinformatics pipeline connected with a data management and clinical interpretation system for interpretation of genomic alterations including CNVs and translocations. This pipeline allows clinical molecular geneticists and molecular pathologists to provide quality clinical laboratory services to oncologists, pathologists, and other clinicians.
A. Scott-Van Zeeland 1 , E. Puffenberger 2 , A. Torkamani 3 , S. Feinberg 1 1 Cypher Genomics, San Diego, CA; 2 Clinic for Special Children, Strasburg, PA; 3 The Scripps Research Institute, San Diego, CA. Introduction: Whole-exome and whole genome sequencing assays are increasingly being utilized in the diagnosis of rare diseases. These disorders frequently present with complex symptom constellations and idiopathic origin, making the search for the causative mutation, and subsequent validation, a time-and labor-intensive process. However, as diagnostic sequencing assays are increasingly adopted there will be a need to incorporate greater automation into the interpretation workflow to enable both time-and cost-effectiveness while maintaining a high level of accuracy. In this study, we evaluated the performance of the Cypher Genomics fully automated interpretation solution (Mantis) against expert manual interpretation of 35 pediatric cases with rare or idiopathic disease previously seen at the Clinic for Special Children (CSC), a clinic that serves a primarily Amish and Mennonite population. Methods: A retrospective analysis was conducted on a total of 108 genomes comprising 35 proband studies submitted to the Cypher Genomics Mantis platform for automated annotation and interpretation. Family-based analyses were pursued for each proband with family structures ranging from canonical trio analyses to multiple affected individuals per family of arbitrary relatedness. Automated variant prioritization and summary generation was driven by selecting the nearest phenotype contained in the Cypher Genomics Disease Ontology. Concordance between previous molecular diagnoses and automated interpretation was investigated. Metrics pertaining to the number of potential disease variants disclosed in each automated summary as well as number of secondary findings were also evaluated. Results: Based on a retrospective analysis of 35 pediatric cases with a confirmed molecular diagnosis there is > 90% concordance for findings in genes previously associated with disease between automated and manual interpretation. Concordance between methods for variants of uncertain significance in novel disease genes will be discussed. Further, it was observed that, on average, each case reported 2.4 top implicated variants for manual review after automated prioritization. Finally, consistent with secondary reporting guidelines, in this population we observed an average of 1.2 reportable events per proband and 1.8 reportable events per affiliated family member. Conclusions: As the clinical adoption of broader genome sequencing assays increases, it will require the utilization of automated workflow components to facilitate greater laboratory throughput and avoid manual interpretation errors. In this retrospective study we examined the concordance between manual processes and an automated interpretation solution to determine the validity of medical experts utilizing the Mantis technology in prospective cases. We found a high level of agreement between the automated variant prioritization methods and manual interpretation. Therefore, the Mantis automated interpretation solution may be an effective tool to allow medical geneticists and clinical laboratories to interpret rare disease cases at scale as genomics is increasingly adopted in this patient population. Introduction: In recent years, significant advancements in PCR technology have enabled the use of a PCR-only workflow for size analysis of the CGG-repeat region in the X-linked FMR1 gene. This genetic locus is directly associated to the transmission and manifestation of fragile X syndrome and a number of related disorders. To date, fragment size analysis of FMR1 PCR products is conducted with significant hands-on manipulation by trained operators, which can be laborious with larger sample sets. We describe the development and implementation of a novel algorithm for fully automated and highly accurate analysis of the FMR1 CGG repeat region. This approach provides integrated QC checks, safeguards interpretative accuracy and consistency, and streamlines data analysis. Methods: Fragment size analysis (.fsa) files or corresponding GeneMapper tables (v4.1, Applied Biosystems), for 273 genomic DNA samples were generated using the AmplideX FMR1 PCR assay (Asuragen, Inc.), and used as input data for algorithm development. Multiple signal processing strategies were employed to enable separation of convoluted information from both the gene-specific and repeat-primed peak profiles produced by the assay. Software performance was then evaluated against a comprehensive set of 1000 previously annotated clinical samples, which spanned normal to full-mutation FMR1 genotypes.Results: Software was designed to allow single-click analysis without the need to select electropherogram peaks for FMR1-related peaks or size standards. A comprehensive set of auto-tuning and quality control steps, including the use of AmplideX FMR1 Control DNA, was embedded into the process to minimize error and increase reliability and robustness of performance across multiple platforms and input variables. Automated sizing of the FMR1 CGG repeat region was 100% concordant (within ±1 repeat) with corresponding manually-analyzed data across a 1000-member test sample set; results were obtained in less than 1% of the time needed for manual analysis. In addition, the software was able to identify the presence or absence of non-convoluted AGG interruptions in samples bearing alleles within the clinically-relevant risk range, in accord with known sample annotations. Finally, proof-of-concept was demonstrated for direct analysis of capillary electrophoresis raw data without an external ladder for size calibration.
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org Conclusions: Results from this study demonstrate the accurate and robust performance of a fully automated FMR1 analysis tool. These results underscore the potential utility of software integration into a comprehensive FMR1 PCR workflow that augments data quality, enhances interpretive reliability, and reduces analysis and reporting bottlenecks, particularly for high-volume testing labs.
C.M. Scott, K.J. Hampel, D. Bruso, N. Sidiropoulos The University of Vermont Medical Center, Burlington, VT. Introduction: Documenting the internal laboratory workflow of library preparation in the clinical setting is complex and challenging. Although some vendor solutions are available, sparse literature has been published to provide guidance on their performance or established benchmarks for laboratory developed tests within a CLIA laboratory. We present our experience with the Sunquest Molecular module to offer insight into the challenges and benefits of integrating a paperless tracking system into an existing laboratory information system. Methods: Following CAP and NYS accreditation requirements, a customized protocol management solution for library preparation is designed. Individual steps in the multi-day workflow are broken down, permitting system utilization by multiple technologists and streamlining the process of repeating steps for insufficient quantity or quality metrics. An existing interface between Sunquest and the electronic health record, Epic, facilitates status reports to the ordering physician and pertinent ancillary clinical staff. Results: The Sunquest Molecular module allows full documentation of workflow for all steps of library preparation from extraction to sequencing. Inventory management reports are established to aid with reagent lot number tracking and lot to lot cross-checks of quality. Barcoded labels aid in the identification of extremely small tubes and reduce data entry errors. Conclusions: Given the complexity of internal laboratory workflow documentation, the Sunquest Molecular module is a valuable tool. It facilitates the documentation of the entire library preparation process and offers features to improve efficiency and reduce errors. There are some configuration items (ie, 2D barcodes, reflex ordering) that need improvement in future versions and the product will need to continually evolve to meet changing regulatory requirements. Overall, this tool meets the current challenges of documenting workflow documentation. Introduction: Next-generation sequencing (NGS) is rapidly becoming a mainstay in clinical research and diagnostics due to its ability to interrogate many loci simultaneously with high sensitivity. A systems approach is required, however, to overcome experimental complexity, sample heterogeneity, and data analysis burdens. Although generic "easy-to-use" and high-performance NGS analysis solutions are often claimed, critical disconnects persist across pre-and postanalytical assay phases, particularly for the most challenging clinical samples. We present Quantidex Reporter, a fully-integrated informatics solution that leverages a systems approach to clinical NGS. Methods: Quantidex Reporter is a push-button bioinformatics analysis suite that was developed and optimized with a companion reagent ecosystem of targeted DNA-Seq and RNA-Seq panels (Asuragen) to ensure integration from sample to results. The software can be installed onto a standard desktop machine, includes built-in visual summaries of targeted NGS QC metrics, and enables simple export of analysis results. Quantidex Reporter includes a proprietary variant calling engine that differs from other callers by directly incorporating pre-analytical QC information (amplifiable DNA template count via the Quantidex DNA Assay) into variant calling. The algorithm was trained on 425 residual clinical specimens, comprised of 171 FFPE and 254 FNA tumor biopsies with independent truth measures.Results: We highlight the utility of an integrated systems approach that tightly couples the assay and the analytics through evaluation of an independent test set of over 80 FFPE specimens. Additionally, variant call concordance was evaluated between 30 FFPE and matched fresh frozen and plasma specimens. Comparisons with generic variant calling algorithms reveals that incorporating sample-specific experimental data enhances mutation detection sensitivity, especially for variants present between 0.5% and 10%, and improves PPV for libraries prepared with fewer than 200 amplifiable DNA template molecules. Additionally, Quantidex Reporter assigns samples to QC risk categories based on the likelihood of false-negative calls. Conclusions: The sensitivity and specificity of variant calling, and the ability to assess false-negative risks on a per-sample basis, can be augmented by utilizing pre-sequencing QC information as part of a systems approach to targeted NGS. The combination of high-quality wet bench assays and novel sample-informed analyses, packaged in a streamlined, simple-to-execute workflow, can provide relief from the requirements of in-house bioinformatics expertise and high-performance computing infrastructures while also improving analytical results for oncology precision medicine applications. F.D. Oakley, Y. Wang, C. Belludi, A.S. Kim, C.L. Vnencak-Jones Vanderbilt University Medical Center, Nashville, TN. Introduction: PCR amplification of highly polymorphic short tandem repeat (STR) loci is used to distinguish between donor and recipient alleles when monitoring for engraftment after stem cell transplantation. Approximately 3500 of these assays are performed annually at our institution. Annotation of the data requires an extensive use of time by the resident or fellow on service to interpret and prepare the results for daily sign out by the attending. To reduce repetitive aspects of annotation, a computer program specialized to the needs and work flow of the Molecular Diagnostics lab (MDL) was designed and compared to the existing manual system. Methods: Automation of STR annotation was implemented using the Perl programming language. The program was designed to accept text formatted ABI electrophoresis data, while also automating data retrieval of previous archived studies for each patient. To evaluate the program, two fellows and two attendings each interpreted 25 assays using the existing manual annotation method. Following a period of at least four weeks, the assays were computer-annotated and returned to the respective individuals for interpretation. The two methods of interpretation were compared by several parameters including analysis time, number of data pages printed, and calculated percentages of donor and recipient DNA in each specimen. Results: The computer program was able to automate the annotation of 96 out of 100 cases. The program accurately identified donor and recipient peaks, fluorescent channel bleed through, forward and reverse stutter, and percent engraftment using appropriate calculations for informative markers. Mean analysis time of the engraftment studies was significantly faster using computer assisted annotation (270.9 s versus 148.6 s; n = 96; p < 0.001, paired T-test) without affecting the final interpretation. The computer annotation method also reduced color printed pages by 82.1% and reduced overall paper usage by 17.9%. Limits to the program included the inability to provide analysis on new patients for which no prior data was archived and difficulties in annotating low levels of signal bleed through. Additionally, the program is limited to assisting with initial annotation and does not provide a final interpretation. Conclusions: Using a computer program to pre-annotate the data generated from our high volume stem cell transplant engraftment studies provides accurate results, reduces expenses, and is applicable in 96% of cases. Further, adopting this workflow into the MDL lab will reduce our overall daily analysis time by 45%. Introduction: VarScan2 is a variant calling tool widely used for the identification of single nucleotide variants (SNVs) and small insertions/deletions (indels) in nextgeneration sequencing (NGS) data. The standard protocol for the use of VarScan2 recommends applying a filter that removes SNVs near indels. Its purpose is to prevent false positive variant calls due to frequent systematic artifacts from local misalignments around indels. However, automatic removal of variant calls to improve accuracy may negatively affect the sensitivity of a clinical assay. Here we assess the impact of disabling the VarScan2's SNVs-near-indel filter on the accuracy and sensitivity of an NGS-based clinical assay for solid tumors. Methods: SAMtools mpileup files of 101 clinical cases were analyzed using VarScan2 with and without the SNVs-near-indel filter. The sequencing data were generated by an in-housedeveloped, 198-gene-panel assay for solid tumors. The bioinformatics pipeline included BWA 0.7.9a-r786 in paired end mode using the bwa-mem algorithm for mapping, and the SAMtools 0.1.19 BAQ adjustment for SNV calling. All additional variant calls generated by the analysis without the filter were reviewed for accuracy on the Integrative Genomics Viewer (IGV) ver 2.3.40(48). Results: The analysis without the VarScan2's SNVs-near-indel filter generated a total of 7 additional SNV calls on different genes and cases. All were confirmed as true calls by manual review calls. The allele frequencies of these variants ranged from 5.19% to 32.77% (mean=18.16%, median=14.85%). Importantly, one of the additional SNV calls was an actionable mutation (NRAS p.Q12C). The allele frequencies of the indels that suppressed the SNV calls ranged from 0.07% to 0.44% (mean=0.20%, median=0.17%), values below our assay's limit of detection of 1%. The indels included 6 single-nucleotide deletions and 1 single-nucleotide insertion, and were present on reads different from those containing the SNVs. Also, some of these indels were detected on quality control samples and were not reproducible. Conclusions: Removing the VarScan2's SNVs-near-indel filter does not increase false positive variant calls in an NGS-based assay for somatic mutations; on the contrary, it increases the sensitivity of the assay by enabling the identification of true SNVs near indels, including actionable mutations. Overall, the findings suggest that current improvements in alignment algorithms along with the BAQ adjustment may eliminate the need for the SNVs-near-indel filter in pipelines for somatic mutations. However, further studies are necessary to determine whether these findings apply to different pipeline architectures.
I22. Use of an Open Source User Interface to Generate Rule-Based Annotations for the Interpretation of Clinical Next-Generation Sequencing W.L. Schulz, R. Torres Yale University, New Haven CT Introduction: Of the approximately 10,000 genomic variants that can be expected to result in an amino acid change compared to the human reference genome, only a very small fraction have known pathogenic, prognostic, or therapeutic significance. Even with these variants of known significance, accurate annotation is a timeintensive process. As a result, clinical annotation of genetic variants remains a key challenge of interpretation. Although clinical laboratories have rapidly adopted next generation sequencing technology, informatics tools that provide user-defined annotations of genetic data are not widely available. Thus, we sought to customize an open-source graphical user interface to create user-defined, rule-based annotations for genetic variants. Methods: The nginx webserver was deployed to a Linux virtual machine running CentOS 7. The jQuery-QueryBuilder package was installed and customized to support fields for a defined set of variant metadata relevant to rule-based annotation. This included chromosomal location, gene, variant allele frequency, and base pair change. The package is able to automatically generate SQL-and JSON-based queries that can be run directly against SQL Server or MongoDB databases. The package was extended to generate Elasticsearch queries and allow for annotation of sequences that match user-generated query parameters. Results: The updated graphical user interface is capable of creating complex, rule-based annotations for genetic variants with simple language input. The generated queries are easily shared as they can be immediately translated into multiple query languages. With the use of a Python script, variants can be accurately annotated in real-time prior to physician interpretation. Conclusions: The implementation of a graphical user interface based on the open-source QueryBuilder library allows users to dynamically and easily create rule-based variant annotations. We expect that this technology will improve accuracy and consistency in variant reporting, as well as decrease the amount of time required for physicians to annotate variants. In addition, if widely adopted, the data sharing features will allow departments and institutions to standardize the interpretation of well-defined genetic variants. 0% JAK2 amp was identified in 79% of the PD-L1/2 amp cases often on the same amplicon. The other most frequently altered co-altered genes were TP53 (75%), CDKN2A (28%), MYC (19%), CDKN2B (18%), KRAS, RB1, and PIK3CA (each in 11%). Other known potentially targetable genes found in the PD-L1/2 amp cases included EGFR (8%), ERBB2 (5%), MET (4%), KIT (4%) and BRAF (2%). In the 206 amp cases, 4% contained HPV-16 and 1% contained EBV viral DNA (32% and 11% in 28 squamous cell carcinomas). Conclusions: Amplifications of PD-L1 and/or PD-L2 genes were distributed across a diverse landscape of tumor types which were more frequently identified in solid tumors than in hematologic malignancies (p <0.005). Finally, PD-L1 and/or PD-L2 amp were significantly enriched in DLBCL compared with other hematologic malignancies (p =0.0001). Clinical correlation with outcomes and other biomarkers is ongoing. . Differences in BI implementation for NGS can change variants (VTs) identified for further interpretive evaluation. Custom BI (versus instrument on-board analysis) can offer advantages, including the ability to update parameters to the analytic pipeline (PL). Updates to the PL necessitate (re)validation, however guidelines for validation have not been established. Here we present an approach to validation of an internal custom-designed BI PL upgrade. Methods: A custom BI PL (v1.0), developed for use with Illumina TruSight Tumor panel, was updated to alter multiple BI processes. These updates include primer trimming, higher stringency in removing reads with soft-clipped bases, and increased removal of off-target reads (v2.2). For validation, FASTQ files (FQ) from 50 previously analyzed samples (v1.0) were analyzed with the updated BI PL (v2.2). FQs were selected to represent a spectrum of clinical case types seen in routine evaluation, including cases with single nucleotide variants, insertions, deletions, samples with poor DNA quality and samples without prior mutational findings. Specific samples with known potential problematic detection issues (e.g. insertions near primers) as well as commercial control samples were included. All VTs identified by PL v1.0 and v2.2 were compared to assess for concordance between the two versions for VT call, VT allele frequency (VAF) and read depth. Results: 79/79 non-synonymous VTs called on v1.0 were identified in v2.2. Two nonsynonymous VTs identified in v2.2 were not identified in v1.0. One of these VTs was a SMAD4 p.W524C variant of unknown clinical significance with low VAF. The second was an EGFR p.T790M mutation at 6.4% VAF (originally identified using an alternate methodology). Both mutations were not identified in v1.0 due to overlap with a region of primer in a different amplicon resulting in signal 'dilution'. Persistent assay artifacts were significantly reduced: average artifact calls 11.94 (v2.2) versus 26.68 (v1.0) (p<0.001). Although v2.2 had increased stringency for eliminating reads based on multiple parameters, read depth was not adversely impacted for clinical variant calling. Conclusions: Alterations or upgrades in BI PLs require systematic validation designed to fully characterize consequences of the changes. Reanalyzing previously evaluated specimens representing the spectrum of clinical findings combined with commercial control samples can form the basis of a PL revalidation. Knowledge of individual BI tools and how they impact data analysis is a critical component of understanding potential intended and unintended consequences of BI PL alterations. Introduction: Targeted amplicon sequencing (TAS) by next-generation sequencing of formalin-fixed, paraffin-embedded (FFPE) DNA has become an acceptable approach for the management of oncology patients. In order for TAS to achieve full clinical utility, informatics tools must provide a standardized and scalable manner of evaluating the biological significance and clinical actionability of somatic variants. Methods: We used QCI to assess variants from 44 FFPE specimens of advanced non-small cell carcinoma, melanoma, and colorectal cancer. These specimens, used for TAS validation in our laboratory, represent biopsy material from patients refractory to conventional therapies that were under evaluation for tyrosine kinase inhibitor regimen. QCI includes curated knowledge from drug labels, professional guidelines, open clinical trials, case counts from clinical literature, and from many public databases such as the Allele Frequency Community. The ACMG guidelines are then used to automatically classify each variant. These computed classifications include a comprehensive evaluation of each individual variant's biological The Journal of Molecular Diagnostics ■ jmd.amjpathol.org significance and treatment information, which is presented to the clinician in a final report. Seven replicate samples were used to measure QCI reproducibility for KRAS, EGFR and BRAF. Our study compares QCI results from 51 variant call format (VCF) files with our manual assessments from queries of standard public databases such as ClinVar and COSMIC to determine the efficacy of QCI for use in a clinical laboratory setting. We also included known sequencing artifacts and variants with low coverage and quality scores to determine the effect that these features might have on QCI performance. Results: A total of 1319 variants were classified by QCI. A significant number (17.1%) of the total variants (N = 226, p = 1.35e-08, CI: 4.73 to 31.1) were classified as "Likely Pathogenic" but not as actionable/reportable; 23% of this class had variant call quality scores less than 50. Artifacts found at amplicon ends were classified by QCI as either "Benign" (91.1%) or "Uncertain" (8.9%). Once poor quality variants were removed (scores < 50), QCI was comparable to our method for annotation (QCI accuracy: 97.6%, VCUHS accuracy: 98.2%). Replicate samples showed good concordance levels (R2 = 0.929, CI: 0.61 to 1.0). Conclusions: QCI's classifier performance is commensurate with our manual method for somatic variant annotation with the added benefit of presenting curated annotations to drive a levels-of-evidence-based approach that quickly assess biological significance and clinical actionability. Importantly, careful screening should still be performed by the operator to remove false positives due to low coverage or quality scoring prior to analysis in QCI. (5):405-23), recommending five-tier classification schema. The guidelines provide a set of detailed and weighted scoring rules based on objective data points and published evidence for classifying variants. We developed an informatics solution to help implement the ACMG 2015 recommendations. Methods: Using the scoring rules from the published guidelines, we developed an algorithm that was implemented as a web-based application. Whenever possible, evidence such as variant type, HGVS nomenclature, minor allele frequency, in-silico prediction scores and disease frequency were automatically calculated or retrieved from local instances of public databases as feeder to the classification algorithm. The algorithm was validated using 50 well-characterized variants from 8 genes (BRCA1, BRCA2, CFTR, RET, SDHB, SDHD, TP53, and VHL), previously classified according to ACMG 2007 guidelines. The application was developed using ASP.NET framework v4.5 and hosted on a Microsoft Windows Sever 2008 server environment. Microsoft SQL Server 2012 hosted the backend databases. Results: Our algorithm classified 43 out of 50 variants in agreement to the 2007 ACMG classification system. For pathogenic, benign, and variants of unknown significance, 23 of 26, 19 of 22, and 1 of 2 variants respectively were classified identically. Discrepancies were found in benign variants with unknown allele frequency (benign-likely benign) and for pathogenic variants that were not truncating and without well-established functional studies supportive of a deleterious effect (pathogenic-likely pathogenic). However, if a change in classification of one level is permitted, all 50 variants were properly classified. Although not unexpected, this was a practical difficulty encountered while validating our algorithm in the absence of a "gold standard." Conclusions: We present an informatics solution for facilitating the implementation of ACMG 2015 guidelines for germline sequence variants encountered in clinical practice. The progressive growth of publically available databases of curated variant interpretation should greatly facilitate implementation of these guidelines. This web application is being actively developed to offer additional features, including a comprehensive knowledgebase, to further assist in variant interpretation. , and joining (J) gene segments to generate unique sequence combinations. Cancers originating from the malignant transformation of individual lymphoid cells share these unique gene rearrangements, resulting in a clonal population. Somatic Hypermutation (SHM) is a cellular mechanism used to increase diversity of B-cell receptors at the somatic level, accruing mutations in the V-region specific to that cell. Minimal Residual Disease (MRD) occurs when a small number of leukemic cells escape treatment and cause a relapse when the cancer cells proliferate to dangerous levels. Next-Generation Sequencing (NGS) assays for detecting these essential disease markers are coming to the forefront of diagnostic methods, and these assays require a new set of bioinformatics tools to confidently assess the results in a straightforward manner. To this end we have developed software capable of accurately analyzing sequencing data of IGH and TRG genes on the Illumina MiSeq and Thermo Fisher Ion PGMTM platforms, for the purpose of measuring the degree of clonality, SHM and MRD. Methods: MiSeq or PGM fastq output was taken from LymphoTrack IGH and TRG assays (Invivoscribe, Inc. San Diego, CA USA) for positive and negative B-cell and T-cell lines, as well as positive and negative clinical samples, for use with the clonality and SHM portion of the LymphoTrack bioinformatics software. For MRD analysis, fastq data from dilution series of cell lines were used as well as diagnostic, post-treatment, and posttransplant clinical samples. Clonal populations in every sample were interrogated after numerous quality control steps, and SHM determination was calculated after alignment of B-cell amplicons to the reference sequence it best matched. MRD assessment was made by searching for the clonal sequence found in the positive control or the diagnostic clinical sample. Results: The LymphoTrack bioinformatics software was able to detect clonal populations in clinical and control cell line samples with a sensitivity of 2.5% cell line dilution in a tonsil background. The SHM analysis was highly correlated with the mutation rate provided by IMGT. Finally, MRD detection was able to find the same or similar sequences to the clonal target at a detection level of at least 1 in 105 cells. The software completes these tasks within minutes per sample, and provides a comprehensive package for analyzing data from the LymphoTrack assay. Conclusions: The LymphoTrack bioinformatics software updated version in development is likely to prove useful to laboratories around the work investigating clonality, SHM and MRD. Introduction: Cancer panels use next generation sequencing (NGS) technology to identify clinically relevant variants in cancer-related genes in order to impact patient care. Validating cancer panels for clinical testing is challenging due to the large number of variant positions covered by the panels and the low variant allelic frequency (VAF) that is often present in somatic cancer. The goal of this project was to utilize GenomAnalytics (GenomOncology, Cleveland, OH) software to facilitate the clinical validation of a 50 cancer-gene NGS panel. Methods: Two cohorts of samples were used in this study: reference standards with precise allelic frequencies Tru-Q HorizonDX DNA Reference Standards ('gold standard set') (Horizon Diagnostics, Cambridge, UK), and formalin-fixed, paraffin embedded (FFPE) tumor DNA from 50 clinical specimens with previously determined single nucleotide variation or indels ('validation set'). DNA libraries were constructed with Ion AmpliSeq Cancer Hotspot Panel v2 kit (ThermoFisher) and sequenced on the IonTorrent PGM platform. BAM and VCF files were analyzed with GenomAnalytics software to determine accuracy, precision, sensitivity, and specificity. Optimal thresholds for run-and sample-level sequencing metrics were identified for total depth, variant depth, quality, and variant allele frequency. These metric thresholds were used for analysis and annotation of VCF and BAM files from FFPE cancer samples using GenomOncology Clinical Workbench. Results: Analysis of the HorizonDx reference standard cohort with GenomAnalytics software indicated that variants with a variant allele frequency at or above 3.5% were reliably detected when the variant had a quality score greater than 6 and the variant location had 300 total reads with 6 or more reads supporting the alternate allele. Applying these sequencing metric thresholds to the patient cohort resulted in identification of 89 true positive variants and no false positive variants. There were 4 expected variants that were not identified by the sequencing (false negatives). Using these thresholds, the assay had an accuracy of 100%, a positive predictive value of 100%, a sensitivity of 96%, and a specificity of 100%. Comparison of replicate samples using GenomAnalytics showed the assay was 99.9% repeatable when comparing calls across all target regions. Conclusions: GenomAnalytics is a robust and powerful software tool to identify and quantify NGS performance metrics and run/sample-level thresholds. Using this tool facilitated our clinical validation of a 50 cancer gene panel allowing us to successfully implement routine clinical NGS testing of FFPE cancer samples. Introduction: Increasing applications of molecular testing in routine oncology care has presented new challenges for specimen handling, test ordering and ensuring appropriate test utilization. In addition, appropriateness of specific test orders need to be determined in the context of diagnosis, treatment stage and prior molecular test results. These result in multiple complex rules and decision algorithms to ensure judicious test utilization. We developed a decision support system based on a custom rule engine to facilitate triaging of tests and specimen accessioning. Methods: A decision support system (PAL, Processing Algorithms) was developed in our laboratory. All existing specimen accessioning algorithms accounting for all possible presentation scenarios and combinations for new and follow-up samples and tests for different hematologic malignancies were compiled into a single "rule" table. A simplified user interface simulating a test requisition form was created to reproduce a test order by the accessioning technologist. Database connections were created to import key decision matrix variables using a barcode scanner into a unified user interface. Once the requested sample and test information were entered, the software would run an internal rule-based report to generate recommendations for DNA/RNA extraction platforms, DNA/RNA tests to be ordered, cancelled or sent for triage. The system auto-generates triage emails to the pathologists. Technical specifications are as follows: Platform: ASP.NET, Language: Visual Basic, Database: MS Access and SQL Server, Script: JavaScript. The system was optimized and implemented after extensive beta testing and end-user training. Results: A total of 2531 peripheral blood and bone marrow samples were processed through the PAL system over 3 months. Training new technologists for sample processing was facilitated by the decision support system, which allowed them to focus on the actual sample log-in and DNA/RNA extraction instead of lookup the required correlative information in multiple different systems and learning complex rules for test orders. For pathologists, triage emails with correlative data made triaging faster and more efficient. From the laboratory management perspective, the data on test volume by shift, day and week provided critical insights. Capturing the DNA/RNA extraction platforms helps fulfil additional regulatory requirements. Overall, the median turnaround time for triage was <24 hours with no effect on overall turnaround time. Conclusions: Implementation of custom automated decision support system for sample accessioning and triaging provides a useful low-cost and high-yield tool beneficial to the technologists, pathologists and laboratory management. Introduction: The level of minimal residual disease (MRD) is the single most powerful independent prognostic factor for lymphoproliferative disorders. Nextgeneration sequencing (NGS) can improve the sensitivity of MRD testing. The standard approach involves the tabulation of each immune receptor gene sequence to determine the distribution of clonal rearrangements within a lymphocyte population. Sequencing errors affect the accuracy of this approach. Here we describe an informatic system that minimizes the impact of sequencing errors by limiting the analysis to the gene junction region. Methods: A series of malignant lymphoma cases consisting of baseline testing with high-tumor burden and a followup sample with low-level disease was collected. T-cell gamma receptor amplicon libraries were created and subjected to Ion PGM sequencing. FASTA files were imported into Excel and unique gene sequences were tabulated using Excel Visual Basic. The dominant clones in the baseline specimen were identified. Junction sequences between the second cysteine codon of the V region and the first phenylalanine/tryptophan of the J segment were extracted. All sequences were realigned to the baseline junction region to determine the relative abundance of the baseline clone among polyclonal sequences. Results: Tabulation of the sequences showed that the majority of baseline lymphoma specimens contained two dominant clonal sequences. One of the two sequences tended to be present at higher levels and this bias was retained in follow-up samples. A sequencing accuracy of 99.6% was observed. At an error rate of 0.4% only 35% to 52% of clonal sequences (133 to 150 bp) were tabulated using the standard method. Limiting data analysis to the junction region increased the detection rate to 66% to 85%, representing an improvement of approximately 1.6 fold. The same junction sequences were observed in less than 0.003% of reads in control samples. Sensitivity can be further increased by truncation of the specific sequence in the junction region. The resulting decrease in specificity can be diminished by restricting the analysis to V/J segments found in the dominant clone. Conclusions: By using the junction sequence of each clonal rearrangement, we can quantify the clones more accurately for MRD levels. The presence of both baseline rearrangements in MRD analysis improves the confidence of a positive result. NGS sequencing error rates, albeit a small percentage, can adversely affect clonality assessment which can be improved with refined informatics strategies. The tertiary component of genome sequence analysis requires identification of informative disease-gene-mutation associations for extraction of clinicobiological meaning from patient data. The accuracy and efficiency of tertiary analysis is limited by inaccessibility and non-uniformity of this information in both public databases and primary articles. Here we introduce MASTERMIND -a suite of novel analytic tools that dramatically reduces the time and effort required to organize and integrate genomic information from any data source including millions of full-text scientific articles and dozens of heterogeneous variant databases. Methods: To comprehensively interrogate genomic data from both structured and unstructured databases, an automated querying architecture was designed using customized open-source analytics engines and a combination of publicly available and customdeveloped APIs. Curated lists of diseases and gene transcripts with synonyms comprising 11.7K and 50.9K total entries were used as initial query parameters. Custom-designed algorithms were used to generate comprehensive mutation query lists comprising 602M total entries sorted by biological outcome and used as secondtier queries. Results: Using titles and abstracts of 24M primary articles, we identified 909K putative disease-gene associations (average 24 articles per association) which we then confirmed by automated scanning of 5.8M full-text articles to comprehensively identify all disease-mutation citations within each article. This information was then organized according to the strength of the association based on the total number and quality of individual citations and on the position of diseasegene-mutation keywords within the text. Integrated metadata for each finding was used to prioritize disease-gene-mutation associations in accordance with the abundance and quality of supporting evidence. These associations were then organized within a singular graphical UI with display of all relevant information from the primary source material used to drive data prioritization including interactive access to annotated full-text articles. This method of rapidly assimilating diseasegene-mutation associations for display was reproduced for several additional structured databases including the genomic variants contained within ClinVar. Conclusions: We have devised MASTERMIND to rapidly and automatically harvest genomic information from disparate data sources including full-text scientific articles and external databases of genomic variants. This tool rapidly and comprehensively interrogates, organizes and displays genome data and has promising applications in expediting tertiary analysis of human genome sequencing data in clinical assays of individual patients. Introduction: Clonality assessment is an integral part of the diagnosis and follow-up of B and T cell malignancies. The emergence of NGS methods for clonality testing has opened the door to a new era in diagnosis and understanding of clonality. NGS assays allow for the full characterization of clonal populations with higher sensitivity and specificity than traditional methods but the optimal bioinformatics pipelines remain to be established. Here we describe our pipeline and tools for immune repertoire analysis for use in routine clinical testing. Methods: Clonality testing was performed using LymphoTrack assays (InVivoScribe) with adapters for Illumina MiSeq. We have developed a bioinformatics pipeline and data analysis tools to identify the diversity and clonality in samples of patients diagnosed with B or T cell malignancies. The pipeline starts with demultiplexing, adaptor trimming and creating FastQ files from the output of MiSeq (Illumina) sequencer. Paired-end reads are combined and fasta files are created. Identical sequences are removed and only a single sequence is kept in the final fasta file, along with the count in the header. Each unique sequence is blasted using igBlast to search local IGH and TCR databases. IGH and TCR reference data are downloaded from IMGT website. igBlast output is grouped based on different VJ combinations. The generated output summary is loaded into a MySQL database, along with hit details for each sequence. Visualization tools were created for easy pathologist review. Results: A total of 249 samples were analyzed. Clonality calls were made and compared to results generated by the proprietary LymphoTrack Software provided by Invivoscribe. The custom pipeline (MSK-LYMPHOCLONE) was able to analyze and assign same clonality calls in all samples. Visualization tools created for easy viewing and mining of results were made available through web interfaces for analysis during clinical signouts. A run page lists all samples in each clinical run with tabbed links to tools for review and data analysis including: 1) Interactive, zoomable, stacked histogram showing different VJ combinations in different colors; 2) Tabular clone summary interface; 3) A detailed igBlast output for each VJ combination sequence. Multiple selected sequences can be downloaded, or aligned through this interface; 4) A BLAST interface to do sequence search against sample sequences. This tool is used to track a clone sequences in minimum residual disease studies. Conclusions: We have developed and validated MSK-LYMPHOCLONE, a custom data analysis pipeline and suite of tools to mine and analyze next generation sequencing data for IGH and TCRG in lymphoid malignancies. This may be used independently or in conjunction with existing software solutions. The diagnosis and disease monitoring of B-cell lymphomas is complicated by morphological features and surface marker profiles that overlap with non-malignant processes. To clarify these diagnostic conundrums, immunoglobulin heavy chain (IgH) clonality testing is often performed using multiplex polymerase chain reaction (PCR), and subsequent capillary electrophoresis (CE)-based detection of the amplicons. However, non-malignant processes can also be clonal and the interpretation of CE-based assays is subjective. In contrast, multiple studies have reported the association of specific mutations in several immune and epigenetic modifier genes with certain subtypes of lymphoma. Importantly, these mutations are not detected in non-malignant processes, such as infection. Here we report the development of an assay that applies next-generation sequencing (NGS) technology for multiplexed detection of B-cell lymphoma-specific mutations and IgH clonality assessment. The quantitative nature of NGS produces results that can be more objectively interpreted than the current CE techniques. The multiplexed assay is designed to be implemented in small clinical molecular laboratories, using PCR primer-based library preparation. We have designed the assay to answer two common diagnostic questions in B-cell lymphoma: likelihood of lymphoma (determination of clonality) and differentiating Germinal Center (GC) type from Activated B-cell (ABC) type DLBCL (mutations in EZH2 versus MYD88). Methods: PCR primers containing an Illumina P5 forward or P7 reverse adapter fused to gene specific primers (MYD88, EZH2, and IgH) were used to amplify genomic DNA isolated from Oci-LY3 and Oci-LY7 DLBCL cell lines. The PCR products were combined and further amplified using a universal primer set containing a single index and the overlapping P5 and P7 adapters. The resulting DNA library was quantified using an Agilent Bioanalyzer and sequenced on an Illumina MiSeq. Output sequence data was analyzed using fastqc, in which adaptor sequences were identified, and subsequently clipped using fastx toolkit. Alignment to the hg19 reference genome was performed with Burrows-Wheeler Alignment Tool using default settings. Target coverage was assessed using Integrated Genomics Viewer. Results: Approximately 17,500 sequence reads mapped to EZH2, 13,700 to MYD88, and 3,100 to IgH targets. Approximate base phred scores was 39 for all targets. VAF for the myd88 L265P mutation was 100%. VAF for the ezh2 R882 mutation was 50%. Conclusions: NGS technology for multiplexed detection of B-cell lymphoma-specific mutations and IgH clonality assessment offers results that can be interpreted more objectively than can be with current CE methods in the differentiation of malignant and non-malignant processes.
A. Karunamurthy, S. Zhong, K. Callenberg, L. Santana Dos Santos, R. Mitchell, G. Burdelski, J. McHugh, D. Maglicco, A. Parwani, L. Pantanowitz, Y. Nikiforov, M. Nikiforova, S. Roy University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: Next-generation sequencing (NGS) based testing in clinical laboratories have led to generation of massive genomic datasets that creates workflow bottlenecks resulting in suboptimal turnaround times and increased propensity for errors. Modern web technology has allowed efficient and reliable access to NGS genomic data across research and clinical community by leveraging the power of Web APIs. The later is a set of protocols that facilitate software development by exposing useful functions and content for consumption by other applications across a network. We aimed at improving our downstream NGS testing workflow by implementing Web API-based interoperability. Methods: Upon completion of primary bioinformatics analysis, the reported DNA variants from NGS data were annotated using a custom web application, VariantExplorer (VE) (Ubuntu Linux, Django framework v1.5). In addition, VE also detected copy number variations (CNV) and gene fusions using novel algorithms from the raw sequences. Subsequently, all genomic alterations and coverage statistics from VE were asynchronously consumed by a downstream web application SeqReporter (SR) (Microsoft Windows Server, ASP.NET framework v4.5) using web API hosted by VE. JavaScript Object Notation (JSON) format was used for messaging data over a secure and high-bandwidth network (1Gb/sec). The received data was then further annotated and prioritized with our in-house knowledgebase in SR. Results: NGS data from a total of 1906 tumor samples across several tumor types and gene panels were analyzed using the new web API based framework. This new framework substantially streamlined the downstream data analysis workflow by removal of manual transfer and input of genomic data. This resulted in reduced human related errors and yielded very valuable hands-off time for technical staff during the entire process. This also optimized the version control of public and internal databases across the two applications and enabled easy tracking of background analytic processes through a real time monitoring interface. One important challenge encountered was re-validation of the API with every upgrade to the host application (VE). In context of this design, API based messaging poses a potential limitation of scalability to large-scale genomic data transfer. Conclusions: Implementing web based API in our molecular laboratory helped streamlining and optimizing the downstream NGS workflow. Web based API enabled seamless communications across both applications in an operating system agnostic manner with improved performance and promising results. Close monitoring, re-validation with upgrades to host applications, network bandwidth and limited scalability need to be considered when implementing a framework described here. K. Ouyang, J. Tahiliani, Y. Ho, M. Sommargren, M. Jen, B. Bunker, B. Johnson, S. Topper, K. Nykamp Invitae, San Francisco CA Introduction: With improvements in technology, the cost of DNA sequencing has dropped precipitously. At the same time, the amount of data available for analysis and review has increased significantly. Communicating all this information in a succinct, cohesive manner to clinicians and patients can be extremely challenging and time consuming. To facilitate this process, we use a unique score-based evidence system for variant classification named Sherloc and created a framework for describing the relevant information in a clinical genomics report. We've divided the variant details (VD) into four main sections: 1) a molecular description of the variant, 2) evidence that is directly applicable to the variant (eg, population data, clinical and functional data), 3) indirect and predictive evidence (eg, computational analysis), and 4) a final summary. Suggested VD texts were created for different scenarios. Informatics solutions support the automation of these details into the report and metrics are routinely gathered as a gauge to optimize text language and increase future efficiency. Methods: We collected usage data and analyzed the overall frequency of suggested text modifications by each VD section, variant type (missense, frameshift, splicing, nonsense), and classification (pathogenic, likely Results: For any suggested text used across all VD sections, 23.2% was unmodified. Indirect and predictive evidence (29% unmodified) was the least modified VD section; evidence that directly applies to the variant (16% unmodified) was the most frequently modified. When the data was analyzed by variant type, text describing missense variants (27.5% unmodified) was the least modified and frameshift variants (15.0% unmodified) were the most modified variant type. A variant classification of VUS (29.9% unmodified) demonstrated the least text modification, while a classification of likely benign (7.7% unmodified) was the most modified. Conclusions: The data indicate that conveying variant analysis results is still generally reliant on user customization. Our analysis provides evidence that improving the efficiency of reporting variant information is a compelling need in genomics. When these findings are incorporated into informatics planning, further elaboration of the framework may decrease the frequency of modified text across various areas. This optimizes the reporting process of each variant as the analysis and description of data can then be reproducible across users.
D. Iakovishina 1,3 , A. Vachnadze 2 , A. Afanasyev 2 1 Ecole Polytechnique, Paris; 2 iBinom, Moscow, Russia. Introduction: Next-generation sequencing (NGS) allows analyzing whole exome sequencing data, providing us with insights into the landscape of somatic mutations (SVs). Exome sequencing became a cost-effective approach for mutation detection in genetic diseases. Copy number variation detection is one of important part of exome analysis. It can be associated with various types of diseases: Charcot-Marie-Tooth disease, hereditary neuropathy with pressure palsies, Smith-Magenis syndrome and various types of cancer. In contrast to SNPs or indels CNV analysis still remains very challenging. Methods: In our study we compared tools which have been recently presented in prominent journals. By brief comparative evaluation, performance of these tools is very contradictory: all in common they give only 10% of concordant variants. This poses a challenge to clinical geneticist to decide which tool's results are more reliable. Previous tools overviews do not take into account the emerged tools, such as ExonDel or GISTIC2. Therefore, our studies cover this gap and gives comprehensive analysis including methods overviews and recommendations according to the data type. In the study we assessed the ability of 5 computational programs ExonDel, ExonCopy, Conifer, EXCAVATOR, GISTIC2 focusing on short and large CNVs regions predictions. To perform in-depth tools' overview, we prepared whole exome sequencing simulation data. It allows embedding artificial CNVs into an exome and tracking the percentage of detected structural variants and excluding false positive predictions. Exome sequencing method entails multiple errors, such as GC-content (amplification rate associated with gc-rich or gc-poor regions), duplicates reads, pcr-artifacts etc. In practice, these constraints generate noise in the data and represent a major complication for CNV detection tools' performance. To approximate our study to practice, we provided our simulation with these types of errors. We estimated CNV tools on the simulated data to evaluate the parameters and range of the results. To prove the results we performed analysis on the real-datasets. Results: The datasets are 5 clinical exomes with different phenotypes processed with the same technology Agilent SureSelect XT. To confirm real-data results we used microarrays. Based on these results, we identified the most precise tools for both short and large CNV regions. The latest version of ExonDel released on April 28, 2015, showed the highest sensitivity for the large CNV regions. Conclusions: Conifier still shows the least of phase positive results. We compared the tool in terms of the break-point position, number of predicted number of alleles and quality distribution. Also we give suggestions about parameters for each tool that can significantly influence on the results.
C. Hong, E. Zmuda, T.L. Grossman, G.A. Toruner Nationwide Children's Hospital, Columbus, OH. Introduction: In recent years, massively parallel sequencing (MPS) technologies have increasingly been used in clinical diagnostic laboratories. Although this technology expands the throughput and resolution for identifying genetic drivers of disease, the complexity of wet lab and bioinformatics procedures utilized by MPS requires equally sophisticated methods for monitoring the overall quality of the data being generated. As a result, BamQ has been developed to automate a process for clinical assessment of target sequencing coverage. Methods: BamQ is a streamline quality control (QC) workflow composing of two steps; preprocessing and reporting quality metrics. For any given test ordered (e.g., either targeted panel or exome), it jmd.amjpathol.org ■ The Journal of Molecular Diagnostics synchronizes the gene symbols and exon coordinates for the target regions defined by the BED file with either NCBI RefSeq or Ensembl GENCODE genes to ensure that all exons associated with gene targets are captured by the BED file. Then, it analyzes the BAM files to summarize the quality metrics of the coverage in a report; mean read depth, a desired minimum read depth per target coverage ratio, reagent capture efficiency, fraction of regional coverage at both gene level and exon level using user defined minimum read depths, and genomic coordinates that fail to meet the requirement. Importantly, the poorly covered regions are annotated with all necessary information to expedite a follow-up Sanger Sequencing. BamQ also generates pictorial representations of the overall coverage to see the uniformity of mapping profile across a chromosome, a distribution of the fragment length in paired-end sequencing, an area under curve (AUC) plot of the coverage, and coverage comparison with intronic regions around exon boundaries. Results: In our preliminary study, we assessed 16 BAM files (mean depth 170x) generated by Agilent Technologies ClearSeq Inherited Disease kit, Illumina HiSeq 2500, and GenomeNext framework (bwa-0.7.12). We anticipated coding regions and up to 20 base-pair long intronic regions from splice junction to be covered by minimum reads 10x. An average 94.9% of the extended exons (41,096 out of 43,284) are covered across the total covered 3,062 genes. For 5.4% of the exons, Sanger Sequencing filling is needed, since at least one base pair is not covered by 10x. Conclusions: BamQ has been actively developed in consideration with the need of clinical laboratories. The comprehensive QC metrics provides multiple orthogonal validation check points before tertiary analysis and it helps to monitor and trend the reagent performance. BamQ supports parallel computation. BamQ is freely available from http://sourceforge.net/nch_childlab/. Introduction: Clinical implementation of next-generation sequencing (NGS) assays requires analytic pipelines and tools for variant review, decision support, and integration into laboratory information systems. Current solutions range from commercially-available genomic workbenches to laboratory-developed software; however, their use can be limited by cost, the need for programming expertise, or the need for multiple tools. We developed a tool based on widely available software to seamlessly perform variant review, interpretation, and clinical reporting within our workflow. Methods: During implementation of an NGS-based cancer mutation panel using the Ion Torrent PGM platform (Life Technologies) and NextGENe analysis software (Softgenetics, LLC), we developed an informatics solution for our CLIAcertified laboratory. Access 2010, software available with the Microsoft Office (Seattle, WA) productivity suite, is widely used as both a database and database management system. The software was semi-customized using Visual Basic for Applications, Structured Query Language, and macros to allow incorporation into our analysis pipeline. It was optimized using 26 sample data sets and functionally validated using an additional 25 data sets. To examine appropriate data transfer, we compared the variant and quality metric data available in the original NextGENe files, the Access database, and files with exported data. Three molecular pathologists independently performed data import and analysis for precision studies. Results: We developed a customizable software tool that allows single-click importing of variant lists and associated quality metrics from NextGENe; a single-window view of each variant and its associated data fields; direct hyperlinked integration with Integrative Genomics Viewer (Broad Institute) for bam review; custom filtering; efficient variant interpretation with linkouts to public databases (including COSMIC, dbSNP, and UCSC Genome Browser) for decision support; semi-automated primer selection for variants; single-click exporting of annotated results to LIS-compatible template reports; password-protection; and error handling. Data was transferred with 100% fidelity and 100% reproducibility. Conclusions: This system can rapidly adapt to NGS pipelines that output bam and vcf files. Due to the use of MS Access, software development requires minimal programming expertise and can be modified based on the NGS platform, bioinformatics pipeline, and laboratory needs. Using single-click, semi-automated processes can increase usability and potentially reduce copy-andpaste-related errors and assay turn-around time. Historical data may be stored to aid in variant review, and a back-end server may be utilized to prevent reduction in speed and stability. Introduction: Cytomegalovirus (CMV) drug resistance develops in 5% to 10% of patients treated with antivirals for 3 months and is primarily associated with mutations in the UL97 phosphotransferase and CMV UL54 DNA polymerase genes. Genotypic testing for CMV resistance is currently done by Sanger sequencing, which has limited ability to detect minor variants below the 20% level, and often requires manual evaluation of chromatogram. Next-generation sequencing (NGS) provides higher sensitivity for rare variant detection compared to Sanger. We have developed an NGS assay for CMV genotypic drug resistance testing using an amplicon resequencing method on the Illumina-MiSeq platform. This is an adaptation of the test previously developed for the 454 GS Junior. A new bioinformatics pipeline was needed to efficiently handle the higher read yield and modified library organization to generate drug resistance information from sequence data. Methods: We divided the analysis process into 4 modules. In the first module, Fastq reads from individual samples were aligned to CMV UL97 and UL54 reference sequences using Burrows-Wheeler Aligner (BWA). The second module identified overlapping aligned read pairs, merged these read pairs based on the overlap region, assorted individual amplicons by their respective primer positions based on start and end positions, and trimmed the primer sequence. This process partially normalizes the amplicons by discarding excess reads. The third module generated a pileup similar to SAMtool's and called raw variants. The fourth module filtered out potential noise variants based on an observed error model threshold and translated true variants to amino acid mutations. This module also annotated each mutation with drug resistance information based on a local mutation interpretation database. Results: The four modules were integrated to run from a shell prompt to queue the analysis steps for each sample loop over all samples from a run. Benchmarking of the pipeline was performed on a desktop computer with 3Gz Intel core i3 and 4GB RAM. A typical test gives an output of 2.5 GB of compressed read data for 24 samples. The analysis takes 12 hours to complete when run in a single thread mode. Conclusions: We successfully developed an integrated data analysis pipeline for CMV drug resistance mutation detection from MiSeq data, which takes the fastq read files as input and provides a drug resistance mutation report as output. The custom modules are platform-independent python scripts, and can be run using standard desktop computing resources. Parallel/multi-threading can further reduce runtime. This pipeline can be adapted for use with other sequence target by providing reference sequence, primer positions and the mutation database. Approximately 7000 tissue samples are processed per year. As in many academic laboratories, the current manual work flow is labor intensive and prone to human errors such as sample contamination and mix-up. In a concerted effort to maximize efficiency and reduce errors, a multidisciplinary team applied industrial engineering, software development, and automation technologies to create an integrated infrastructure designed to improve patient safety and eliminate non-value added work. Method: Industrialization of the tissue workstation involved three interoperating processes: 1) Lean process engineering, 2) custom designed sample management software applied at the gaps between existing laboratory information systems, and 3) robotic platforms for sample tube labeling and DNA extraction. Industrial engineering methods applied for workflow optimization included defining process flow maps and color coded designated workspaces to create a synchronized sample flow system. Time studies were conducted to establish baselines for performance evaluation. Batch manual sample printing and label confirmation processes were replaced by an on-demand real-time 2D barcode based label printing solution. Software approaches included custom application development as well as barcode-driven, macro-based enhancement of existing process worksheets to replace manual buddy checks. Manual nucleic acid purification was replaced by a Freedom EVO robotic workstation (Tecan, Männedorf, Switzerland), using a custom script designed to eliminate handson pipetting bottlenecks. Results: A total of 10 critical error-prone manual steps out of 14 (71%) were eliminated using an industrialization-driven workflow. This included seven manual steps from sample receipt to microdissection, as well as three manual sample transfer steps. Total sample time from log-in to completed product was also reduced from 600 to 510 minutes, for a 15% reduction. Conclusions: Clinical molecular diagnostic laboratories are under severe time constraints to provide accurate diagnostic results. Sample management and the pre-analytic processes of tissue preparation and nucleic acid extraction are key rate-limiting steps for many labs. A multidisciplinary approach involving process engineering, software implementation or development, and barcode-driven automation has the potential to simultaneously improve safety and productivity in the clinical laboratory.
S. Kadri, I. Mujacic, M. Wurst, B. Long, C. Zhen, N. McDonald, Z. Jiang, L.V. Furtado, J.P. Segal University of Chicago, Chicago, IL. Introduction: Clinical laboratory adoption of next-generation sequencing (NGS) technology has dramatically increased the production of large-scale genomic information. However, because most NGS clinical assays now routinely generate large lists of genetic variants for each patient, laboratories are increasingly faced with a burdensome volume of interpretive and reporting tasks. With many variants to triage for each patient, it is critical that each variant be properly annotated in order to speed review and prevent mis-categorization. Three types of variants are at highest risk for mis-classification: 1) variants with differing functional effects depending on transcript (alternative isoforms), 2) potential splice variants annotated solely as "intronic", and 3) complex variants that are frequently mis-annotated as multiple individual variants. Here, we introduce a toolkit to identify and address these three problematic variant types. Methods: The REMEDY software suite has two components, 1) PlexVar: A tool for reviewing variant positions and pre-emptively identifying complex mutations and 2) TransVar: A decision tree-based tool for streamlining the analysis of individual variants annotated across multiple transcripts. PlexVar analyzes variant lists to detect putative complex mutations present (based on inter-variant distance) and evaluates aligned data to detect true complex mutations with components present in cis. TransVar utilizes ranked lists of transcripts and functional effects to prioritize data presented to reviewers, ensuring that functional consequences are highlighted and that mutations affecting rare transcripts are not missed. Results: To assess performance, we used this toolkit to review variant lists and annotations from 300 Illumina-based NGS clinical cases originally annotated via Alamut Batch software in single longest-transcript mode. PlexVar identified 16 unique complex somatic mutations in this set (100% sensitivity based on manual review). Of these mutations, PlexVar correctly annotated 15 as being present in cis and one as sub-clonal, creating correct new VCF entries. TransVar correctly picked up all notable functional consequences regardless of affected transcript, including 72 potential splice variants (exon boundary +/-6 bp) not adequately flagged by Alamut. Conclusions: Proper annotation is critical in order to support correct reporting and interpretation of variants. The incorporation of these two algorithms into our annotation pipeline has led to a notable reduction in mis-annotation error risk. By generating increased confidence in the accuracy of variant calls and annotations, the software reduces manual review work and allows for high confidence automated filtering of variants based on functional effects.
P. Khalili 1 , J. Gale 2 , M.A. Vasef 1 1 University of New Mexico, Albuquerque, NM; 2 Tricore Reference Laboratories, Albuquerque, NM. Introduction: Targeted next generation sequencing is currently adopted by many CLIA-certified laboratories for detection of clinically significant mutations. The majority of laboratories rely on the primary data analysis software included in the sequencing platform for mutation detection. Recommendations for sequence data analysis are provided by professional organizations; however, currently there is no established guidelines mandated by the regulatory organizations concerning data analysis procedures. During our validation studies for the 50 Gene Cancer Hotspot panel v2 (Life Technologies), we noticed that a small subset of mutations were initially missed by either the software included with the sequencing platform (Integrative Genomic Viewer variant caller) or the second software that we had employed as part of our validation(NextGene). Methods: The sensitivity data for detection of the large (more than 15 base pairs) indels from our validation studies were retrieved for this study. The analytic sensitivity was determined by dilution of known indels of KIT exon 11 from our archived patient samples. Samples with mutated allele frequencies ranging from 1% to 20% were subjected to massive parallel sequencing using the Ion Torrent PGM platform. The data were analyzed by the Integrative Genomic Viewer (IGV) variant caller and NextGene data analysis software simultaneously. Results: The 36 base pair (bp) insertion mutation in KIT exon 11 was only detected in 1 out of 15 samples by the IGV variant caller at allele frequency of 3.6%. However, the NextGene software was able to detect the mutation in all 15 samples with allele frequencies ranging from 3 to 8%. In contrast to the 36 bp insertion, the 54 bp deletion mutation was detected in 10 out of 10 samples by variant caller with allele frequencies of 2 to 17%, whereas NextGene software failed to detect this large deletion in any of the aforementioned samples. Additionally, the 33 bp insertion mutation was detected in 2 out of 15 samples by NextGene at allele frequencies of 16% and 17%, and 0 out of 15 samples by variant caller. Conclusions: Based on our results, no single software could reliably detect all the potential large indels. Therefore, addition of a second software appears to improve the detection sensitivity for large indels.
Farsight Genome Systems, Inc., Sunnyvale, CA. Introduction: Sequencing will soon be an essential tool in the diagnostic workup of solid tumors. Of the more than 700 oncology drugs in the clinical development pipeline, 73% are expected to require a biomarker. Improved software systems are needed to manage the complexity of multiple-marker testing. We set out to build a software system that would reliably deliver concordant results across variations in cancer type, tissue preservation, and target enrichment with high-performance, medical-grade analytics that could be readily validated and integrated into the solid tumor workflow at most pathology laboratories. Here we report findings from an initial verification study. A larger study now underway will be included in the data presented at the meeting. Methods: Fifty four samples, from 5 different laboratories' published data, were chosen to represent a diverse mix of processing conditions and tumor types. The criterion for selection was the presence of one or more actionable variants in AKT, ALK, BRAF, BRCA1, CDKN2A, EGFR, KRAS, NRAS, PIK3CA, PIK3R1 or PTEN. Thirty seven samples were from patient tumors, including lung, colon, esophageal and cancer of unknown primary, of which 18 were FFPE. Nine samples from circulating tumor cells (CTCs) were included, along with a dilution series of 8 cell line samples commonly used for laboratory validation. This study was performed using tumor-only data. The New Software System under evaluation was developed independently, configured with a pre-defined Test Panel of 156 variants, then locked for the duration of the study. Identity-masked FASTQ files were processed as a single batch. The results were unmasked for comparison to the original published source. Results: The New Software System identified all actionable variants in 36 of 37 patient tumors, missing only 1 of 2 variants in a single sample. All of the cell line dilution series were correctly reported. Five of the 9 samples were correctly reported in the CTC series, the remaining samples had 1 missed variant. With read depth below 30x, the missed calls in the CTC series point to inconsistent read depth as the cause for uneven performance in this specimen type. Across all patient tumor samples, successful calls had read depths of 50x to 2800x, suggesting a functional limit of detection of 50x. Conclusions: In this initial verification study, the New Software System demonstrated high concordance with cell line and patient solid tumor samples, both FFPE and frozen. Although a larger validation study is planned, these initial results show promising performance with a diverse mix of processing conditions and tumor types. Here, we employ an alternative PT method to assess BI by exchanging raw data files (FASTQ (FQ)) among multiple institutions all using Illumina's TruSight Tumor (TST) gene panel kit and MiSeq instrument. Methods: Six sites that had validated the TST assay participated in the study. Each site submitted FQ files from 4 previously tested specimens. The 24 FQ datasets (including 3 commercial calibrators) were de-identified and redistributed to 5 of the centers via secure ftp server (one site could not reanalyze datasets). Concordance in reporting of variants and variant allele frequencies (VAF) were determined for the 5 centers. Results: Various alignment and variant calling tools were used across sites including BWA-MEM, GATK, GSNAP, FreeBayes, Clinical Genomicist Workstation, and NextGENe. Excluding calibrator samples, 48 variant calls (16 clinically significant variants (csV) and 32 variants of unknown significance (VUS)) were reported across 21 samples. Overall, 68.8% of total variants (33 of 48) were concordant between all sites; this included csV (11 of 16) and VUS (22 of 32). There was 100% concordance in reporting single-nucleotide csV (cs-SNV).
% to 33.5%) as determined by digital droplet PCR were also analyzed. All reported values from the calibrators had VAFs within 10% of the target value; 78% of calls were within 5%. No sites reported VAF <5% in the calibrator samples, consistent with each site's reportable range. In contrast to SNV reporting, the reporting of csV insertions/deletions (indels) only had 28.6% concordance (2 of 7); discordant indels included two EGFR exon (ex) 19 deletions (del), one EGFR ex20 insertion (ins), one KIT ex11 ins, and one ERBB2 ex20 ins. Three FQs datasets had an identical EGFR ex19 del; 2 were reported with 100% concordance; the third was only identified by 2 of 5 sites at VAFs of 14 and 18%. The same 2 concordant sites also reported the KIT ins not detected by the other 3 sites. Conclusions: This report highlights the need for BI PT for clinical laboratories. The concordance for reporting cs-SNV was acceptable for the first implementation of this PT program; however, there was low concordance in reporting csV indels. The source of these discrepancies requires further evaluation to determine how best to implement corrective action. This alternative PT program assessing BI identifies areas for laboratory quality improvement in NGS. Introduction: Bioinformatics analysis and the quality control process in Precision Genomics goes beyond the basic guidelines set by the American College of Medical Genetics (ACMG). Our bioinformatics quality control results validate the upstream wet bench process and indicate a high percentage of clinically actionable mutations. Intermountain Precision Genomics is a full service laboratory that provides cutting edge quality metrics in clinical reports delivered to ordering physicians. Methods: At Intermountain Precision Genomics we sequence all exons and flanking introns of 96 genes that play a role in known cancer pathways. Samples are sequenced along jmd.amjpathol.org ■ The Journal of Molecular Diagnostics with a cell line control. After samples are sequenced the raw data are captured and the bioinformatics quality control is performed. All samples are pre-processed by removing the longest primer sequence from the 5' end to prevent false negatives. A quality trimming of Q-30 is performed on the paired end reads. Paired reads are aligned using the Burrow Wheeler Aligner (BWA) and aligned reads are merged using the SAMTools suite of algorithms. For each patient sample, four quality metrics are evaluated. Sensitivities and specificities are calculated by comparing variants found in the controls to known variants captured by external databases. Results: The minimum threshold for the base calling accuracy and the clusters passing filters is set at 85%. The mean depth of coverage and the quality threshold are generated using the Picard toolkit. The minimum thresholds for mean depth of coverage is 70X and the average depth of coverage across validation samples is 286X. The quality threshold for exons covered at 10X was at 90%; therefore, adhering to ACMG standards. These thresholds were calculated by statistically evaluating our validation samples which included 100 patient samples and cell line controls, including NA18507 and NA12878. If minimum thresholds for bioinformatics quality are met, 80% of our clinical reports contain actionable mutations. Conclusions: At Intermountain Precision Genomics, we are using a novel approach of generating quality control reports that validate the clinical laboratory protocol. The precision genomics test offers higher sensitivity, coverage, and a superior ability to identify actionable mutations. Introduction: Identification of key variants using WES in patients suffering from rare Mendelian genetic disease is a complex task. Although WES enables detection of an enormity of variants without a prior suspicion of particular disease-causing genes, there remains a necessity to focus on relevant regions of the genome to facilitate timely data review. To do this, one may focus on specific genetic targets informed by the patient's disease or phenotype. There are a number of emerging tools that automate this task, and this study assesses the strengths and weaknesses of these tools in the context of a diagnostic odyssey patient review. Each tool was evaluated with phenotypes from 4 patients previously studied with WES. Methods: We identified tools that accepted phenotype terms and outputted a list of associated genes. Tools were assessed for underlying method, ease of use, data type/source, and if based on a curated subset of data, how recently it was updated. Tools were classified by the underlying data source for gene list generation and subsequent phenotype-to-gene results compared within and between tool classes. The tool similarity was assessed using unsupervised hierarchical clustering with the metric of Euclidean distance for the gene list and average linkage criteria for each phenotype term. We also evaluated the general performance of multi-phenotype verses single phenotype queries. Results: We identified 11 tools able to generate gene lists from phenotype terms. Phenomizer and Phenolyzer use the Human Phenotype Ontology and return known disease-causing genes using OMIM or other curated databases. GLAD4U, Geneprospector, SNPs3D, and Genie use literature mining as the primary methodology. FindZebra, Biograph, Polysearch, DisGeNet, GeneWeaver, and Phenolyzer use a more integrated data mining approach using varied sources for evidence. We observed extreme variability in the number of genes generated per phenotype and clustered the tools by gene list similarity. Based on these results and observations of single verses multi-phenotype querying, we suggest a generalized framework for tool integration in a clinical diagnostic setting. Conclusions: The plethora of publically available tools able to identify genes relating to phenotype terms differ in search methodology, input terms, output data, and user friendliness. Despite this variability, when used in concert these tools can identify relevant gene lists for variant interrogation associated with diagnostic odyssey cases. This study should provide end users with an understanding of tool function, how these tools could be used for generalized or specific phenotype queries, and guide future integration into analysis strategies for diagnostic sequencing testing using a structured tiered approach.
BioDiscovery, Inc., Hawthorne, CA. Introduction: Targeted panel Next-generation sequencing (NGS) has been used in research and clinical testing labs worldwide. This technology takes advantage of the single base pair resolution of NGS and provides a cost effective way to focus results on clinically actionable genes by sequencing a limited number of targeted regions in genes that have a clear diagnosis, are actionable with prescription drugs or compounds in clinical trials, or have known impacts on prognosis or outcome. Until recently, DNA microarray technology has been the best technology of detecting and analyzing copy number events; labs with this technology usually run the same samples simultaneously on targeted NGS to detect sequence mutations and on DNA microarray to obtain the coordinating copy number information. This requires extra cost, labor, and time; as a result, many labs do not use microarray technology anymore. Getting copy number information from targeted panel NGS and combining it with sequence mutations is thus extremely valuable. Methods: We have developed a method called BAM Consensus which only requires the loading of BAM files and targeted panel design files to generate copy number results. In combination with the VCF files that contain the sequence mutations, an integrated analysis of both events can be easily carried out. Results: BAM consensus was applied to targeted sequencing panels in cancer and constitutional sample sets. Differences in overall read-depth resulted in variable sample quality across the cohorts; however most sample quality was adequate for copy number estimation and a quality threshold was assessed. Results indicate that relative copy number can be estimated for targeted regions comparable to the results achieved with microarray for the same targeted regions. Additionally, some arm length changes can also be estimated in genomic areas with wider targeted coverage. Integration with VCF files identified several samples with co-occurring copy number and sequence variant changes in targeted regions of interest. Conclusions: Targeted panel NGS is especially useful in detecting somatic mutations in known oncogenes and tumor suppressors in cancer and has been the major application of this method in the field so far. However, especially in cancer cases, combining the somatic mutations and copy number events in the same genomic regions or genes are crucial to a comprehensive diagnosis and creating an effective treatment plan. In constitutional samples, this provides a second tier of analysis for genes of interest, especially when no clear sequence mutation is detected. We have tested BAM Consensus in a variety of cancer and constitutional samples and it proves a very easy and straightforward way for targeted panel analysis. Introduction: Laboratories with limited resources may lack an adequate bioinformatics pipeline to analyze and report NGS results. Our study aimed to develop an affordable solution for reviewing clinical NGS data. Methods: Variant caller (VCF) files were generated by the Variant Caller (VC) plug-in provided with the Ion PGM sequencer (high stringency) and 3rd party software, NextGene (NG) by Softgenetics (low stringency). VCF files, patient demographics, and NGS QC metrics were transferred to software developed in Excel Visual Basic. Analysis consists of two automated macros, prompted by user input. The first script checks rolling databases to annotate and sort raw data based on various filters, including mutation frequency (<1%), coverage (<500x), known SNPs, and sequencing artifacts. VCF files from VC and NG are compared and common variants are identified. A manual review of nucleotide sequences is performed for a predefined subset of genes based on tumor type. Technologists evaluate the annotated data and verify the mutations to report. The second script then compiles a comprehensive report, based on mutation status and patient demographics, from lab developed, user accessible databases. To ensure technical accuracy, 131 known positive cases were analyzed using the bioinformatics pipeline and clinically relevant findings were validated via Sanger Sequencing. Cases consisted of pancreatic neoplasm (34%), lung carcinoma (15%), tumor cell line (15%), desmoids tumors (10%), myeloid neoplasm (9%), melanoma (7%) colon carcinoma (4%) and GIST (3%). Results: All clinically relevant mutations (100%) were detected by the NGS bioinformatics pipeline and by Sanger Sequencing. On average, VC reported 22 calls per case, whereas NG reported 86 calls per case. An average of 3 variants per case was common to both VC and NG VCF's. On average 2 variants per case (67%) were identified as clinically relevant somatic mutations and 1 variant (33%) were rare hereditary germline single nucleotide polymorphisms (SNP). These results account for 9% of total VC findings and 2% of total NG findings. Of the total variants per case, an average of 24% were identified as common SNPs, 8% as confirmed sequencing artifacts, 42% as low frequency artifacts (<1% allele frequency), and 2% as variants with low coverage (<500x), ultimately filtering approximately 77% of the raw data. Conclusions: This bioinformatics pipeline is an affordable and accessible alternative to commercially available products or bioinformatics team. The combination of two software packages into one pipeline reduces the likelihood of error and ensures accurate reporting. Furthermore, automation of processes such as data sorting and report compilation reduces resources needed to analyze clinical specimens. Introduction: Next-generation sequencing (NGS) has rapidly replaced more traditional molecular techniques for somatic mutation detection in formalin-fixed, paraffin-embedded (FFPE) tumor tissue. Somatic mutation detection covering an increasing number of cancer-related genes helps identify targeted therapies that are likely to work against a cancer patient's specific tumor. Although these types of NGS panels often identify targetable mutations, no actionable mutations are identified in many of the cases tested. To increase the yield of actionable findings, somatic copy number changes can be assessed by targeted methods (FISH) or genome-wide testing such as chromosomal microarray (CMA). We evaluated the feasibility of using NGS amplicon depth of coverage data analysis to identify changes in copy
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org number in routine clinical specimens subjected to a 50-gene NGS panel. Methods: Following routine clinical sequencing of FFPE DNA from a variety of tumor types with the 50-gene Ion AmpliSeq Cancer Hotspot Panel v2 (CHPv2) (Thermo Fisher Scientific), NGS amplicon coverage data was generated using the Torrent Suite (v.4.02) Coverage Analysis plugin (v4.0). For each chip sequenced, a coverage summary file was generated detailing the depth of coverage for each of the 207 amplicons. Coverage data of 50 samples from six separate CHPv2 runs was combined. For each amplicon of each sample the percent of total coverage was calculated and normalized to the mean percent coverage for each amplicon so that a normal copy number at a given locus would have a normalized value of ~1.0. Various methods were evaluated for establishing normalized coverage value thresholds above and below normal. Samples with predicted copy number changes were subjected to OncoScan FFPE Assay (Affymetrix) and/or FISH for confirmation. Results: Copy number gains (gene amplifications) and/or losses (likely homozygous deletions) of genes included in the 50 gene NGS panel were predicted in 10/50 (20%) of samples included in this analysis. Six of these samples with DNA isolated from tissue with >50% tumor content and at least 80 ng available were selected for confirmation with OncoScan. Predicted copy number gains were confirmed in ERBB2, EGFR (n=2), 4q12 (including KIT and PDGFRA) and FGFR3. Predicted copy number losses were confirmed in PTEN and CDKN2A (n=2). Amplification of ERBB2 was also confirmed by FISH. Conclusions: Amplicon-based NGS panels such as CHPv2 that are designed to detect point mutations and small insertions and deletions may also be used to detect larger somatic copy number changes such as amplifications and homozygous deletions spanning one or several genes. Additional validation would be required before introducing this type of analysis in routine clinical testing. Introduction: Interpretation of missense variation is a central challenge in the use of whole exome sequencing for the diagnosis of inherited disease. Genes accumulate missense variation at different rates, making it difficult to interpret whether the presence of a rare missense mutation in a known disease-causing gene is a likely explanation for a patient's condition. We have investigated using measures of the tolerance of each gene for missense variation as a tool to inform variant scoring for clinical exome sequencing. Methods: We re-analyzed variants of uncertain significance (VUSs) identified in genes associated with human disease, as defined by the Human Genome Mutation Database (HGMD) in ~500 patients who have undergone whole exome sequencing as part of the NCGENES (North Carolina Clinical Genomic Evaluation by Next-generation Exome Sequencing) study. We developed a model incorporating four factors: 1) the Residual Variation Intolerance Score (RVIS) calculated for each gene based on data from the Exome Aggregation Consortium (ExAC), 2) the ratio of rare to common missense variation in the gene, 3) the ratio of non-synonymous to synonymous variation in the gene, and 4) the proportion of pathogenic variation due to missense changes in each gene in HGMD. Results: We found that the probability of a variant call being pathogenic or benign in a given gene was well correlated with that gene's tolerance for missense variation. A model considering multiple factors was more predictive of NCGENES benign or pathogenic calls than RVIS alone, rare to common missense variant ratio alone, or the ratio of non-synonymous to synonymous changes alone. Using the probability score generated for each gene by this model, we re-assessed all variants that had been called VUS by manual review in the NCGENES cohort. We determined that 223/1951 (11%) of VUS calls had a greater than 95% chance of being pathogenic based on our logistic regression model, whereas no VUS calls were made for missense variants with a less than 5% chance of being pathogenic. The number of VUS calls potentially influenced by considering missense tolerance varied greatly by indication for testing. Indications with a high proportion of variants with informative missense tolerance scores include dysmorphology (31%), cancer (19%), and intellectual disability/autism (16%). Conclusions: Using these criteria, this model provides support for re-categorization of 11% of VUS calls in the NCGENES cohort, all with increased evidence of pathogenicity. When combined with other scoring criteria, a model of missense variation tolerance such as this one can provide information to aid variant calling in the context of whole exome or more directed gene sequencing, and might be useful to prioritize variants for review. Introduction: TruSeq amplicon-based targeted sequencing assays running on the MiSeq mid-range sequencing platform (Illumina, San Diego, CA) have been successfully deployed in many clinical oncology laboratories. The HiSeq platform can sequence more samples at a time, potentially giving much higher throughput than the MiSeq platform for high volume laboratories. However, this platform does not currently provide software for analyzing HiSeq data generated using TruSeq chemistry. Here, we describe the development of a pipeline for running TruSeq based cancer panels on HiSeq sequencers that can enhance scalability and also serve as a complementary pipeline for new panels in development. Methods: A bioinformatics pipeline using open-source software including BWA-MEM, Picard, Samtools, bamUtil, GATK, Pindel, and in-house scripts was developed. Thirty nine patient samples previously run on a MiSeq-based clinical assay were tested using a HiSeq sequencer, using challenging clinically reported mutations as the targeted gold standard. Parameters in each step of this pipeline were fine-tuned in order to optimize concordance with clinically reported mutations. Testing was performed on a high performance computing cluster (HPCC), with workflows built in IBM Platform Project Manager and published on IBM Platform Application Center interface (IBM, Armonk, NY). The workflow was deployed on a Redhat Enterprise Linux cluster with Intel 2.6 GHz compute nodes of 24 CPU-cores and 384 GB of RAM each (Red Hat, Raleigh, NC). Results: Use of HPCC and the customized pipeline resulted in approximately a 10-fold increase in computational throughput. 83% (39/47) of expected variants were called in the initial custom HiSeq pipeline. Parameter optimization resulted in an additional 11% (5/47) expected variant calls, with 6% (3/47) of expected variants remaining uncalled. The veracity of the uncalled variants is being investigated using additional orthogonal methods. Conclusions: The HiSeq pipeline we developed was able to call variants from TruSeq libraries with high but not complete concordance compared to previously run clinical MiSeq assays. Moving to a high-performance computing cluster along with HiSeq sequencing offers substantial gains in terms of throughput, computing power, and increased storage capacity.
Introduction: As next-generation sequencing (NGS) gains acceptance as a powerful tool for the management of cancer patients, many laboratories struggle with how to efficiently evaluate and report NGS variants. Manually filtering and annotating variants can be challenging and time consuming. Alternatively, the cost of developing and implementing a custom bioinformatics pipeline and/or purchasing a commercial variant management tool can be prohibitive. Although a do-it yourself (DIY) approach to a genomic knowledgebase can be more economical, this advantage may be obviated if time is squandered with a haphazard approach. Furthermore, for the resulting data to be useful for efficiently reporting molecular results, the data must be organized and stored in a way that is rapidlyretrievable and is easily exported into a molecular report. Methods: Here, we outline a step-by-step approach to building a custom genomic knowledgebase for annotating and reporting NGS variants using data obtained from freely available online resources. We also provide specific examples of how this process has been applied to the management and reporting of NGS-derived somatic mutation data from a custom-designed myeloid malignancy 43-gene hotspot panel. We have found the following 7 steps to be useful in developing and maintaining our DIY knowledgebase: 1) Compile a list of genes to be included in the knowledgebase; 2) Determine which data elements are relevant for the gene(s) of interest -for example, protein function/domains, putative mechanism of tumorigenesis, targeted therapies, diagnostic/prognostic significance, etc; 3) Decide the preferred software format (ie, spreadsheet, Word document, database, etc.); 4) Create a template for organizing the required data elements; 5) For each data element, identify an appropriate public resource from which to extract the applicable data -for example, protein domain map from UniProt.org, reported variants from Ensembl.org, etc; 6) Incorporate the information into the applicable data fields of the reporting template; 7) Make a plan for (and a commitment to) data management and timely updates. Results: This approach can be used for reporting variants detected by NGS, and is easily scalable for single-gene assays or for large targeted NGS panels. Conclusions: Although we designed our knowledgebase to address somatic mutations in cancer, this approach can be used to manage microbiologic or inherited disease NGS data, or be expanded to include data from other multi-gene analytic methods (eg, comparative genomic hybridization, expression profiling). The highly multiplexed nature of current NGS platforms may lack the analytic performance to achieve accurate sequencing in certain genomic regions giving rise to erroneous variant calls (VC). In our experience, there are certain positions where artifacts occur at greater than the critical 5% alternative allele frequency (AAF) sensitivity threshold. Here we establish a procedure to identify these problematic positions. Methods: For this study, we include solid tumor samples sequenced using the Illumina TrueSeq Amplicon -Cancer Panel. VCs included had a minimum of 250X coverage, quality filter. VCs were aggregated by amplicon and position to obtain the number of samples in which that variant occurred (variant occurrence fraction -VOF) and the median AAF (MAAF) per position. To identify potential artifacts we identified variants jmd.amjpathol.org ■ The Journal of Molecular Diagnostics that had MAAF below the MAAF for mutations reported with high confidence (HCRM) and a VOF above that of the HCRM with highest VOF. Results: There were 315 cases available for analysis. A total of 666 different variant calls were identified of which 207 (31.08%) consisted of HCRMs. The median (IQR) number of variant calls per sample was 2 (1 to 6). Excluding common population variants, the remaining hese, 7 were present in >12.38% of the samples (maximum VOF of a HCRM [KRAS and below the median of HCRM (28.19%) and were considered likely artifacts (Figure 1 ). The median AAF of these VCs was 6.26 and included calls in the following positions: ABL1 (chr9:133750487), ATM (chr11:108225628), GNA11 (chr19:3114893), GNAQ (chr9:80343587), HNF1A (chr12:121432117), RB1 (chr13:48955657), STK11 (chr19:1220518). Conclusions: NGS technology allows for reliable high throughput sequencing of solid tumors. However, interpretation of variant calls, especially at AAF near the specified assay sensitivity has proven challenging due to the common finding of recurrent sequencing artifacts. We believe that the presence of these false positive calls may be related to the panel design and primer set used and less likely due to the sequencing platform or bioinformatics pipeline, and thus identification of these potential artifacts should be panel specific. Cataloguing these potential artifacts may aide in discerning real variants from potential artifacts when rendering an interpretation of the mutations found in a cancer sample. Introduction: In the clinical laboratory setting, interpretation of next generation sequencing (NGS) findings can be challenging, especially for equivocal variant calls that do not meet clinical laboratory-established criteria for a true mutation or an artifact. With experience and parallel testing using orthogonal methods such calls can be accurately categorized as being real versus a platform/sequencing artifact. We summarize our single institution experience with confirmatory Sanger and/or pyrosequencing testing for clinical NGS testing. Methods: A total of 11,000 solid tumor samples (FFPE, FNA smears) with >20% tumor that received clinical molecular testing were included in the study. Library preparation was performed using the Ion AmpliSeq Cancer Panel covering 46 genes (04/2012~08/2013) and the Ion AmpliSeq Cancer Hotspot Panel v2 covering 50 genes (08/2013~current). Sequencing was performed on the Ion Torrent Personalized Genome Machine (PGMTM). Human genome build 19 was used as the reference. Alignment to the reference and variant calling was performed by Torrent Variant Caller. Laboratorydeveloped Sanger sequencing (covering all 207 amplicons on 46/50 gene panels) or pyrosequencing (selected hotspots) were used as confirmation methods. A sequencing coverage of 250× (bidirectional) was regarded adequate. Results: A total of 426/11,000 (3.9%) cases with equivocal variant calls were also tested by orthogonal tests. The indication for most (368/426, 86.4%) of the cases was to confirm suspected variant calls such as yet unreported sequence variants, large deletions, duplications, and indels in homopolymer regions. Sequence variants were confirmed in 231/368 cases (true positive, 62.8%). Orthogonal methods did not show suspected sequence variants in 113/368 cases (false positive, 30.7%). These included single base-pair deletions in homopolymer regions, sequence variants detected predominantly in one strand and cases with mutant allelic burden <20%. In 24/368 cases, sequence variants were missed by the variant caller program, but detected by manual inspection and confirmed by orthogonal methods (false negative, 6.5%). The remaining (58/426, 13.6%) cases were due to inadequate NGS coverage in genes of interest, which showed no mutations upon orthogonal testing. The rate of requests for confirmatory testing declined with increasing user experience. The artifacts were annotated in the laboratory-developed software for future use. Conclusions: Confirmation of equivocal calls using orthogonal methods, especially early in the implementation of NGS testing in the clinical setting, allows accurate identification and cataloging of true mutations versus artifacts. These studies provide a useful resource for future training, sign-out and competency assessment. Introduction: A 32-gene-all-exon myeloid panel has been designed and validated in our lab using Illumina Miseq platform. The data has been analyzed using Miseq Reporter and Variant Studio. The validation strategy, test performance, and the lessons learned during the validation are shared here. Methods: 1) A 32-gene panel was created by working with clinicians specialized in myeloid diseases.The panel is developed using Illumma's TSCA technology and consists of 1275 amplicons. 2) 22 patient samples previously tested by a reference laboratory, A mixture of cell line DNAs from Coriell Institute, Horizon Diagnostics with known variants and 21 samples with known indels from Moffitt Cancer Center TCC Archive were tested.The DNA was extracted using QiaCube. The DNA was measeured on NanoDrop and the libraries were quantitated using TapeStation (Agilent Technologies).A total of 124 samples were tested over 31 runs on Miseq instrument (reagent kit v2) with 4 samples per run. 3) The data has been analyzed using Miseq Reporter and Variant Studio. The coverage of all amplicons, variants, and variant allele frequencies (VAF) were evaluated. Results: 1) Coverage: Among 1275 amplicons, 1239 amplicons have a median depth of coverage greater than 200X; 1205 greater than 1000X. In over 90% of the samples, the depth of coverage is within the range of 0.5 to 1.5 median coverage; 2) Sensitivity and Specificity; 3) Cell line results: The cell line mixture has 500 variants within the targeted area of this panel. 500 variants (VAF from 5% to 100%) are expected to be detected by this panel. Overall sensitivity = 92.94%. Largest deletion detected correctly is 23 nucleotides in ASXL1 gene, and largest insertion detected correctly is 20 nucleotides in NPM1 gene; 4) Patient Sample Results: Our in-house test results of 22 patient samples were compared with the results of a reference lab. Both sensitivity and specificity are greater than 99%; 5) Variant Allele Frequencies: Overall VAF correlation is 98.717% and R2= 0.8959 for the variant detected respectively. Conclusions: Our panel covers all exons of 32 genes. It consists of 1275 amplicons with a total size of 236066bp. When it was run using Miseq flow cell with 4 samples per flow cell, it offers a median coverage of 7300X. The VAF comparison analysis shows that variant call results are reliable when the coverage is 200x or more and the allele frequency is 5% or higher. The lesson learned in this validation is that all exon coverage may not be necessary. Some exons are not important, but covering these exons drains the resource that may be used to cover the clinically relevant exons. Introduction: The NextSeq 500 sequencer (Illumina) is advantageous for tumor molecular profiling by Next Generation Sequencing due to increased output enabling high read depth, yet with run times similar to lower volume sequencers such as MiSeq. The NextSeq 500, however, uses a new two-channel technology where two filters are used for image acquisition with a mixture of dyes used to detect all four nucleotides. We present an evaluation of the NextSeq 500 for suitability for tumor molecular profiling, by comparison to MiSeq. Methods: Performance of the NextSeq 500 was assessed by comparing results of 155 libraries prepared from formalinfixed, paraffin-embedded (FFPE) tumor or matched blood samples using the TruSeq Amplicon Cancer Panel (TSACP, Illumina) sequenced on both the NextSeq 500 and MiSeq. Assay-specific quality metrics included reads on target, coverage >500x in regions of interest, aligned reads, and variant comparison. NextSeq 500 performance was also assessed by run quality metrics (cluster density, clusters passing filters, data output, read quality) using control samples (PhiX). Results: Assessment of assay-specific quality metrics from 155 libraries tested by both NextSeq 500 and MiSeq (TSACP) showed comparable reads on target (NextSeq: 96.9+0.9% SD; MiSeq: 97.5+0.7% SD), aligned reads (NextSeq: 95.5+2.2%; MiSeq: 96.5+2.6%) and regions of interest coverage meeting lab-defined threshold of >500x in >85% of the library (153/155 samples for both NextSeq and MiSeq). Two hundred eleven variants identified by MiSeq (from 93 positive cases out of 155 libraries total) were also detected on NextSeq 500. Two variants at allele frequency of ~10% were detected on NextSeq 500 but not MiSeq, likely due to higher read depth. Variant allele frequencies were comparable between the NextSeq 500 and MiSeq (R2 > 0.98, n=211). For assessment of NextSeq 500 run quality metrics, use of PhiX demonstrated acceptable cluster density (target: 129 k/mm2 to 165 k/mm2, achieved: 111 k/mm2 to 227 k/mm2), clusters passing filter (target >80%, achieved >85%), data quality scores (target: >75% bases >Q30 with 2x150bp, achieved: >80% bases >Q30), and data output (mid-output mode target: 32.5 to 39 Gb with 2x150bp, achieved: 24 Gb to 42 Gb; high-output mode target: 100 Gb to 120 Gb, achieved: 80 Gb to 120 Gb). Conclusions: The NextSeq 500 sequencer's twochannel sequencing by synthesis technology provides similar quality metrics for assay and run-specific measures for tumor molecular profiling, as compared to the four-channel sequencing technology of the MiSeq, and comparable variant results from tumor molecular profiling. Introduction: With increased demand for molecular testing for mutations in multiple genes associated with treatment response to targeted therapies, prognosis, and diagnosis in solid tumor and hematologic neoplasms, we have previously validated all 26 genes of the Illumina TruSight Tumor (TST) and selected regions of 21 genes on the Myeloid (TSM) panels within a CLIA-certified laboratory, with customized frameworks for reporting results to clinicians. Since sequencing reagents comprise a large portion of per-test costs, we examined the effect of pooling both library types on sequencing coverage and quality to determine whether it could be used to reduce costs. Methods: Archived libraries from previously tested samples for routine clinical testing (16 TST and 6 TSM) were selected for validation. Each previously prepared library had a unique index pair. Libraries were sequenced using V3 chemistry on an Illumina MiSeq instrument. Two diluted amplicon libraries (DAL) were prepared for each run, 1 for TST samples and 1 for TSM samples, and combined according to an Illumina-supplied protocol. The first run combined 8 TST libraries (4 samples) and 4
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org TSM libraries for proof of concept. The second run combined 16 TST libraries (8 samples) and 6 TSM libraries to simulate an average volume of samples received a week within our laboratory. Analysis of samples was performed by first running analysis based on a sample sheet including only TSM samples, using the Custom Amplicon analysis workflow and a 2x150 bp read length. Analysis was then requeued using a sample sheet including only TST samples, using the Amplicon-DS analysis workflow and a 2x121 bp read length. Results: Overall, 100% of clinically validated amplicons had depth of coverage of at least 1000 reads for TST samples on both runs, and >99% of clinically validated amplicons on the TSM panel had depth of coverage of at least 500 reads. The average coverage on the proof of concept run was 42246 reads per amplicon (range 3607-160247) for TST samples and 5314 reads per amplicon (range 74 to 14966) for TSM samples. The second run simulating clinical throughput demonstrated an average coverage of 261846 reads (range 1393 to 163197) per amplicon for TST samples and 5514 (range 74 to 2010) reads per amplicon for TSM samples. Variant calls demonstrated 100% concordance with previously run samples for clinically validated regions. Running libraries together saved $1442 per v3 cartridge not used. Conclusions: Pooling of TruSight Tumor and Myeloid Panel libraries is an effective way to reduce sequencing costs and allow low to medium throughput labs using these kits to provide somatic mutation testing in a more cost effective manner. Introduction: HIV-1 drug resistance testing to guide optimal antiretroviral therapy relies on up-to-date interpretative algorithms and user-friendly software applications available to clinical laboratories performing such testing. Methods: Thirty (30) clinical plasma specimens belonging to a cohort of treatment-experienced, HIV-1 infected patients were tested with the FDA-approved ViroSeq HIV-1 Genotyping System, version 2.0 (VQ; Abbott Molecular, Inc.) for genotypic mutations in the protease (PR) and reverse transcriptase (RT) regions of the HIV-1 genome. Sequences generated were analysed and compared for resistance interpretation between ViroSeq HIV-1 Genotyping System Software version 3.0.1 and a FDA-registered data processing module (DPM) software application (Advanced Biological Laboratories S.A.) which includes the HIV-1 drug resistance (DR) interpretive databases such the Stanford HIVdb and the virtual-phenotypic-based interpretative system of Geno2Pheno (G2P). Results: A total of 570 individual drug resistance interpretations were generated from the 30 specimens. On an average, HIVdb showed resistant (R) results (27.2%) more frequently than VQ (25.8%). VQ was more likely (64.4%) to show susceptible (S) results than HIVdb (53.7%) and G2P (56%). Compared to VQ results as the reference, HIVdb yielded 73.3% "positive" minor discordance (S versus intermediate or I, or I versus R), 10% negative" minor discordance (I versus S, or R versus I), 3.3% of "positive" major discordance (S versus R), and no "negative" major discordance (R versus S). Among specimens showing same interpretation results between SG and G2P, 50% showed different results for VQ: 50% "positive" minor discordance (S versus I, or I versus R), 6.7% "negative" minor discordance (I versus S, or R versus I), 3.3% of "positive" major discordance (S versus R), and no "negative" major discordance (R versus S). Differences were observed for abacavir (45%), atazanavir (17%), didanosine (10%), tenofovir (7%), stavudine (3%), zidovudine (3%), efavirenz (3%), nevirapine (3%), saquinavir (3%), amprenavir (3%), darunavir (3%), and nelfinavir (3%). Conclusions: Different interpretations were observed among the interpretative software applications and databases currently available for HIV-1 DR testing. User-friendly interpretive software applications and continually updated databases are required to reliably report DR results for optimal antiretroviral therapy. Introduction: Barcoding and automated labeling solutions represent powerful tools for the busy laboratory to reduce technologist work and human pattern recognition errors. A limiting factor in implementation can be the lack of easily accessible or customizable software to support existing laboratory workflow approaches. For a major sample automation project in our laboratory, we identified a number of software development needs which could be facilitated through a common, scalable toolkit for sample management. Methods: Hardware components of our solution include single sample readers (Symbol bar code scanners, Motorola, Chicago, IL), 96-well rack readers (Perception Rapid Rack Reader, FluidX, Nether Alderley, UK), and an automated rack sample storage tube labelling system (XTL-96, FluidX). Custom software libraries were developed to provide an easily accessible application programming interface (API) to these three device classes. Wrapper classes for devices were written in C# using Visual Studio 2010 (Microsoft, Redmond, WA). These wrapper classes hide most of the underlying complexity of writing device level drivers, which for this set of devices included USB human interface device class calls, TCP/IP remoting, serial port manipulation, and printer stream dumps. A custom software application, MDL Sample Manager, was written in C# to fill gaps in our laboratory informations systems and provide walk-away derivative sample tube generation from previously scanned parent tube manifests, and to automatically map laboratory-specific sample identifiers to bulk-rack scannable preprinted tube level DataMatrix codes. Results: Initial software design testing identified a number of important desiderata for sample tube labelling and management. These include 1. Ability of software to transform a linear array of samples into a two-dimensional rack output using variable ordering strategies, 2. Ability to perform barcode based sample order validation either at the time of initial manifest generation or subsequently as a quality check, 3. Support for keyboard/mouse/display free operation via clear auditory feedback and maximizing barcode drivability, and 4. Ability to provide scalable asset management via easy definition of storage locations and rack level "check-in/check-out" of samples.Conclusions: We are releasing our device level application programming interfaces as open source software, as well as example code showing prototype implementations and usages. We believe wider spread implementation of these technologies in the clinical lab will be associated with improved workflow and sample fidelity. Introduction: Advances in our understanding of the genetic basis of human diseases has led to increased demand for molecular genetic testing. This demand has led to skyrocketing reference laboratory costs that are a challenge for hospital laboratory budgets, as reimbursement rates cover a fraction of the cost. Our institution has also experienced a dramatic rise in volume and cost of reference laboratory testing over the past decade. In fiscal year (FY) 2008 our reference laboratory volume comprised 8.0% of total laboratory test orders and increased to a peak of 13.8% in FY2012. Reference laboratory costs rose from 4.9 million dollars to 7.9 million dollars annually during that same period. To manage costs and improve quality of care, we developed a reference laboratory utilization management review process. Methods: To address the significant rise in volume and cost, we used a three-pronged approach consisting of reference laboratory consolidation, internalization of testing that could be performed in-house for a lower cost, and providing educational resources and test selection guidance via a multidisciplinary utilization review team. Results: Review of our send-out tests allowed for consolidation to fewer laboratories and negotiation of lower pricing, with no reduction in send-out laboratory quality or service. Test internalization included chromosomal microarray, requiring capital investment, and additional high volume molecular tests that could be internalized using existing equipment. A multidisciplinary team of 3 PhD Pathology faculty members and an MD Medical Geneticist was formed to review molecular genetic test requests for appropriateness. Initial review of genetic test requests is done by a pathology resident in consultation with one of the PhD faculty. Complex clinical presentations or large gene panels elicit a second review by the Medical Geneticist. Appropriate requests are sent to the lowest cost laboratory able to provide the service, minimizing the financial burden for the patient and the hospital system. Education to increase awareness of our mission includes presentations at Grand Rounds, section meetings tailored to subspecialty testing, and development and utilization of algorithms. Using these strategies we realized a dramatic reduction in annual reference laboratory costs from 7.9 million dollars (FY 2012) to 5.2 million dollars (FY 2014). Conclusions: Although this decrease included reductions due to internalization of genetic and non-genetic testing, a significant proportion was the result of education and test selection guidance through our team. We continue on our mission to find a lower cost alternative for genetic testing so we can benefit from accurate diagnoses while being stewards of our health care dollars. Introduction: Patient safety is a key healthcare goal spanning the entire spectrum of patient care activities including laboratory testing. Rapid expansion and increasing applications of molecular diagnostics in patient care necessitates development of patient safety goals and practices specific for molecular diagnostics. Although CAP and CLIA guidelines provide an extremely valuable framework and practice requirements for quality management and patient safety, the clinical laboratories can further enhance patient safety by developing innovative practices and workflows. Our goal was to implement a focused patient safety committee to proactively evaluate, monitor and improve patient safety within our clinical molecular diagnostics laboratory. Methods: We created a patient safety committee (PSC), comprised of technologists and supervisors with differing levels of laboratory experience. The committee is charged with random monthly spot checks of various areas in the lab. These areas include reagent labeling, quality control, storage, instrument calibration and maintenance, proficiency testing documentation, testing personnel competency records and new test validation records. Additionally, members of the jmd.amjpathol.org ■ The Journal of Molecular Diagnostics committee participate in root cause analysis of errors and quarterly process checks to identify areas of improvement within the testing workflows. New Quality and patient safety indicators are placed as needed and existing indicators continuously optimized. Monthly reports are provided to the laboratory leadership and discussed in monthly staff meetings. Results: Since the inception of the PSC in our clinical laboratory, we have seen a marked decrease in deficiencies in 4/5 (80%) of the areas that are checked monthly. We were able to identify areas of improvement as related to reagent labeling and storage, which coincidently are among the common most cited deficiencies during CAP inspections. The PSC monitored compliance with documenting regular personnel meetings to continuously improve communication. The PSC also implemented a universal "buddy check" system to minimize specimen identification errors during critical procedural steps. This process has empowered the entire laboratory staff with active and meaningful participation in patient safety. Conclusions: As patient safety is a visible and important quality metric for hospitals, molecular diagnostics laboratories can benefit from the implementation of a dedicated patient safety committee to achieve the patient safety goals by increasing awareness and regulatory compliance.
OTH03. How to Save $9,000 in One Day: The Financial Impact of Test Utilization Management for Molecular Genetic Testing J.A. Kaylor 1 , R. Levy 2 1 Arkansas Children's Hospital, Little Rock, AR; 2 University of Arkansas for Medical Sciences, Little Rock, AR. Introduction: Healthcare costs are rising and a substantial factor is laboratory tests. Molecular genetic tests are costly and are routinely ordered at many hospitals by a variety of clinicians. We report the significant financial impact of a test utilization management program in a pediatric institution for molecular genetic testing, and we describe the protocol used to achieve such results. Methods: With a data collection window from July 2012 through May 2015, all molecular genetic tests sent to reference laboratories by Arkansas Children's Hospital were reviewed by the laboratory genetic counselor. Tests were reviewed by evaluation of medical charts and phenotypic data. Literature searches and cost analysis were used to determine the appropriateness of a test order. Tests were categorized as appropriate or inappropriate, and contact with the ordering physician was made and documented to discuss testing. Following a discussion with the ordering physician, with their approval documented, a test was considered for modification. Test modifications were classified as changed or cancelled. A test would be cancelled if it was decided that a test would not to be performed at that time. A test would be changed if the original order was altered to facilitate more appropriate testing or laboratory. Tests were only changed or cancelled with the ordering physician's approval. Results: A total of 4,349 genetic tests were reviewed by the Laboratory Genetic Counselor. Two hundred sixty eight (268) tests were modified (6%), either changed or cancelled. On average each test modification yielded $1581 in savings. Cost avoidance from modification of 268 tests ordered from July 2012 through May 2015 totals $423,784. The genetic counselor in the role of test utilization management had an average financial impact of $145,297 yearly. Case example: Three siblings of a 17 year old female with Von Willebrand disease were evaluated at ACH and full gene sequencing of VWF ($3050) was ordered for each child. The original proband had no molecular confirmation of her disease. The cases were reviewed and following a consult with a pathologist, the provider was contacted to discuss the limited utility of VWF gene testing in the siblings without a known familial mutation. The tests were cancelled, and the proband was tested instead. VWF sequencing was normal and the siblings are monitored clinically, as is the proband. The cost savings as a result of this case review was $9,150. Conclusions: A pediatric institution can achieve significant cost avoidance with implementation of a test utilization management program for molecular genetic testing. Genetic counselors are equipped with the knowledge and communication skills to engage providers in discussions when test modification may be needed. Introduction: Molecular pathology has requirements which are considered unattainable or esoteric given the current general state of affairs in Nigeria. Systemic challenges include widespread infrastructural deficits, including utilities, and a definite knowledge gap compared to other parts of the world where molecular diagnostics is established. We present the tentative steps taken to establish a molecular pathology laboratory in a large teaching hospital, the challenges encountered, and the steps taken to resolve them. Methods: Focus group discussions by a technical committee charged with the responsibility for setting up the laboratory. Participants are all pathologists with different levels of training and interest in molecular pathology. Larger participation was invited from an architect, quantity surveyor and a legal practitioner. Results: A building to provide for 2000 sq. feet laboratory space (capable of BSL 3 operations), bio-banking and storage space, office space and a conference room was designed. Challenges identified include a lack of staff with adequate knowledge and training in molecular pathology, a currently non-existent laboratory quality management culture, weak policy framework, paucity of funds, a lack of strong institutional support outlining commitments and organization, and weak interpersonal relationships between key decision makers and implementers. The activities were unremunerated. Conclusions: Significant challenges are present on the way to achieving molecular pathology capacity in Nigeria. A concerted effort by pathologists, administrators and policy makers will lead to its attainment and contribute to the provision of quality health care. We also look forward to building enduring productive partnerships.
S. Hendrickson, K. Walker, A. Rao Baylor Scott and White Health, Temple, TX. Introduction: Applying lean principles with continuous quality improvement can improve patient care with laboratory results produced at the lowest cost, within the shortest time frame. The molecular laboratory at Baylor Scott and White Hospital has experienced a growth of greater than 15% of routine high volume testing in addition to introduction of complex labor intensive oncology testing in the last 10 years. The current lab space was originally designed for a test volume of approximately 25,000 tests and was inadequate for current needs. The annual number of tests ran per technologist showed a 48.9% increase from 2013 to 2014 and continues to trend in a positive direction. Labor intensive methodologies including next-generation sequencing continue to increase and in addition, lack of automated data entry continues to be burdensome. Methods: We hypothesized that applying lean principles including 5S, work cell processing and visual work place would help address these challenges. Primary, secondary and balance measures were assessed prior to and after lean implementation. Balance measures included percentage of overtime hours worked by technologists and employee turnover within the molecular laboratory. Results: Results of value stream mapping saved employees an average of 1680 steps per day or just over 4 miles per week. Repeat testing decreased from 4% to 0.8% and has remained at 0.8% in the 6 months following reorganization. A secondary process measure, turnaround time, yielded a decrease of 10 minutes in STAT CSF enterovirus testing, or a 2.7% decrease. Overtime hours decreased by 15.46 hours per pay period, equating to a 56.7% decrease. Employee turnover went from an average of 1 to 2 employees per quarter in the year prior to reorganization to zero employee turnover in the year following lean implementation. Conclusions: After implementation, some positive outcomes included a more organized work area, decrease in extraneous supplies, and significant reduction in employee motion. Additionally, it was realized that individual value stream mapping could be implemented for continuous quality improvement. Prostate-Specific Antigen (PSA) test is a widely used test for screening men for this cancer. However, several advisory groups recommend against the use of PSA because of its high false-positive rate, unsubstantiated outcome, and small benefit. Thus, there is an urgent unmet need for novel and more accurate diagnostic and prognostic biomarkers for prostate cancer early detection in men. Methods: Recently, we have performed genome-wide target capture of miRNAs and target mRNAs followed by deep-sequencing to identify a group of PCa-specific single -binding sites loc the expression of many protein-coding genes, we hypothesize that PCa-specific SNVs in miRNAs or miRNA-binding sites in mRNAs may play pivotal roles in PCa development. Results: Preliminary SNVs identification was based on genome-wide target capture of miRNAs and miRNA binding sites followed by deep-sequencing. Capture probes were designed against 42Mb of 3'UTR of coding genes and 124Kb of miRNA (mature and pre-mir) genomics sequences. We have identified SNVs in miRNA and miRNA binding sites using next-generation DNA sequencing, and some of these SNVs were confirmed by the digital PCR. We have also developed a focused SNVs panel (Ampliseq) for high throughput screening of SNVs in classically defined disease staging of prostate cancer in men. Conclusions Here we present an efficient method by combining SNVs in miRNA and miRNA target binding sites that will provide multidimensional screening assay for prostate cancer early detection and classification in men. Introduction: Fusion genes are well-recognized oncogenic drivers in cancers. We assessed the feasibility and clinical utility of whole transcriptome sequencing (RNAseq) for detecting diagnostic and targetable fusion genes in childhood sarcomas of uncertain diagnosis. Methods: A RNA-seq pipeline for fusion gene detection was trained using a test set of 9 soft tissue sarcomas with clinically diagnosed fusion genes. A validation cohort of an additional 28 sarcomas or mesenchymal tumors of uncertain diagnosis were analyzed for gene fusions. Total RNA from fresh-frozen tumors was used to prepare poly-A+ stranded Illumina libraries, generating ~89 x 106 (2 x 100bp) reads/sample. Fusion genes were detected using deFuse (v.0.6.1). Predicted fusions were ranked on clinical utility using disease-specific clinical guidelines (WHO), COSMIC database, as well as published literature and classified as category 1 (pathognomonic fusions of established clinical utility), category 2 (fusions of potential utility; e.g., targetable fusions), category 3 (fusions in cancer genes), and category 4 (other fusions). Category 1-3 fusions were confirmed by PCR. Results: A total of 7304 fusions (1452 interchromosomal, 5852 intrachromosomal including 4075 read-through events) were identified, including 233 fusions retaining an open reading frame (ORF). Analysis of the 233 ORF-retaining fusions revealed a median of 6 fusions per tumor (range 0-17). In the test set (n=9), RNA-seq correctly identified 8 of 9 expected fusions (EWSR1-WT1, 3 EWSR1-FLI1, EWSR1-ERG, PAX3-FOXO1, ASPSCR1-TFE3, 1 BCOR-CCNB3), failing to call a BCOR-CCNB3 fusion that was detected but filtered. In 4 of 28 validation cases with one partner gene known to be rearranged by FISH (EWSR1-, CIC-, EWSR1-, CIC-), RNA-seq identified the corresponding partner genes (ETV1, FOXO4, FLI1, DUX4, respectively). Six category 1 or 2 fusions were identified in the remainder of the validation cases (n=24), including two activating fusions in unexpected tumor types that are potentially targetable (KIAA1549-BRAF: chest wall sarcoma; FGFR1-ADAM32: embryonal rhabdomyosarcoma), a pathognomonic fusion in a novel tumor type (MLL-MLLT10: round cell sarcoma), and in two cases, diagnostic fusions (FUS-CREB3L2: low-grade fibromyxoid sarcoma, and EWSR1-ERG: atypical Ewing sarcoma). In total, therefore, fusion genes of established or potential clinical utility were identified in 10 of 28 (35.7%) validation cases. Conclusions: RNA-seq is a powerful tool for detecting fusions of clinical utility in sarcomas, and may be especially relevant for unbiased detection of therapeutic targets, as evidenced by the unexpected identification of targetable fusions in the analyzed sarcoma cohort. Introduction: Studies suggest that tumor infiltrating lymphocytes (TILs) in cancer tissues has potential in prevention and inhibition of tumor. The presence of CD8+ effector T cells and the ratio of CD8+ /FoxP3+ regulatory T cells seem to correlate with improved prognosis and long term survival in solid tumors. Blocking of individual inhibitory receptors of T cells or their ligands like PD-1 (Keytruda, Nivolumab), CTLA-4 (Ipilimumab) PD-L1 (MPDL3280A), is now becoming a promising approach for cancer therapy. To evaluate the effect of these therapies on TILs and understand their role in cancer therapy, we have established the quantitative immunohistochemistry (IHC) assays for CD3, CD8, and Foxp3 for immunophenotyping of lymphocytes in tissues. Methods: Approximately 40 formalin fixed paraffin embedded (FFPE) tissue blocks of various types of tumors were sectioned and immunolabelled with anti-CD3, anti-CD8 and anti-FoxP3 antibodies using IHC autostainers. The tumor area (including tumor invasive margins) on stained slides were identified, marked up and scored using two scoring methods: 1: pathologist (semi-quantitative: percentage of immunolabelled cells versus total cells) and 2: Definiens automated image analysis software (quantitative: percentage of immunolabelled cells versus total cells and cells / mm2 area in a tumor area were evaluated.). The CD3, CD8 and FoxP3 IHC assays were validated for various (renal cell carcinoma, lung carcinoma, hepatocellular carcinoma, melanoma and endometroid carcinoma) tumor tissues to analyze the inter-run, intra-run, interanalyst and inter-pathologist concordance using both scoring methods. A direct comparison of traditional pathologist scores versus scores from automated analysis (percentage of immunolabelled cells versus total cells) was performed to validate the quantitative scoring method. Results: The results indicate that automated CD3, CD8 and Foxp3 quantitative IHC assays for various FFPE tumor tissues have met the validation acceptance criteria for intra-run, inter-run, inter-analyst variations and quantitative scoring method (percentage of immunolabelled cells versus total cells) shows a good correlation with pathologist scores. Conclusions: Quantitative scoring method is more reproducible than semi-quantitaive scoring method as semiquantitative method could be inherently subjective, and might produce ordinal rather than continuous variable data. These quantitative CD3, CD8 and Foxp3 IHC assays can be utilized to enumerate the different T cell populations in tumor area including tumor invasive margins to evaluate the effect of cancer therapy on TILs in patients. Introduction: Gonadoblastoma is a rare tumor that can arise in dysgenetic gonads containing Y chromosome material. The occurrence of gonadoblastoma in phenotypically and apparently chromosomally normal females is very rare with the presence of Y chromosomal material usually excluded by peripheral blood cytogenetics. Gonadoblastoma is considered a benign neoplasm but can transform into a germ cell tumor, typically dysgerminoma. Here we present a case of an 11 year old pre-menarchal phenotypic female with large bilateral ovarian masses and normally developed uterus and fallopian tubes. Methods: At the time of surgery, tumor was processed by multiple methods, including air-dried touch preparations, frozen, and formalin fixed. Hematoxylin and eosin stained slides and standard immunohistochemistry was performed on paraffin embedded tissue. Interphase FISH using probes for X (DXZ1) and Y (DYZ3) centromeres, as well as other chromosomes was performed on touch preparations. High resolution genome wide single nucleotide polymorphism (SNP) microarray analysis was performed on the DNA extracted from the frozen tissue using Illumina Infinium CytoSNP-850K BeadChip with >831,000 loci across chromosomes 1-22, X, and Y. The copy number alterations were visually inspected and manually determined according to the change of the logR ratio and the B allele frequency. Peripheral blood cytogenetic analysis was carried out on metaphase cells from PHA stimulated lymphocytes following G-banding. Results: Histopathologic analysis of the tumor revealed bilateral dysgerminoma arising in a background of gonadoblastoma with areas of age-appropriate ovarian follicle development. Peripheral blood cytogenetic analysis disclosed an apparently normal female karyotype, 46, XX in 60 cells. Tumor tissue from the right ovary analyzed by FISH showed mosaic trisomy X in 67% of the cells. Although the changes seen on the tumor SNP array were very subtle, the copy number variation profile and genotyping pattern were suggestive of mosaic pseudopolyploidy, with 3 to 6 copies for each chromosome of 1 to 22 and X and with most chromosomes showing tetraploidy. Additional confirmatory FISH studies were performed using probes to several chromosomes which were consistent with the SNP array findings. Overall, no Y material was identified by FISH study on ~200 cells or by SNP array. The FISH study also rules out low mosaicism (<2%) with 95% confidence. Conclusions: Gonadoblastoma is exceedingly rare in phenotypic and genotypic 46, XX patients and to the best of our knowledge, this is the first report of a case with SNP array and FISH data showing absence of Y chromosome material in this tumor. The underlying mechanism for tumorigenesis still remains to be further characterized.
A.A. Lo, C. Jie, M. Rao, G. Yang, N. Beaubier Northwestern University, Chicago, IL. Introduction: Inflammation is a known risk factor for cancer and inflammatory bowel disease (IBD) is associated with increased risk of colorectal cancer (CRC), often with signet ring (SR) morphology. Molecular studies have demonstrated differences between sporadic and IBD-associated CRC. However, no investigation of IBDassociated CRC has been performed using a whole exome sequencing (WES) approach on IBD-associated SRCRC. Methods: WES was performed on IBDassociated SRCRC in 5 patients (ages 46-67; 3 M, 2 F) using Agilent SureSelect exome capture and Illumina HiSeq 2000. Analysis included removing low quality reads and adapter sequences, then aligning to sporadic CRC in HGA19 with BWA software. Single Nucleotide Polymorphism (SNP) calling was done with GATK software. SNPs were annotated with NCBI RefGene and compared to NCBI dbSNP 141, 1000 Genomes and UCSC genome. Bioinformatic analysis was performed with Ingenuity Pathway Analysis (IPA) after concordant exonic SNPs were identified. Results: Sequence data contained 137-172 million reads with 91% surviving cleanup. Alignment showed at least 99.8% coverage of targeted regions with average depth of 120X to 155X. Among the five SNP data sets, four had a dbSNP rate of 98% whereas one had a rate of 93.4%. We found 9586 concordant SNPs with the same variant allele and genotype. These included 1412 missense, 10 stopgained and 11 stop-loss exonic variants in 1077 genes. Bioinformatic function enrichment analysis of concordant SNP genes shows significant association with CRC (p-value<0.0001). IPA analysis shows significant association with known CRC pathways including MSP-RON signaling, dsDNA break repair and inhibition of matrix metalloprotease pathways (p-value<0.05). Four of these exonic SNPs were novel variants localized in IGFN1 (chr1:201178904 A>T), MUC4 (chr3:195512186 T>C), MUC16 (chr19:9002519 G>A) and PRSS1 (chr7:142460369 G>C), of which IGFN1, MUC4 and MUC16 are associated with CRC according to function analysis. Though all four are missense SNPs, the novel MUC4 SNP is a homozygous polymorphism in the longest isoform, verified by NCBI Refseq, whereas the other three are heterozygous alleles. Conclusions: We present the first pilot study of WES in IBDassociated SRCRC. In addition to common CRC mutations, there is evidence of Mucin gene family mutations. Interestingly, MUC4 mutations have been associated with poor prognosis in sporadic CRC and we provide the first molecular evidence jmd.amjpathol.org ■ The Journal of Molecular Diagnostics that IBD-associated SRCRC may harbor mutations associated with prognostic significance. Therefore, confirmation in a larger cohort study is warranted.
A. Brown 1 , K. Lim 2 , G. Corpus 2 , M.T. Hustek 3 , T. Tran 2 , C. Chang 2 1 University of Central Florida College of Medicine, Orlando, FL; 2 Florida Hospital, Orlando, FL; 3 Center for Diagnostic Pathology at Florida Hospital, Orlando, FL. Introduction: Papillary thyroid cancer (PTC) is the most common type of thyroid cancer and the BRAF mutation has been found in 44% of PTC. Not only does the BRAF mutation provide a source of identification of PTC, but it also provides valuable prognostic information, as its presence has been correlated with more aggressive phenotypes. Fine-needle aspiration (FNA) with cytological analysis is widely used as the initial step for evaluation of thyroid nodules. Currently, molecular tests, such as the BRAF mutation, using cellular DNA from FNA specimens are routinely used to support the diagnosis of PTC when the morphologic findings of FNA is equivocal. The goal of this study is to evaluate the diagnostic yield of detecting BRAF mutation in the cell-free supernatant fluid from cytocentrifugation of the FNA specimen. Methods: A total of 91 consecutive FNA samples of thyroid lesions performed at Florida Hospital (FH) were evaluated for BRAF mutations using both cellular DNA and cell-free (i.e. supernatant). Cellular DNA was analyzed at a reference laboratory using allele-specific PCR with sensitivity of detecting 2% mutant alleles. Cell free DNA was analyzed at FH Molecular Diagnostic Lab using pyrosequencing with a limit of detection of 10% mutant alleles. Additionally, BRAF mutation data was correlated with both pathology and FNA findings. Results: Of the 91 samples evaluated, 77 (85%) samples had amplifiable DNA from supernatant for BRAF mutation. Two (3%) of the 77 samples were positive for BRAF mutations. Of interest, these 2 samples showed wild type BRAF in their cellular DNA counterparts. Five samples were positive for BRAF mutations using cellular DNA and all of them revealed wild type DNA in the acellular components. Among the 13 samples showed morphologic findings suspicious for or diagnostic of PTC, 6 (46%) samples (one supernatant and 5 cellular) were positive for BRAF mutation. This indicates that testing cell free DNA in the FNA specimens may increase the diagnostic yield by 1/13 (8%). The remaining sample showing BRAF mutation on cell-free DNA was from a case of FNA showing atypical morphology of unknown significance. Conclusion: Our preliminary results indicate that the vast majority of routinely discarded FNA supernatants contain amplifiable DNA. Additionally, profiling the mutations of BRAF and other genes using cell-free DNA may provide valuable diagnostic information to assist the diagnosis of PTC in a substantial percentage of patients, particularly if morphologic findings are equivocal and no mutations are identified in cellular DNA samples.
K. Damjanovich-Comenares 1,2 , C. Willmore-Payne 1,2 , P. Wilfahrt 1 , E. Downs-Kelly 1,2 , K. Geiersbach 1,2 1 ARUP Laboratories, Salt Lake City, UT; 2 University of Utah Health Sciences Center, Salt Lake City, UT. Introduction: The updated ASCO/CAP (2013) Guidelines for interpreting HER2 in situ hybridization (ISH) on breast cancer indicate that reflex testing is useful for resolving equivocal HER2 (ERBB2) FISH results. FISH with an alternative control probe and real time PCR are compared. Methods: Thirty breast cancers previously tested with an FDA approved HER2 FISH assay (10 amplified, 10 non-amplified, and 10 equivocal) were studied. All cases were tested two ways: 1) Dual color FISH using a control probe targeting RAI1 (17p11.2) and a probe targeting ERBB2 (17q12), and 2) real time PCR amplification of ERBB2 and a control locus at 2q11.2 (EIF5B). Reflex FISH and PCR amplification status was based on HER2/control ratios of either >2.0 for amplified or <2.0 for non-amplified. Results: Reflex FISH results for non-amplified and amplified tumors were 100% concordant with the results of the FDA approved FISH test. No equivocal cases were re-classified as amplified by PCR (0/10), whereas 4/10 equivocal cases were re-classified as amplified using the reflex FISH assay. The 4 cases that were classified as nonamplified using PCR and amplified using the reflex FISH assay tended to have a lower tumor percentage (60-80% for discordant results versus 70% to 90% for concordant results). Conclusions: Cases with co-amplification of HER2 and the centromere may be classified as equivocal with the FDA approved assay, and by current guidelines, these cases can be re-classified as amplified if the copy number gains are limited to discrete segments of chromosome 17 including HER2 and the pericentromeric region. With the new ASCO/CAP definition of HER2 equivocal FISH results, optimal analytic sensitivity of the reflex assay has become critical to avoid false negative results. Our data show that a reflex FISH assay is more sensitive than real time PCR for detecting low level amplification of ERBB2.
Sensitivity and Specificity R. Chandramohan, J. Sadowska, C. O Reilly, M. Ladanyi, L. Borsu Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: To address the need for a clinical NGS platform for low input DNA samples and the requirement for a quick turnaround time in some clinical settings, we developed and validated a solid tumor custom MSK-AmpliSeq panel in formalinfixed, paraffin-embedded (FFPE) tumor and matched normal DNA samples and benchmarked this amplicon-based Ion Torrent sequencing assay against our hybridization capture-based clinical NGS clinical assay, MSK-IMPACT (Cheng D, et al 2015) . Methods: Given the tendency of Ion Torrent technology for higher basecalling error rates within homopolymer stretches, we first focused our validation on samples with at least one insertion/deletion (INDEL) in combination with one or more single nucleotide variants (SNV) in the exons covered by the MSK-AmpliSeq panel and considered for NYS DOH submission. DNAs from 68 specimens (34 tumor/normal pairs) previously run on MSK-IMPACT were selected. MSK-AmpliSeq library construction was performed with 20ng of FFPE tumor and matched normal DNA by multiplexed amplification of 891 regions tiled by 2126 amplicons. MSK-AmpliSeq covers 94 genes including all clinically actionable genes present in the larger MSK-IMPACT panel. Sequencing was done on the Ion Torrent PGM and the data were analyzed with a custom pipeline developed at MSK. Aligned BAM files generated by Torrent Suite were primer trimmed and variant calling was performed to detect SNV and INDEL variants using VarScan2, which were annotated using ANNOVAR. SNVs and INDELs were filtered for variant allele read depth and variant allele frequency (AF) at 15X and 5% and 40X and 10%, respectively. Genotyping of historical (unmatched) normal blood BAM files at variant loci was used to remove recurrent systemic artifacts. Results: By comparing the MSK-AmpliSeq variants detected in the 34 tumor/normal pairs with a stringent custom analysis pipeline to the variants previously identified by the MSK-IMPACT assay, 72/77 (93.5%) were detected by the MSK-AmpliSeq assay in the validation exons, of which 41/44 (93.1%) and 31/33 (93.9%) were SNVs and INDELs, respectively. The five missed exon variants (6.4%) were present at lower coverage upon reviewing the sequencing raw files. Two SNV and two INDEL variants were discarded by the pipeline due to our filtering thresholds and one SNV due to an amplicon failure. One new variant detected by the MSK-AmpliSeq pipeline analysis but not originally called by MSK-IMPACT was found at sub-threshold coverage upon manual re-review of the MSK-IMPACT sequencing reads. Conclusions: MSK-AmpliSeq Ion Torrent sequencing data analyzed through a custom pipeline identified the correct variants in 93.5% of the calls detected with our MSK-IMPACT clinical NGS assay. Specifically, we were able to confidently detect 93.1% SNVs and 93.9% INDELs calls present at more than 5% and 10% AF respectively. To eliminate potential false positive calls, our analysis thresholds missed 6.4% of real variants present in the sequencing raw files. Our results demonstrate that combining a stringent filtering for somatic variants detected in tumor/normal pairs with the exclusion of targeted primer sequences and recurrent sequencing artifacts results in a robust AmpliSeq-based assay with improved sensitivity and specificity. Its mutational profile is known to be similar to small cell carcinoma, but genomic aberration of LCNEC is not well identified. Methods: In this study, 22 LCNEC samples was analyzed to investigate genomic aberrations including single nucleotide variation (SNV) and small indel mutations, copy number variation (CNV) and structural variation under targeted next generation sequencing (NGS) system. Using genomic DNA isolated from tumor and matched normal FFPE tissues, DNA libraries were generated using the SureSelect XT kit with TruSeq adapter, and 505 targeted genes were enriched by capturing with RNA baits (Agilent), and followed by sequencing with MiSeq system (Illumina). Mutect, somatic indel locator, Breakmer, and Recapseq algorithms were used for identification of SNV, indel, Translocation and CNV, respectively. Results: Under 2.9 Mb of whole target size, mean target coverage and % more than 30x coverage were 168x and 94. 14q12-q13 (NKX2-1, FOXA1 ) and 8p22 (DLC1). Specific copy number alteration was associated with tumor recurrence. One patient showed a well known druggable target, ALK-EML4 fusion, and which is validated by immunohistochemistry. Conclusions: Our data suggest that targeted NGS could be used for detection of druggable targets and genome based prognostication of LCENC of the lung. M.B. Durso, S. Zhong, A.I. Wald, L.M. Kelly, K.M. Callenberg, Y.E. Nikiforov, M. Nikiforova University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: Transcriptome sequencing (RNA-Seq) serves as a discovery tool for detection of novel fusion types and quantifying gene expression levels in research laboratories. It is not routinely applied in the clinical setting due to difficulties in both wet sequencing and bioinformatics analysis. We have developed and validated an RNA-Seq approach that can be effectively used in the clinical laboratory for detection of gene fusions and gene expression in oncology samples. Methods: RNA-Seq analysis was performed with sequencing parameters optimized for use in the clinical laboratory and validated on 200 ng of RNA from 13 frozen tumors with known fusion types. Transcriptome libraries were generated with the TruSeq Stranded Total RNA kit (Illumina) and sequenced using rapid run on the HiSeq 2500 using on-board clustering and the HiSeq Rapid SBS Kit v2 (Illumina). A custom bioinformatics pipeline was developed that uses both existing bioinformatics tools including Tophat and Chimerascan, and also in-house developed scripts and filters for quantifying gene expression levels as well as improving the specificity of fusion detection. The pipeline was validated on 26 tumors with previously known gene fusions. Results: RNA-Seq analysis was optimized to sequence 5 tumor samples in a 27 hour run. An average of 192x106 total reads per sample were generated with 97% of >Q30 quality score. Between 2 and 50 reads spanning break points was detected per driver fusion. The developed bioinformatics pipeline accurately detected different types of fusions including fusions in ALK, BRAF, PPARG, RET, NTRK1, and NTRK3 genes in all samples analyzed. Custom filters narrowed down the fusion candidate list to 1-3 clinically important fusions per tumor sample that were further confirmed by evaluation of break points and expression pattern of fused genes. The cost of sequencing reagents was $832/sample and library preparation, sequencing, and analysis running time was 59 hours per sample. Conclusions: Whole transcriptome (RNA-Seq) analysis can be used in clinical setting for accurate detection of gene fusions and gene expression in oncology samples. Its clinical utility may be the highest for those tumors where no targetable mutations had been detected by less expensive and more rapid targeted NGS analysis.
M. Kibukawa, S. Zhang, M.J. Marton Merck & Co., Inc, Rahway, NJ. Introduction: Hdm2 is the major negative regulator of the tumor suppressor p53. It represses p53 transcriptional activity through binding to p53. MK8242 is a potent and selective small molecule inhibitor of the Hdm2-p53 interaction. The purpose of this study was to assess whether RT-qPCR assays for p21 (CDKN1A) and PHLDA3 (which are transcription targets of p53) could serve as quantitative, highly reproducible, and sensitive pharmacodynamic biomarkers of MK-8242 inhibition of p53, and to determine whether PK/PD modeling could be used in the early clinical strategy of accelerating Phase I dose-escalation studies. Methods: P21 (CDKN1A) and PHLDA3 mRNA expression levels in peripheral blood were measured by RT-qPCR assays. Analytical validation of these assays included the selection of suitable housekeeping genes, determination of qPCR input mass, assay linear dynamic range, QC standard quality control limits, assessment of RT-qPCR amplification efficiency, and analytical precision of the RT-qPCR assay. Based on the assay determination, we analyzed pharmacodynamic data following single dose administration of MK-8242 in whole blood of 66 healthy male volunteers (HMV).The pharmacokinetic and pharmacodynamic effect of MK-8242 was assessed and the time-weighted average PD effect over the first 6 hours (TWA0-6hr) was calculated as the pre-specified PD endpoint in this study; this data was used as part of the Go/NoGo decision in subsequent clinical trials. Results: GUSB and G6PDH genes were selected for MK8242 Phase I studies in blood to normalize the RT-qPCR data. For the clinical study (66 HWM samples), there were statistically significant differences in the TWA0-6hr of relative gene expression between placebo and 80mg and 200mg MK-8242 treatment for both p21 and PHLDA3 mRNA (p < 0.001 and p< 0.002, respectively). PK/PD modeling indicated that induction of p21 and PHDLA3 mRNAs were positively correlated with MK8242 exposure. Conclusions: The assay, based on the biology of the p53/HDM2 system, has provided valuable PD information in a PK/PD study, which has been used in PK/PD modeling to rapidly demonstrate proof-of-concept.
C. Ionescu-Zanetti 1 , R. Brobey 2 , K. Rosenblatt 2 , M. Dehghani 3 , M. Schwartz 1 , R. Amato 2 1 Fluxion Biosciences, South San Francisco, CA; 2 University of Texas Health Sciences Center, Houston, TX; 3 Companion Dx, Houston, TX. Introduction: Next-generation sequencing (NGS) of blood-derived nucleic acids is an emerging paradigm for determining the mutational status of cancer patients over time. Both circulating tumor cells (CTC) and cell-free circulating DNA have been proposed as possible sample types for extracting tumor DNA. Here we present data from a CTC enrichment modality that results in tumor cell purities of >10% and a high sensitivity NGS data analysis workflow that enables the use of standard amplicon panels typically used for primary tissue. This study is aimed at urological cancers (kidney, prostate), and adds to previously published data presented for bladder cancer patients using this technique. Methods: Blood samples from prostate and kidney cancer patients were enriched for CTCs using the IsoFluxTM System, from a starting blood volume of 7.5ml to 14ml. Matched samples were enumerated to determine the CTC load, where CTCs were defined as CK+, CD45-nucleated cells (DAPI+). Cells were lysed and DNA was amplified by whole genome amplification (WGA) using the NGS Kit (Fluxion Biosciences), and quantified via qPCR. Targeted libraries were sequenced using the PGM (ThermoFisher) sequencing instrument; data was analyzed using a customized variant calling/ filtering pipeline based on standard Ion Reporter alignment tools and VarSeqTM for variant filtering and functional interpretation. Results: Multisite analytical validation data, based on spiking of cells into whole blood, and a matched molecular and bioinformatics approach demonstrates a detection limit down to 10 cells from a blood draw with a false positive rate of below 0.1 calls per sample. Clinical data from two different urological cancer pilot studies (prostate and kidney) demonstrates the detection of somatic variants for a majority of samples, and significant overlap between detected mutations and known somatic mutation sites. For example, in prostate patients, we detect common mutations (ie, TP53, PTEN, and APC genes) that are similar to the population distribution of mutation rates in tissue biopsies. Conclusions: This assay makes possible the detection of somatic variants from urological cancer patients without the need for a tissue biopsy. Introduction: Currently, histopathological techniques comprise the core of clinical neuropathology. Though there is an increasing role for molecular studies in clinical neuropathology, present methods are largely limited to single gene/target assays. We sought to establish a multiplexed, cost-efficient platform to improve molecular analysis and clinical management. Methods: To enhance diagnostic precision and molecular analysis at NYU neuropathology, we validated a molecular diagnostic platform using genome-wide methylation array profiling (Illumina Infinium HD Human Methylation 450k) with DNA derived from both fresh tissue and formalin-fixed paraffin embedded tissue. The internal workflow included morphologic review, sample preparation, molecular profiling and bioinformatic analysis. Methylation profiles were compared to a reference cohort (German Cancer Research Center) of 2150 cases from 77 tumor entities previously profiled and analyzed using a random forest algorithm and customized bioinformatic packages, shared between our institutions. Select copy number variants (CNV) and mutations were confirmed by Fluorescence in situ Hybridization (FISH) or sequencing and mutation specific immunohistochemistry. MGMT status was confirmed by methylation specific PCR and 1p/19q status by PCR LOH. Results: Two hundred eighty two (282) in-house and consult adult and pediatric brain tumors were profiled. There was a high degree of concordance with morphologically rendered diagnosis (100%) and standard molecular tests such as 1p/19q LOH (87.5%), EGFR amplification (87.5%), BRAF duplication (100%), MGMT promoter methylation (91.7%) or IDH1 R132H (100%) testing. In 60 diagnostically difficult cases, methylation profiling provided additional, relevant information in 30 of 60 (50%) cases, leading to a change of diagnosis in 9 (15%), clarification of the diagnosis in 7 (12%) cases, and further molecular subgroup refinement in 14 (23%) of cases. Conclusions: The 450k methylation array demonstrates potential for application in a clinical neuropathology setting by showing high concordance with currently performed clinical tests and exhibiting the capacity to cost-efficiently replace multiple molecular tests with a single assay. Furthermore, it provides additional molecular substratification and identifies biologically relevant diagnostic subgroups, thereby improving diagnostic accuracy, and helping inform appropriate clinical management decisions. Introduction: Cancer continues to be one of the leading causes of death in children ages 1 to 14 in the US, with approximately 1 in 5,000 children developing cancer every year. For patients with high stage/grade soft tissue tumors, such as Ewing's sarcoma and Rhabdomyosarcoma, 5-year survival rates are only 30%, with relapse patient survival less than 15% to 20%. The aim of this study was to develop a clinically-validated, RNA-sequencing (RNA-Seq) assay of formalin-fixed, paraffinembedded (FFPE) tumor tissue, as no such clinical assay exists to date upon abstract submission. This assay is to be used in conjunction with a commercially available DNA sequencing assay in order to detect clinically actionable mutations, including gene over-expression, in pediatric cancer relapse patients who have a <20% chance of event-free survival. Method: We analyzed FFPE tissue from 12 tumors previously identified to have gene amplification and/or over-expression by clinically validated methods (IHC or FISH). RNA was reverse transcribed to cDNA, and libraries prepared with adapters compatible with the Illumina sequencing platforms. Samples were sequenced on the Illumina MiSeq platform using v3 chemistry, single-read, 150 cycles. Precision was assessed by performing the assay on RNA extracted from 2 paired normal-tumor samples, one month apart. An inhouse bioinformatics pipeline (MapSplice and RSEM) and manual analysis of overexpression were independently performed of normalized reads in order to validate results using two separate analysis methods. Manual analysis of within-sample normalization of genes of interest was accomplished by selecting 11 housekeeping genes previously shown to have uniform expression across multiple tissues. When matched normal tissue was unavailable, data from The Cancer Genome Atlas (TCGA) database was used for comparison with our results. Results: Overexpression of previously validated genes was detected in 9 out of 12 samples. Discrepant results from IHC/FISH results were due to no normal control available, tumor heterogeneity, sequencing failure due to poor library quality, and poor normal versus tumor tissue determination; however, when applicable, discrepant RNA-Seq results were validated by qPCR. Reproducibility of concordance of the top 20 overexpressed genes involved in cancer pathogenesis were shown to be dependent upon the percent of mapped reads to the transcriptome. Conclusions: We describe here a proof-of-principle for a clinically validated whole-transcriptome RNA-Seq assay on FFPE tissue in order to detect over-expression of clinically relevant genes in cancer patients. However, as with all clinical assays, there are limitations to the RNA-Seq assay which are determined by the specimen type, normal-matched controls, and tumor heterogeneity. Introduction: Melanoma is a genetically heterogeneous disease with a high rate of somatic mutation as compared to other cancers. BRAF gene mutations are considered to be the most common genetic anomaly identified in melanoma cases, with rates of detection between 36% to 50%. Other commonly detected mutations occur in the NRAS (18% to 21%), TP53 (13% to 16%), PTEN (5% to 10%), and KIT (4% to 8%) genes. Identifying specific mutations has therapeutic implications, such as using vemurafinib in patients with BRAF V600 mutations or imatinib in patients with KIT mutations. Furthermore, melanomas often harbor multiple mutations, and recent studies suggest that performing a cancer panel to identify multiple mutations may be cost effective and increase a patient's quality-adjusted life years, compared with single-site mutation testing. The molecular laboratory at our tertiary care hospital recently implemented a panel to sequence 26 cancer-associated genes in solid tumors. The aim of our study was to compare the data from our next generation sequencing (NGS) cancer mutation panel to that of recently published data. Methods: Retrospective review of our laboratory information system identified melanoma cases evaluated with our molecular laboratory's CMP-26 panel. A PubMed literature search was performed to identify publications reporting rates of various mutations in melanoma. A search of the catalogue of somatic mutations in cancer (COSMIC) was also performed to determine the rates of mutations found in melanomas. Results: Forty-eight melanoma cases were identified that had CMP-26 performed between March 2014 to April 2015. Nine cases were excluded, including 2 cases of non-cutaneous melanoma (uveal and vaginal) and 7 cases with insufficient sample. In the 39 remaining cases, the following high-frequency mutations were identified: BRAF-36% (14/39); NRAS-28% (11/39); TP53-13% (5/39); PTEN-3% (1/39); and KIT-10% (4/39). Two or more mutations were identified in 36% (14/39) of cases, including one case with both BRAF V600E and NRAS mutation. Clinically significant mutations were not identified in 18% (7/39) of cases. Conclusions: Similar rates of the high-frequency gene mutations, BRAF, NRAS, TP53, PTEN, and KIT, were identified by our melanoma panel, as compared to rates previously reported in the literature. In addition, more than one-third of our cases harbored multiple mutations, which is again similar to what has been published. Using a panel to test for multiple mutations will not only help guide clinicians in treatment options, but also further our understanding of melanomagenesis. Children's Hospital, Boston, MA; 2 Brigham and Women's Hospital, Boston, MA; 3 Columbia University, New York, NY; 4 Columbia, New York, NY; 5 Dillon, Boston, MA; 6 University of Miami, Miami, FL; 7 Massachusetts General Hospital, Boston, MA; 8 Charles University, Plzen, Prague, Czech Republic; 9 Dana Farber Cancer Institute, Boston, MA. Introduction: The ETV6-NTRK3 chromosomal translocation has been described in cases of infantile fibrosarcoma, congenital mesoblastic nephroma, secretory breast carcinoma, papillary thyroid carcinoma and mammary analogue secretory carcinoma. The chimeric protein functions as a homo-or heterodimerized constitutively active tyrosine kinase, with downstream activation of PTK, the Ras-MAPK and PI3K pathways, elevation of cyclin D1 and increased cell cycle progression. Presence of this translocation is currently tested at the time of diagnosis using karyotype and/or ETV6 breakapart FISH. The EML4-ALK fusion is functionally similar, and has been described in non-small cell lung carcinoma. Here we describe three cases with a novel EML4-NTRK3 translocation. Methods: As part of the multiinstitutional iCat study profiling 100 pediatric solid tumors, review and molecular analysis of slides from a two-year old boy with a history of sarcoma of the forearm and metastatic recurrence in the lung was performed, including array comparative genomic hybridization (aCGH, Agilent 4x180k), followed by RT-PCR and Sanger sequencing. A custom dual fusion FISH probe (Agilent SureFISH) was designed to further confirm the rearrangement in the index case and to screen an additional 47 archival cases: infantile fibrosarcoma, congenital mesoblastic nephroma, secretory breast carcinoma, salivary gland neoplasms including mammary analogue secretory carcinoma and selected undifferentiated sarcomas. All testing was done on archival formalin-fixed paraffin embedded (FFPE) tissue. Results: The rearrangement was suspected on the basis of transitions in copy number gain at EML4 and NTRK3, and confirmed with RT-PCR and Sanger sequencing. Custom FISH probes revealed the EML4-NTRK3 fusion in 3 of 48 tumors. The index case, initially diagnosed as undifferentiated sarcoma, is clinically, morphologically and immunophenotypically indistinguishable from infantile fibrosarcoma. Two additional cases were identified, one congenital mesoblastic nephroma, and a second undifferentiated sarcoma which is also indistinguishable from infantile fibrosarcoma. The breakpoints for the index case, confirmed with RT-PCR, show a fusion of exon 2 of EML4 to exon 13 of NTRK3. Each of the 3 tumors had tested negative by ETV6 breakapart FISH or karyotype at the time of diagnosis. Conclusions: The novel EML4-NTRK3 fusion is present in both infantile fibrosarcoma and congenital mesoblastic nephroma. These results suggest that a revision in the testing strategy at the time of diagnosis is warranted since ETV6 breakapart FISH will not identify the EML4-NTRK3 fusion. A revised strategy would screen tumors with either NTRK3 breakapart FISH or with sequencing assays which can robustly identify and characterize translocations. Introduction: It has been known for several years that patients with colorectal adenocarcinoma (CRC) that demonstrates a mutation in KRAS will not benefit from EGFR inhibitor therapy. Many laboratories offer KRAS mutation testing, but it may be restricted to detection of mutations in KRAS exons 2 and 3. Evolving literature indicates that other RAS mutations are also clinically significant, not just in KRAS, but also NRAS, including some in exons 2, 3, and 4. In this study we identified 190 archived samples from histologically confirmed CRC cases previously tested for RAS mutations, and retested using an expanded RAS mutation panel that included codons 12, 13, 59, 61, 117 and 146 of both NRAS and KRAS. Methods: Archived frozen deoxyribonucleic acid (DNA) samples were searched for cases previously tested for RAS mutations on one of two platforms: the Signature KRAS Mutation test that will detect 7 common mutations in exons 2 and 3 (Asuragen Inc., Austin, TX, using a Luminex 100 System); and the OncoFOCUS Panel v1.0 with KIT extension using the MassARRAY system (Agena Bioscience, San Diego, CA). 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, specimens were assessed for DNA integrity using the iPLEX Pro Sample ID Panel (Agena Bioscience), and all specimens with adequate amplifiable DNA were then interrogated with a new mutation panel that includes extended RAS mutations, OncoFOCUS + KITv2 (Agena Bioscience, in development), both using the MassARRAY platform. Results: Twenty one (21) of 81 samples tested with the Signature PCR assay were positive for KRAS mutations (26%). When retested using the new extended RAS OncoFOCUSv2, 11 additional mutations were detected (14%), including KRAS Q61H and A146T; and NRAS G12D/E, G12V, G12C/Y, and Q61H, for a total of 40% RAS mutations in this group. Forty five (45) of 109 samples tested with the OncoFOCUSv1 kit were initially positive for KRAS or NRAS mutations (41%). When retested using the extended RAS OncoFOCUSv2, no additional mutations were detected. Conclusions: 1) Laboratories testing only KRAS exons 12 and 13 may miss up to 14% of extended RAS mutations in patients with CRC. 2) The OncoFOCUSv2 kit and MassARRAY system detected extended RAS mutations in about 40% of 190 archived CRC tissue samples from a US Midwestern patient population. M.B. Wachsmann, J. Balani, G. Ewing, P. Gopal, S. Yan, D.H. Oliver The University of Texas Southwestern Medical Center, Dallas, TX. Introduction: KRAS mutations are frequent in NSCLC and colorectal carcinoma (CRC). The purpose of this study is to categorize the type and frequency of KRAS mutations in our lung and colon cancer patient population and identify signatures, trends, and similarities/differences between the sites. Studies have demonstrated that specific mutations may be a consequence of exposure to environmental mutagens. Specifically lung tumors have been shown to harbor a high proportion of G to T transversions related to exposure to polycyclic aromatic hydrocarbons whereas colorectal adenocarcinomas have a high frequency of transitions at CpG dinucleotides. Methods: Specimens included FFPE biopsies and resections, and were all evaluated for diagnosis and adequacy by a surgical pathologist prior to microdissection. DNA was extracted using Qiagen QIAamp FFPE Tissue kits. Mutation analysis was done by Sanger sequencing or Sequenom mass spec. For Sanger, PCR and BigDye Terminator v3.1 cycle sequencing reactions used labdeveloped primers. Forward and reverse sequences were run and manually read on a 3130xl Genetic Analyzer. Sequenom PCR and extension primers were lab developed using Sequenom software. Mass spectrometry peaks were manually read from the MassARRAY Analyzer 4 system. Results: In NSCLC samples we identified 111 KRAS mutations out of 387 total cases (28.7%), and in CRC sixty-one KRAS mutations were identified out of 145 total cases (42.1%). In lung the two most frequent mutations were c.34 G>T (39/111, 35.1%) and c.35 G>T (18/111, 16.2%) . In contrast, KRAS c.34 mutations were infrequent (4.9%) in colon. The two most frequent mutations in CRC were c.35 G>A (22/61, 36.1%) and c.38 G>A (14/61, 23.0%). Codon 13 mutations were rare (6/111, 5.4%) in lung, and in contrast to colon, no c.38 mutations were identified. KRAS c.37 mutations were rare in both NSCLC and CRC. In our small population codon 61 mutations were nearly twice as frequent in lung (9.0% of mutations) as in colon (4.9%). Conclusions: In our patient population, the frequency and signature of KRAS mutations in lung and colon cancer are generally similar to that cited in their separate literatures, with the exception of codon 61 mutation frequency. The mutational pattern of KRAS is not the same in colon and lung, as would be expected based on their environmental exposures. The high frequency of the specific mutations, G>T transversion in lung and G>A transition in colon appear to be significant attributes for tissue-specific KRAS mutations. Ultimately, the different pattern of KRAS mutations in lung and colon and their role in tissue-specific oncogenesis warrant further investigation. H. Dubbink, P. Atmodimedjo, R. van Marion, J.M. Kros, M.J. van den Bent, W.N. Dinjens Erasmus University Medical Center, Rotterdam, Netherlands. Introduction: Cancer cells are genomic instable and accumulate tumor type-specific molecular aberrations, which may represent hallmarks for predicting prognosis and targets for therapy. For example, co-deletion of chromosomes 1p and 19q marks gliomas with an oligodendroglioma component and predicts a better prognosis and response to chemotherapy. Current diagnostic methods for showing 1p/19q codeletion or loss of heterozygosity (LOH) are fluorescent in situ hybridization (FISH), microsatellite analysis and multiplex ligation-dependent probe amplification (MLPA). We here describe a novel method to detect large chromosomal aberrations in a diagnostic setting that can be combined with simultaneous mutation detection.
The assay is based on single nucleotide polymorphism (SNP) analysis and next generation sequencing (NGS) on an Ion Torrent platform. Highly polymorphic SNPs evenly distributed over chromosome 1p and 19q were selected. The sensitivity and specificity of targeted SNP analysis was experimentally determined with DNAs extracted from 49 routine formalin-fixed, paraffin-embedded (FFPE) glioma tissues and the results were compared with diagnostic microsatellitebased LOH analysis and calculated estimates. Results: We show that targeted SNP analysis allows reliable detection of 1p and/or 19q deletion in a background of 70% of normal cells, is more sensitive than microsatellite-based LOH analysis, and requires much less DNA. This assay can be easily extended for analysis of additional genomic regions and mutational analysis of any gene of interest, all with the same amount of DNA input. Conclusions: Combined NGS-based LOH and mutation detection is perfectly suited to become standard practice for routine glioma diagnostics and other diagnostic molecular pathology applications. (NENs) . The developmental role of Insm1 varies in different subtypes of endocrine tissue (eg, pancreatic A-cells versus B-cells). As the functional status of NENs often carries clinical and prognostic significance, we use molecular and in silico techniques to correlate INSM1 expression with hormone production in functional NENs. Methods: We used qRT-PCR to assess INSM1 mRNA expression in gastrointestinal neuroendocrine neoplasms (GI-NENs), pancreatic neuroendocrine neoplasms (Pan-NENs), and pituitary NENs. INSM1 transcription factor binding sites were predicted using public databases and software. 2.5 kilobases of the upstream promoter region of selected genes were downloaded from the UCSC genome browser (http://genome.ucsc.edu) and analyzed for predicted INSM1 binding sites using the JASPAR online software (http://jaspar.genereg.net/). Conservation of the binding sites was estimated using conservation tools in the UCSC genome browser. Expression levels do not correlate with serum chromogranin (CGA) levels. We find putative INSM1 binding sites in the promoter regions of a number of genes related to neuropeptide hormone production. Higher numbers of putative binding sites correlate with increased expression of INSM1 in neoplasms with a corresponding secretory profile. Conclusions: In hormone producing neuroendocrine neoplasms, we find that INSM1 expression level correlates with hormonal subtype, and that loss or diminished expression of INSM1 often correlates with loss or diminished production of functional hormone. Loss of hormone production by neuroendocrine neoplasms also often correlates with dedifferentiation and worsened clinical outcomes. Quantitative evaluation of INSM1 may add valuable prognostic information in the assessment of NENs.
A. Gartel, Z. Zheng, D. Qin Moffitt Cancer Center, Tampa, FL. Introduction: Adenocarcinoma with neuroendocrine differentiation is a common pathology diagnosis, indicating that part of adenocarcinoma differentiates into neuroendocrine carcinoma. Adenocarcinoma and neuroendocrine collision tumor are two separate tumors that happen to arise in vicinity of each other anatomically and grow into each other. It is difficult to differentiate 'adenocarcinoma with neuroendocrine differentiation' from 'adenocarcinoma and neuroendocrine collision tumor' since both are morphologically similar. However, molecular test can be helpful in the differential diagnosis. Here, we report three interesting cases to illustrate the importance of molecular test in differential diagnosis of collision tumor. Methods: Three tumors with adenocarcinoma and neuroendocrine components are tested for KRAS, EGFR, BRAF, PIK3CA and NRAS mutations. The three tumors are labeled as tumor A, B and C. Pyrosequencing was performed on each nodule according to manufacture instructions. Results: All three tumors have adenocarcinoma and neuroendocrine components. The molecular test results are listed in adenocarcinoma and neuroendocrine components. The two components are intermingled together. The differential diagnosis includes 'adenocarcinoma with neuroendocrine differentiation' and 'adenocarcinoma and neuroendocrine collision tumor.' Traditionally, pathologists will look for morphological evidence of transition from adenocarcinoma to neuroendocrine. If there is such evidence, the diagnosis goes to 'adenocarcinoma with neuroendocrine differentiation.' If otherwise, or with adjacent distinct neuroendocrine primary tumor, then diagnosis goes to 'adenocarcinoma and neuroendocrine collision tumor.' Such criteria are difficult to apply when two components are intermingled. In such situation, molecular tests could be very helpful and more objective. The two components of tumor A and C bear different genotypes, indicating these are from different clones. Therefore, these are more likely to be 'adenocarcinoma and neuroendocrine collision tumor.' The two components from tumor B have the same mutation, therefore it is more likely from the same clone, hence 'adenocarcinoma with neuroendocrine differentiation.' ST22. Detection of Somatic Mutations at 0.5% Frequency from cfDNA and CTC DNA Using a Multiplex Next-Generation Sequencing Assay D. Brinza, D. Dhingra, C. Scafe, R. Chien, F. Hyland ThermoFisher Scientific, South San Francisco, CA. Introduction: Availability of effective blood screening for tracking of recurrence and resistance of tumors may improve outcomes in the future. Research studies suggest that virtually all tumors carry somatic DNA mutations, and these may serve as biomarkers that may be tracked from blood. The two well-characterized sources of tumor DNA in blood are circulating tumor cells (CTC) and cell-free tumor DNA (ctDNA). The abundance of CTC and/or ctDNA in blood may be very low at critical stages such as early recurrence or development of resistance. Hence there is great interest in being able to detect biomarkers at very low frequency from blood, and in characterizing the relationship between somatic mutations present in the tumor and those in CTC or ctDNA. Methods: We present a research use only analysis workflow for peripheral monitoring that enables detection of low frequency variants in blood. We developed an analysis algorithm, using statistical modeling of next generation sequencing reads, and optimizing parameters and filters to enable sensitive and specific detection of somatic mutations to 0.5% allele ratio. We demonstrate the analysis on a blood sample split into 3 sub-samples comprising white blood cells, CTC enriched, and cell-free DNA (cfDNA) samples.We used lysis to isolate white blood cells (germline), centrifugation to extract plasma DNA (cfDNA), whereas CTC cells were isolated using Cynvenio LiquidBiopsy platform a fully automated antibody-based solution. We barcoded 3 sub-samples and run them on a single Ion 318 sequencing chip using Ion AmpliSeq Cancer Hotspot Panel (CHPv2), that enables very deep (~10,000x coverage) and accurate sequencing. This panel allows interrogation of ~2800 relevant biomarkers from COSMIC and FDA actionable databases, and de-novo variant detection at ~20,000 genomic positions. Mutations were annotated using the Oncomine database in Ion Reporter software. The research assay requires a small amount of input DNA (~10ng), and has a fast turn around time from extracted DNA to variants of less than 24 hr. Results: We tested the limits of variant detection in a dilution series; in CTC; and in cfDNA. First, we diluted an engineered DNA sample from Acrometrix into a control sample (NA12878) down to 0.5% frequency. The Acrometrix sample contains ~500 common cancer mutations from COSMIC and FDA actionable databases. We achieved >99% sensitivity and specificity for variants present at frequency above 0.5%.Next, we spiked CTC cells from cancer cell lines into normal blood samples at ratio 1:1,000,000, obtaining 40% purity of CTC after enrichment, and demonstrated > 99% sensitivity and specificity of variant detection. Finally, we performed analytical validation of variant detection performance in cfDNA using a dilution series of two normal blood samples, and we detected all 20 variants present in either sample in cfDNA when their frequency was above 0.5%. Conclusions: The analysis workflow may facilitate researchers to study biomarkers in DNA from germline, CTC, and cfDNA, all available from a single blood sample, and to explore the relationship between biomarkers observed in solid tumor and those in blood. C.I. Dumur, P. Anderson, M. Sabato, C.N. Powers, A. Ferreira-Gonzalez Virginia Commonwealth University, Richmond, VA. Introduction: Next-generation sequencing (NGS) is widely being adopted in clinical settings for a variety of applications, including the detection of actionable somatic variants in tumor specimens. Library preparation is a critical, hands-on and timeconsuming step in the NGS workflow. During library preparation, each library is prepared in an independent well of a 96-well plate, encompassing several washes and magnetic bead-binding steps. This format increases the number of technical hours as more samples/libraries are prepared, while increasing the risk of humanintroduced error. Automation of library preparation is a valid approach to reduce these issues. Here, we present the validation and implementation of an open liquid handling platform for medium to high-throughput library preparation for routine utilization with the Ion AmpliSeq Cancer Hotspot Panel v2 (CHP2) assay on FFPE clinical specimens. Methods: The VERSA 1100 Gene automated liquid handling workstation was evaluated and compared against the manual library preparation method. Three different checkerboard experiments, using mutation-positive samples and either mutation-negative or no template control (NTC) samples alternated on a 96-well plate were performed to assess for possible cross-contamination during automation. Library DNA yields were measured by Qubit to assess efficiency, and samples other than NTC were sequenced to assess accuracy in variant calling compared to previously called variants from paired manually prepared libraries. Reproducibility was assessed by running a total of 16 technical replicates for mutation-positive and mutation-negative samples, 8 replicates each, in the same plate, and in multiple days. In addition, previously tested patient samples, including a patient sample that had previously failed to yield good quality library, were used to prepare libraries on the VERSA 1100 Gene. Results: All checkerboard experiments showed no evidence of cross-contamination by showing no DNA on the NTC wells, or no variants called on negative samples after sequencing using the CHP2 assay. Also, high reproducibility was observed in both, library yields and variants called across all technical replicates. All patient DNA samples yield good quality libraries, including the one that had previously failed using the manual library preparation method. All patient variants were called with highly correlated (Pearson's r>0.990) frequencies to those obtained with the manual method. Conclusions: Overall, our results show that the performance of the VERSA 1100 Gene automated liquid handling workstation is very robust and might eliminate human-introduced errors, when compared to the manual library preparation method for the CHP2 assay. S. Puetz, K. Schilter, N. Harding-Jackson, S. Suster, A.C. Mackinnon Medical College of Wisconsin, Milwaukee, WI. Introduction: Myxoinflammatory fibroblastic sarcoma (MIFS) is a rare, low-grade, soft tissue malignancy predominately occurring in young or middle-aged adults. The tumors are mainly limited to the distal extremities with a high rate of local recurrence and low metastatic potential. The lesions tend to be small and often present as slowgrowing, poorly-defined, painless masses. Varying proportions of inflammatory, myxoid, and large atypical spindle and epithelioid cells make the differential diagnosis of MIFS challenging. Molecular cytogenetic assays, such as fluorescence in situ hybridization (FISH) are effective diagnostic aids by identifying translocations specific to soft tissue tumors. Previous studies described t(1;10) translocations in a subset of MIFS resulting in the rearrangement of two genes, TGFBR3 and MGEA5. To further investigate the utility of detecting TGFBR3 and MGEA5 rearrangements as a potential diagnostic marker, we performed FISH on a large cohort of MIFS. Methods: The cohort consisted of 4 control cases demonstrating the t(1;10) translocation by conventional karyotype analysis and 49 tumors diagnosed as MIFS by histopathological review. FISH analysis was performed using a custom-designed, locus-specific probe for MGEA5 according to the manufacturer's protocol (Empire Genomics). For each case, a total of 40 nuclei were scored by fluorescence nuclei showed loss of the 5'-MGEA5 probe. Results: All 4 control cases demonstrated the t(1;10) translocation by FISH. 37 of 49 tumors could be confidently scored by FISH analysis, and none demonstrated the t(1;10) translocation. Conclusions: The t(1;10) translocation, previously described as a common cytogenetic alternation in MIFS, was not identified in our cohort of MIFS cases. Additional studies interrogating larger cohorts and other described aberrations, such as amplification of chromosome 3, are necessary to cytogenetically characterize this complex sarcoma. Introduction: Histopathological diagnosis of diffuse gliomas is subject to interobserver variation and correlates modestly to major prognostic and predictive molecular abnormalities. We investigated a series of patients with locally diagnosed anaplastic oligodendroglial tumors included in the EORTC phase III trial 26951 on PCV chemotherapy to explore the diagnostic, prognostic and predictive value of targeted next-generation sequencing (NGS) in diffuse glioma, and to assess prognostic impact of FUBP1 and CIC mutations. Methods: Mostly formalin fixed paraffin embedded samples were tested with targeted NGS for mutations in ATRX, TP53, IDH1, IDH2, CIC, FUBP1, PI3KC, TERT, EGFR, H3F3A, BRAF, PTEN and NOTCH, and for copy number alterations of chromosome 1p, 19q, 10q and 7. TERT mutations were also assessed with PCR. Results: Material was available from 139 cases, in 6 results were uninformative. One hundred twenty six (126) tumors could be classified: 20 as astrocytoma, 49 as oligodendroglioma and 55 as glioblastoma; 2 as childhood glioblastoma, leaving 7 unclassified (total 91% classified). Molecular classification was of clear prognostic significance and correlated better with outcome than classical histopathology. In 1p/19q co-deleted tumors outcome was not affected by CIC and FUBP1 mutations. MGMT promoter methylation remained the most predictive factor for survival benefit of PCV chemotherapy. Conclusions: Targeted NGS allows a clinically relevant classification of diffuse glioma into groups with very different outcome. The diagnosis of diffuse glioma should be primarily based on a molecular classification, with the histopathological grade added to it. Future discussion should primarily aim at establishing the minimum requirements for molecular classification of diffuse glioma. M.L. Leung, Y. Wang, C. Kim, R. Gao, E. Sei, D. Maru, S. Kopetz, N. Navin University of Texas MD Anderson Cancer Center, Houston, TX. Introduction: Colorectal cancer (CRC) remains to be a detrimental disease worldwide because of our incomplete understanding on cancer's two elusive topics, metastasis and Intratumor heterogeneity. Although TCGA studies have used nextgeneration sequencing (NGS) to characterize mutations in primary CRC tumors, few studies have identified mutations in metastatic tumors. The central problem is that
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org conventional NGS technologies are limited to reporting genomic information on million of cells, and therefore cannot resolve genetic intratumor heterogeneity and identify subclones that play an important role in metastasis. Methods: To investigate genomic heterogeneity in CRC and trace metastatic lineage, we have developed a highly-multiplexed single cell DNA sequencing method that combines flow-sorting of single nuclei, time-limited multiple-displacement-amplification, low-input library preparation, library barcoding, targeted capture and NGS. Specifically, we perform targeted sequencing of a 200-cancer-gene panel using single cells from the primary tumor and liver metastasis of a CRC patient. Results: We sequenced 45 tumor cells from each primary and metastatic tumor, as well matched populations of tumor cells. Our data achieve 85% coverage breadth and 100x coverage depth on average. We have identified nonsynonymous mutations that are shared in both primary and metastatic tumors. We have also found a small portion of mutations that are only present in the liver metastasis, which may play an important role in metastatic dissemination. We used the single cell data to construct phylogenetic trees, which identified subpopulations at both organ sites. Conclusions: We found that this CRC patient followed a late-dissemination model, in which tumor cells co-evolved for a long period of time before disseminating to remote tissues. By sequencing single cells from primary and metastatic tumors, we resolve intratumor heterogeneity and identify subclones and mutations involved in metastasis in CRC patients. Our single cell sequencing tools have higher sensitivity compared to bulk sequencing methods, and allow us to identify potential targets for therapy to inhibit the metastatic dissemination of tumor cells.
T. Devos 1 , D. Kottwitz 2 , A. Schlegel 2 , R. Tetzner 2 1 Epigenomics AG, Seattle, WA; 2 Epigenomics AG, Berlin, Germany. Introduction: Despite lung cancer (LC) being the leading global cause of cancer death, there are few recommended screening methods and their worldwide adoption is very limited. As a consequence, LC patients are frequently symptomatic at time of presentation and are often diagnosed with advanced stage disease associated with poor prognosis. The recent recommendation of imaging based screening using low dose computed tomography (LDCT) in the United States only addresses a limited segment of the screening population and suffers from a high false positive rate. Thus, there is a growing need for LC detection at early stages by non-invasive high sensitivity methods to follow up indeterminate LDCT results. Aberrant DNA methylation detected in liquid biopsies has been shown to be a clinically useful biomarker and the detection of methylated SHOX2 in plasma has been reported for LC. Here, we report on the combination of the SHOX2 biomarker with two additional methylation markers, FOXL2 and PTGER4, to develop a highly sensitive marker panel for the detection of lung cancer in plasma. Methods: Quantitative real time PCR assays were developed to determine the presence of methylated DNA fragments of the three marker genes SHOX2, FOXL2 and PTGER4 in bisulfite converted DNA. DNA was extracted from plasma samples from lung cancer patients and age-matched healthy individuals. Different combinations of the three markers were investigated in a training study. A testing study was conducted to verify the marker panel performance. Results: In the training study comprising 30 patients, the performance of combinations of the three markers reached very high sensitivity (70% to 100%) with few false positive results (5% to 25%). This clinical performance was confirmed in a substantially larger testing study of 240 patients. Lung cancer specimens were identified by the presence of at least two of the methylation markers. However, the highest sensitivity was reached if test positivity was based on detection of any of the three marker genes. Conclusions: The combined analysis of the three DNA methylation tumor markers SHOX2, FOXL2 and PTGER4 can be used to detect lung cancer in plasma samples with very high sensitivity. The combined analysis of all three markers yields up to a 100% detection rate. These findings were derived in a training study and confirmed in a testing study. The marker panel is being used as the basis for the development of a blood-based lung cancer diagnostic tool with very high sensitivity. This new diagnostic tool may ultimately show clinical utility in combination with current imaging techniques by reducing the false positive rate of such methods and has the potential for broader applications in the early detection of LC.
Vanderbilt University Medical Center, Nashville, TN. Introduction: Elevated microsatellite alterations at selected tetranucleotide repeats (EMAST) has been observed in up to 60% of colorectal cancers (CRCs). Multiple reports have suggested EMAST as a prognostic marker; however, its underlying mechanism and association with MSI status is unknown. In this study, we aimed to evaluate EMAST in CRCs regarding its relationship to MSI status and pathologic features, and to identify sensitive EMAST marker sequences with frequent alterations. Methods: We studied a pilot cohort of 26 MSI-H and 17 MSS paired normal and CRC samples. The stability of a panel of 13 tetranucleotide (including 3 with AGAX repeats) and 3 pentanucleotide repeat markers (AAAGA repeats) was assessed using PCR. EMAST status was determined as EMAST-High (H), EMAST-Low (L) or stable depending on the number of markers demonstrating instability and loss of heterozygosity (LOH) status of TP53 using PCR amplification of a dinucleotide repeat marker. These results were correlated with histological findings when available and MSI status using a Chi-square test. Results: EMAST-H was observed in 63% of 43 CRCs tested, with a frequency of 96% in MSI-H and 12% in MSS tumors. There were similar numbers of EMAST-L (47%) and stable (42%) in MSS tumors. Among the 16 markers examined, tetranucleotide repeat markers containing a complex TCTA repeat at the VWA locus (78%), and GATA at D7S820 (70%), had the highest frequency of instability in EMAST CRCs. Other markers that were unstable at a frequency over 50% included a complex tetranucleotide repeat of TTTC (FGA), AGAT (D5S818), TATC (D13S317) and a complex TCTA (D8S1179), and a pentanucleotide sequence of AAAGA. Loci containing AATG at loci TPOX and THO1 have the least instability at 14% and 6%, respectively. Over 60% of EMAST-H tumors showed MSI at TP53 and the majority of the rest did not exhibit LOH at TP53. None of the MSS CRC displayed MSI at TP53. In MSS CRCs, EMAST-L was associated with lower grade, whereas EMAST stable CRCs tended to be higher grade. Perineural invasion (PNI) was rare in EMAST-H CRCs, whereas stable tumors always displayed PNI. Interestingly, Crohns-like response was significantly associated with the presence of EMAST and was specific for tetranucleotide versus pentanucleotide repeat markers. Conclusions: EMAST, including H and L, are present in all MSI-H CRCs, and 59% of MSS CRCs. EMAST-sensitive markers are not limited to those containing AAAG and AGAX sequences. Further, EMAST-H tumors were associated with MSI rather than LOH at TP53, the absence of PNI overall, and Crohns-like response in those CRCs demonstrating only tetranucleotide instability. J.J. Plyler, J. Houghton, B. Printy, G.J. Latham Asuragen, Inc., Austin, TX. Introduction: In oncology, commercially available targeted NGS gene panels are increasingly used to identify clinically-actionable variants and help guide targeted therapies and individualize patient management. Two of the most commonly used panels are based on AmpliSeq and TruSight chemistries. In this study, we compared and contrasted the analytical performance and time-motion workflows of these two targeted NGS methods using 60 FFPE tumor samples. We also compared a recently commercialized technology, the Quantidex Pan Cancer Kit. Methods: Residual clinical FFPE biopsies (n=60) from 6 different tissues, along with defined mixtures of FFPE DNA, were analyzed. Extracted DNA was quantified using 4 different methods (spectrophotometry, fluorescent dye binding, and two distinct qPCR assays). Functional DNA quality was assessed by qPCR. DNA samples were qualified for enrichment based upon the suppliers' instructions, and processed using the AmpliSeq Cancer Hotspot Panel v2 (ACHP, Thermo Fisher), TruSight Tumor sequencing panel (TT, Illumina) and Quantidex Pan Cancer Kit (QPC, Asuragen). NGS data were generated on an Illumina MiSeq (TT and QPC) and Ion Torrent PGM (ACHP). Bioinformatic analyses were performed using each vendor's pipeline. Timemotion analyses were derived from a batch size of 9 samples for ACHP, 8 for TT, and 20 for QPC. Results: Median values of 62.8 ng/μl (spectrophotometry), 10.6 ng/μl (fluorescent assay), and 1901 amplifiable DNA copies/μl (Quantidex DNA Assay) were obtained across 60 FFPE tumor DNA samples. The median percent of amplifiable templates was 9.6. Using the suppliers' prescribed DNA QC criteria, 57/60 samples could be analyzed with ACHP and QPC, and 29/60 with TT. The median A260 DNA input for qualified DNA into targeted enrichment was 14.3 ng for QPC, 58.5 ng for ACHP and 2000 ng for TT. Passing DNA samples were enriched using each kit, and sequenced to a median read depth of >1000x. The sample-level agreement in variant calls across all 3 kits was >95% across shared gene loci. However, differences in the detection of low-level variants from low-copy DNA inputs were revealed using defined FFPE DNA mixtures. Finally, time-motion analyses -on time and time-to-result for QPC compared to ACHP and TT. Conclusions: This study describes a detailed accounting of sample QC measures, analytical performance, and time-motion workflows for 3 commercially available pan-cancer NGS enrichment kits using 60 FFPE tumor biopsies. The results support high levels of variant call accuracy for all 3 methodologies using vendor protocols, but exposes marked differences among the kits in DNA input amount, the fraction of samples passing QC, and hands-on and overall turn-around time. C.M. Vanderbilt, K.L. Jones, M. Geraci, B. Gao, W.A. Franklin University of Colorado Denver, Anschutz Medical Campus, Aurora, CO. Introduction: Whole exome sequencing (WES) has been suggested as an alternative to current piecemeal approaches to predictive testing for lung cancer. Because it provides a measure of mutational load, it may also be predictive of response to anti-PD1 treatment, a therapy recently proven effective in lung carcinoma. However, many issues remain to be resolved before WES can be applied routinely in a clinical setting. These include selection of a proper method, platform, and testing protocol. In the present study we evaluate tissue coring as a method for obtaining DNA from FFPE tumor tissue, calculated the minimum flow cell area required to achieve high level exome coverage from a commercially available library, and assessed the relationship between coverage and mutation load. Methods: jmd.amjpathol.org ■ The Journal of Molecular Diagnostics Tissue cores were obtained from tumor rich regions of paraffin blocks and from normal lung tissue using a Beecher Manual Tissue Arrayer. Quality of DNA extracted from the cores was assessed using both Bioanalyzer and Qbit protocols. A sequencing library was created using the Agilent Sure Select XT5 (v5) library kit. Sequencing was performed using an Illumina Hiseq 2500 ultrahigh throughput sequencing system. Two flow cells were used for each of 4 samples to obtain a high level of coverage and to estimate the effect of reducing coverage on mutation detection by computational methods. DNA from non-tumoral regions was used to exclude genomic polymorphisms. DNA from both tumors was also tested by a clinically validated NGS panel. Results: DNA yield from two cores from each tumor was 17 and 13 micrograms. Average DNA strand lengths were approximately 6000. Sequencing of each specimen was consistently successful with 8e+8 raw sequences obtained for each tumor, 93% of which could be mapped. Overall average coverage for each probe was 2.4e+04. Fifty seven (57) mutations were identified in the adenocarcinoma (AC) and 302 in the squamous cell carcinoma (SCC). All mutations identified on clinical NGS were seen on WES. By calculation reduction of coverage by half the number of calls (and thus mutation load) was significantly reduced on each tumor. Conclusions: Tissue coring can be used to obtain DNA for WES of tumors. The sequenced tumors fell into the two categories of high and low mutation load. Distinction between high and low mutation load may be prognostically significant and may predict response to immunological checkpoint inhibition with anti-PD1 therapy. WES is a feasible method for evaluating this potentially important biomarker in FFPE, and coverage plays a crucial role in calculating an accurate/consistent mutation load. Guseva, A.A. Stence, R. Sompallae, J.C. Schade, A.D. Bossler, D. Ma University of Iowa Hospitals and Clinics, Iowa City, IA. Introduction: Sarcomas are uncommon malignancies of mesenchymal cell origin, which can be difficult to diagnose based on morphology or traditional ancillary studies. Genomic rearrangements which result in aberrant fusion proteins and transcriptional dysregulation are one of the three major mechanisms for sarcomagenesis. Identification of tumor type-specific translocations is not only important for diagnosis but also for prognosis and targeted therapies. Traditionally, gene fusions are detected by fluorescent in situ hybridization (FISH) or RT-PCR, which requires prior knowledge of the fusion partners. We evaluated the ability of a next generation sequencing (NGS) based assay utilizing Anchored Multiplex PCRTM enrichment to detect gene fusions involving 26 genes associated with soft tissue tumors. Methods: Thirty-nine cases of sarcomas including alveolar soft-part sarcoma (ASPS, 3 cases), alveolar rhabdomyosarcoma (ARM, 2 cases), synovial sarcoma (SS, 5), Ewing sarcoma (EWS, 5), dermatofibrosarcoma protuberans (DSFP, 4), solitary fibrous tumor (SFT, 13), extraskeletal myxoid chondrosarcoma (EMS, 2), extraskeletal myxoid chondrosarcoma (EMC, 2), endometrial stromal sarcoma (ESS, 3) were included in this study. Two cases of lung adenocarcinoma were also tested. Total RNA or total nucleic acid was extracted from formalin-fixed, paraffin-embedded (FFPE) tissue. Sequencing analysis for gene fusions was performed using the Universal RNA Fusion Detection Kit (ArcherDX, Boulder, CO), the ArcherTM FusionPlexTM Sarcoma Panel (ArcherDX) and the Ion Torrent personal genomic machine (Life Technologies, Carlsbad, CA). One positive and two negative (RNAs from normal blood) controls were included in each run. Data were analyzed using the Archer Analysis Pipeline 3.0. The NGS results were correlated with morphologic, immunohistochemical (IHC) and FISH findings. Results: Expected gene fusion products were identified in each tumor type: ASPSCR1-TFE3 (ASPS), PAX3-FOXO1 (ARM), SS18-SSX1 (SS), EWSR1-NR4A4 (EWS), COL1A1-PDGFB (DFSP), NAB2-STAT6 (SFT), EML4-ALK (lung adenocarcinoma). Specific fusion break points were identified and the fusion variants were confirmed by RT-PCR followed by Sanger sequencing. There was 100% concordance of NGS findings with RT-PCR, Sanger, FISH, and IHC results. Conclusions: This new NGS-based gene fusion assay could detect sarcoma-related fusions in a single assay from various tumor types. It could differentiate fusion partners at single-nucleotide resolution, detect the break points, and identify the specific fusion products without prior knowledge of the partners. The assay is reliable and works well on FFPE tissue. It is extremely useful for the diagnosis of difficult cases, and has prognostic as well as therapeutic value. L.J. Whiteley, M. Cankovic, B. Shaw, D. Chitale, K. Mosey Henry Ford Health System, Detroit, MI. Introduction: Implementing Next-generation Sequencing (NGS) is a project that clinical labs will have to integrate into workflow to maintain standard of care for patients at their facilities. The task of brining on NGS can be daunting when there isn't a dedicated group working exclusively on the project. Lab administration is also faced with the undertaking of justifying the upfront cost of validating this technology. The aim of this study was to assess integration of NGS into a mid-sized clinical lab for oncology application. The focus was on the analytical bench processes of sample and library preparation and sequence generation. Although an important component to NGS validation in a clinical setting, the bioinformatics process of sequence alignment, annotation, variant calling and reporting are not covered in detail.
N.V.
Methods: Lab techs assessed 3 commercially available NGS cancer-targets panels on Illumina's MiSeq Dx. Kits were evaluated utilizing a validation set consisting of samples previously received and tested with non-NGS technologies. Validation set 50 FFPE (resections, biopsies and cytology materials with varying site source) and 26 blood/bone marrow samples. Technologists appraised each kit's process using a checklist consisting of the following 10 parameters: DNA input, cost per sample, cost of initial set up, peripheral equipment needs, reagent needs, hands on steps, turnaround time, ease to abide by clinical lab guidelines and ease of transfer to bioinformatics. Techs were asked to give general list of brief pros and cons of each kit from bench standpoint also. Results: Kits yielded data with 90% of reads with Q and variant analysis was performed with software from vendors. Kits yielded consistent results with variation in total reads, percent reads on target and depth of coverage. Variant analysis yielded similar results and were concordant with results from previous methodology. Each kit had strengths and weaknesses with bench workflow when evaluated with outlined checklist of parameters. Team evaluation of each checklist with specific parameters carrying more weight generated one kit ranking superior to others. Conclusions: Validating NGS in a clinical lab is a challenge that when strategically planned can be successful even in a mid-sized lab. With many commercial kits offered our lab found it best to outline specific parameters and evaluate kits with those parameters. The capacity of our facility to handle peripheral needs of each process whereas adhering to CLIA, CAP and ISO guidelines carried most weight. We found that the most bench level hands on and costly the large vendor's kit worked best in initial launch of NGS into routine lab operation. N.V. Guseva, A.A. Stence, R. Sompallae, J.C. Schade, M.R. Tanas, A.D. Bossler, A.M. Bellizzi, D. Ma University of Iowa Hospitals and Clinics, Iowa City IA Introduction: Although initially recognized as a tumor involving the pleura only, solitary fibrous tumor (SFT) has been shown to occur essentially at any site within the body. 10% to 15% of SFTs metastasize and metastatic tumors are uniformly lethal with no effective medical therapy. The behavior of SFT is difficult to predict based on morphology. In some cases, it is difficult to differentiate SFT from other tumors with a staghorn-like vascular pattern. Recently, a gene fusion between NAB2 and STAT6 was identified as the defining molecular signature of SFT and different fusions correlated with histology and behavior. Due to the proximity of NAB2 and STAT6 on chromosome 12, this fusion is not amenable to detection by FISH. We evaluated 12 cases of SFT using a next generation sequencing (NGS)-based assay which allows simultaneous detection of multiple fusions and specific breakpoints at single-base resolution. The NGS results were correlated with immunohistochemical (IHC), clinical and pathologic findings. Methods: Twelve tumors from 10 patients with a pathologic diagnosis of SFT were selected. Total nucleic acid was extracted from formalin-fixed, paraffin-embedded (FFPE) tissue. Targeted RNA sequencing for gene fusions was performed using the Universal RNA Fusion Detection Kit (ArcherDX), the ArcherTM FusionPlexTM Sarcoma Panel and the Ion Torrent PGM (Life Technologies). Data were analyzed using the Archer Analysis Pipeline 3.0. The fusions detected were confirmed by RT-PCR followed by Sanger sequencing. IHC for STAT6 were performed using a polyclonal antibody (sc-621, Santa Cruz) at a dilution of 1:1000. Results: NAB2-STAT6 fusion was identified in all tumors. Five types of fusions were detected: NAB2ex4-STAT6ex2 (4 cases of malignant and 1 case of benign SFT), NAB2ex2-STAT6ex5 (2 metastatic SFTs from the same patient), NAB2ex6-STAT6ex16 (2 cases of malignant hemangiopericytoma (HPC)/SFT), NAB2ex6-STAT6ex17 (1 malignant HPC/SFT and 1 SFT with malignant features) and NAB2ex3-STAT6ex18 (1 malignant HPC/SFT). The latter three gene fusions were reported to be associated with more aggressive tumors. When 2 tumors from the same patients (2 different patients) were tested, fusion products with the same breakpoints were detected. The presence of all gene fusions was confirmed by RT-PCR followed by Sanger sequencing. IHC for STAT6 showed strong nuclear staining in all tumors. Conclusions: NAB2-STAT6 fusions with various breakpoints can be reliably detected by this NGS-based assay. NAB2ex6-STAT6ex16/17 and NAB2ex3-STAT6ex18 fusions were associated with tumors with HPC morphology and more aggressive behavior, consistent with report in the literature. The identification of specific fusions and breakpoints can help with diagnosis and may predict tumor behavior.
M. Abedalthagafi 1,2 1 Brigham and Women's Hospital, Boston, MA; 2 King Fahd Medical City, Riyadh, Saudi Arabia. Introduction: The appropriate use of adjuvant therapy in patients with gross totally resected atypical meningioma requires an accurate assessment of recurrence risk. We sought to determine whether cytogenetic/genetic characterization may facilitate better estimation of the probability of recurrence. Methods: We first analyzed our clinical database including high resolution DNA copy number data to identify 11 common copy number aberrations in a pilot cohort of meningiomas of all grades. We
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org summed these aberrations to devise a cytogenetic abnormality score (CAS) and determined the CAS from archived tissue of a separate cohort of 32 patients with gross totally resected atypical meningioma managed with surgery alone. Propensity score adjusted Cox regression was used to determine whether the CAS was predictive of recurrence. Results: An association between higher CAS and higher grade was noted in our pilot cohort with heterogeneity among atypical tumors. Among the 32 patients who underwent gross total resection of an atypical meningioma, the CAS was not significantly associated with age, gender, performance status, or tumor size/location but was associated with the risk of recurrence on univariable analysis (HR per aberration, 1.52, 95% CI 1.08-2.14, p=0.02). After adjustment, the impact of the dichotomized number of copy aberrations remained significantly associated with recurrence risk (HR 4.47, p=0.05) . Conclusions: The number of copy number aberrations is strongly associated with recurrence risk in patients with atypical meningioma following GTR and may inform the appropriate use of adjuvant radiation therapy in these patients or be useful for stratification in clinical trials. Introduction: Many patients with pancreatic adenocarcinoma have metastatic disease at presentation. Prognosis is poor, with a median survival of only a few months. Understanding the progression of metastatic disease is essential to the development of screening and treatment methods that may enhance patient survival. In this study, our goal was to investigate the genetic relationship of metastatic pancreatic tumor to its primary counterpart using next-generation sequencing technology. Methods: Fifteen patients with pancreatic adenocarcinoma and metastatic disease were selected. Specimens included whipple resections, excisional biopsies and fine needle aspirates (FNAs). An attending pathologist reviewed H&E stained slides, and selected paired blocks containing >10% tumor content of primary and metastatic tumor. Of the 15 cases, seven cases had no tissue remaining in one of the paired blocks; therefore these cases were not sequenced. Sequencing was performed on both primary and metastatic tumor of the eight remaining cases. DNA was extracted from 8 FFPE slides, and quantified using the PicoGreen method. Barcoded libraries were prepared and sequenced on the Ion Torrent PGM using 318 chips. Variants were annotated using GoldenHelix SVS software 8.2.1. Paired genomic profiles of primary and metastatic tumor were then compared. Results: One lung biopsy and 1 FNA of the pancreas failed qPCR. Of the remaining cases that we sequenced, 4 contained metastasis to the liver, one contained metastasis to the lung and to the skin/soft tissue, and one contained metastasis to the omentum. All cases contained KRAS and TP53 variants in both the primary and metastatic tumor. One case gained an APC E1317Q mutation in a metastatic liver lesion, whereas all of the remaining cases retained the same genetic profile between primary and metastatic tumor. A CDKN2A mutation was present in one case and SMAD4 was lost in one case. Conclusions: The majority of cases in our study retained their genomic profile between pancreatic primary and metastatic tumor, with the exception of one case gaining an APC mutation in the metastatic tumor. Larger studies may yield more information about the distribution of variants in tumors that have metastasized to similar sites. Further research of the genetic framework of metastasis will lead to a better understanding of the pathogenesis of pancreatic cancer and its impact on patient prognosis.
University of Nebraska Medical Center, Omaha, NE. Introduction: Next-generation sequencing (NGS) cancer panels promise to more broadly and efficiently deliver clinically actionable mutation information to either direct targeted therapy or suggest potential clinical trials for which a patient may be eligible. Current guidelines recommend testing colorectal tumors for extended RAS and BRAF mutations to direct targeted therapy or assist in prognostication and Lynch Syndrome stratification, respectively. A five gene (KRAS, NRAS, HRAS, PIK3CA, BRAF) clinical colorectal cancer panel was developed using the Ion AmpliSeq Cancer Hotspot Panel v2 on the Ion Torrent PGM System and clinically instituted at Nebraska Medicine in January 2015. Methods: Fourteen (14) formalin-fixed, paraffin-embedded colorectal cancer specimens were evaluated to ensure >20% tumor tissue. DNA was extracted and subjected to library preparation using the 50 gene Ion AmpliSeq Cancer Hotspot Panel v2, clonal amplification and Ion Sphere Particle enrichment on the Ion One Touch 2 and One Touch ES, and NGS on the Ion Torrent PGM System. Data was analyzed using Variant Caller Software v.4.0 and NextGENe software v.2.3.4. The original target region BED file for the Ion AmpliSeq Cancer Hotspot Panel v2 was modified to include only the 21 amplicons (covering over 200 hotspots) for KRAS, BRAF, NRAS, HRAS and PIK3CA. The BED files were then uploaded to the Torrent Server and NextGENe Software for reanalysis of solely those regions. Coverage was >400X for all specimen amplicons evaluated and averaged approximately 3000X. The GenomOncology Clinical Workbench was used for mutation annotation and reporting of targeted therapy and clinical trial information. Results: Analyzed tumor percentage ranged from 25% to 70% for the fourteen cases. Of these fourteen cases, 10 (71%) had at least one mutation identified and 4 (29%) had extended RAS mutations (two c.436G>A;p.A146T KRAS non-exon 2 mutations and two NRAS mutations). Five PIK3CA mutations were identified, three of which were concomitant with KRAS mutations (50% of KRAS-mutated cases). Neither BRAF nor HRAS mutations were identified. In addition to targeted therapy recommendations, clinical trial information (within a 500 mile radius) was reported for all 10 positive cases (average 6.7 clinical trials reported per case). Conclusions: The two extended KRAS mutations would be missed by current in vitro diagnostic KRAS assay design. KRAS and PIK3CA mutations occur concomitantly in a subset of colorectal cancers. Panel testing by NGS efficiently identifies extended RAS and other important mutations in genes involved in colorectal cancer pathogenesis. This information can be useful in both directing targeted therapy and identifying clinical trials for which patients may be eligible.
Carcinomas Using the MethyLight Assay S. A. Turner, F.B. de Abreu, E.Z. Liu, G.J. Tsongalis, J.A. Lefferts, L.J. Tafe Dartmouth-Hitchcock Medical Center, Lebanon, NH. Introduction: Defective DNA mismatch repair (MMR) is a common contributor to tumorigenesis, and is well characterized in colorectal (CR) and endometrial (EM) cancers. Around 3% to 5% of CR tumors carry heritable mutations in MMR genes, MLH1, PMS2, MSH2, or MHS6 resulting in reduced expression and a diagnosis of Lynch Syndrome (LS). EM carcinomas represent the most common extracolonic malignancy in these patients. To distinguish patients with LS from those with sporadic cancer, screening guidelines suggest immunohistochemistry (IHC) staining for MMR gene expression in all CR and EM tumors. The majority of tumors with loss of MLH1 expression contain 'somatic' methylation of the MLH1 promoter. By determining methylation status, further genetic testing including MMR gene sequencing can be avoided. Here, we describe the validation of an assay to determine MLH1 methylation status. Methods: gDNA was isolated from 15 CR and 5 EM formalin-fixed paraffin-embedded tumors (FFPE) using Gentra PureGene Kit (Qiagen). 50ng of sample DNA, CpGenome Universal Unmethylated DNA control and CpGenome Universal Methylated DNA control (Chemicon International) were bisulfite converted using the MethylEdgeTM Bisulfite Conversion System (Promega, Madison, WI). Methylation was assessed using a quantitative methylation assay that utilizes fluorescence-based real-time qPCR technology, MethyLight. qPCR was performed using primers and probes for the methylated 'C region' of MLH1 gene promoter and the -actin (ACTB) reference gene. A methylation standard curve for both genes was obtained using a converted methylated control DNA to determine the relative methylation levels of each sample. The percentage of methylated reference (PMR) was calculated by dividing the MLH1sample/methylated control ratio by the ACTB sample/methylated control. A PMR value of greater than 4 was defined as methylated and was previously shown to correlate with loss of MLH1 expression. Results: Of 20 FFPE tissues we identified 14 positive and 6 negative for MLH1 methylation resulting in 100% concordance with external lab testing. All positive samples had a PMR value greater than 5 with 11 samples having a PMR value of greater than 10. All negative samples had a PMR value of less than 0.2. A 5% methylated control sample run with each assay showed a PMR range of 3.5 to 6.5. Primer efficiencies for MLH1 and ACTB ranged from 90% to 110% and 95% to 115%, respectively, for all runs. Conclusions: MethyLight is a quantitative and highly accurate assay capable of determining MLH1 promoter methylation status. MLH1 promoter methylation is used in screening algorithms for LS and should be available in the workup of IHC MLH1 negative tumors. D.L. Duncan, K. Greene, A.L. Treece, S. Elmore, M.O. Meyers, N. Patel, M.L. Gulley University of North Carolina, Chapel Hill, NC. Introduction: Gastric adenocarcinoma is not one disease but rather is comprised of four molecular classes: EBV-positive, microsatellite instability (MSI), genomically stable, and chromosome instability. We mined data from The Cancer Genome Atlas to create a 28-gene predictor of molecular class and to inform options for clinical trial enrollment among patients with metastatic adenocarcinoma of the stomach or gastroesophageal junction (GEJ). Methods: Macrodissected tumor from 26 formalin fixed paraffin embedded tissues was tested using massive parallel sequencing of selected regions of 26 human cancer genes (Illumina TruSight Tumor reagents on a MiSeq). Non-synonymous mutations and small indels were cataloged if they had allele frequency >5% with population frequency < 1%. Q-rtPCR was used to measure 5 Epstein-Barr virus (EBV) encoded microRNAs reflecting the EBV-positive class of tumors. MLH1 methylation was tested in a subset of tumors to assist, along with evidence of hypermutation, in identifying the microsatellite instability class. Results: By histology, 6 tumors were intestinal, 8 were diffuse (suggesting the genomically stable class), 1 was mixed, and 11 were not specified. Ten were located in antrum, 7 in body/fundus, 5 were gastroesophageal/cardia, and 4 anatomic sites were not specified. Six patients had no detectable mutations, whereas the remaining 20 patients had between 1 and 4 mutations per tumor specimen. The most commonly mutated genes were TP53, APC, FBXW7, KRAS and CDH1. Less commonly mutated were SMAD4, PIK3CA, GNAS, ALK, CTNNB1, and MSH6. Average read depth was >10,000, and average mutant allele frequency was jmd.amjpathol.org ■ The Journal of Molecular Diagnostics 15%. EBV microRNA was expressed in 3/25 cancers tested, and EBER was localized to malignant cells in one of those tumors. Nine tumors had TP53 mutation characteristic of, but not specific for, chromosome instability class. 8/26 patients had results considered actionable for purposes of clinical trial enrollment (5 KRAS, 2 PIK3CA, 1 EBV positive). Conclusions: The spectrum of genomic findings shed light on the intrinsic biology of each tumor. Actionable genomic variants were detectable in nearly a third of gastric adenocarcinomas, which adds options for targeted therapy beyond Her2 targeted therapy for which ERBB2 overexpression/amplification is standard of care in recurrent/metastatic cancer. In addition to their value for selecting clinical trial options, KRAS or PIK3CA mutation is considered prognostic in certain gastric cancer subtypes. Introduction: The landscape of somatic mutations in the common human cancer has recently been characterized using exploratory next generation sequencing (NGS) on high quality DNA from fresh frozen surgical cancer specimens. However, in order to determine the mutation spectra of patients in routine clinical practice, diagnosed by small FFPE biopsies with limited tumor yield, we need target enrichment assays that can deliver a high sequence depth from small amounts of partially degraded DNA. Methods: Here we describe our work to optimize gene panels for mutation analysis of solid tumor biopsies using the HaloPlex targeting system, addressing small DNA input, short fragmented DNA targets, and sequence artifacts due to oxidation and deamination. We have optimized and validated gene panels for clinical diagnostic use, by comparison to pyrosequencing and quantitative PCR (qPCR) in a large set of clinical samples with known mutational status of hotspots in KRAS, BRAF, NRAS, PIK3CA and EGFR. Results: Our results show a complete concordance between the previously defined pyrosequencing and qPCR genotypes and the corresponding variants detected using NGS. The variant allele frequencies estimated by NGS and pyrosequencing correlated well, with few outliers. Both point mutations and smaller indels (<25 bp) were successfully identified. The technical sensitivity of mutation detection was determined to at least 2%. Conclusions: We believe that the established cancer gene panels for targeted enrichment and NGS can replace pyrosequencing and qPCR for molecular diagnostics in solid tumors and will be useful for screening of unselected populationbased prospective and retrospective cancer patient cohorts in clinical research. Introduction: MET proto-oncogene on chromosome 7q31 encodes receptor tyrosine kinase, c-MET, for the hepatocyte growth factor (HGF). Constitutive activation of c-MET in cancer promotes uncontrolled cell proliferation that leads to aggressive tumor behavior and poor clinical prognosis. Increased MET copy numbers can be resulted from MET amplification and polysomy chromosome 7. Clinically, fluorescent in situ hybridization (FISH) on formalin fixed paraffin embedded (FFPE) tissue is the standard test used to detect MET amplification, using the ratio of MET/centromere 7 >2.0. When the ratio is <2.0 with increased copy number gains of both MET and centromeric region or with polysomy 7, FISH is often reported as negative for MET amplification. The array based technologies such as array comparative genomic hybridization (array CGH) and single nucleotide polymorphism array (SNP array) enable the assessment of copy number aberrations in tumor genome. In this study, we used OncoScan SNP array to re-assess lung adenocarcinoma cases reported as negative for MET amplification by FISH but have increased MET copy numbers. Methods: FFPE tissues from selected 14 patients were obtained from the archives at the U.T. M.D. Anderson Cancer Center (2013 to 2015). Cases were separated into 3 groups based on FISH results: positive for MET amplification, increased MET copy numbers, and negative for MET amplification. OncoScan (Affymetrix) was performed and Nexus Copy Number software (BioDiscovery) was used to obtain MET and genome-wide copy number profiles. Results: MET copy number by OncoScan showed good concordance with FISH in MET amplified and negative groups. When copy number=4 cut-off is applied in the increased group, 3 out of 7 cases were re-interpreted as positive for MET amplification. In addition, MET amplified and increased MET copy numbers groups showed similar patterns of gains and losses that were distinct from the negative group. On average, the percent genomes changed in MET amplified, increased, and negative cases were 59.00, 48.41, and 27.37%, respectively. Furthermore, t-test showed that percent genome changed in positive and increased groups were significantly different from the negative group (p=0.025, p=0.046), whereas there was no significant difference between positive and increased group (p=0.378). Conclusions: Our data show that OncoScan might be useful as a reflex test for cases with increased MET copy numbers but are reported as "negative for amplification" by current FISH criteria. In addition, similar genomic profiling in the increased MET copy numbers and MET amplified groups suggests that the current FISH reporting criteria could be refined to potentially help identifying more patients eligible for MET targeted therapy-tyrosine kinase inhibitors.
V. Sero 1,2 , C. Forcato 2 , G. Buson 2 , P. Tononi 2 , C. Bolognesi 2 , F. Fontana 2 , G. Medoro 2 , N. Manaresi 2 , F. Bischoff 1 1 Silicon Biosystems Inc., La Jolla, San Diego, CA; 2 Silicon Biosystems S.p.A., Bologna, Italy. Introduction: We have previously shown reliability in isolating pure populations of cells from complex tissues using the DEPArray. Macrodissection, the gross manual dissection of FFPE samples guided by a histologic section, is used to isolate areas of interest within a specimen for optimal downstream analysis. Though better defined than a non-dissected tissue sample, the actual percentage of targeted cells obtained by macrodissection is dependent on the degree to which tissue architecture and admixed cellular components allow. Here we demonstrate preliminary results showing the degree of stromal cell contamination following macrodissection and the concurrent recovery of 100% pure tumor cell populations using the DEPArray. Methods: FFPE macrodissected sections (n=4;) originating from prostate; breast, pancreatic and lung primary tumors were evaluated. Each was processed using a disassociation and staining procedure followed by DEPArray sorting based on cytokeratin (Ker), vimentin (Vim) and nuclear staining (DAPI) to verify DNA content. Recovered cell populations were then directly lysed in the collection tube prior to PCR-based target enrichment for next generation sequencing using Ion AmpliSeq CHPv2. Results: DEPArray analysis allowed identification of well separated cell populations, including tumor (Ker+/Vim-) and stromal (Vim+/Ker-) cells in 3 out of 4 samples. We were thus able to estimate the % of tumor cells (mean 11% range 4% to 18%), demonstrating an unexpected low frequency of tumor cells remaining following macrodissection, In fact, 89% (82% to 96% range) of the cells analyzed were of stromal origin. For subsequent NGS analysis, groups of pure cells (mean 88 cells, range 37 to 183) for each population were recovered. Among only the tumor cells isolated from the breast cancer specimen, we observed three homozygous somatic variants (FGFR2, PTEN, and TP53 genes), all annotated in COSMIC database and with missense/stop-gain effect in the coding sequence of related genes. However, the unsorted (prior to DEPArray) population shows a frequency of approximately 30% for all three variants, whereas only the wild-type allele is present in the recovered pure stromal cells. This implies that the unsorted population contains the wild-type alleles, coming from the stromal cells. The amount of this allelic contamination seems to be approximately 70%, consistent to DEPArray analysis. Conclusions: DEPArray technology can be used to isolate pure tumor cells from heterogeneous FFPE samples that have undergone macrodissection. Thus, the DEPArray platform brings digital precision to detection, quantification and recovery of pure target cells for subsequent downstream molecular analysis that can improve cancer diagnosis and treatment decisions. Introduction: Over the last several years, there has been an increase in the implementation of next-generation sequencing (NGS)-based assays in clinical laboratories. Availability of multiple sequencing platforms offers opportunities for workflow optimization and improvement in laboratory efficiency. Here, we compare the performance of the two currently available semi-conductor-based NGS platforms, Ion PGM and Ion Proton, using the commercially available Ion AmpliSeqCancer Hotspot Panel v2 (Thermo Scientific). Methods: Ten (10) ng of DNA from 31 manually, micro-dissected formalin-fixed paraffin-embedded (FFPE) patient tissue specimens with >20% tumor and one FFPE treated cell-line control was used to generate libraries as per manufacturer's instructions using the Ion AmpliSeq Library Kit 2.0 and unique barcodes from the Ion Xpress Barcode Adapters 1-96 Kit (Thermo Scientific). After the libraries were quantified using the Ion Library TaqMan Quantification Kit, they were normalized and pooled. 20 pM of pooled libraries were loaded on the Ion PGM (13 samples/chip) and the Ion Proton (32 samples/chip). Sequencing on the Ion PGM and Ion Proton was performed using Torrent Suite 4.2.1 and the Ion Sequencing 200v2 kit on a 318v2 chip and using Torrent Suite v4.4 and the Ion PI HI-Q Sequencing 200 Kit on a PI v3 chip, respectively. Comparison of the mutation calls was performed, along with cost analysis and time study. Results: Seventy-three mutation calls resulted from the Ion PGM sequencing. There was a 96% (70/73) concordance in mutation calls between the Ion PGM and Ion Proton sequencing analysis. Three TP53 mutations (c.532C>G p.H178D, c.536A>T p.H179L, and c.524G>A p.R175H) (NM_000546.5) were called as delins by the TS4.4 caller. Comparison of variant frequencies showed a 0.93 correlation coefficient. The AQ20 reads on the Ion Proton were on the average 10 times higher than on the Ion PGM indicating sample number/chip can be doubled. There was approximately a 10% cost saving in running the 50-gene panel on the Ion Proton (32 samples/chip) versus the Ion PGM (12 samples/chip) due to higher equipment maintenance associated with the Proton. On average, the hands-on time per sample was the same for each platform. Conclusions: Our study shows feasibility to run the 50-gene panel on the Ion Proton. Cost savings per sample can
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org be improved by automating wet bench processes and increasing the sample batch size on Proton. Factors such batch size, scheduling, and turn-around-time should be considered prior to implementation of higher throughput Ion Proton in a clinical laboratory.
Z. Kolar, A. Burdova, J. Bouchal, K. Smesny-Trtkova, P. Rulisek, M. Klimesova Faculty of Medicine and Dentistry, Palacky University, Olomouc, Czech Republic. Introduction: TMPRSS2-ERG fusion in prostate cancer cells is determined by double-strand DNA breaks induced by androgen signaling and transcription stress. The ERG oncogene can induce and modulate prostate inflammation which is associated with increased risk of prostate cancer. However, role of factors such as polymorphism of the androgen receptor (AR) gene on androgen signaling in this process has not been adequately investigated. Methods: We examined the infiltration patterns of CD204 and CD3 positive cells in areas of benign hyperplasia (BPH), intraepithelial neoplasia (PIN) and carcinoma of the prostate (PCa) and correlated these with ERG rearrangement and expression as well as with clinicopathological data including AR gene polymorphism (SNP rs6152) in a cohort of 100 patients with prostate cancer who had not undergone androgen ablation therapy. In this pilot study we used immunohistochemical methods for evaluation of CD204, CD3 and ERG expression, FISH for analysis of fusion status, direct PCR-RFLP on blood and tissue samples for the evaluation of polymorhism status and relevant statistical methods. Results: TMPRSS2-ERG positive patients had more predisposition to infiltration malignant structure of the prostate by CD204+ and CD3+ cells than patients without gene rearrangement, higher expression ERG and more inflammatory "hot spots." Patients with fuse gene and type G polymorphism had significantly lower CD204+ infiltration than patients with type A polymorphism. The different roles of CD204+ and CD3+ cells are suggested by the fact, that the most significant changes in infiltration between patients with fusion and without fusion were found in PIN and PCa areas for CD204+ whereas in BPH areas for CD3+. Conclusions: The greater abilitiy of inflammatory cells with CD3+ and CD204+ phenotype to infiltrate malignant and premaligant regions in the prostate may be a consequence of androgene signaling mediated by an activated ERG oncogene and influenced by AR polymorphism. L. Ma, L. Mazur, A. Sung, B. Baltadjieva, M. Mihalov ACL Labs, Rosemont, IL. Introduction: Extended RAS testing of metastatic colorectal carcinoma (mCRC) patients is now standard of care because mutations in the RAS gene are associated with poor response to targeted therapies and assay results determine therapy selection. The most commonly reported mutations are in codons 12 and 13 of exon 2 of the KRAS gene, and less commonly at several codons of exons 3 and 4 of KRAS and 2, 3 and 4 of NRAS. In this report, we describe several novel RAS mutations, detected in our laboratory. Methods: Two hundred and three (203) formalin fixed, paraffin embedded (FFPE) samples from mCRC patients were tested. Tumor tissue was manually micro dissected, and DNA extraction was performed using PinPoint extraction kit (Zymo Research). 1) PANClamp KRAS and NRAS mutation detection with real-time PCR kit (PANAGENE INC) performed on ABI 7500. Sensitivity is 3% allele mutant present. 2) Pyrosequencing with RAS extension Pyro Kit (QIAGEN GmbH), performed on Pyromark Q24 instrument. Sensitivity is 5% allele mutant present. 3) Laboratory Developed Assays (LDA) based on Sanger Sequencing for KRAS and NRAS exon 2/3/4, performed on 3130 XL Genetic Analyzer (Life technologies/ ThermoFisher). Sensitivity is 15% allele mutant present. Results: From the 203 tested specimens 61(31%) were positive for KRAS mutations (codon 12 and 13). From the remaining 142 specimens 23 (16.2%) were found to harbor other KRAS or NRAS mutations including: KRAS 61 (0.7%), KRAS 117 (0.7%), KRAS 146 (4.9%), NRAS 12 (2.8%) and NRAS 61(4.2%). Each mutation was detected by all three methods. In addition 6 new mutations (4.2%) were detected; including three point mutations: KRAS codon 14, KRAS codon 149, NRAS codon 135; and three novel complex mutation on KRAS codons 60 and 61. These novel mutations were detected only by Sanger sequencing and were missed by the other methods. Five of these mutations were not registered in COSMIC database. Conclusions: Our study has identified a number of RAS genomic variants not routinely detected in traditional hot-spot screening assays. Although the precise functional deficits introduced by these new RAS mutations are still unclear and need further study, they could potentially affect therapy selection. We conclude therefore that choosing the appropriate screening technique is crucial for thorough extended RAS mutation analysis. Morphologically these tumors show osteoid and/or chondroid matrix and other features that overlap with their primary bone counterpart. Herein, we investigate the clinicopathological and molecular features of ESOSA in a series of cases and explore potential morphological and molecular parameters that may affect outcome. Methods: 32 cases were retrieved and reviewed. Clinical history and followup was obtained through electronic record review. DNA from FFPE tissue was extracted and processed from 27 cases. Genome-wide DNA copy number alterations (CNA) and allelic imbalances (AI) were analyzed by SNP-array using Affymetrix OncoScan FFPE Assay (Affymetrix, CA, USA). Massive high throughput deep parallel sequencing was performed using a customized panel targeting 410 cancer genes. Log rank, fisher exact test and Cox were used for statistical analysis. Results: Our series includes years (19-93) and a median follow up of 24 months (6-120 months). Age more than 60 and positive margin status were associated with worse outcome overall survival and local recurrence, respectively (p=0.0017, p= 0.0279). Frequent genomic alterations included copy number losses in Tumor suppressor genes including CDKN2A (70%), TP53 (56%) and RB1 (49%). Mutations affecting methylation/demethylation, chromatin remodeling and WNT/SHH pathways were identified in 40%, 27%, and 27%, respectively. No evidence of Kataegis or structural rearrangements affecting TP53 was seen. 11% of the cases showed PIK3CA and TERT promoter variant mutations previously reported in other sarcomas. Unsupervised clustering analysis of genetic events showed that cases harboring simultaneous TP53 and RB1 biallelic CN losses were associated with worse OS and LR (p=0.04, p=0.02). CDKN2A losses were associated with worse -survival (p=0.002). Conclusions: Our findings suggest that age>60, positive margin status, simultaneous biallelic TP53 and RB1 losses and CDKN2A loss are associated with a worse outcome in ESOSA. Although both, conventional pediatric OS and ESOSA share genetic similarities, there are notable dissimilarities such as absence of kataegis and mechanistic differences through which these pathways are involved in ESOSA. Introduction: Tumor-related microRNA profiles are promising diagnostic tools, but finding the most specific tumor signature in the complex network of inter-dependent microRNAs is challenging. Prior work identified microRNAs differentially expressed in carcinoma compared to normal gastric mucosa, but almost nothing is known about expression of these factors in premalignant lesions adjacent to cancer. Methods: MicroRNA expression profiling was performed by rtPCR on 107 FFPE samples that were macrodissected to enrich for adenocarcinoma, dysplasia, gastritis, and non-inflamed gastric mucosa. All samples were obtained from gastrectomy or diagnostic biopsy specimens for gastric adenocarcinoma. Each specimen was analyzed with a panel of 15 microRNAs that were selected based on previous work identifying relevant microRNAs in gastric cancer and reliable normalizer microRNAs. One-way analysis of variance (ANOVA) was performed to identify microRNAs with differing expression across the 4 sample types, and subsequent Tukey-Kramer post-hoc analysis was utilized to characterize expression profiles. Results: ANOVA identified 9 microRNAs with significantly different expression among the four sample types: miR-196a, miR-196b, miR-25, miR-19a, miR-106b, let-7i, miR-30e, miR-23a , and miR-185 (all adjusted p<0.003201). Posthoc analysis identified two distinct expression patterns for these microRNAs. Although all of the microRNAs were differentially expressed between tumor tissue and mucosa (all p<0.05), only miR-196a and miR-196b also showed differential expression between tumor tissue and gastritis samples (p<0.05). The remaining 7 microRNAs were differentially expressed in gastritis versus mucosa (p<0.05), but no difference in expression was identified between tumor tissue and gastritis. Dysplasia and adenocarcinoma samples showed no significant difference in expression across all microRNAs. Conclusions: These results strengthen prior evidence that miR-196a and miR-196b are key markers of gastric adenocarcinoma, with consistent overexpression in dysplasia and invasive cancer compared to gastric mucosa and inflamed mucosa. Furthermore, this analysis suggests the presence of a discrete microRNA expression profile that may be attributable to tumor-associated inflammation in lesions adjacent to malignancy. It is critical to acknowledge the influence of tumor microenvironment and "field effect" on expression of microRNA when attempting to select the most relevant and specific analytes for clinical assay development.
Sorting and Recovery of 100% Pure Tumor Cells for NGS, Using the DEPArray System V. Sero 1 , C. Forcato 2 , G. Buson 2 , C. Bolognesi 2 , F. Fontana 2 , G. Medoro 2 , N. Manaresi 1,2 , F. Bischoff 1 1 Silicon Biosystems Inc., La Jolla, San Diego, CA; 2 Silicon Biosystems S.p.A., Bologna, Italy. Introduction: In earlier studies, we have demonstrated reliable recovery of pure populations of (rare) cells from complex tissues using the DEPArray system. Fine Needle Aspiration (FNA) is often an outpatient and safe procedure routinely used to examine tissue or bodily fluid from a lesion or cyst helping to make a diagnosis or rule out conditions such as cancer. Although FNA is also used to assess the effect of jmd.amjpathol.org ■ The Journal of Molecular Diagnostics treatment, the procedure may often appear to have failed due to insufficient number of target cells in the sample and contamination with normal cells. Here we provide preliminary results showing 100% efficiency in recovering pure tumor cell populations from FNA samples known to have low tumor burden using the DEPArray platform. Methods: FNA paraffin embedded sections (n=5; one 50 microns each) from metastases originating from pancreatic (n=2) and breast (n=3) primary tumors were evaluated. Each was processed using a gentle tissue disassociation and staining procedure followed by DEPArray sorting based on cytokeratin (Ker), vimentin (Vim) and nuclear staining to verify DNA content. The recovered cell populations were directly lysed in the collection tube prior to PCR-based target enrichment for next-generation sequencing using Ion AmpliSeq CHPv2. Results: DEPArray sorting allowed identification of 3 distinct cell populations representing tumor (KER+), stromal (VIM+) and putative EMT (KER+/VIM+) cells. Overall, only 21% (4.3% to 42.7% range) of the total (mean of 6335) cells analyzed were of tumor (KER+) origin. Groups of pure cells (mean 105 cells, range 15-200) for each population were recovered for sequence analysis. In one of the breast cancer FNA samples analyzed, TP53 loss of heterozygosis was observed but only in the sorted tumor (KER+) cells and not in the unsorted, stromal (VIM+), or EMT (KER+/VIM+) populations. In addition a PIK3CA missense somatic heterozygous variant was detected in both tumor and EMT populations but notably was absent in stromal cells confirming it as a somatic mutation. Conclusions: DEPArray resolves two pressing problems associated with FNA samples obtained for genomic analysis: too few target cells and unwanted admixture of normal cells. DEPArray allows for phenotypic distinction between the sorted cells. Moreover, this approach allows for sequence analysis that is suitable for detecting genomic aberrations such as CNVs and LoH, which cannot be evaluated as precisely in the unsorted sample. Thus, the DEPArray platform brings precision to detection, quantification and recovery of pure target cells for subsequent downstream molecular analysis that can improve cancer diagnosis and treatment decisions. Introduction: Targeted deep sequencing application in clinical routine is hindered by the necessity of using formalin-fixed, paraffin-embedded (FFPE) specimens, characterized by contamination of normal cells, minute amount of tissue, and DNA damage. This results in lack of confidence in variant calling. Confirmation with alternative sequencing platforms is recommended, but often impossible due to insufficient sample availability. Here we show how, independently of sequencing platform, only analysis of digitally-sorted, pure, cell subpopulations yields all and only true-positive actionable somatic variants. Methods: Using DEPArray digital sorter, we recovered precise numbers of cells (range 140 to 300) from 100%-pure, cancer cells and normal stromal cells subpopulations within FFPE tissues (distinguished by Keratin/Vimentin immunofluorescence and DNA content) from lung (n=2) and pancreatic ductal adenocarcinoma (n=1) patients. After lysis, libraries from cell recoveries as well as unsorted samples were prepared for both IonTorrent PGM and Illumina MiSeq sequencing, using AmpliSeq Cancer Hotspot Panel v2 and Swift Biosciences Accel-amplicon 56G Oncology panel, respectively. Both PCR-based target enrichment kits can work with small amount of DNA as required by minute clinical specimen. We then carried out a comparative analysis of the variant calls obtained by both sequencers, retaining only positions included in both panel designs. Results: Variant interpretation from DEPArray sorted cells (tumor versus stromal) clearly identified -in homozygosis-4/4 (100% specificity) true-positive nonsynonymous somatic variants (in TP53 and STK11 genes), 3 out 4 annotated in COSMIC database, 100% concordant across platforms. By contrast, following variant interpretation from only unsorted samples, specificity decreased sharply to 4/24 (16%) for both PGM and MiSeq, still with good 22/26 (85%) concordance and 100% sensitivity, due to the increase of false positive somatic heterozygous misinterpretation of actual germline variants. Interpretation of unsorted variants versus matched normal (deduced from sorted stromal cells) improved specificity to 4/5 80% for both platforms, with decreased 4/6 (66%) concordance. In addition, DEPArray sorting evidenced Loss-of-heterozygosis (LOH) events (n=8), in both PGM and MiSeq, otherwise undetectable in unsorted samples (alone or versus normal). Conclusions: Our data shows that i) both AmpliSeq-PGM and Accel-Amplicon-MiSeq workflows provide consistent results on low DNA inputs (0.9ng to 2ng), ii) comparison with matched normal is mandatory for correct interpretation of actionable somatic mutations, iii) only sorting cells by DEPArray guarantees the utmost accuracy required by clinical actionability of NGS results, including LoH events.
Adenocarcinoma and Relationship to Other Gene Expression Prognostic Panels G. Mayhew 1 , N. Hayes 2 , C. Perou 2 , M. Lai-Goldman 1 , H. Faruki 1 1 GeneCentric Diagnostics, Durham, NC; 2 University of North Carolina, Chapel Hill, NC. Introduction: The Lung Subtype Panel (LSP), a previously validated 52-gene expression signature for classifying FFPE lung tumor samples into Adenocarcinoma (AD), Squamous Cell Carcinoma (SQ), and Neuroendocrine (NE) (NE comprising small cell carcinoma and carcinoid), was evaluated for its prognostic value in multiple publically available datasets. Prognosis by LSP was compared to several other prognostic gene expression panels, developed for Non Small Cell Lung Cancer (NSCLC) or Adenocarcinoma (AD). Survival predictions, and prognostic strength, as compared to LSP, was examined. Methods: The LSP 3-class nearest centroid predictor developed in array data was applied to stage I and II AD samples in TCGA (RNAseq, n=399), the Director's Challenge (Affy array, n=371), and Tomida et al (Agilent array, n=92) datasets. Samples were predicted as AD, SQ, or NE. Kaplan Meier plots and logrank tests were used to assess and compare 5-year survival in two gene expression groups, namely histologically-defined AD predicted AD (AD-AD) and histological AD predicted SQ or NE (AD-notAD). Using Cox models and controlling for T and N stage, survival differences in LSP and several other lung cancer prognostic panels were evaluated. In all cases, research-based versions of these other lung cancer Risk Scores were calculated as weighted averages of each panel's genes in the three microarray datasets. Results: The LSP predictor called cases AD-AD in 81% of stage I and II AD samples and called samples SQ and NE in 7% and 12% of cases, respectively. The AD-notAD group (AD by histology and SQ or NE by LSP) had worse survival than the AD-AD group (AD by both histology and LSP) in each data set (logrank p-values <0.002). Pooling the 3 data sets and using a stratified cox model that allowed for different baseline hazards in each dataset, the hazard ratio (HR) comparing AD-AD to AD-notAD was 2.13 (95% CI 1.60-2.84, p<0.001) after adjusting for T and N stage. In similar pooled analyses, all three other genomic predictors tested were prognostic after adjustment for T and N stage (p<0.003). However, exclusion of AD-notAD samples resulted in reduced prognostic strength of all predictors, with 2 becoming not significant. When all 4 genomic predictors plus T and N stage were included in a pooled analysis, LSP remained prognostic (p=0.008, HR=1.59 (95% CI 1.13-2.25)) whereas two others lost significance and the third was reduced to p=0.0314. Conclusions: The LSP gene expression tumor subtyping provides valuable clinical information identifying a subset of histologically-defined AD samples with poor prognosis and biological features more similar to SQ or NE tumors. The prognostic strength of other gene expression signatures evaluated was reduced by exclusion of AD-notAD samples. Introduction: Tumor molecular profiling by next-generation sequencing (NGS) aims to identify clinically-relevant somatic variants actionable for patient management. Although commercial targeted NGS gene panels examine more genomic regions than smaller non-NGS genotype assays, it is not known if the more extensive genomic analysis improves clinical utility of the test. In this study we compared clinical actionability of variants identified by testing of 1619 tumor samples using either a commercial NGS targeted gene panel or a non-NGS MALDI-TOF MS mutation-specific custom panel. Methods: DNA was extracted from 1619 formalin fixed, paraffin embedded tumor samples from various tumor sites from participants in the IMPACT/COMPACT clinical trial at Princess Margaret Cancer Centre (NCT01505400). Samples were tested by either: 1) NGS, using the TruSeq Amplicon Cancer Panel (TSACP), testing 48 genes and 212 amplicons (MiSeq, Illumina); or 2) MALDI-TOF MS, using a custom mutation panel testing 279 mutations within 23 genes (MassARRAY, Agena Bioscience). Blood samples were tested in parallel for variant comparison to tumor. Classification of variants for actionability used the somatic variant classification scheme of Sukhai et al, Genet Med, 2015, with Class 1 and 2 variants with highest impact on clinical actionability. Results: Of 792 cases tested by NGS 74% (583 cases) had one or more somatic variants (333 with 1 variant, 250 with >1 variant, 209 with no variants). By comparison, of 827 cases tested by MALDI-TOF MS only 41% (340 cases) had one or more somatic variants (297 with 1 variant, 43 with >1 variant, 487 with no variants). When considering clinical actionability, however, the NGS method identified 19.7% (156/792) of patients with Class 1 variants and 16% (128/792) as Class 2, compared to the MALDI-TOF MS method which identified 27.7% (229/827) of patients with Class 1 variants and 8% (69/827) as Class 2. Overall both methods identified approximately 36% of cases with variants in the actionable Class 1 or 2 categories. Conclusions: Although NGS testing with TSACP identified more somatic variants in our patient cohort than mutation-specific testing (TSACP 74% versus MALDI-TOF MS 41%), there was no difference in the detection of clinically actionable variants detected (36% for both methods). The use of NGS panels
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org enriched for clinically actionable variants will increase the amount of clinically useful information derived from using NGS methodology.
H. Zhu 1,2 , J. Shelton 1,2 , A. Kincaid 1 , R. Zeigler 1 , D. Ilsley 1 , A.D. Simmons 2 , G.J. Latham 1 1 Asuragen Inc., Austin, TX; 2 Clovis Oncology Inc, San Francisco, CA. Introduction: Molecular resistance to targeted therapies is a common consequence of first-line cancer treatments. In non-small cell lung cancer (NSCLC), a frequent source of acquired resistance is the EGFR T790M mutation. Methods that can assess low-abundance T790M mutations are needed to evaluate the emergence of resistance and assess the efficacy of new therapies to advance drug development. We utilized targeted next-generation sequencing (NGS) and Droplet Digital PCR (ddPCR) to quantify EGFR mutations, insertion/deletions (indels), and copy number variations (CNV) to compare and contrast the mechanisms of acquired resistance to erlotinib, a reversible tyrosine kinase inhibitor (TKI), and rociletinib (CO-1686), a novel, oral, irreversible TKI. Methods: Mice bearing PC-9 (del746-750 EGFR) human NSCLC tumors were chronically dosed with vehicle, erlotinib, or rociletinib (10 animals/group). At progression, the erlotinib-treated group was split into two cohorts. Erlotinib dosing was continued in 3 animals. In the remaining 7 animals, erlotinib dosing was discontinued and rociletinib dosing was initiated and continued until resistance emerged. Tumor DNA at progression was assessed using the QuantidexTM DNA Assay (Asuragen). Targeted NGS analysis was conducted using the SuraSeq 500 cancer panel (Asuragen) and sequencing on a MiSeq (Illumina). All NGS data analyses were performed using the SuraSight bioinformatics pipeline (Asuragen). ddPCR assays for EGFR T790M, EGFR del746-750, and EGFR CNVs were run on a QX200TM ddPCR System and analyzed using QuantaSoft Software (Bio-Rad). Results: Emergence of EGFR T790M was detected in all erlotinibresistant tumors but not in rociletinib-resistant (RR) tumors by targeted NGS and ddPCR. Both methods reported T790M mutations from <1% to 79% abundance; correlation coefficients for variant quantification were >0.99 for both high (>10%) and low-abundance (<10%) variants. EGFR del746-750 mutations were identified consistent with the known genotype of the PC-9 cell line. Importantly, NGS analysis reported a novel 11-15-fold copy gain of MET that was observed in RR tumors, indicating both a molecular path to resistance and a possible therapeutic strategy to combat this resistance using MET inhibitors. By comparison, no change in EGFR copy number was detected using NGS, a finding that was confirmed by two distinct ddPCR EGFR CNV assays. Conclusions: High-sensitivity molecular assays such as targeted NGS and ddPCR are powerful tools to interrogate base-substitution mutations, indels, and CNVs. These methods can be combined to provide detection, quantification, and confirmation of low-level "driver" mutations and help unravel mechanisms of resistance for novel therapies with precision medicine applications in oncology.
Biomarker for Patients with Lung Cancer T.A. Boyle 1 , H. Yu 2 , K. Ellison 1 , A.A. Kowalewski 1 , L. Rozeboom 1 , C. Rivard 1 , D. Chan 1 , F.R. Hirsch 1 1 University of Colorado, Aurora, CO; 2 Shanghai Pulmonary Hospital, Shanghai, China. Introduction: Tumor cell programmed death ligand 1 (PD-L1) protein expression is an emerging biomarker associated with response to anti-PD-L1 immunotherapy in patients with lung cancer. The identification of PD-L1 protein expression by immunohistochemistry is complicated by the use of different antibodies, staining platforms, scoring systems and interpretations. Detection of PD-L1 mRNA expression by in-situ hybridization (ISH) on slides may be a more reproducible method to complement detection of protein expression by standard immunohistocytochemistry (IHC) for identifying patients with lung cancer who are most likely to respond to immunotherapy. Methods: Unstained slides from formalinfixed, paraffin-embedded lung cancer cell line microarrays (N=71) were stained for PD-L1 protein expression by IHC using the SP142 antibody (Monoclonal Rabbit anti-PDL-1/CD274 Clone SP142, 1:100, Spring Bioscience) and the Benchmark XT automated stainer (Ventana). PD-L1 mRNA expression was assessed by mRNA ISH with recommended probes (Probe-Hs-CD274) and reagents from Advanced Cell Diagnostics (ACD), Inc. PD-L1 protein expression was defined as positive if more than 5% of tumor cells were positive. Positivity for mRNA expression was defined as a staining score of 2, equivalent to 20X magnification. PD-L1 protein and mRNA positivity were compared with Spearman correlation analysis and the two-sided Fisher's Exact Test. Results: PD-L1 protein expression was positive in 32.4% (23/71) of lung cancer cell lines with a higher, though not significantly different, prevalence in non-small cell lung cancer (NSCLC) compared to small cell lung cancer (SCLC) cell lines (37.7% (20/53) versus 16.7% (3/18), p=0.15). PD-L1 mRNA expression was positive in 38.0% (27/71) of lung cancer cell lines with a significantly higher prevalence in NSCLC than SCLC cell lines (45.3% (24/53) versus 16.7% (3/18), p=0.048). 20 cell lines were positive for both protein and mRNA expression, 41 cell lines were negative for both, 3 were positive for protein expression but negative for mRNA expression, 7 were positive for mRNA but not protein expression. PD-L1 protein and mRNA expression were positively correlated (r=0.64, p=2.0 E-09) by Spearman correlation analysis. Comparison of PD-L1 protein and mRNA data dichotomized into positive and negative results demonstrated a strong association (p<0.0001) with the Fisher's Exact Test. Conclusions: PD-L1 mRNA and protein is expressed in approximately 1/3 of lung cancer cell lines with a higher prevalence in NSCLC than SCLC cell lines. PD-L1 mRNA expression is mirrored by PD-L1 protein expression and mRNA ISH may useful in conjunction with protein IHC to develop, characterize, and standardize PD-L1 as a biomarker for immunotherapy in patients with lung cancer. P.M. Rindler, M.L. Wallander, R. Margraf, D. Nix, D.N. Baker, C.P. Vaughn, W.S. Samowitz, A. H. Grossmann ARUP Laboratories, Salt Lake City, UT. Introduction: For translocation-associated mesenchymal neoplasms, identification of gene rearrangements is crucial for diagnosis, yet there are limited assays available for testing these tumors. Next-generation sequencing (NGS) can provide a comprehensive diagnostic assay. A challenge to using DNA-based NGS is that breakpoints typically occur within low complexity intron sequences. Capture array design must maximize breakpoint region coverage while minimizing off target reads. We developed two array design strategies to detect sarcoma translocations.Methods: To cover all possible breakpoints, we designed a SureSelect capture array targeting 38 sarcoma-associated genes where the entirety of each gene was captured. Libraries were prepared from a Ewing sarcoma, a low grade fibromyxoid sarcoma (LGFMS), and two non-tumor samples. Sequencing was done using an Illumina HiSeq 2000, 2x100 paired-end reads. Alternatively, to maximize the number of target genes in our panel, a literature search of ~120 sarcoma-associated genes was performed to identify the specific exons and introns involved. A second SureSelect capture array was designed to these locations. Both arrays were subjected to vendor probe design algorithms to identify regions potentially problematic for sequencing. Results: Both capture arrays have a footprint of approximately 3 Mb. The 38 gene panel detected both the Ewing sarcoma t(21;22)(q22:q12) EWSR1-ERG and LGFMS t(7;16)(q33:p11) FUS-CREB3L2 translocations. Our false discovery rate was 0%; however the percent on target reads was less than 15%, consistent with non-specific capture by probes designed to sequences of low complexity. Sequencing data for the 120 gene panel was not available at the time of abstract submission. Nevertheless, subjecting our capture arrays to vendor probe design algorithms resulted in nearly all repeat sequences being masked including regions with the highest UCSC mappability score of 1. This data suggests that vendor probe design tools may be unnecessarily stringent.
We have developed two unique approaches to translocation detection that maintain a footprint size close to 3 Mb providing sufficient read depth while allowing for sample multiplexing in order to reduce sequencing costs. Finally, our study raises important considerations for capture array design to low complexity sequences.
A. Goodman 1 , O. Rouhi 2 , L. Buckingham 1 1 Rush University Medical Center, Chicago, IL; 2 Orlando Health, Orlando, FL. Introduction: Lung cancer, the second most common type of cancer in the United States is responsible for 27% of all cancer deaths nationally. In the non-small cell lung cancer (NSCLC) subgroup, this high mortality rate is likely due to a lack of early disease detection and resistance to current treatments. Tumor cells undergo a specific metabolic shift from the endogenous cell's metabolism in order to support the unregulated proliferation characteristic of cancer, which creates a gene expression profile unique to the tumor. This study investigated two metabolicallyrelated transcription factors, FOXM1 and KRAS, and their possible predictive and prognostic value in early stage NSCLC patients. Methods: DNA and RNA were isolated from archival tissue of 120 early stage NSCLC patients. DNA was subjected to pyrosequencing to identify KRAS and NRAS mutations in the cancerous cells. Quantitative PCR was used to measure gene expression in normal and tumorous lung tissue. This data was then compared to patient demographics, time to recurrence (TTR) and overall time of survival (OS). Results: FOXM1 expression had no effect on overall survival time (p=0.875), but it did show a negative trend on time to recurrence (p=0.274). Out of 104 samples, 16 KRAS and 0 NRAS mutations were found. KRAS mutations were found in 1/9 non smokers and 14/64 smokers. KRAS mutations included G12C (n=6), G12D (n=3), G13D (n=3) and other G12 mutations (n=4). Expression of FOXM1 had no effect on outcome in patients with WT KRAS, but negatively affected TTR in patients with any mutant KRAS. G12D mutations had whereas G12C ation had a significant effect on both TTR and OS with respect to tumor stage and the different KRAS mutant subtypes. Median OS were 79.1 mos, median not reached, 29.2 mos, and 145 mos for wild type, KRAS G12C, G12D, and G13D, respectively. Conclusions: FOXM1 expression showed a negative association with TTR in early stage NSCLC patients with KRAS mutations in this patient group. The varying effects support different KRAS mutant subtypes as prognostic factors for NSCLC and suggest that metabolic changes of tumor cells could play a role in patient outcome. Activation of different jmd.amjpathol.org ■ The Journal of Molecular Diagnostics pathways by the mutant subtypes provides a new group of possible targets for future treatments. B. Das, F. Ahmad, S. Bisht SRL Ltd, Mumbai, Maharashtra, India. Introduction: Colorectal carcinoma is one of the most common cancer associated with high mortality and morbidity across the globe. Accumulation of several mutations in the RAS-RAF-MAPK and PI3K-PTEN-AKT signalling pathways play crucial role in promoting CRC. Molecular evaluation of KRAS, BRAF, PIK3CA mutation as well as MSI status has become an important part in CRC assessment and their alterations may determine the therapeutic response towards the therapies. Methods: In this study KRAS, BRAF, and PIK3CA mutations were determined using direct sequencing in 204 samples. MSI status were evaluated in 70 samples which had corresponding adjacent normal tissue by fragment analysis for 5 loci NR27, NR21, NR24, BAT 25 and BAT 26. Results: In our study, the frequency of KRAS, BRAF and PIK3CA mutations were 23.5%, 9.8% and 5.9% and MSI status were observed as 20.6%, 1.4% and 78.5% for MSI-High, MSI-Low and MSS respectively. We detected, five different substitution mutations at KRAS codon 12 (G12S, G12D, G12A, G12V, and G12C), and one substitution type at codon 13 (G13D). KRAS mutations were significantly higher in patients >50 years, and were associated with moderate/poorly differentiated tumors and adenocarcinomas. Unlike KRAS mutations, BRAF V600E mutations were more frequent in patients < 50 years and associated with well differentiated tumors and right sided tumors. PIK3CA E545K was the most recurrent mutation whereas other mutations detected were T544I, Q546R, H1047R, G1049S, and D1056N. Five cases showed concurrent mutation of KRAS and PIK3CA mutation. In addition mutations of KRAS, BRAF and PIK3CA coexisted with MSI high status. Conclusions: This is the first study to evaluate the PIK3CA mutation in Indian CRC patients. The frequency of KRAS, BRAF, PIK3CA and MSI were similar to worldwide reports. Furthermore, identification of molecular markers has unique strengths, and can provide insights into the pathogenic process and help optimise personalised prevention and therapy. Introduction: Pancreatic adenocarcinoma is often identified at advanced stage and is associated with a dismal prognosis and typically a poor response to conventional chemotherapeutic agents. Recent clinical trials using targeted therapy have shown some promise for survival in patients with pancreatic adenocarcinoma. The identification of additional molecular pathway alterations will be important for potential future targeted therapy, especially in patients that respond poorly to current therapies. The aim of our study is to use an expanded massively parallel nextgeneration sequencing panel to identify additional clinically actionable molecular pathway mutations in pancreatic adenocarcinoma. A clinically actionable gene variant is defined as a loss of function or, gain of function that is associated with an FDA/NCCN therapy or with a therapy in an actively recruiting clinical trial. Methods: Formalin-fixed paraffin-embedded (FFPE) blocks from pancreatic resection specimens of 15 patients with pancreatic adenocarcinoma were submitted for DNA extraction using the QIAamp FFPE tissue kit (Qiagen). The DNA quality and quantity were analyzed using Nanodrop 200 (Thermo Scientific) and E-Gel EX agarose gel, 1% (Invitrogen). Greater than 200 ng of DNA is required to progress to the library preparation. The expanded massively parallel sequencing panel analyzed 358 genes which included 188 clinically actionable genes run on Illumina HiSeq platform. Sequence analysis was performed by submitting FASTQ files generated from Illumina's CASAVA software to a Clinical Genome Analytics (CGA) data analysis pipeline to perform automated read quality assessment, alignment, and variant calling. SNP's, indels and CNV's were called using GATK, Pindel, and CONTRA, respectively. Actionable somatic variants were identified using a manually curated Clinical Knowledgebase (CKB). Results: The age of the patients ranged from 57 to 81 years with a M:F ratio close to 1:1. Three of 15 tumor samples failed the QC for poor DNA yield. 5 of the remaining 12 samples were wild type with no mutations identified. The 8 non-wild type tumor samples showed KRAS mutations at codon 12 including G12D (n=4), G12R (n=2) and G12V (n=2) substitutions. Germline variants of AURKA F31I were also identified in 3 of the KRAS mutated cases. No copy number variations were detected in any of the tumor samples analyzed. Conclusions: Current practice at our institution includes neoadjuvant chemotherapy and radiotherapy for all patients diagnosed with pancreatic adenocarcinoma. Interestingly, 67% (8/12) of the tumors analyzed harbored KRAS gene mutations in codon 12 that were still detected after neoadjuvant chemo/radiation. No additional somatic mutations considered clinically actionable were identified in the 358 gene panel. Variants of unknown significance were not addressed in this analysis.
Breast Cancer Prognostic Gene Signature Assay L. Cai, J. Riojas, G. McDowell, Y. Wang, J. Sebastian Laboratory Corporation of America Holdings, Research Triangle Park, NC. Introduction: The Prosigna Breast Cancer Prognostic Gene Signature Assay is a FDA-approved assay, based on the PAM50 gene signature, which provides a risk category (low, intermediate, high) and a numerical score (1 to 100) for the assessment of distant recurrence of disease at 10 years for post-menopausal women with early stage, hormone receptor positive (ER+/PR+), invasive breast cancer. The Prosigna Score gave the strongest prognostic information in years 5 to 10 when compared with the Oncotype DX Breast Cancer Assay (Genomic Health, Inc) recurrence score (RS) and immunohistochemical markers (IHC4) in a retrospective analysis of more than 1000 patient samples from the TransATAC study. The Prosigna assay also provides more prognostic clarity by categorizing more patients as high risk and fewer as intermediate risk when compared to other predictive breast cancer assays. In this study, we have evaluated the clinical and analytical performance features of this assay. Methods: To evaluate the assay's accuracy, repeatability and reproducibility, the Prosigna assay (NanoString Technologies, Inc) was performed on RNA isolated from patient breast FFPE tissue with known patient outcome and the synthetic RNAs. It simultaneously measures the expression levels of 50 genes used in PAM50 classification algorithm in a single hybridization reaction on the nCounter Dx Analysis System (NanoString Technologies, Inc) using nucleic acid probes designed specifically to those genes. Surgical gross tumor size and lymph node status are used in the Prosigna assay classification algorithm. Results: Of the specimens tested for validation, there was 100% concordance between the Prosigna risk category and patient outcome. Repeatability and reproducibility were also determined to be 100% based upon the Prosigna risk category. The Prosigna Breast Cancer Prognostic Gene Signature Assay is offered at Laboratory Corporation of America Holdings (LabCorp) for clinical testing for patient who meet diagnostic criteria. From the 156 specimens tested so far in this study, in the node-negative patient subset (N=142): 24.8% were classified in high risk category, 36.9% in intermediate risk category and 38.3% in low; in the node-positive patient subset (N=14): 64.3% were classified in high risk category and 35.7% in low risk category. In this date set, about 35.5% had >2cm gross tumor size s <1% which was mainly due to low RNA yield from small tumor. Conclusions: The Prosigna Breast Cancer Prognostic Gene Signature assay is a robust and reproducible assay using FFPE samples. The observed percentage of risk category is comparable to published findings. Introduction: Therapy for triple negative breast cancer, a subtype of breast cancer defined by lack of ER, PR, and HER2 expression, remains limited. There are currently no proven targeted therapies for triple negative breast cancers. Nextgeneration sequencing (NGS)-based molecular diagnostic assays are able to detect clinically actionable gene alterations. The aim of this study, therefore, was to analyze the gene alterations of triple negative breast cancers using a recently validated, expanded NGS assay, the JAX Cancer Treatment Profile (JAX-CTP). Methods: Twenty triple negative breast cancers were retrospectively identified from the pathology database. Slides were centrally reviewed and the most representative tissue block was selected for NGS testing. DNA was extracted from formalin-fixed, paraffin-embedded (FFPE) sections with a thickness of 10 um each using the QIAmp DNA FFPE Tissue Kit (Qiagen). Using hybrid capture, the genes of interest were enriched and sequenced on the Illumina HiSeq 2500 or MiSeq sequencers followed by variant detection and functional and clinical annotation. The JAX-CTP detects actionable variants, in the form of single nucleotide variations and small insertions -CTP is also validated for the detection of clinically actionable gene Results: 17/20 (85%) of triple negative breast cancers contained at least one somatic mutation detected by the JAX-CTP. MYC amplification was the most common alteration, present in 75% of tumors. Six other genes demonstrated amplification including CCDN1, HSF1, NTRK1, CXCR4, FGFR2, and EGFR. TP53 was the second most commonly mutated gene (4/20, 20%). Mutations were also present in PALB2, MET, PIK3CA, JAK3, AURKA, FGFR4, JAK, and KDR. Conclusions: The JAX-CTP assay identified a variety of clinically actionable gene alterations in triple negative breast cancers with MYC amplification being present in the majority of tumors. Further studies investigating MYC as a potential therapeutic target in triple negative breast cancer are warranted. mutation data from India. The aim of the present study is to determine the frequency and distribution of the KIT and PDGFRA mutations in Indian GIST cases. Methods: This is a retrospective study on histologically proven GIST cases (n=100). Molecular analysis for KIT exon 9, 11, 13 and 17 was performed in 100 cases and PDGFRA exon 12 and 18 was performed in 24 of these cases by Sanger sequencing from formalin-fixed, paraffin-embedded tissues. Sequencing data was analyzed. Cases with nonreadable/ noisy data were excluded from the study (n=20). KIT/PDGFRA mutation status was correlated with clinicopathological parameters. Results: The age range of the study group (n=80) was 23 years to 79 years (median 57 years) with majority of cases (n=53, 66.3%) in the age group of >50years. The male:female ratio was 3.2:1. Stomach was the most common primary site followed by small intestine. Spindle, epithelioid and mixed types accounted for 73.7% (n=59), 21.2% (n=17) and 5% (n=4), respectively. Direct DNA sequencing revealed activating KIT mutations in 66.3% (n=53) and PDGFR mutations in 8.3% (2/24) cases. KIT mutations seen were: inframe deletions (n=29, 54.7%), missense mutations (n=9; 17%), complex mutations (n=8; 15.1%), internal tandem duplications (ITD) (n=6, 11.3%) and insertion (n=1; 1.9%). Exon 11 was predominantly affected in 84.9% (45/53), followed by exon 9 in 11.3% (6/53) and 1.9% each in exons 13 and 17 (1/53, each). Exon 11 mutations spanned codons 550-560. Exon 9 mutations showed duplication of codons 502 to 503 in 5 cases, whereas one showed a novel mutation of c.1509_1510insACCTAT. Exon 13 (K642E) and exon 17 (N822K) point mutations were seen. Both PDGFR mutations (2/24) affected exon 18; inframe deletion (4.16%) at codon 843 to 846 and point mutation (4.16%) D842V was observed. Exon 12 was wild-type in all the cases. No statistically significant correlation was observed between the KIT/PDGFR mutation status with the histological parameters. Conclusions: The current series presents the trend and frequency of KIT/PDGFRA mutations in GIST cases from a tertiary cancer centre of India. KIT exon 11 mutations are the commonest. This study reports a hitherto undescribed novel mutation in exon 9. Also, it is the first study from India to report mutations in exon 13, exon 17 and heterozygous deletion at PDGFRA exon 18.
S. Menon, I. Arora, O. Shetty, T. Pai, M. Gurav, G. Bakshi, S. Desai Tata Memorial Centre, Mumbai, Maharashtra, India. Introduction: Renal cell carcinoma (RCC) resulting from gene fusions of TFE3 gene, located on chromosome Xp11.2, is a distinct aggressive type of renal cancer predominantly reported in children and young adults. These tumors are often misdiagnosed as clear cell or papillary RCC owing to their morphological overlap. Methods: We analyzed 25 morphologically and clinically suspected Xp11 translocation RCCs by fluorescence in situ hybridization (FISH) to detect TFE3 gene rearrangement. Results: Most patients presented with hematuria (56%) and with advanced stage at diagnosis (Stage III and Stage IV; 60%). Mean age was 23.8 years (ranging from 6 to 70 years), with a female: male ratio of 2.5:1. Mean tumor size was 6.5 cm. Morphologically, these tumors had admixture of cells with clear and eosinophilic cytoplasm with nested or papillary architecture with/without psammomatous calcification. TFE3 immunohistochemistry (IHC) was done in 14 out of 25 cases. TFE3 IHC was strongly and diffusely positive only in 5 cases. However, molecular translocation studies (done in all cases) for TFE3 by FISH (ZytoLight SPEC TFE3 dual color break apart probe, Germany) detected translocation of TFE3 gene and confirmed the diagnosis in 12/25 cases. Lymph node metastasis was seen in 5 out of 12 cases. Follow-up was available only in 11/12 TFE3 translocation positive patients, with median follow up period of 0.5 years (range from 0.25 to 7 years). Four out of 11 patients developed distant metastasis 6 months after diagnosis and 2 out of these 4 patients died of disease within 2 years of diagnosis. Conclusions: Translocation associated renal carcinomas are aggressive tumors which require molecular confirmation for accurate diagnosis as immunohistochemistry (IHC) using TFE3 antibody is found to be of limited utility. N. Kumari, N. George, P. Shukla, R. Vishwakarma, A. Agarwal, N. Krishnani Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow, Uttar Pradesh, India. Introduction: Papillary thyroid carcinoma (PTC) is the commonest malignancy of follicular epithelial cells in thyroid. BRAF V600E is the commonest genetic alteration in PTC reported with a very wide range from 18% to 87% in different studies world over. The association of BRAF V600E with clinicopathological features in PTC is also quite variable with some studies showing significant association of BRAF V600E with aggressive features whereas others have shown no association. The present study was done to evaluate the prevalence of BRAF mutation and its association with clinicopathological features in north Indian population. Methods: Ninety consecutive cases of PTC were reviewed histologically. DNA was extracted from formalin fixed paraffin embedded tissue. BRAF was amplified and digested overnight with restriction enzyme at room temperature. The digested product was then run on 2% gel to check the bands. Wild-type BRAF yields 3 bands of product size 117, 87 and 33 base pairs. Mutant BRAF gives an undigested product of 237 base pairs. Results: Overall BRAF mutation was found in 63.3% of PTC with 70.3% in classical variant, 66.7% in tall cell variant, 38.9% in follicular variant and 60% in poorly differentiated variant. There was no significant association of BRAF V600E with clinicopathological features; however recurrence rate was significantly higher in PTC with BRAF mutation (p value = 0.04). BRAF mutant PTC cases also developed recurrence at a significantly faster rate (median 15 months) compared to wild PTC (median 20 months) (p value = 0.006). Conclusions: The prevalence of BRAF V600E was seen in 63.3% cases of PTC. PTC with BRAF V600E showed increased rate of recurrence with a shorter time interval post-surgery compared to BRAF wild PTC.
Understanding Intratumoral Heterogeneity Using NGS Y. Ding 1,2 , M. Kluk 1 , H. Fernandes 1,2 1 Weill Cornell Medical College, New York, NY; 2 New York Presbyterian Hospital, New York, NY. Introduction: The advances in precision oncology to recognize intratumoral heterogeneity is an important barrier to the success of cancer treatment. Intratumoral heterogeneity is often observed when two or more tumor foci exist within a common spatial location or in the setting of metastases which may be in close proximity or at an anatomically distant site. Inter-and intratumoral heterogeneity pose a challenge to staging of the cancer by the pathologist and for the management of cancer by the oncologist, because histologically it may not be clear whether these foci represent separate primary tumors, simultaneous subclones or metastasis of a single primary tumor. The complete genomic landscape may not be captured in a single specimen. To further investigate the extent of intratumoral heterogeneity in our oncology specimens, we retrospectively compared genomic profiles of anatomically distinct tumor foci that were subject to targeted next-generation sequencing (NGS) analysis. Methods: From January 2015 to May 2015, a total of 19 patients with multiple tumor foci were included in this study. Of these, eight patients were diagnosed with papillary thyroid carcinoma (42.2%), eight were lung adenocarcinoma (42.2%), one with angiomyolipoma of the kidney (5.2%); one with craniopharyngioma (5.2%) and with a histological diagnosis of adenocarcinoma of the colon (5.2%). In 2 out of 8 lung adenocarcinoma cases, NGS testing was performed on three anatomically distinct foci, whereas in the rest of the 17 patients, two different foci were tested. In all 40 individual specimens, DNA was extracted from the tumor-enriched area of FFPE tissue and sequenced on the Ion Torrent Personal genome Machine (PGM) after library preparation and enrichment using AmpliSeqCancer Hotspot Panel v2 (Life Technologies). Results: Six of the eight (75%) lung adenocarcinoma cases tested showed varied genomic profiles among the foci tested. Individual foci from theses 6 cases harbored a different actionable variant (EGFR or KRAS) respectively. Additional cancer-relevant variants present in 3 of the 6 pairs of lung adenocarcinoma specimens also differed among the foci tested. On the other hand, individual foci tested from papillary thyroid carcinoma showed varying genomic signatures in only 3 of the 8 (38%) cases. A TP53 variant was identified in one of the two tumor foci from colon adenocarcinoma. In the 2 remaining cases, 100% concordant results were obtained from different foci of the tumor. Conclusions: Of the tumor types tested, intratumoral heterogeneity was most commonly detected in lung adenocarcinomas. The heterogeneity within papillary thyroid carcinomas was not as pronounced. NGS provides a reliable platform for addressing intratumoral heterogeneity. Introduction: To overcome replicative senescence, two specific and highly recurrent point mutations, G228A and G250A, in the promoter of the human telomerase reverse transcriptase (TERT) gene appear to generate de novo binding sites for the ETS transcription factor GABP and increase TERT expression in multiple types of cancer including hepatocellular carcinoma (HCC) and glioblastoma (GBM). The purpose of this study is to determine the performance of laboratory-developed sequencing assays to detect TERT promoter mutations in childhood cancers. Methods: M13 sequence-tagged primers were designed to amplify 235 bp of the TERT promoter. Genomic DNA was extracted from frozen and formalin-fixed, paraffin-embedded specimens from 25 hepatoblastoma (HB) patients (median age: 2 years; range: 0.2 years to 11 years), 7 GBM patients (median age: 47 years; range: 16 years to 70 years), 6 cell lines, and 2 tonsil specimens. Bi-directional Sanger sequence analysis was conducted with Mutation Surveyor, version 4.0.4 (Softgenetics, State College, PA). Pyrosequencing was performed using the same primers for PCR, a sequencing primer near the G228A and G250A mutations, and a Qiagen PSQ 96MA Pyrosequencer with PSQ 96 SNP Software AQ (Uppsala, Sweden). Results: Thirty eight (38) of 40 specimens produced informative results with both assays; no amplification was detected in 1 GBM and poor Sanger sequence quality was obtained in another GBM. The concordance rate between the two assays was 35/38 (92%). The 3 discordant specimens included 2 presumednegative specimens, tonsil and Jurkat cells, which tested positive by Sanger, and 1 GBM which tested negative by Sanger. G228A point mutations were identified in 3 HB patients, 3 GBM patients, HepG2 cells, and SK-N-SH cells. No G250A mutations were observed by either assay. The limit of detection, determined by mixing heterozygous G228A mutant HepG2 cells with wild-type HEK-293 cells, was between 5% to 10% by pyrosequencing and 10% to 20% by Sanger. Inter-run jmd.amjpathol.org ■ The Journal of Molecular Diagnostics reproducibility of amplification and sequencing was 100% with a subset of wild-type and mutant specimens. All 3 mutation-positive HB specimens were from patients older than 8 years of age and histologically showed pleomorphic features consistent with transitional liver cell tumors (TLCT) known to have genetic features overlapping with HCC. Conclusions: A robust TERT promoter pyrosequencing assay is presented with greater sensitivity and specificity than Sanger sequencing methods. Additional analysis with an in-house Agilent SureSelect QXT targeted nextgeneration sequencing panel is ongoing. Consistent with previous studies, TERT promoter mutations appear frequently in pediatric TLCT and in adult GBM. Introduction: The capability of NGS to interrogate a broad range of DNA mutations in a single assay has precipitated a paradigm shift in precision medicine from singletarget assays to highly multiplexed NGS panels. Current NGS panels have increased the breadth of content but do not streamline the analysis of RNA and DNA markers into a unified assay. We present a comprehensive approach for targeted clinical NGS that enables simultaneous quantification of DNA and RNA, a streamlined workflow compatible with low-input total nucleic acid (TNA), and specimen compatibility that includes FFPE, FNA, and liquid biopsies. Methods: Sample QC was performed using a novel qPCR assay that quantifies and partitions functional DNA and RNA from TNA isolations. PCR-based target enrichment was conducted using Quantidex targeted NGS reagents and sequenced on the MiSeq (Illumina) or the PGM (Thermo Fisher). Library sequences were analyzed using Quantidex Reporter, a bioinformatic analysis suite that directly incorporates pre-analytical QC information to improve the accuracy of variant calling, fusion detection and RNA quantification. Results: We present two unified RNA/DNA cancer panels: 1) a thyroid cancer panel that covers 56 DNA targets and 90 RNA targets, including 50 gene fusions and 40 mRNA targets; and 2) a lung cancer panel that covers 55 DNA targets and 131 RNA targets including over 100 gene fusions and over 20 mRNA expression markers associated with clinical actionability. Both panels interrogate RNA and DNA events from a single TNA sample in one sequencing run. The thyroid panel was evaluated on 123 FFPE thyroid lesions and 65 FNAs and revealed >98% agreement with independent methods. A diagnostic classifier that was migrated from FFPE to FNA increased diagnostic sensitivity from 74%, using DNA markers alone, to 98%, using RNA and DNA markers, while maintaining high specificity. The lung panel was assessed with cell-line and synthetic controls as well as 100 NSCLC FFPE specimens. Analytical concordance between matched FFPE, fresh frozen and liquid biopsies was also performed. Integration of a customized bioinformatics pipeline and variant caller with wet-lab QC enhanced mutation call sensitivity for variants present at less than 10% mutation, improved PPV for low-input specimens, and achieved absolute quantification of RNA. Conclusions: The Quantidex NGS system features optimized sample QC and enrichment chemistries that unify RNA and DNA targets and enhance quantitative variant calling and expression analyses to enable reliable, accurate and comprehensive molecular characterizations of tumor specimens. These approaches can be utilized with multiple types of cancer biopsies and NGS platforms to advance diagnostic, prognostic and theranostic applications. S. Epari, H. Kurani, A. Dutta, P. Shetty, A. Moiyadi, J.S. Goda, T. Gupta, R. Jalali Tata Memorial Centre, Mumbai Maharashtra, India. Introduction: BRAF gene is one of the key gene altered in most paediatric lowgrade gliomas, which includes BRAF fusion and BRAF V600E mutation. BRAF fusion is commonly seen in Pilocytic astrocytomas (PCAs) and pilomyxoid astrocytomas (PMAs), whereas BRAF V600E mutation is more frequent in non-PCA paediatric low grade gliomas (PLGG), including pleomorphic xanthoastrocytomas (PXAs) and gangliogliomas (GG). This is a study on evaluation of occurence of BRAF V600E mutation in different spectum of non-ependymal low grade glial tumors of children and young adults. Methods: Formalin-fixed, paraffin-embedded (FFPE) tissues of all non-ependymal low grade/variants of glial tumors diagnosed in to 2013 were evaluated for BRAF V600E mutation by sequencing and subsequently correlated with different clinicopathological features using appropriate statistical methods. Results: A total of 136 cases formed the study sample with M:F=1.7:1 (85:51) and age range from 2 years to 30 years. No cases were of infantile age. Most common location was cerebral hemisphere (n: 52; 38.2%), followed by posterior fossa (n: 42 cases, which included 34 cases in cerebellum, 2 in cerebellar peduncle and in the rest 6, no specific site could be known), diencephalon (n:23 cases -sellar/suprasellar: 10; thalamus: 6, hypothalamus: 4, optic chiasma: 2 &3rd ventricle:1), optic nerve and spine:7, lateral ventricle:3 and 2 in brainstem. PCA (n:83; classic:61; atypical:22) was the commonest histological subtype, whereas others include infiltrating low grade astrocytoma (n:13), PMA (n:1); PXA (n:9 -low grade:4; with anaplasia tumors (n:14), astrocytomas-NOS (n:8), glioneuronal tumors (GNT; n:7) and subependymal giant cell astroctyoma (n:1). 111 cases were yielded interpretable results for BRAF V600E and mutation of the same was in 13 cases (11.7%) only 9 of them were PCA; 2 were aPXA and one each is of GNT and infiltrating glial tumor respectively. One case of PCA was in the setting of neurofibromatosis, which was of wild BRAF type. However, presence of BRAF V600E mutation did not showed any statistical correlation any of the histological subtypes, age of the patient and locations of the tumor. Conclusions: Thirteen cases (11.7%) exhibited the BRAF V600E mutation, including 28.6% of PXAs, 6.7% of LGA, 12.7% of PCAs and 14.3% of GNTs with no significant stastiscal correlation for patient's age, sex, location of tumor and major histomorphological features. No mutations were detected in the ODGs and OAs. BRAF V600E mutation is an uncommon but a distinct genetic alteration in the paediatric LGG and LGGNTs, can be identified across all different histological types with rarity in the pediatric infiltrating gliomas.
Intestinal Stromal Tumor: First Report from Indian Patients F. Ahmad, B. Das Super Religare Laboratories Ltd., Goregaon (West) , Mumbai, India. Introduction: Mutation of the c-KIT and PDGFRA are one of the most important molecular changes observed GIST and are useful in predicting the responsiveness to imatinib. The current study evaluated these gene mutations in Indian GIST patients. Methods: We investigated the frequency and distribution pattern of c-KIT (exons 9, 11 and 13) and PDGFRA (exons 12 and 18) by direct sequencing in a series of 70 Indian GIST cases. Results: C-kit and PDGFRA gene mutations was observed in 27 (38.5%) and 4 (5.7%) of the seventy cases respectively. Most of the c-KIT mutations involved exon 11 (85.7%), followed by exon 9 (14.3%) whereas none of the cases showed exon 13 mutation. Most frequent mutation in exon 9 mutations was duplications resulting in Ala503-Tyr504, whereas one had novel point mutation S476G. In contrast to exon 9 mutations, exon 11 mutations were mostly inframe deletions (19/24) at codons 550-560, whereas other exon 11 mutant cases were point mutations at codons 559, 560, 568, 573, and 575. To our knowledge, P573T, Q556_V560delinsH, Q575H and Q575_P577 were novel variations and have never been reported earlier. The PDGFRA mutations were seen predominantly in exon 18 at codon 842 (D842V), while, exon 12 showed a novel indel variation (V561_H570delinsT). No significant correlation between c-KIT/PDGFRA mutation and clinicopathological data were observed. Conclusions: In conclusion, this study highlights the frequency and distribution pattern of c-KIT/PDGFRA mutation in Indian cohort and identified novel variations that added new insights into the genetic heterogeneity of GIST patients. To the best of our knowledge, this is the first report on PDGFRA gene mutation from Indian subcontinent. Several studies have revealed that particular anti-tumor agents are exported out by specific ABC transporters. Based on this information it could possible to personalize treatment of cancer patients. Methods: Tumor tissues for this study were obtained from 31 patients (age 28 yrs to 80 yrs) with a primary diagnosis of breast cancer (BC) treated at the St. Elizabeth Cancer Institute between years 2012 and 2013. Patients with all stages of BC and treated with a standard of care FEC, FAC or AC used for cDNA reverse transcription reaction. Expression of seven selected ABC genes (ABCB1, ABCC1, ABCC2, ABCC3,. ABCC4, ABCC5, ABCC11) was were used as a reference (GAPDH, B2M, IPO8, EIF, PPIA). The relative quan baseline expression in sensitive tumor cells) was considered as overexpression and was used as a surrogate value of cell resistance to treatment. Results: Of the 31 studied patients 6 (19,4%) had physiological expression of all 7 ABC genes, 5 (16,1%) had over-expression of only 1 ABC gene, 13 patients (41,9%) had overexpression of 2 to 4 ABC genes and 7 (22,6%) patients over-expressed 5 to 7 ABC genes. The disease progression was assessed by continuous measurement of oncomarkers -CEA, CA125, CA15-3, TPS, ROMApost, ROMApre, HY4. Until today, 12 (38,7%) patients have been in remission. Progression was observed in 19 (61,3%) patients. Conclusions: The frequency of over-expressed ABC genes revealed that ABCC11 is the most frequently overexpressed gene in patients in remission and that ABCC1 is the most frequently overexpressed gene in patients with disease progression. It is likely that 5-FU -the substrate of ABCC11 -was actively exported out of the tumor cells in this group of patients. We therefore speculate that the remaining two compounds from the FAC/FEC regiment could be sufficient to prevent disease progression. The ABCC1 gene, on the other hand, is responsible for the export of cyclophosphamide and epirubicin or adriamycin out of the tumor cells and results in a treatment resistance. Based on our findings we conclude that assessment of expression of specific ABC genes involved in efflux of 5-FU, cyclophosphamide and epirubicin or adriamycin could be considered as a
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org predictive biomarker of therapeutic response in patients with breast cancer treated with FAC, FEC or AC regiments.
Circulating Free Tumor DNA S. Hellwig 1 , N. Welker 1 , C. Vaughn 1 , W. Akerley 2 1 ARUP Laboratories, Salt Lake City, UT; 2 University of Utah, Salt Lake City, UT. Introduction: Activating mutations of the epidermal growth factor receptor (EGFR) tyrosine kinase drive malignancy in a number of cancers. In non-small cell lung cancer (NSCLC), EGFR-activating mutations (EGFRm+) have an overall prevalence of ~15%. Treatment with EGFR-targeted tyrosine kinase inhibitors (TKI) has become standard of care for these patients. A major challenge of TKI therapy is the rapid development of resistance, causing progression in most cases within a year. Although several mechanisms of TKI resistance exist, the most prevalent at 50% to 60% of cases, is mutation of EGFR at amino acid 790 from threonine to methionine (T790M). To address this challenge, next generation TKIs, with improved response rates to EGFR T790M+ tumors are currently in clinical trials (AZD9291; CO-1686). Early detection of acquired T790M mutations after initiation of first-line EGFRtargeted TKI therapy will be a vital element in redirecting treatment during disease progression. Repeated biopsy of patients with NSCLC to monitor tumor evolution is highly invasive, technically challenging, expensive, and delays therapy. Here, we demonstrate that simple blood testing of cell-free circulating tumor DNA (ctDNA) by digital droplet PCR is a viable alternative, capable of detecting EGFR T790M with high sensitivity and accuracy. Methods: A digital droplet PCR (ddPCR) assay was designed using internally quenched hydrolysis probes to detect both wildtype and T790M EGFR, and samples were analyzed using the RainDrop system (RainDance Technologies). Sensitivity, reproducibility and specificity of the ddPCR assay were determined using reference standards and circulating free DNA isolated from healthy donors. ctDNA isolated from a cohort of 10 NSCLC patients and 100 healthy controls was analyzed for T790M mutational status. Results: Background of our ddPCR T790M assay of circulating free DNA was determined from a large cohort of healthy donors, demonstrating zero to five T790M positive droplets, representing 0% to 0.1% total droplets analyzed. In ctDNA isolated from NSCLC patients' blood plasma, we reliably detected the presence of EGFR T790M at allele frequencies from <0.5% to >30%, well above background. In cases with established EGFR T790M mutational status from tumor biopsies, 5 with mutation and 5 wildtype, 100% concordance with ctDNA ddPCR results was observed. Conclusions: These results demonstrate that our ddPCR assay is a reliable and accurate method to detect EGFR T790M mutations in plasma specimens from NSCLC patients. Given the high concordance with tissue biopsy results, this assay represents a rapid, non-invasive "liquid biopsy" approach that should greatly improve lung cancer patient care.
M. Barbi de Moura, A.I. Wald, A. Priddy, M. Miller, K. Cieply, S. Chiosea, M.N. Nikiforova University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: Copy number (CN) changes are commonly found in many cancers, and knowledge about gains or losses of whole chromosomes or chromosomal regions is helpful in cancer diagnosis, prognosis, and personalized medicine approach to treatment. With the advance of next-generation sequencing (NGS) for cancer detection it is important to verify the reliability of CN estimation from NGS data. In this study, we evaluated the utility of the Ion AmpliSeq Cancer Hotspot Panel v2 NGS data for detection of CN changes in clinical samples. Methods: Forty-four tumor samples were analyzed by fluorescence in situ hybridization (FISH), loss of heterozygosity (LOH), and by targeted NGS. The Ion AmpliSeq Cancer Hotspot Panel v2 (Life Technologies) was used for library generation and sequenced on Ion Torrent platforms. The bioinformatics pipeline (Grasso et al J Mol Diagn. 2015 Jan;17(1):53-63) was used with few modifications to determine CN changes. The NGS CN results were compared with FISH and LOH analysis. Results: First, 10 normal blood and 10 normal tissue samples were used to establish the normal CN baseline. Next, 44 samples including liver, brain and lung tumors were sequenced and analyzed against the normal baseline and results were compared with FISH and LOH analysis. At least 2 amplicons were sufficient to make a CN call. Our analysis showed that CN ratios of < 0.1 and > 3 (where CN=0 and CN=6, corresponding to copy number loss and gain, respectively), accurately identified CN changes in 100% of calls. Concordant results were obtained for many genes including CDKN2A, EGFR, and ERBB2. CN ratio of > 0.5 and < 2.0 correctly called normal (CN=2) in 86.5% of calls. CNR=0.1-0.5 and CNR=2-3 were considered as indeterminate or "grey zone" calls. Overall, the method demonstrated high analytical sensitivity (100%) and specificity (87%) for detection of homozygous deletion (CN=0) and high level of amplification (CN>6). It was highly reproducible with all CN changes consistently detected at similar level gains or losses. Conclusions: Copy number changes can be successfully detected in the data generated from Ion AmpliSeq Cancer Hotspot Panel v2 without modification of the panel design. It demonstrated high analytical accuracy for detection of high level of amplification or deletion in clinical samples. C.H. Edgerly, D.L. Duncan, M.O. Meyers, S. Elmore, M.L. Gulley University of North Carolina at Chapel Hill, Chapel Hill, NC. Introduction: Plasma microRNA is thought to be derived from apoptotic or necrotic cells, virions, exosomes and other microvesicles. Modern molecular technology facilitates quantification of multiple microRNAs to yield expression profiles that may reflect disease status. Methods: The GastroGenus miR Panel was designed to measure gastric cancer-related microRNAs and spiked controls. Expression of 48 microRNAs was measured by rtPCR on 52 plasma specimens from patients with active gastric adenocarcinoma (n=33), patients with gastric adenocarcinoma following surgical resection (n=6), and healthy adults (n=11). Results: Plasma of active cancer patients had upregulated hsa-miR-375, -185-5p and -486-5p, and downregulated hsa-miR-199a-3p and -191-5p , as compared to healthy controls (all adjusted p<0.00135). Plasma of cancer patients following surgical resection had upregulated hsa-miR-375 and -25-3p, and downregulated hsa-miR-199a-3p, as compared to healthy controls. Five of six resection patients remained in remission at least 2 months following surgery (range 2 to 44 months). One patient relapsed with metastatic carcinoma 5 months following surgery. Conclusions: The findings demonstrate significant dysregulation of circulating cell-free microRNAs in gastric adenocarcinoma patients. Furthermore, pilot data suggest how microRNA profiles change after intent-to-cure therapy. The findings provide novel insights into cancer biology and reinforce the potential for non-invasive measures of tumor status that could facilitate management of affected patients. A.I. Wald, M. Melan, W. Ernst, S. Roy, M. Nikiforova University of Pittsburgh Medical Center, Pittsburgh PA. Introduction: The accurate identification of genetic changes in central nervous system (CNS) tumors is critical for the appropriate clinical management of patients. Until recently, simultaneous screening for multiple genetic alterations in small brain biopsies was difficult, especially in formalin-fixed, paraffin-embedded (FFPE) tissues. We evaluated performance of in-house developed NGS panel (GlioSeq) for detection of various types of genetic alterations occurring in adult and pediatric CNS tumors. Methods: GlioSeq targeted NGS panel is designed to detect point mutations/indels in 30 CNS tumors related genes, 24 genes for copy number changes, 14 types of rearrangements in EGFR (EGFRvIII), BRAF, and FGFR3 genes, and expression of 3 housekeeping genes. Sequencing of 2 DNA libraries and 1 RNA library was performed on Ion Torrent PGM or Ion Proton (Life Technologies) and analyzed with Torrent Suite with Variant Caller plug-in and an in-house developed bioinformatics pipeline. Results: Forty four clinical FFPE tumor samples were analyzed, representing all major types of adult and pediatric brain tumors. All samples were successfully sequenced for mutations and copy number changes and 41/44 (93%) for gene fusions and gene expression. Alterations were detected in 100% tumors and were appropriate for each tumor class and subtype (eg, KIAA1549/BRAF fusions in pilocytic astrocytomas, point mutations in IDH1/2, CIC, and TERT in oligodendrogliomas, EGFR amplification and EGFRvIII mutation in glioblastomas, AKT1 and NF2 mutations in meningiomas, and MYCN and CDK6 amplification in medulloblastoma). Mutations and gene fusions were detected down to 3% to 5% of allelic frequency. Cost of reagents for GlioSeq was 10 times lower as compared to combination of conventional methods. Conclusions: GlioSeq NGS panel has demonstrated accurate and sensitive detection of a wide range of point mutations, copy number changes, and gene fusions. It allowed for rapid, cost-effective screening of routine brain tumor samples providing valuable information for diagnosis, prognostication and treatment of these patients. R. Maglantay, M. Cabanero, R. Gupta, G. Salame, Y. Lee, B. Mize, N.N. Chen State University of New York Downstate Medical Center, Brooklyn NY. Introduction: Uterine carcinosarcomas are rare but aggressive entities with pathological and molecular heterogeneity. Due to its relatively low incidence (4 per 100,000), few studies have been performed to investigate the status of KRAS mutation and its relation to survival. Previous reports have identified somatic mutations of KRAS in primary uterine carcinosarcomas in 10% to 30% of cases. Mutations in KRAS have been shown to have therapeutic implications in other tumors, in particular, the response to anti-EGFR treatment in colorectal carcinoma. In addition, molecular studies characterizing endometrial carcinomas have shown KRAS to be associated with endometrioid-type morphology, with less aggressive course as opposed to serous-type morphology. The purpose of this study is to identify KRAS mutations in uterine carcinosarcomas and its correlation with morphology and survival. Methods: A total of 32 cases of uterine carcinosarcoma from institutional archive in the last 20 years were evaluated for mutations in codons 12 and 13 of the KRAS gene using the TrimGen KRAS Mutation Detection Kit (Mutector, TrimGen, US). Histologic slides were reviewed for diagnosis and tumor components. Patients' clinical data were obtained from the medical records. Mean survival was calculated for patients in each stage and Kaplan-Meier test was performed to compare survival statistics. Results: A total of 5/32 cases (15.6%) have mutations in KRAS codons 12 (2/5) or 13 (3/5). They were classified as: stage I: 2/5; stage II: 0; stage III: 2/5; and stage IV: 1/5. Serous morphology was present in 4/5 cases compared to 1/5 for endometrioid. There was no difference in survival in jmd.amjpathol.org ■ The Journal of Molecular Diagnostics stage I (KRAS+ tumor, 30 months; 31 months) . In stage III tumors, the overall survival in the KRAS+ tumors versus KRAS-tumors are 13 months and 32 months (p value 0.433) respectively. In stage IV tumors, the overall survival are 6 months and 11.5 months (p value: 0.433) respectively. Conclusions: Based on our preliminary study, KRAS positive tumors show a trend of lower survival in the advanced stages of uterine carcinosarcoma. In our setting, KRAS mutation correlates more with serous morphology which has a poorer prognosis relative to endometrioid in contrast to what has been previously reported in other studies. To our knowledge, nothing has been reported in particular about KRAS mutation status and its correlation with survival in carcinosarcoma. Further studies with larger cohorts are needed to confirm our findings. were used in the validation. RNA were extracted from cell lines using the Qiagen RNeasy Mini kit and total nucleic acid were extracted from formalin-fixed, paraffinembedded (FFPE) tissue using Agencourt FormaPure Kit. Before library construction, the RNA concentrations were determined by Qubit. Custom fusion panels were designed for target enrichment using the Archer Universal RNA Reagent Kit v1 protocol. Libraries were constructed through sequential transfer of 200 ng of input total nucleic acid into tubes containing lyophilized pellets of all of the enzymes and buffers required for library creation. Purified libraries were quantified using the KAPA Biosystems Library Quantification Kit diluted to 2nM. All barcoded samples were sequenced on the MiSeq using the MiSeq v2 Reagent Kit 300 cycles. Bioinformatic analyses of raw sequence data were performed using the Archer Analysis v3.1.1. Results: Ten of the 20 samples (8 commercial cell lines and 2 clinical FFPE samples) with known fusion genotypes and 1 negative sample were used in the accuracy study. The assay accurately detected 100% of the fusion genes. Sanger sequencing confirmed detected fusion genes. The remaining 8 clinical samples with unknown fusion genotype were screened resulting in one sample with RET-CCD6 fusion genotype. Sanger sequencing confirmed the fusion gene. Analytical sensitivities established a lower limit of detection of 3% mutant tumor in a background of normal. Inter-and intra-run precision studies showed 100% concordance. Conclusions: The validated NGS assay allows for an alternative approach to FISH, IHC, and RT-PCR detection of clinically relevant and actionable fusion genes.
Suggests STK11 and NOTCH1 as a Candidate Biomarkers E. Ryzhova, D. Marchion, F. Abbasi, L. Nong, A. Gartel, Y. Yin Xiong, A. Magliocco H. Lee Moffitt Cancer Center & Research Institute, Tampa FL. Introduction: The standard treatment of high grade serous ovarian cancer (HGS-OvCa) with platinum-taxane chemotherapy is followed by the development of chemoresistant disease in a significant proportion of patients. Molecular mechanisms underpinning chemoresistance remain obscure. In this study we performed mutation profiling of patients with varying responses to chemotherapy to investigate whether platinum-resistant phenotype can be linked to a particular mutation signature. To this end we used a highly sensitive masspectrometric approach with a focus on lowfrequency alleles. Methods: Archival DNA samples from 37 patients with HGS-OvCa, who were classified as either complete responders (CR, n=18) or incomplete responders (IR, n=19) to primary platinum-based therapy were subjected to mutation profiling using the Agena BioScience Oncopanel. The Oncopanel relies on the iPlex MassArray methodology and permits parallel screening of 214 mutations in 26 oncogenes and tumor suppressor genes. Results: Frameshift or nonsense mutations in STK11 gene were found in 22% of the CR group suggesting a significant association with the disease phenotype (Fisher's exact test, p=0.046). Further, a mutation in NOTCH1 gene was identified in 16% samples regardless of platinum-taxane response status (R2328W, n=3 CR, n=3 IR). NOTCH has been previously associated with certain hematologic malignancies and solid tumors; however it was not described in the context of HGS-OvCa. Lastly, well-established actionable mutations in KRAS (G12D, n=2; G13D, n=1), EGFR (E746_A750del, n=1), and BRAF (V600E, n=1) genes were detected in patients demonstrating CR and IR. Conclusions: We have identified STK11 mutations as possible biomarkers for platinum-therapy response. Additionally we have found that NOTCH1 could be a candidate for therapeutic targeting in HGS-OvCa. Furthermore, we have identified several HGS-OvCa specimens that harbor known actionable mutations. These data suggest that routine multiplexed mutation profiling of patients with HGS-OvCa may facilitate the proper selection of available therapeutic options and/or clinical trials. A.M. McDonald, B. Kipp, J. Jen, W.E. Highsmith, K. Rumilla, S. Kerr, U. Aypar, R. Jenkins, W. Sukov, C. Botz, R. Graham, J. Voss, B. Crusan, X. Wang, L. Holtegaard, X. Wu, L. Peterson Mayo Clinic, Rochester, MN. Introduction: Next-generation sequencing (NGS) of the human transcriptome enables detection of gene fusions which may allow for targeted treatment in cancer patients. We have developed and assessed the performance characteristics of an automated RNA Seq assay for the detection of gene fusions in solid and hematologic tumors. Methods: Total RNA was extracted from solid tumor tissue and whole blood using a Qiagen miRNeasy Micro and Mini kit respectively. RNA quality [RNA Integrity determined with a Qubit 2.0 fluorometer. Using a TruSeq RNA Sample Preparation v2 Kit (Illumina), isolation of polyadenylated mRNA with oligo-dT beads and second strand cDNA synthesis were performed on a JANUS Automated Workstation (Perkin Elmer), while NGS libraries were prepared on a Biomek FXp Liquid Handler (Beckman-Coulter) via custom-built protocols. Paired-end, 101 bp sequencing was completed on a HiSeq 2500 (Illumina) in Rapid Run mode. Data was analyzed using MapRSeq, a Mayo Clinic-developed suite of alignment, fusion detection and filtering programs. Eighteen tumors with 16 different known fusions, 18 tumors with unknown fusion status, and 7 normal tissue samples were analyzed. One sample containing an FGFR2-TACC3 fusion was run in triplicate and again on a separate run to assess reproducibility. Also, the BCR-ABL and FUS-DDIT3 fusions were analyzed at dilutions ranging from 100% to 25% and 98% to 3.125%, respectively. Control RNA was chemically degraded and analyzed at RIN values ranging from 10 to 3.9. Results: Gene fusions were detected in 14/18 (78%) samples. Fusions were confirmed by an independent method (ie, FISH, RT-PCR, or RNA Seq in another lab). Of the undetected fusions, all were >5 kb from the 3' end of the transcript and had limited coverage in at least one of the gene partners. No fusions of clinical significance were found in the additional tumor or normal tissues. The FGFR-TACC3 fusion was detected in all replicates between and within runs. BCR-ABL was detected at a lower limit of 25% and FUS-DDIT3 at 12.5%. All control RNA fusions were detected down to a RIN of 8.4. However, fusions with a breakpoint within 1 kb of the 3' end of the fusion transcript were detectable at a RIN as low as 3.9. Conclusions: We have assessed the performance characteristics of an RNA Seq assay for gene fusion detection. The method is amenable to automation and allows for detection of gene fusions in tumor tissue with high specificity and good reproducibility. However, when using polyA selection for mRNA, there are several factors that may affect sensitivity of fusion detection including tumor percentage, RNA quality (RIN value), and distance of the breakpoint from the 3' end of the fusion transcript. A. Popa, P. Anderson, D.A. Surve, M. Sabato, A. Ferreira-Gonzalez, C.I. Dumur Virginia Commonwealth University, Richmond, VA. Introduction: The vast majority of samples clinically used for the Ion AmpliSeqTM Cancer Hotspot Panel v2 (CHP2) assay are formalin-fixed, paraffin-embedded (FFPE) tissue specimens, which require deparaffinization prior to DNA extraction using silica-based columns. The whole DNA isolation process is performed manually, which can prove very laborious, especially with increasing number of samples. To partially automate and increase the throughput of DNA isolation from FFPE samples, we assessed the performance of the Qiagen EZ1 Advance XL instrument for DNA isolation to be used on the CHP2 assay. Methods: DNA isolation using the EZ1 DNA Tissue Kit and the EZ1 Advanced XL DNA Paraffin Section Card was compared to the manual deparaffinization and subsequent silica-based column DNA manual extraction method. Forty-eight different FFPE specimens were extracted directly from unstained slides containing 10 μm sections of tissue areas ranging from 3 mm2 to up to 570 mm2, by both the manual and automated methods. The DNA yield per slide was determined in each case. The double-stranded (ds) DNA concentration obtained with the QubitTM 2.0 fluorometer were additionally correlated with the tissue area and cellularity. Furthermore, library preparation for the CHP2 assay and sequencing were performed on 15 of the 48 tested samples, and the variants identified were compared with those from paired samples subjected to manual DNA isolation. Results: We found an excellent correlation between the automatic and manual methods (Pearson's r = 0.82) for the DNA yield obtained per slide. In addition, the dsDNA concentrations highly correlated with tissue cellularity (Pearson's r = 0.73). Of the 15 patient samples that were used for library preparation for the CHP2 assay, 11 yield good quality library material and were sequenced, resulting in a perfect correlation with variants identified using the manual DNA isolation method. The 4 samples that failed to yield sufficient library material using the automated method, were isolated from small fine needle aspiration (FNA) samples, using less than 4 slides per case, leading to significantly lower dsDNA concentrations (3.4 ± 2.6 ng/μl) than the other 11 samples (p = 1.8e-03). These 4 samples had previously been extracted from more than 5 slides each, using the manual method and the cell blocks were exhausted at the moment of performing this The assay provides a PAM50 risk of recurrence score (Prosigna score), which correlates with the probability of distant recurrence-free survival at ten years for postmenopausal women with hormone receptor-positive, early stage breast cancer. The PAM50 gene expression signature measures the expression levels of 50 genes plus eight constitutively expressed normalization genes to report the Prosigna Score, which is used along with the patient's nodal status to assign a risk classification defined by pre-specified cutpoints. Methods: Fifteen samples (7 reference specimens and 8 clinical samples) were used in the validation. Prior to RNA extraction, pathologists circled the region of viable invasive breast carcinoma on the H&E slides, estimated the tumor surface area, and provided the nodal status and the clinical history associated with each sample. RNAs were extracted using the Roche FFPE RNA kit and analyzed by the NanoString nCounter Dx Analysis System. The system generates a Prosigna Score on a 0 to 100 scale using a proprietary algorithm, and categorizes each case based upon pre-specified thresholds as High, Intermediate or Low Risk. Results: All 15 samples passed the quality control metrics of the assay. Expected risk categories were derived from the 7 reference samples. Eight clinical samples were assigned to risk categories. Precision studies from reference and the clinical samples showed consistent results between inter-and intra-runs, meeting the performance specifications defined by the assay: 100% concordance within 2.9 SD of Prosigna score unit and 100% precision with average bias of +/-1 Prosigna Score unit. Conclusions: The Prosigna Breast Cancer Prognostic Gene Signature Assay was analytically validated and implemented in our CLIA-certified hospital laboratory environment. Its workflow was proved to be straightforward and easy to follow by clinical laboratory scientists, confirming the simplicity of the assay and its suitability as an in vitro diagnostic test.
J. Bacher 1,2 , R. Zhao 1,2 , D. Storts 1 , C. Sievers 2 , R. Halberg 2 1 Promega Corporation, Madison, WI; 2 University of Wisconsin, Madision, WI. Introduction: Lynch syndrome (LS) is caused by mutations in the DNA mismatch repair genes which increases risk of colorectal cancer, endometrial cancer and other cancers. The National Comprehensive Cancer Network recommends screening for LS at time of diagnosis for all CRC patients using microsatellite instability (MSI) testing and/or immunohistochemisty (IHC), followed by BRAF V600E testing, followed by genetic testing. A recent cost-benefit assessment for identifing LS by Snowhill et al (2015) found that the greatest health benefit was obtained by using the strategy of MSI, followed by BRAF, followed by genetic testing. BRAF V600E mutations occur in up to 80% of sporadic MSI-high tumors with MLH1 promoter methylation, but rarely occurs in LS. We investigated the use of a single multiplex PCR assay for simultaneous analysis of MSI and BRAF V600E mutations. Methods: An allele specific PCR assay for detection of the BRAF V600E mutation was developed and assayed simultaneously with the MSI Analysis System (Promega). The MSI+BRAF multiplex was used to evaluate 74 adenomatous polyps and 22 sessile serrated polyps, which were characterized for MSI and mismatch repair protein expression by IHC. Mixing experiments were conducted to determine the limit of detection. Specificity for detection of the BRAF V600E was assessed using synthetic templates for all BRAF V600 mutations. Confirmation of BRAF positive samples was done by Sanger sequencing. Results: 9.4% of adenomas were MSH-High and 6.8% were positive for BRAF V600E mutations (all MSI stable). Of the MSI-high cases, 3 had MLH1/PMS2 loss of expression, 1 MSH2/MSH6 and 3 had normal expression for all four mismatch repair proteins. There were no adenomas with BRAF mutations that also exhibited either loss of MLH1 or were MSI-High. In contrast, a significantly higher proportion, or 82% (18/22), of sessile serrated polyps were positive for BRAF V600E mutations. Among these cases, 2 had loss of MLH1 expression (both MSI stable) and another was MSI-High with normal MLH1 expression. The limit of detection for the BRAF V600E mutation was around 1%. Conclusions: Combining MSI and BRAF testing into a single assay was found to be an effective and sensitive method and could greatly simplify the initial screening step for LS. Tumors that are MSI-High and have BRAF V600E mutaions may be excluded from further testing, potentially saving time and reducing costs. Tumors that are MSI-High but lack a BRAF V600E mutation are candidates for genetic testing. If a next-generation sequencing gene panel containing all 4 mismatch genes is used for genetic testing, this would eliminate the need for IHC testing and make screening for LS a simple two-step strategy. Introduction: Mutations affecting codons 132 of IDH1 and 172 of IDH2 are frequent in low grade gliomas and associated with favorable outcome. The common methods of IDH1,2 testing are characterized by the limited spectrum of detected mutations (immunohistochemistry, IHC) or low sensitivity (Sanger sequencing). To overcome these limitations we developed a high throughput, sensitive and multiplexed GliomaPanel test using Agena Bioscience's MassARRAY technology. Methods: GliomaPanel was designed to interrogate all the essential mutations in IDH1 and IDH2 genes. Also included were 116 hot-spot targets in an additional 17 genes with potential clinical or research utility as determined by the Moffitt Cancer Center neurooncology group. The method of detection involves simultaneous DNA amplification in 8 multiplex PCRs, followed by a single nucleotide extension using a probe adjacent to a targeted site. The identity of the incorporated nucleotide is determined by the mass of extended probe using MALDI-TOF mass spectrometry. The IDH1 and IDH2 mutation profiling was validated in accordance with the CAP guidelines and included the total of 78 FFPE diagnostic samples. For the majority of samples, the mutation status was pre-determined by IHC. Results: GliomaPanel performance characteristics such as accuracy, specificity, sensitivity, precision, and limit of detection were established and/or validated. 1) The test accuracy as to IDH1 and IDH2 mutation detection was 99%.
2) The analytic specificity and sensitivity were 100% and 97% respectively. 3) All mutations under analysis were detected with high precision and the lower limit of detection was 10%. Notably, the rare IDH1 R132C mutation was detected in several glioma patients that were IDH1 negative in IHC tests. The discrepancy was resolved by Sanger sequencing that confirmed the presence of the mutation. Additionally, R132C mutation at low allelic frequency (20%) was identified in one patient, whose IDH1 status was negative in both IHC and Sanger sequencing. Finally, we assessed the GliomaPanel DNA quality jmd.amjpathol.org ■ The Journal of Molecular Diagnostics requirements and demonstrated the 70% of DNA samples, that failed Illumina NGS TruSight panel due to either low concentration or low amplificability, were accurately processed by GliomaPanel. Conclusions: GliomaPanel offers cost-effective highthroughput solution for highly reproducible and accurate detection of IDH1 and IDH2 mutations. The method detects a broad range of mutations, is highly sensitive, has relaxed DNA quality requirements, and represents a valuable alternative to conventional methods for targeted mutation profiling. Cornell Medical College, New York, NY; 2 New York Presbyterian Hospital, New York, NY. Introduction: Minimally invasive sampling by cytology or core needle biopsy is often used for initial diagnosis, and confirmation of recurrence in patients with lung cancer. Needle core biopsies, cytology specimens and resected specimens have all been used for somatic mutational analysis of non-small cell lung carcinomas (NSLC). However, the correlation between the mutation detection using next-generation sequencing (NGS) from fine-needle aspiration (FNA) cell blocks, CT-guided needle core biopsy and corresponding resection specimens has not been adequately documented. Methods: Twelve samples from patients with NSLC, including 10 cell blocks from CT-guided FNA and 2 needle core biopsies, and the corresponding resected FFPE tissue specimens were identified for the study. The mean neoplastic cellularity of the FNA specimens was 22.9% (range 15% to 70%). This was significantly lower than the biopsies at 65% (range 50% to 80%) and resections at 67.7% (range 30% to 80%). All specimens were sent to the Molecular Pathology laboratory where DNA was extracted and subject to NGS using the Ion Torrent AmpliSeq Cancer Hotspot Panel v2 with the Personal Genome Machine (PGM). Results: Three sensitizing EGFR mutations and 5 KRAS variants were detected in the FNA/biopsy specimens. DNA from the corresponding twelve surgical resections had 3 sensitizing EGFR mutations and 6 KRAS variants. The discrepant specimen had a KRAS Q61H variant that was present in the resection but not detected in the FNA. A total of 8 additional variants (TP53=5; PIK3CA=1; APC=1; STK11=1) were identified in FNA/biopsy specimens. The corresponding surgical resection specimens had 9 variants including TP53=4; PIK3CA=1; APC=1; STK11=1 and CDKN2A=1. There was 100% correlation between the needle biopsy specimens and the resections. The concordance between the actionable variants present in FNA and resections was 89% and the overall variant concordance was 90%. Conclusions: Cytology specimens and needle core biopsies provide adequate material for somatic mutational analysis performed on the PGM with the Ion AmpliSeqCancer Hotspot Panel v2. Excellent concordance of variants detected in cytology specimen/needle core biopsies and resected tumor suggests that NGS results obtained from minimally invasive cytology specimens can be used for therapeutic decision making in patients with lung cancer.
Characterize cfDNA C. A. Schumacher, J. Laliberte, C. Couture, T. Harkins, J. Irish, L. Kurihara, S. Sandhu, V. Makarov Swift Biosciences Inc., Ann Arbor, MI. Introduction: Liquid biopsy is a non-invasive tool to assess cancer burden by examining the tumor-derived fraction of circulating, cell-free DNA (cfDNA) from plasma. We have tested two assays to monitor cancer burden by examining cfDNA: a global methylation sequencing assay and a targeted amplicon sequencing assay to identify mutations across 56 oncology related genes. Genome-wide hypomethylation has been demonstrated as a surrogate biomarker for cancer and can detect cancer independently of tumor-specific mutations. Amplicon-based detection of specific point mutations, however, provides a window into tumorigenesis and potential therapeutic resistance. Amplicon sequencing costs are 10-times lower than methylation sequencing with a faster turnaround time, but it is only effective if the cancer has a mutation covered by the panel. This study utilizes both assays to probe the same diverse dataset as a means of characterizing their efficacy across a broad spectrum of cancers. Methods: cfDNA was extracted from eight cancer patients and five normal controls. To monitor methylation status, whole genome bisulfite sequencing was performed using 5 ng of bisulfite-converted cfDNA. A minimum threshold of greater than three standard deviations below the methylation level of the normal controls in 1.1% of the genome analyzed was considered significant as described by Lo et al 2013. To detect tumor-specific mutations, 10 ng of cfDNA was used for the multiplexed amplicon sequencing panel. Samples from both assays were sequenced on an Illumina MiSeq. Results: Six samples demonstrated significant hypomethylation in cfDNA, ranging from 2% to 40% when compared to healthy controls. One negative sample originated from a subject with high-grade fallopian tube serous adenocarconima, whereas the most hypomethylated cfDNA came from a subject with metastatic adenocarcinoma of the colon. Other tumor types with intermediate cfDNA hypomethylation included invasive breast carcinoma and pancreatic ductal carcinoma. Interestingly, point mutations in the covered genes were not detectable in all samples; for the remaining samples, 1 to 4 mutations per sample were detected and had mutational allele frequencies of less than 5% to 14%. Conclusions: These methods provide the basis for using liquid biopsy to monitor tumor burden. When looking at a single time point, we have demonstrated that each assay has varying utility based on cancer type and severity, thereby highlighting the necessity to carefully consider these factors before choosing a characterization method. To further evaluate these techniques, we will now apply them to a longitudinal study to follow mutation and hypomethylation status of cfDNA before, during, and after treatment.
L. Borsu, J. Intrieri, Z. Momin, H. Yu, G. Riely, M. Ladanyi, M. Arcila Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: Next-generation sequencing (NGS) represents a highly attractive system to assess EGFR mutant lung cancers with secondary resistance to EGFR tyrosine kinase inhibitors (TKI). This allows the upfront assessment for both known and yet undefined resistance markers which may be relevant for treatment. However, the detection of the often subclonal EGFR T790M mutation, the most common acquired resistance mechanism, may be limited with this approach, and higher sensitivity methods with shorter turnaround time are often required. Here we describe our approach to testing resistance samples using a combination of a hybridization capture-based NGS assay for targeted deep sequencing of cancerrelated genes (J Mol Diagn. 2015; 17:251-64 ) and a rapid, highly sensitive assay for T790M mutation detection using digital PCR technology (dPCR) on the NGS libraries. Methods: Resistance samples received for routine mutation analysis were selected. DNA was extracted, sheared and pre-enrichment sequencing libraries were prepared. 40ng aliquots of the NGS libraries were used for rapid testing using RainDrop dPCR system whereas part of the extra barcoded libraries from patientmatched tumor and normal samples libraries was pooled captured and sequenced with the MSK-IMPACT assay, followed by a custom analysis pipeline identifying the somatic alterations. High sensitivity T790M testing by Sanger sequencing in the presence of LNA-based suppression of wild type was performed for confirmation in all cases. Dilution series of mutant with normal libraries were utilized to establish analytical sensitivity and LOD. Results: A total of 24 resistance samples were tested (20 pos, 4 negative) with 100% concordance between the digital PCR and high sensitivity method. Quantitative results by dPCR on positive samples ranged from 1-65% and showed high correlation with variant frequency (VF) obtained by the NGS method. Serial dilution studies based on 40ng DNA inputs, demonstrate an LOD of 1x10-4. Higher LOD could be achieved with higher inputs. Turnaround time for dPCR results was 1.5 days versus 3 weeks for NGS. Lower inputs down to 1ng of library still allow the detection of T790M even if present in 1% VF. Conclusions: The use of digital PCR on NGS libraries is a rapid, highly sensitive and efficient approach for the detection of EGFR T790M on EGFR TKI resistance samples. It maximizes the amount of tissue retained for comprehensive NGS assays while establishing T790M status with a rapid turnaround time and higher sensitivity, in keeping with established guidelines.
Target-Selector Assays V. Alexiadis, T. Watanskul, V. Zarrabi, V.M. Singh, L.J. Arnold, L.J. Arnold Biocept, Inc., San Diego, CA. Introduction: Determining the presence of mutated circulating tumor DNA sequences in the blood is an emerging technology termed "liquid biopsy" and is used to monitor tumor burden and to guide personalized treatment. Examples of critical mutations for lung cancer patients are activating mutations L858R and Del19 on chromosomes 21 and 19 respectively, as well as the resistance mutation T790M on chromosome 20. Finding the status of these sequences in clinical lung cancer cases is essential, but can also be a great challenge due to the presence of excess wildtype sequences arising from normal necrotic and apoptotic cell material that has been shed into the bloodstream. To overcome these limitations, Biocept has developed real-time PCR based assays ("Target-Selector") for the detection of EGFR mutations T790M, L858R and Del19 even when a large excess of wild-type background DNA sequences is present. The Target-Selector assays rely on the blocking of wild-type amplification with a proprietary blocker while allowing mutant DNA templates to be amplified normally. In combination with Sanger sequencing Target-Selector assays can potentially identify any point mutation occurring in a short stretch of target DNA (8 or 13bp for L858R or T790M respectively), or in the case of deletions up to 24bp (spanning EGFR K745 to T751). Data from the clinical validation of these EGFR Target-Selector assays will be presented. Methods: Circulating nucleic acid was extracted from blood plasma and used in Target-Selector assays specific for the amplification of T790M, L858R, Del19 or EGFR region on chromosome 20 as a control. Amplified mutant sequences are quantified using a standard curve. They are then purified and used in Sanger sequencing reactions with an oligonucleotide specific for the target sequence. Results: We used the Target-Selector assays for clinical validation of lung cancer plasma samples. The assays can reliably detect ~7 copies mutant T790M, L858R and Del19 in >2000-fold excess of wild-type background. We were able to routinely detect T790M mutant patients using the T790M Target-Selector assay. Likewise, using the Del19 Target-Selector assay, we were able to identify del747-750 insP, del747-752 and del746-750 deletion mutations in various clinical lung cancer cases. We were also able to identify L858R mutations in clinical lung cancer cases with both A>C and AG>CT transversions. Of the 74 samples tested, we found 69 concordant cases (93.2%) with tissue biopsy. Conclusions: The Target-Selector assay is an effective tool to identify EGFR mutations T790M, L858R and Del19 in the presence of a large excess of wildtype sequences. The Target-Selector assays using patient blood closely matched tissue biopsy results. D.C. Corney, E. Pribitkin, C. Goswami, E.S. Johnson, M. Tuluc, S.C. Peiper, C. Solomides, Z. Wang Thomas Jefferson University, Philadelphia, PA. Introduction: A next-generation sequencing (NGS) panel was designed for the comprehensive detection of somatic mutations in 23 genes frequently mutated in thyroid cancer. The aim of this study was to use the panel to determine the ability of genomic analysis to enhance the diagnostic power of pathologic examination of fine needle aspirates (FNAs) with an indeterminate cytopathology diagnosis (Bethesda III, IV, V). Methods: A custom NGS panel targeting 23 genes frequently mutated in thyroid cancer was designed using multiple databases including TCGA and COSMIC. Seventy-eight FNAs from 72 patients were analyzed including assay validation specimens (n=20) and consecutive specimens for NGS diagnostic studies (n=58). Cytopathology diagnoses included Bethesda category (BC) III (n=57), IV (n=17) and V (n=4). Results: Mutations with known or likely pathogenic significance were detected in 43.6% (34/78) of total FNA specimens analyzed: 33% of BC III (19/57), 47% BC IV (8/17), and 50% BC V (2/4), including recurrent NRAS (n=11), BRAF (n=6), PTEN (n=5), EIF1AX (n=3) and HRAS (n=3) mutations. Surgical followup was available for 30 patients, of which 27 were found to be malignant. Of the patients with malignant disease, 1 was BRAF V600E and TP53 co-mutated, 2 contained BRAF K601E mutations alone, 2 had co-mutation of NRAS Q61 with either TERT promoter mutation or EIF1AX splice site mutation, and 9 had NRAS Q61 mutations alone. A further 5 patients contained pathogenic mutations in other genes, 2 contained variants of unknown significance, and 6 did not contain mutations detected by the panel. Three patients had benign disease upon surgery; 1 with no mutations and 2 with variants of unknown significance. Consistent with the low mutation rate in thyroid cancer, a single pathogenic mutation was detected in the majority of malignant specimens and only 4 cases of co-mutation were detected, including that of NRAS/EIF1AX co-mutation suggesting that EIF1AX mutation is not mutually exclusive with MAPK pathway mutations. Overall, the positive predictive value of the full panel was 100% with 70.4% sensitivity, compared to 100% with 51.9% sensitivity when considering BRAF/NRAS mutation status alone. The extra information obtained by this panel played a major role in the decision to follow-up with repeated FNA, lobectomy or total thyroidectomy. Conclusions: This customdesigned thyroid cancer-specific sequencing panel significantly increased the probability of identifying somatic mutations in cytologically indeterminate or suspicious specimens. By increasing detection sensitivity without compromising the positive predictive value this tool adds genomic evidence to the cytologic findings and has the potential to play an important role in the surgical decision making process.
P. Ward 1 , L. Dubeau 1 , P. Jonas 1 , T. Long 1 , F. Juan 1 , G. Kim 1 , L. Aye 1 , J.W. Bacher 2 1 University of Southern California, Los Angeles, CA; 2 Promega Corporation, Madison, WI. Introduction: Lynch syndrome accounts for 2% to 4% of endometrial cancers. This familial cancer predisposition disorder is caused by DNA replication errors due to germline mutations in genes encoding mismatch repair enzymes. Diagnosis is suspected from loss of immunoreactivity for mismatch repair enzymes and presence of replication errors in genomic microsatellite sequences that are especially prone to such errors (microsatellite instability). Half of affected women with endometrial cancer harbor germline MSH6 mutations. Several cases with loss of MSH6 expression show no instability in the most commonly used panel of microsatellite sequences. We sought to test the hypothesis that microsatellites biomarkers with longer polyA tracts could be more sensitive than traditional biomarkers for the detection of subtle microsatellite instability as typically associated with MSH6 mutations. Methods: Formalin-fixed, paraffin-embedded tissue sections of immunohistochemistry for MSH6, MLH1, MSH2 and PMS2 expression. Antibody probes showing < 5% immunoreactivity in tumor cells but immunopositivity in normal cells were scored as negative. Lack of expression of either MLH1+PMS2 or MSH2+MSH6 was scored as consistent with microsatellite instability. Genomic DNA was tested for microsatellite instability using a novel set of longer polyA tract biomarkers and the commonly used shorter biomarkers in the Promega MSI analysis kit. Microsatellite alterations were scored as microsatellite stable (MSS), equivocal, subtle (St-MSI; 1-2 bp shift), or unstable (MSI; > 3 bp shift). Tumors with alterations at >30% loci were scored as MSI-high. Results: The short polyA biomarkers detected 42 microsatellite alterations (43% MSI; 45% St-MSI and 12% equivocal) and 8 cases with MSI-high. The long polyA biomarkers detected 51 microsatellite alterations (88% MSI; 8% St-MSI and 4% equivocal) and 11 cases with MSI-high, 4 with MSI-low, and 2 with equivocal microsatellite alterations. Immunohistochemistry identified 8 cases compatible with MSI, all of which showed MSI-high with both sets of PCR biomarkers. Of the three MSI-high cases detected with the long polyA biomarkers only, one showed loss of MLH1+PMS2 expression and the others showed normal protein expression. Eleven MSS cases based on PCR showed loss of MSH6 expression, 4 also with loss of PMS2 expression. Conclusions: The long polyA biomarkers improve detection sensitivity and facilitates analysis of MSI phenotypes in endometrioid adenocarcinomas. Immunohistochemistry for MSH6 may be unreliable in the evaluation of potential for Lynch syndrome, as a significant number of cases that lack MSH6 expression show no evidence of microsatellite instability. Introduction: The development of advanced molecular platforms, bioinformatics tools and the rapidly growing number of biomarkers that are the potential targets for new therapies have contributed to a rapid increase in cancer testing. Somatic mutations in multiple genes can be reliably detected concurrently by next generation sequencing (NGS) at levels well below traditional sequencing methods. Our IntelliGEN assay, a NGS "hot spot" panel, provides an assessment of targeted mutations using a panel of 50 known cancer genes. In this study, we have evaluated the clinical and analytical performance features of this assay. Methods: The IntelliGEN assay was developed based on the Ion AmpliSeq Cancer Hotspot Panel v2 (Life Technologies). A single pool of 207 primer pairs is used to perform multiplex PCR to prepare amplicon libraries from genomic "hot spot" regions. DNA from various cancer specimens and cell lines were used to evaluate the assay's accuracy, repeatability, reproducibility and analytical sensitivity. Identified reportable mutations were confirmed by a secondary method. Results: Of the specimens tested during validation, for 4 cell lines with known mutational profiles results were 100% concordant at all covered hot spot locations; mutations (N=150) identified in tumor specimens were confirmed by a secondary method and showed 98% concordance and 2% discordance (due to somatic mutation below NGS detection limit or the insufficient coverage of confirmation method). Repeatability (intra-assay precision) and reproducibility (Inter-assay revision) were 100%. This assay can detect 5% of mutant DNA in a background of wild type genomic DNA when the input DNA is 5ng or more. The IntelliGEN assay has been offered as a clinical test based on the successful performance features. In a series of clinical specimens, the number of mutations identified in each specimen ranged from 0 to 4 with 29.6% having no mutation identified, 36.7% 1 mutation, 20.4% 2, 10.2% 3 and 3.1% 4 mutations. Sixt four percent (64%) of specimens with mutations had one or more actionable treatment or drug resistance information. Each cancer type had a set of common genes that harbored mutations. For example, in lung cancer specimens, BRAF, CDKN2A, EGFR, KRAS, MET, PIK3CA and TP53 were the most common genes that harbored mutations. Small tumor area, low DNA yield and DNA degradation were the major causes of assay failure. Conclusions: The IntelliGEN assay is a robust and reproducible assay using a variety of tumor sample types (tissue block, bone marrow, thyroid FNA). The molecular alterations provided by this assay can assist in making cancer treatment decisions involving targeted therapies in a clinically relevant turn-around time. Introduction: Next-generation sequencing of cancer tissue is becoming a mainstream technique in clinical laboratories because of its potential to contribute to the design of patient-specific therapies. Here, we describe the validation of the ThunderBolts (TB) Cancer Panel from RainDance Technologies for the detection of sequence variants in DNA from FFPE tissue from patients with a variety of solid tumors. Methods: DNA was extracted from FFPE tissue and the quality and quantity of DNA determined by UV absorbance and fluorescence, respectively. A minimum of 10 ng DNA in a total volume of 40 μl was analyzed using the TB assay. PCR libraries from up to 16 samples were pooled and sequenced on the Illumina MiSeq instrument. DNA from three well-characterized human cell lines was used to validate the TB panel. The sequence variants in the TB panel are the same as those in the Life Technologies Ion AmpliSeq Cancer Hotspot Panel v2 so we used this panel to confirm, in a subset of the samples, the sequence variants identified by the TB panel. Nucleotide sequence data were analyzed using the MiSeq Reporter Analysis Pipeline and associated software. Results: We analyzed DNA from 59 FFPE samples and three cell lines. Using cell line mixtures, for expected allele frequencies of 25%, 10%, and 5%, the average observed allele frequencies were 26%, 10%, and 7%, respectively. DNA from two cell lines was sequenced for the determination of Introduction: The N-ras was one of the first oncogenes discovered and is a member of the Ras gene family that consists of over 150 members. Most NRAS mutations are concentrated around two hotspots -codon 12 and codon 61. Of these about 60% are found at Q61, approximately 24% at G12 and 12% at G13. Mutations of the NRAS gene have been found in multiple cancers, including melanoma (13% to 25%), myeloid leukemias (14%), hepatocellular carcinomas (10%), thyroid cancers (7%), colorectal cancers (1-6%), and lung cancer (1%). Methods: As part of the validation testing process, the analytic sensitivity (lower limit of detection) was determined using DNA reference standards from Horizon Diagnostics. The standards used were codon Q 61, codon 59, codon G12/13 and codon 117. Testing was performed using the Qiagen Ras Extension Pyro V2 kit. In addition, DNA from five wild-type NRAS samples was tested at several dilutions, to determine the highest dilution at which results could be obtained. Results: Multiple dilutions of the DNA reference standards from Horizon Diagnostics (4%, 3%, 2%, 1% and 0.5%) were used, using a wild-type DNA as a background. Three identical runs were performed and in each run, testing for each codon was done in triplicate. Consistent and reproducible results of the appropriate mutations were noted down to 2% limit of detection for all four codons tested. The dilution of the wild-type DNA at which results were obtained was a concentration of 0.5ng/μL. Conclusions: We currently offer the NRAS mutation at our lab as part of the next-generation sequencing (NGS) panel. However, the assay requires a minimum DNA concentration of 30ng/μL to perform the NGS assay. Given the observed lower DNA concentration requirement of 0.5ng/μL this limitation of DNA concentration can be overcome using the Qiagen Ras Extension Pyro V2 kit, thus enabling us to perform the testing on small biopsies, fine needle aspirates and cell transfer specimens without compromising sensitivity. Further data will be collected around precision, accuracy, specificity and linearity testing. Introduction: Tumor mutation profiles can inform diagnosis, prognosis as well as treatment options. We have previously published the methodology for detecting molecular alterations in formalin-fixed, paraffin-embedded (FFPE) samples using MSK-IMPACT, a targeted sequencing assay encompassing several hundred cancerassociated genes (Cheng DT et al, JMD, 2015) . Here, we describe our experience following the implementation of prospective MSK-IMPACT testing for patients with solid tumors at Memorial Sloan Kettering Cancer Center since January 2014. Methods: DNA was extracted from tumor (FFPE) and matched normal (blood) samples. Ninety seven percent (97%) of tumors were accompanied with matched normal DNA. DNA samples were sheared and libraries were prepared. Custom DNA probes were used to capture targeted exons and introns from 341 or, more recently, 410 genes and selected introns. Pooled libraries were sequenced on Illumina HiSeq 2500 as 2x100bp paired end reads. The data were analyzed using a custom bioinformatics pipeline following best practices. Results were viewed on an in-house developed web application, which allows manual review of mutations as well as automated clinical report generation. Results: Since the implementation of MSK-IMPACT in the Molecular Diagnostics Service, we have received a total of 5,359 specimens from 4,800 patients (currently >150 tumor samples per week). A proportion of these samples were insufficient due to low tumor content (3%) or low DNA yield (7%). 4,823 tumor samples passed our pre-sequencing QC. 47% were resections, 50% were biopsies and 3% were other specimens. 8% of samples failed sequencing QC and were repeated after requesting new recuts. 4,422 samples were successfully sequenced and produced high quality results. MSK-IMPACT achieved a median coverage depth of 587X in tumors and 372X in matched normal blood DNA. The inclusion of matched normal DNA enabled reliable and specific somatic mutation calling. We identified an average of 7 non-silent mutations per sample (median: 4, range: 0 to 445). The most commonly mutated genes were TP53 (40%), TERT (13%), PIK3CA (13%) and KRAS (11%). We also detected an average of 3 copy number alterations per sample (median: 2, range: 0 to 36). The most common alterations were CDKN2A/B deletion (6.6%), and CCND1, ERBB2 and MDM2 amplifications (4.7%, 4.5% and 4.1% respectively). Conclusions: Our team has successfully sequenced a total of 8,711 clinical samples from tumor and patientmatched blood. From the date at which both the tumor and normal samples are received to the date at which the clinical report is issued, the average turnaround time for MSK-IMPACT assay is approximately 21 days. We have demonstrated the feasibility of running a high content clinical NGS assay in a high volume clinical setting. S.J. Hsiao, A.N. Sireci, V.S. Aggarwal, A.T. Turk, P.L. Nagy, M.M. Mansukhani Columbia University Medical Center, New York, NY. Introduction: To facilitate the identification of mutations in a broad range of solid tumors types with the potential to affect clinical management, therapies, and eligibility for clinical trials, a 467-gene panel (Columbia Combined Cancer Panel, CCCP) developed in collaboration with pathologists and oncologists at this institution and validated in our laboratory, was utilized in the analysis of 65 solid tumor cases. Methods: From July 2014 to April 2015, 65 solid tumors (21 lung, 8 melanoma, 7 pancreatic, 6 CNS, 5 colon, 5 liver, 3 ovarian, 2 esophageal, 2 breast, 2 sarcoma, 1 thyroid, 1 biliary tract, 1 bladder, and 1 testicular tumor) were evaluated by a pathologist for adequacy and enriched for tumor by microdissection if necessary. Targeted exonic and intronic sequence was obtained from DNA purified from FFPE tissue using Custom Agilent Sureselect capture and Illumina HiSeq2500 sequencing. Variants were filtered by an in-house developed bioinformatics pipeline and reviewed by a molecular pathologist before reporting. Results: Across the 65 solid tumors tested by the CCCP assay, 44 potentially actionable mutations were detected (5 of these cases harbored >1 actionable mutation). Lung tumors, which made up the largest tumor type tested, showed a high proportion of potentially actionable mutations (18 mutations/21 cases, 85.7%) versus all other tumor types (26 mutations/44 cases, 59.1%). For a subset of these tumors (lung, colon, pancreatic tumors and melanoma), the results were compared to those that would have been obtained by the standard single gene testing (EGFR exon 18, 19, 20, 21 mutation and KRAS exon 2 mutations by Sanger sequencing, and BRAF V600E mutation by real-time PCR) available in the laboratory prior to implementation of the CCCP assay. Utilization of the CCCP assay resulted in increased detection of potentially actionable mutations from 38.1% to 85.7% in lung cancers, from 57.1% to 85.7% in pancreatic cancer, from 12.5% to 37.5% in melanoma, and from 40.0% to 80.0% in colon cancer. Conclusions: Expanded mutational testing detected a larger percentage of potentially actionable mutations across all tumor types. Our results suggest that greatest benefit may be seen in some tumors types such as lung and colon cancers. However, an important caveat is that the majority of the potentially actionable mutations detected by our 467 gene panel would be detected by most smaller, commercially available targeted NGS panels. Thus, further study and analysis is required to determine the size and scope of expanded mutational testing that yields the best cost to diagnostic yield ratio.
Cancer Panels J. Pettersson, L. Du, T. Long, L. Dubeau, P. Ward University of Southern California, Los Angeles, CA. Introduction: Digital droplet PCR allows precise quantification of molecular targets without standard curves across a wide dynamic range. This technology enables clinicians to precisely monitor treatment responses and early relapse in liquid biopsies, but has not been validated in primary tumors. We sought to investigate the accuracy and dynamic linear range of BRAF V600E mutation quantification by digital droplet PCR and to compare the quantification of mutant BRAF V600E alleles in clinical melanomas using RainDance digital droplet PCR versus sequencing of products of the RainDance Thunderbolt cancer panel on the MiSeq platform. Methods: TaqMan primers and probes targeting BRAF V600E and the normal allele were used for digital droplet PCR. Emulsified reaction mixes were amplified on a PTC-200 cycler. Serial dilutions of genomic DNA from the RKO cell line, harboring 2 alleles of BRAF V600E and one normal allele, were spiked into wildtype genomic DNA to establish the sensitivity of BRAF V600E detection in 500ng DNA. Absolute counts of droplets positive for BRAF V600E and BRAF V600 were captured to determine accuracy and linearity of the assay. Formalin fixed, paraffin embedded sections of primary melanomas were reviewed to estimate the amount of tumor cells within areas marked for manual dissection. Genomic DNA recovered from the macrodissected material was quantified on qPCR and approximately 150ng of amplifiable DNA was used for digital droplet PCR. Another 20-75ng DNA was used to generate libraries for NexGeneration sequencing using the RainDance Thunderbolt cancer panel. Libraries were sequenced on a MiSeq instrument using MiSeq Reagent kit v2 (500 cycles) (Illumina). The percentage of mutant BRAF V600E alleles detected in products of the Thunderbolt cancer panel was compared to that detected by digital droplet PCR. Results: The absolute number of BRAF V600E and V600 droplets in serial dilutions of RKO DNA ranging from 66.67% to 0.11% generated observed results of 66.62% to 0.13% with a r2 value of 0.9998. A BRAF V600E positive melanoma sample tested by digital droplet PCR showed the tumor DNA to contain 90.2% V600E alleles. When this same melanoma was amplified using the Thunderbolt cancer panel and sequenced on the MiSeq, the DNA was found to contain 89.14% BRAF V600E (depth of read 40519:4786). Conclusions: Digital droplet PCR provides a high degree of linearity across a wide dynamic range for the accurate detection of BRAF V600E alleles in solid tumors. The high level of reproducibility of both digital droplet PCR and NexGeneration sequencing of products of the Thunderbolt cancer panel supports the merit of these approaches for accurate quantification of mutations in primary tumors. J. Pettersson, T. Long, L. Dubeau, P. Ward University of Southern California, Los Angeles, CA. Introduction: Glial tumors are the most common primary malignancy of the central nervous system. Standard treatment for glioblastoma, the most aggressive glial cancer, consists of surgery followed by radiotherapy and chemotherapy with the alkylating agent temozolomide. Although outcomes are generally poor, survival is improved in patients whose tumors show methylation of the O6-methylguanine DNA methyltransferase (MGMT) promoter compared to those in whom this promoter is unmethylated. Thus, accurate quantitation of MGMT promoter methylation is key to exploit the predictive and prognostic value of this biomarker. MethyLight is a quantitative real-time PCR assay commonly used to measure extent of promoter methylation including at the MGMT locus. Quantification relies on the generation of a standard curve using varying quantities of fully methylated DNA compared to an endogenous unmethylated control. Accurate quantification is only possible for samples that fall within the linear range of the standard curve. We sought to compare the performance of RainDrop digital droplet PCR to that of MethyLight for the quantification of MGMT promoter methylation. Method: Fully methylated, SssI treated genomic DNA was de-aminated by bisulfite using a Zymo EZ DNA methylation kit. Serial dilutions of the de-aminated DNA were analyzed for MGMT promoter methylation with either MethyLight or digital droplet PCR (RainDrop) protocols. The same primers and fluorescent probes were used in both assays. Real-time PCR assays were performed in triplicate and the average Ct values were used to generate a linearity curve. For digital droplet PCR assays, the absolute number of positive droplets was used to generate the linearity curve. The curves generated from both platforms were compared. Results: The real-time PCR MGMT methylation assay (MethyLight) showed linearity between 100 and 600 normalized MGMT copies per reaction. Above and below this range the assay could only generate a qualitative result. In comparison, the digital droplet PCR assay showed linearity across the entire range of the dilutions (1 to 1:3125) with an r2 value of 0.99. Conclusions: Digital droplet PCR can provide accurate quantification of MGMT promoter methylation. The advantages over MethyLight, which we used as gold standard in these studies, include 1) increased accuracy over a larger range of methylation values, and 2) the fact that the digital droplet approach provides absolute quantification data without the need to generate a standard curve. Should these data be reproduced with other promoters for which DNA methylation levels are clinically relevant, such as hMLH1 and others, digital droplet PCR may become the platform of choice for promoter methylation quantification in general. K.E. Muller, J.D. Marotti, M.D. Chamberlin, G.J. Tsongalis, L.J. Tafe Dartmouth-Hitchcock Medical Center, Lebanon, NH. Introduction: Metastatic breast cancer is a genetically heterogeneous disease and identifying effective treatment for advanced stage disease often poses a challenge. In this study, we used a clinical next-generation sequencing (NGS) hotspot mutation panel to investigate targetable genetic mutations in metastatic breast tumors. Methods: Distant metastases of twenty FFPE breast cancer samples were sequenced using the Ion Torrent PGM and the 50 gene AmpliSeq Cancer Hotspot Panel v2. DNA was extracted from unstained FFPE slides with a minimum tumor cellularity of 10%. DNA was quantified using the PicoGreen method. Barcoded libraries were prepared from up to 10 ng of extracted DNA and multiplexed for sequencing (318 chips). Data analysis was performed using the Ion Torrent Variant Caller Plugin and reference genome hg19. Golden Helix's SVS software was used for annotation of the variants, as well as prediction of the significance of the variants. Histopathologic data was extracted from records when available. Results: Nineteen of 20 samples were successfully sequenced. Specimens consisted of 12 surgical biopsies and small excisions, 7 cytology samples, and one bone marrow biopsy (failed sample). All patients were female with a median age of 64 years (range 43 to 85). The most common metastatic sites included brain (4, 21%), skin (4, 21%), lung (3, 16%), and liver (3, 16%). Overall, 28 variants in 11 genes were observed. Three samples showed no alterations and 16 (84%) showed at least one potentially biologically significant variant (BSV) defined as having FDA approved drugs or clinical trials evaluating their significance. Potential BSVs included mutations in the following genes: TP53 (9), APC (5), PIK3CA (4), MET (2), ERBB2 (1), AKT1 (1), CDKN2A (1), SMO (1) and FGFR3 (1). Data on how these results influenced subsequent treatment decisions is still being obtained. The predominant histologic subtype of the corresponding primary tumors was invasive ductal (10/17, 59%), invasive lobular (3/17, 18%), or mixed subtypes (3/17, 18%). Twelve of the primary tumors were high grade (92%). The predominant molecular subtype of the primary tumors was ER positive, luminal B tumors (6/15, 40%), 3of which had HER2 amplification. Six of the primary breast tumors were triple negative. Three cases had a change in biomarker expression from the primary to metastatic tumor; 2 lost HER2 expression and 1 lost ER and PR expression. Conclusions: Potentially actionable mutations were identified in a majority of breast cancer metastases. Evaluating metastatic breast tumors using a NGS platform provides better understanding of the mechanisms behind tumor progression and evolution and may be useful in directing therapy and determining eligibility for clinical trials.
Adenocarcinomas: A Correlation Study with Targeted Next-Generation Sequencing L. Wang, A. Drilon, M. Rao, R. Aryeequaye, L. Cao, N. Islamdoust, R. Shah, A. Zehir, R. Benayed, J. Sadowska, J. Casanova, M. Berger, M. Hameed, M. Ladanyi Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: RET and ROS1 fusions have been identified as driver alterations in lung adenocarcinoma, both of which are actionable using tyrosine kinase inhibitors (TKI). Clinical trials are ongoing to investigate the efficacy of several TKIs in the treatment of RET or ROS1 fusion lung adenocarcinomas. We present here our experience in RET and ROS1 FISH analysis of lung adenocarcinomas with a summary of patterns of FISH abnormalities and a correlation with targeted nextgeneration sequencing (NGS). Methods: RET and ROS1 FISH studies have been conducted in a clinically and molecularly enriched group of patients (never-or lightsmokers, pan-negative for other known drivers) since 2012. FISH analysis was performed on FFPE tumor samples using dual-color (G-green and R-red) breakapart probes for RET and ROS1. A minimum of 100 tumor cells were examined per tissue section from 6 or more distinct tumor areas. A subset of patients' samples was also analyzed by a hybrid capture-based targeted NGS (MSK-IMPACT, J Mol Diagn. 2015; 17:251-64) to identify RET and ROS1 fusion partners. Results: RET rearrangement was detected in 20 cases by FISH analysis, of which fusion partners were identified by MSK-IMPACT in 13 (KIF5B=7, NCOA4=2, CCDC6=2, KIAA1468=1, TRIM33=1); ROS1 rearrangement was detected in 24 cases by FISH and fusion partners were identified in 12 (CD74=10, EZR=1, SDC4=1). For RETjmd.amjpathol.org ■ The Journal of Molecular Diagnostics fusion cases, rearrangement signals were detected in 30% to 90% (average=63%) of cells scored across the cohort, and the predominant FISH pattern (16 of 20, 80%) was split 5' and 3' RET plus G/R overlapping signals. "Single 3' RET" plus G/R overlapping signals were observed as the dominant signal pattern in 3 cases (19%). In addition, one positive case showed only split 5' and 3' signals (no G/R overlapping signal). For ROS1 positive cases, rearrangement signals were detected in 30% to 87% (average=64%) of cells scored. In contrast to RET FISH, the "single 3' ROS1" and the "split only" (no G/R overlapping signal) patterns were much more common in ROS1 rearrangements, and were observed in 10 of 24 (42%) and 8 of 24 (33%) cases, respectively. The patterns of RET or ROS1 FISH abnormalities had no correlation with specific fusion partners. In addition, all RET-and ROS1-fusion positive samples identified by MSK-IMPACT so far in our study were positive by our FISH assays. Conclusions: FISH patterns of RET fusions are similar to that of ALK fusions; in contrast, FISH patterns of ROS1-fusion are more complex. Introduction: Personalized cancer treatment can benefit from targeted deepsequencing of multiple genes using NGS. The StrandAdvantage test panel assays 212 amplicons (48 genes) spanning hotspot mutations with highest relevance to the treatment of cancer. It is performed using the Illumina TSACP panel. We present results from the analytical validation of this panel performed in 2 different Strand's clinicl reference laboratories located in Bangalore, India and Denver, USA using DNA extracted from fresh tissue, FFPE, commercially available controls and patient samples. Methods: The analytical validation framework addresses accuracy, sensitivity and specificity, limit of detection, precision and reproducibility, using 20 to 30 independent sequencing runs of about 10 control samples each performed in 2 separate laboratories (Bangalore and Denver). All of the above mentioned StrandAdvantage test performance characteristics were evaluated using DNA and FFPE reference standards from Horizon Diagnostics (Cambridge, UK) and cell lines from ATCC, USA. Additionally, SNP concordance was evaluated across 7 HAPMAP samples from a Utah family, and evaluated against Illumina's data on the same samples. The assay was further evaluated for its technical performance on normal and tumor FFPE samples. A limited number of mutations detected in FFPE patient samples were confirmed using Sanger sequencing and ARMS-Scorpion qPCR assay. Limit of detection experiments were performed using the Horizon DNA and FFPE QMRS standards, and various dilutions of cell lines, fresh tissue and FFPE samples. Clinical validation and utility were further established with over 100 independent FFPE samples. Results: We achieved equivalent performance across different operators, sequencing runs and laboratories in 2 independent laboratories (Bangalore, India and Denver, USA) of Strand Life Sciences. The analytical performance shows 98% accuracy, sensitivity and specificity for this assay independently in both laboratories. Precision and reproducibility were established at 99%. The limit of detection of SNPs was set at a conservative 10%, even though the assay detected accurately SNPs at (7% to 8%) lower frequencies. Conclusions: These data thus demonstrate robust technical performance of the StrandAdvantage solid tumor NGS test and its suitability for adoption in clinical oncology practice.
Y. Ding 1,2 , P. Zhang 1 , H. Fernandes 1, 2 1 Weill Cornell Medical College, New York, NY; 2 New York Presbyterian Hospital, New York, NY. Introduction: Microsatellite instability (MSI) is a useful marker for risk assessment, prediction of chemotherapeutic responsiveness and prognosis in patients with colorectal cancer (CRC). Recent guidelines recommend that deficient mismatch repair/microsatellite instability (dMMR/MSI) testing must be performed in all colorectal cancers for prognostic stratification and identification of Lynch syndrome. Ancillary testing algorithms for detection of mutations in the MMR genes is recommended for patients who are MSI-H and BRAF wildtype. The guidelines also emphasize RAS mutation testing of colorectal carcinoma tissue for patients who are being considered for anti-EGFR therapy.The proposed expanded panel includes testing for variants in KRAS and NRAS codons 12, 13 of exon 2; 59, 61 of exon 3; and 117 and 146 of exon 4. The mutational spectrum of RAF, RAS and MMR gene families are collectively addressed using targeted NGS. Methods: Paired normal and tumor DNA from Formalin Fixed Paraffin Embedded tissue of 10 CRC specimens were used for analysis. Immunohistochemistry (IHC) for DNA mismatch repair (IHC-MMR) proteins and NGS using the 50 gene Ampliseq hotspot panel were performed on tumor specimens. DNA from the 10 paired specimens was assessed for Microsatellite instability (MSI). The normal/tumor paired specimens were also subject to NGS on the Proton using a 400 gene comprehensive cancer panel (CCP) (Life Technologies). Results: All 10 patients had tumors with microsatellite instability (MSI-H). Five of these were BRAF V600E positive and indicative of sporadic origin. These tumors showed loss of MLH1 and PMS2 as determined by IHC. A total of 16 extra variants were detected using the 50 gene panel, whereas the CCP detected the same 16 variants plus 18 additional variants. Eleven of these variants were present in the MMR genes and classified as non-pathogenic. Of the 5 MSI-H patients with wildtype BRAF status, 3 exhibited loss of MLH1 and PMS2 by IHC and were referred to genetic counseling. One patient was found to have a somatic variant in MLH1 that was likely pathogenic. The other 2 MSI-H patients with wildtype BRAF did not show loss of MMR genes by IHC, but 1 was shown to harbor a somatic MSH2 variant that was likely pathogenic. The total number of variants in MSI-H BRAF mutant (N=6.8) and MSI-H BRAF wildtype (N=6.7) tumors were comparable. The larger panel detected 50% more tumor-relevant variants than the smaller targeted panel. Conclusions: Testing for MMR and MSI followed by NGS using a comprehensive cancer panel has utility for simultaneous risk assessment, prediction of chemotherapeutic responsiveness and prognosis in patients with colorectal cancer. Additional variants detected with larger panels can stratify patients to appropriate clinical trials. BRAF testing can accurately identify sporadic colorectal cancer in MSI-H cases irrespective of IHC MLH1 status. New screening methods to improve comprehensiveness, sensitivity, throughput and turn-around time are needed. Here we report further assessment of the Archer targeted sequencing technology in detecting fusions in different cancer types. We also describe the ongoing analytical validation of the Archer assay by assessing its precision, reproducibility and sensitivity. Methods: Previously characterized fusionpositive samples (10 cell lines, 30 patient FFPE samples) were tested for accuracy using Archer FusionPlex ALK/RET/ROS and Sarcoma (26 genes) assays. Of these, 5 samples were selected for assessment of assay reproducibility and sensitivity. 250ng of RNA, extracted using standard methods, was added for each sample. For sensitivity, serial dilutions of samples were prepared by mixing fusion-positive and normal cell line RNA. Libraries were prepared using Archer's Anchored Multiplex PCR based enrichment and sequenced on an Illumina MiSeq. Analysis was performed using Archer automated analysis pipeline. Results: Thirteen samples were analyzed with ALK/RET/ROS assay (4 cell lines, 9 FFPE tumors), and 27 with the sarcoma assay (6 cell lines, 21 FFPE). The Archer assay confirmed all fusions except for one EML4-ALK fusion, a discrepancy attributed to significantly lower tumor content in deeper recut sections used for this study. Interand intra-assay reproducibility and precision was demonstrated in the tested samples by consistent detection of expected fusions (SS18-SSX1, EWSR1-WT1, EWSR1-ERG and NAB2-STAT6). One EWSR1-FLI1 cell line was used for sensitivity, and the fusion was detected in all tested serial dilutions samples (100% to 3%). Conclusions: The Archer assays provide a viable alternative to the existing cytogenetic and PCR-based methods for fusion gene detection in sarcomas and lung adenocarcinomas. It allows the concurrent screening for all fusions involving the genes included in each panel regardless of fusion partners or breakpoints and in a single reaction. This enables comprehensive testing with optimal allocation of tissue and expedited results in a clinically actionable timeframe. In addition, interpretation in the 1% to 10% positivity range in the luminal A and B types is prone to error and ESR expression data may help with stronger cohorting of true negative cases. To further improve specificity we compared results using 2 different technologies, RT-PCR and digital color-coded barcode technology that is based on direct multiplexed measurement of gene expression. Methods: We compared the ESR expression and ER protein levels in a cohort of 27 triple negative cancer patients with minimum follow up of 2 years using formalin-fixed, paraffin embedded, blinded tumor samples. They were evaluated for the epithelial to mesenchymal transition (EMT) gene expression as also gene expression of the PI3K-AKT signaling pathway components using RT² Profiler PCR Array expression kits containing 84 genes each. In addition, 14 patients with >50% ER positivity and 8 patients with 1% to 10% ER positivity by IHC were used as controls. These samples were then analyzed with the nCounter Human Breast Cancer ER assay and results compared with the Profiler and immunohistochemistry results. Results: Eighty-nine percent of triple negative patients showed down regulated ESR (true negative) status. One hundred percent of patients in the 1% to 10% range showed down This analysis demonstrates early analytical validation for the incorporation of AFPSTP into the JAX-CTP test system. The Archer fusion detection assay enables the incorporation of a more comprehensive somatic tumor profiling assay with ease and an acceptable turn around time for a clinical laboratory.
K.E. Muller 1,2 , S. Mockus 2 , S. Patterson 2 , G. Ananda 2 , T. Mitchell 2 , V. Spotlow 2 , E. York 1 , S. Palisoul 1 , G.J. Tsongalis 1,2 , C.C. Black 1 1 Geisel School of Medicine at Dartmouth, Dartmouth Hitchcock Medical Center, and Norris Cotton Cancer Center, Lebanon, NH; 2 Jackson Laboratory for Genomic Medicine, Farmington, CT. Introduction: Genetic analysis plays a critical role in effective therapy selection for non-small cell lung cancer. Although outcomes have improved with adenocarcinoma, largely based on mutation-driven targeted therapy, the molecular characterization of squamous cell carcinoma of the lung (SCC) is more elusive, and thus treatment options are still limited. In this study we used a clinical next-generation sequencing (NGS) hotspot mutation panel to investigate new potential targetable genetic mutations in lung SCC. Methods: SCCs of the lung were retrospectively identified from the pathology database. Slides were reviewed for tissue block selection and determination of percent tumor cell content. DNA was extracted from unstained formalin fixed paraffin embedded (FFPE) sections using the QIAmp DNA FFPE Tissue Kit (Qiagen). The JAX-CTP uses hybrid capture to enrich and sequenced exons of 358 genes of interest on the Illumina HiSeq 2500 or MiSeq followed by variant detection and functional and clinical annotation. The JAX-CTP is validated for the detection of clinically actionable variants in the form of single nucleotide variations, small insertions and deletions and amplifications. Results: Nine of 14 lung SCC samples were successfully sequenced. Specimens consisted entirely of pneumonectomy and wedge resections. The majority of patients were male (male:female ratio 3.5:1) with a mean age of 67 years (range 59 to 80) at diagnosis. All tumors were morphologically SCC with a range of differentiation from low, moderate, and poor (1, 11%; 4, 44%; 4, 44%, respectively). Mean tumor size was 4.9 cm. Overall, 14 variants in 7 genes were observed. Three samples showed no alterations, and 6 (67%) showed at least 1 potential significant variant, with a mean of 2.8 variants per tumor. Mutations in the following genes were identified: CCND1 amplification (4), MYC amplification (4), PIK3CA E545K (3), AURKA F31L (2), FGFR1 amplification (1), TP53 R158L (1), AKT1 amplification (1) and RET G691S (1). One VUS was identified in TP53 Y163C in 1 case. Conclusions: Massively parallel sequencing using a NGS hotspot panel identified at least 1 potential clinically actionable variantin the majority of lung SCC samples. Routine NGS of lung SCC can further advance the understanding of the molecular profile of lung SCC, assist in selection of targeted-therapy or clinical trials for patients, and progress the development of novel therapies for these tumors. Introduction: Sarcoma is a heterogeneous malignancy that arises from cells of the mesenchymal origin, compromising connective tissues, such as bones, cartilage, muscle, blood vessel, fat, peripheral nerves, fibrous, or related tissues. It is classified into 2 groups: bone and soft tissue sarcomas. From a molecular perspective, some sarcoma subtypes are associated to either a genetic alteration (translocations) or specific activating mutations. The Archer FusionPlex Sarcoma Panel is a targeted sequencing assay that detects and identifies fusions of 26 genes associated to sarcoma. It enables rapid preparation of multiplexed next generation sequencing libraries for targeted capture of mRNAs produced from fusion genes, allowing the detection of both known recurrent fusions as well as previously unidentified fusions at key breakpoints in target genes. In this study, we evaluated the Archer FusionPlex Sarcoma Panel. Methods: A total of 8 samples were evaluated, 2 controls (negative and positive for the EWSR1/FLI3 fusion) and 6 clinical samples. Total nucleic acids from 6 FFPE samples were extracted using the Agencourt FormaPure TNA Kit, and RNA was quantified using the Qubit HS RNA Assay. One hundred and fifty nanograms of RNA were converted into cDNA, and 7 of 8 samples passed Archer's PreSeq RNA QC that measures the amount of RNA that is of sufficient length to permit fusion calling. After end repair/dA-tailing, adapter ligation, and PCRs, libraries were quantified using KAPA Library Quantification Kit. Library quantification was higher than 70nM, except for the sample that failed the first QC (2nM). Libraries were diluted to 2nM, pooled and sequenced on the Illumina MiSeq. Data analysis was performed using Archer Analysis Software. Results: One hundred percent accuracy was observed. Positive and negative controls were correctly called. Five samples were also correctly called wild-type or positive for the following fusions: FUS/DDIT3, and PAX/FOXO1. The sample that failed the first QC did not pass sequencing QC metrics was decalcified, which explains failure. Conclusions: The Archer FusionPlex Sarcoma Panel, can identify sarcoma related fusions in a single assay, reducing turn-around time, and eliminating reflex testing. It can also contribute to the detection of novel sarcoma subtypes, which may lead to an early diagnosis of the disease, as well as to an appropriate treatment and better clinical outcomes. The Archer FusionPlex Sarcoma Panel is a comprehensive, detailed, innovative, and streamlined test that could be used in diagnostic laboratories. Massively parallel sequencing (MPS) is one way to assess multiple therapeutic gene targets in a patient. Here we surveyed our first year of experience with a 26-gene solid tumor cancer panel to assess clinical utility and cost effectiveness. Methods: During 2014, we assessed 226 cancer patients for variants in therapeutic gene targets using the TruSight Tumor kit on the Illumina MiSeq instrument. We determined positivity rates for variants in each gene in our 3 main tumor types tested (lung, colorectal, and melanoma). We assessed clinical utility by classifying the variants as unknown, therapeutically informative (known) and variants potentially useful for clinical trial inclusion (potential). Finally we calculated the direct cost and turn-around-time (TAT) for the 7 most therapeutically informative genes for solid tumors (EGFR, KRAS, NRAS, BRAF, KIT, PDGFRA, and PIK3CA). These calculations were compared to our previous reflexive algorithmic approach of testing multiple genes using real-time PCR analysis and reference laboratory testing. Results: Approximately 45% of our patients were identified to have a therapeutically informative variant in the patient's tumor type, whereas an additional 11% of patients had a variant classified as therapeutically informative in another tumor type. In 44% of patients, the cancer panel did not inform therapeutically. The positivity rates of the variants identified were consistent with those in the literature. With the previous reflexive testing algorithmic approach the estimated annual direct cost was greater than $250,000, and the estimated TAT (including referral lab testing) was 18 to 33 days. The direct cost of the cancer panel by MPS was approximately $140,000. The TAT for MPS was 6 to 11 days. Conclusions: Using the cancer panel, variants with known or potential therapeutic impact were identified in 56% of patients. Our firstyear experience with an MPS solid tumor cancer panel resulted in improved TAT and estimated direct costs savings. This is likely due to performing the MPS locally and a reduction in costs associated with referral lab testing. and molecular inversion probe (MIP) array (Affymetrix, Santa Clara, CA) were included in the study and compared. Pre-designed TaqMan assay (Applied Biosystems, CA) was selected for at least 2 exons in the gene. RPP40 was used as an endogenous copy number reference in multiplex reactions. Input genomic DNA amount was tested in a range from 2 to 20ng. TaqMan quantitative Assay was performed according to manufacturer instructions. Limit of detection (LOD) of assay was determined by serial dilution SKBR3 cell line DNA (with amplified ERBB2) in wild type DLD1 cell line in 3 independent. Results: Input amount of 5ng FFPE DNA per assay was found to be adequate without limiting its sensitivity for detection of 2 copies in endogenous control. TaqMan assay detected each of the expected 66 CNAs in the 19 positive samples and SKBR3 cell line. The LOD studies showed qPCR assay to be highly sensitive with LOD upto 3 copies consistently. Low level CNAs (4-9 copies) in 6 out of the 19 cases were confirmed also by qPCR. Conclusions: Quantitative PCR-based single gene assay was found to be completely concordant with high throughput CNA detecting platforms like NGS and MIP arrays. The compatibility with low FFPE DNA input, high sensitivity, and shorter turnaround time makes qPCR-based assays a valuable and economically viable option, not only for CNA detection in individual genes but also as confirmatory assays for high throughput CNA screening assays.
ST109. Molecular Alterations Identified in 88% of Lung Adenocarcinoma Specimens P. Choppa 1 , J. Cooc 1 , K. Cogdill 1 , D. Chi 1 , J. Li 1 , K. Bloom 2 1 Thermo Fisher Scientific, West Sacramento, CA; 2 GE Healthcare, Aliso Viejo, CA. Introduction: Non-small cell lung cancer (NSCLC) is the leading cause of cancerrelated deaths worldwide. Two-thirds of patients present with advanced disease and have an average survival of less than 1 year with standard chemotherapy. Over the past several years targeted therapies have been approved showing improved survival and progression free survival in many cases. Identification of biomarkers that predict the efficacy of a targeted agent is essential in proper treatment selection. In addition to the well-established biomarkers there have been a number of potential targets identified over the past decade with the majority occurring at low frequencies. Comprehensive sample profiling is becoming increasingly challenging with small tissue samples having limited tumor content using a sequential testing strategy. Many low yield samples may not get tested for clinically relevant biomarkers without the ability to assess multiple targets in parallel. We applied a targeted 25 gene next-generation sequencing panel with a low DNA input requirement to 794 consecutive lung adenocarcinoma specimens to comprehensively profile real-world samples for current predicative biomarkers as well as emerging targets with potential actionability. Methods: A total of 794 FFPE tissue samples were sectioned at 7uM. Tumor content was determined for all samples through pathology review of an H&E stained adjacent section. DNA and RNA were isolated from the sections using RecoverAll Total Nucleic Acid Isolation kit for FFPE. The RNA was converted to cDNA and both fractions were assessed for yield using QPCR. The sequencing library was generated using Ion AmpliSeq sequencing technology. Sequencing data was generated for each sample using an Ion Personal Genome Machine System. Although the significance of many of these alterations is still under investigation this study demonstrates that targeted next-generation sequencing provides a reasonable testing strategy to comprehensively profile multiple targets on samples that may not provide sufficient content to perform sequential testing. Introduction: Next-generation sequencing (NGS) has been rapidly adopted in clinical laboratories due to its ability to detect a variety of mutation types, such as point mutations, CNVs, INDELs, translocations, in many samples and many genes simultaneously. One significant challenge of this technology is the number of different platforms available, each unique in its chemistry and library preparation workflow. For this reason, there is an urgent need to implement a quality control standard that could be used as a positive control for NGS assays in clinical laboratories. SeraSeqTM (AF20) Reference Material is a quality control material manufactured from purified human genomic DNA as well as biosynthetic DNA with a 20% variant allele target for each mutation, SNPs and INDELs (small and large). It has been designed for use with targeted NGS assays that detect mutations in key oncogenes and tumor suppressor genes. It has been formulated to monitor library preparation, sequencing, and variant allele detection. In this study, we evaluated the SeraSeqTM (AF20) Reference Material. Methods: A total of 3 samples were tested. Samples 1, 2, and 3 had allelic frequencies of 30%, 25%, and 20% respectively. Library preparation, sequencing and data analysis for those samples followed the DHMC clinical sequencing pipeline using the AmpliSeq Cancer Hotspot Panel v2. Samples were quantified using PicoGreen Quant-iTTM PicoGreen dsDNA Assay Kit, and 10ng of gDNA was used to generate barcoded libraries. Libraries were sequenced on the Ion Torrent PGM using a 318 chip. Data was analyzed using the Torrent Suite analysis package (version 4.0.2). Results: Two samples had all 26 mutations (18 point mutations, 2 small INDELs, 2 large INDELs, and 4 point mutations in homopolymer region) as expected. One sample missed a point mutation in the TP53 gene. Almost all mutations detected showed the expected allelic frequency. Only 1 mutation, H1047R in the PIK3CA gene, showed a lower frequency in all 3 samples. Conclusions: The SeraSeqTM (AF20) Reference Material contains a variety of clinically significant mutations in genes associated with hematologic and solid tumor malignancies. In addition, these mutations include point mutations as well as small and large INDELs. For this reason, it can be used as a positive control during the validation process and in routine clinical sequencing of NGS cancer hotspot panels.
Patients with Non-Small Cell Lung Cancer: A Quality Improvement Study V. Padmanabhan, H.S. Currens, H.B. Steinmetz, E.J. Rizzo, A.J. Erskine, T.L. Fairbank, F. de.Abreu, G.J. Tsongalis, L.J. Tafe Dartmouth-Hitchcock Medical Center, Lebanon, NH. Introduction: Non-small cell lung cancer (NSCLC) has a high incidence world-wide and the discovery of acquired genetic alterations in genes such as epidermal growth factor receptor (EGFR) and anaplastic lymphoma kinase (ALK), pharmacologically targetable tyrosine kinases, has changed the way lung cancers are diagnosed and treated. At Dartmouth-Hitchcock Medical cytopathologists perform rapid on-site evaluation for adequacy of specimens for diagnosis. For the past 4 years, the adequacy assessment has included collecting material for molecular testing. When separate PCR assays were used to test for mutations in individual genes, more than 95% of cytology samples were adequate for testing. For the past 2 1/2 years however, next generation sequencing (NGS) analysis has been performed using the Ion Torrent PGM and the 50 gene AmpliSeq Cancer Hotspot Panel v2. With this, we noticed an increase in the failure rate of cytology samples and, following complaints from clinicians about inadequacy of cytology samples for molecular testing, this QI project was initiated. Methods: After mapping the process, making a fishbone diagram and collecting baseline data on adequacy of cytology specimens for molecular testing, change #1 was initiated which concentrated all the material in one block. This did not produce the desired results and was followed by change #2 where molecular testing was ordered at the time of collection and the block was cut only once; a total of 20 sections were cut, the initial 12 sections were used for molecular testing when appropriate (with multiple sections on each slide) and subsequent slides for H&E and immunohistochemistry. Data were collected in an Excel sheet and adequacy rate was determined. Results: At baseline, 51 cases collected by CT-guided needle core biopsy (NCB) and touch imprint cytology between January 2014 and August 2014 were analyzed by NGS, 68.8% of specimens were adequate for molecular studies. Following change #1, 5 of 10 (50%) samples were adequate for molecular testing. Following change #2, 6 of 6 (100%) samples were adequate. During the same period, 28 cases were collected by fine needle aspiration (FNA) and 25 (89.2%) were adequate for molecular studies. Following change #1, 13 of 16 (81.2%) samples were adequate for molecular. Following change #2, 7 of 7 (100%) samples were adequate. Conclusions: This study focused on factors that are controllable in the pathology department laboratory and on maximizing the use of scant tissue. Concentrating all cores and FNA material in one block and precutting slides markedly improved adequacy of material for molecular testing using NGS. Introduction: The use of targeted therapies in personalized cancer treatment has become routine practice in the clinical environment. Therapeutic targets are selected based on the patient's individual somatic mutational profile. Patients diagnosed with advanced stage of colorectal cancer (CRC) can be treated with VEGF targeted drugs, which blocks angiogenesis, or EGFR targeted drugs, which inhibits cell growth. The aim of our study was to analyze a subset of patients diagnosed with CRC using a large next generation sequencing (NGS) panel that could potentially reveal other possible treatment targets. Methods: A total of 15 FFPE colon samples were screening using the JAX Cancer Treatment Profile (JAX-CTP), which was designed to sequence coding exons of 358 cancer-related genes. DNA was purified from FFPE tumor samples using the QiaAMP DNA FFPE Tissue Kit (Qiagen) and library preparation was performed with Agilent SureSelect XT. The JAX-CTPcustom capture library was then hybridized to the prepared DNA and sequenced on the Illumina HiSeq 2500. Data analysis was performed using an internal Clinical Genome Analytical pipeline (JAX-CGA). Results: A total of 7 patients (44%) showed BRAF V600E mutation, followed by RAS mutations (37.5%), PIK3CA (25%), and TP53 (19%). Only 2 samples did not show a mutation in 1 of those genes. Point mutations in AURKA, RET, FGFR4 genes were detected as well as amplifications in MYC, MYCN, and CCND1 genes. Conclusions: The use of a large NGS panel, JAX Cancer Treatment Profile (JAX-CTP), allows simultaneous detection of numerous mutations in clinically actionable genes (BRAF, RAS, and PIK3CA) as well as of genes with potential future clinical significance.
R. Margraf, J. Durtschi, C. Paxton, C. Vaughn, K. Geiersbach ARUP Institute for Clinical and Experimental Pathology, Salt Lake City, UT. Introduction: A common method of NGS data copy number variation (CNV) detection uses allelic imbalance information. This works well for whole genome, exome, and large gene panels, which typically have many variants across each chromosome to analyze for allelic imbalance. But for smaller targeted gene panels, the number of SNPs available for analysis per gene is often limited. We present a method for designing additional probes to increase the potential of CNV detection by allelic imbalance for smaller, focused gene panels. Methods: DNA was isolated from jmd.amjpathol.org ■ The Journal of Molecular Diagnostics 6 sets of tumor tissue (FFPE or frozen) and paired normal FFPE tissue or blood. These samples harbored known somatic CNVs previously detected by either FISH or SNP microarray. The gene-specific CNV probe sets were designed for 5 genes (EGFR, BRAF, PTEN, P53, ERBB2) using 120bp probes (IDT). In addition to exonspecific probes typically included in NGS capture assays, these additional probes were designed to maximally capture potential variation to use for the allelic imbalance analysis. First, the entire gene region was used in the IDT probe design tool to identify intronic or nearby intergenic regions where probes could be designed. Probes were then designed to cover a variety of known SNP population frequencies (NCBI dbSNP database) at multiple positions along the gene. Regions of SegDups (UCSC), polymer tracts, repeats, A/T or G/C rich were avoided. DNA was enriched for the regions of interest using these probes and sequenced with 2x150 reads on the Illumina Nextseq500. Results: Between 16 and 82 additional 120-bp probes were designed in the intronic and intergenic regions of the 5 genes, creating the potential for 42 to 300 additional SNPs to be detected. Sequencing of the 6 tumor/normal paired samples captured with the designed probes yielded 65, 13, 15, 8, and 8 average SNVs per intron for EGFR, BRAF, PTEN, P53, and ERBB2, respectively. In the 2 samples with previous SNP microarray data, allele status was in agreement using the allelic imbalance analysis for all 5 genes. In the 4 tumor samples with known gene amplifications by FISH, 3 of the samples showed the same amplification using this strategy. The discrepant sample had intra-tumor heterogeneity with 10% of the tumor population showing low level amplification of ERBB2 by FISH. A reflex with the SNP microarray agreed with the NGS result. Conclusions: This method of probe design for CNV detection by allelic imbalance allowed CNV analysis for a smaller, more focused gene panel. The additional data from the designed probes may also be used to increase the data points for read count analysis. Low level amplification and sub-clonal CNVs may not be detectable with this method, but such abnormalities may not be as clinically useful.
Evaluated with a Targeted Next-Generation Sequencing Panel J. Cooc 1 , P. Choppa 1 , K. Cogdill 1 , D. Chi 1 , J. Li 1 , K. Bloom 2 1 Thermo Fisher Scientific, West Sacramento, CA; 2 GE Healthcare, Aliso Viejo, CA. Introduction: Over the years a number of studies have demonstrated that EGFR exon 19 deletions and the L858R substitution are powerful predictive biomarkers in patients treated with Erlotinib or Gefitinib. In addition to exon 19 deletions and L858R there have been a number of EGFR alterations identified which may have potential actionabilty.There are kits currently approved to assess the presence of EGFR mutations in lung cancer samples. Both kits are PCR based methodologies and are limited to detection of alterations that are targeted in the assay design. Sanger sequencing may be used to evaluate samples for alterations that are not targeted in the kit but it may lack the sensitivity to identify mutations occurring at low frequencies. Next-generation sequencing provides an alternative methodology for EGFR mutational profiling with analytical sensitivity comparable to the PCR based kits and the ability to indiscriminately profile EGFR as with Sanger sequencing. We profiled 781 consecutive lung adenocarcinoma samples using a targeted NGS panel to determine the prevalence of EGFR mutations not detectable by PCR based kits. Methods: A total of 781 FFPE tissue samples were sectioned at 7uM. Tumor content was determined for all samples through pathology review of an H&E stained adjacent section. DNA and RNA were isolated from the step sections using RecoverAll Total Nucleic Acid Isolation kit for FFPE. The RNA was converted to cDNA and both fractions were assessed for yield using QPCR. The sequencing library was generated using Ion AmpliSeq sequencing technology. Sequencing data was generated for each sample using an Ion Personal Genome Machine System. Results: A total of 103 samples harbored 1 or more alterations in exons 18 to 21 of EGFR. There were 35 different variants identified in this sample set contributing to 124 total detected mutation events. Eleven of the 35 variants are targeted in the PCR based kits and are responsible for 85 (68.5%) of the identified mutations. The remaining 39 (31.5%) mutations identified in the sample set are comprised of mostly one-off mutations from 24 different variants not detectable with the PCR based kits. Conclusions: The majority of EGFR exon 18 to 21 mutations occurring in NSCLC samples are detectable with the PCR based kits. However there is a subset alterations occurring in exons 18 to 21 that would not be detected by current PCR based methodologies. Although the significance of these alterations is still unclear, there are a number of studies that describe therapy response in patients harboring EGFR mutations other than the most commonly detected variants. Next-generation sequencing provides comprehensive characterization of samples to identify EGFR alterations that may be missed using a targeted PCR based approach. E.N. Ferreira, F.M. Melo, J.D. Ribeiro, M.T. Pimenta, R. Stabellini, F.A. Soares, I.W. da Cunha, D.M. Carraro A.C. Camargo Cancer Center, Sao Paulo, Brazil. Introduction: Identification of somatic mutation is at the basis of personalized medicine, where the genomic information is used to guide individual treatment options for patients. Colorectal tumors harboring mutations in RAS genes (KRAS or NRAS) are unlikely to respond to anti-EGFR antibody therapy, either as monotherapy or in combination with chemotherapy. In addition, the fact that almost 40% of colorectal tumors present mutations in KRAS gene, especially at codons 12 and 13, accounting for 80% and 15% of the cases, respectively, resulted in KRAS screening as a mandatory preliminary test for anti-EGFR therapy indication. Our institution has for long been testing KRAS mutations based on optimized pyrosequencing technique (Macedo et al, 2015) achieving high rates of conclusive results in KRAS testing for FFPE samples, which usually yields low quality DNA. Recently, with the advent of NGS technologies we have developed a screening method based on targeted-sequencing that is able analyze in a single sequencing run not only codons 12 and 13 of KRAS, but also other less common codons of KRAS and NRAS gene, contributing to patient therapy indication by providing expanded genetic information. Methods: We designed a tailored cancer gene panel, composed of 14 genes for sequencing at Ion PGM platform. We have evaluated the presence of KRAS and NRAS somatic mutations in a series of 305 unselected colorectal carcinoma (CRC) patients from our institution, from as little as 40ng of FFPE extracted DNA. Results: In total, we have identified mutation in RAS genes in 51% of 305 CRC tumor samples evaluated by NGS. KRAS mutation was observed in 45.2% of cases. Considering mutations detected in codon 12 and 13 only, our detection rate was similar to the previously obtained by pyrosequencing, 38.4% compared to 33%, respectively. Although the majority of the mutations occurred at the hotpot region in exon 2, a significant proportion of KRAS mutations, 13.5%, occurred in less common codons at exons 3 (codon 61) and exon 4 (codon 117 and 146), showing the importance of comprehensive testing. In addition, we detected NRAS mutation in 6% of cases, the majority of them at codon 61 (78%). Less than 1% of cases (3 cases) resulted in inconclusive results. Conclusions: We have developed a tailored gene panel that showed high sensitivity in detecting mutations and requires only small amounts of DNA with lower rates of failure. This panel is suitable for clinical testing and can be used routinely used for screening clinically actionable somatic mutations in RAS genes for CRC patients. Introduction: Recent discoveries have enabled us to identify common targetable genomic alterations in cancer. Application of molecular inversion probe based SNP microarray technology (OncoScan, Affymetrix) has enabled genome-wide copy number aberrations (CNAs) analysis of solid tumors with limited and degraded DNA purified from formalin-fixed paraffin embedded tissue (FFPE). We previously demonstrated that OncoScan can provide accurate and quantitative assessment of CNAs of the oncogenes HER2 and FGFR1 in breast cancer. In this study, we explored the genome-wide CNAs status including ESR1, ESR2 and PGR genes that encode estrogen receptor (ER) and progesterone receptor (PR) respectively. Methods: We selected 42 resection specimens of high grade, invasive mammary adenocarcinoma from our departmental archive collected between 2011 and 2014. ER and PR overexpression was tested by immunohistochemistry (IHC). ESR1/2 and PGR gene amplification was evaluated by OncoScan using genomic DNA extracted from FFPE. Data analysis was performed using the OncoScan Console Analysis Software (Affymetrix) and Nexus Express for OncoScan (BioDiscovery). A cut-off of 4 for high copy number gains by OncoScan was used for gene amplification based on preliminary concordant results between OncoScan and fluorescence in situ hybridization (FISH) for HER2 copy number (CN) analysis. Results: Six of 29 (21%) ER+ cases by IHC showed ESR1 and/or ESR2 amplification (3 with ESR1 and ESR2 amplification, 2 with ESR1 amplification and 1 with ESR2 amplification). One of 13 (8%) ER-cases showed ESR1 amplification. One of 17 (6%) PR+ cases and 1 of 25 (4%) PR-cases showed PGR amplification. There was no concordance observed between ER and PR overexpression by IHC and ESR1, ESR2 and PGR gene amplification. Further genome-wide analysis showed 39 of 42 cases (93%) had >20% genome with CNAs (range, 22% to 90%; total CN aberrations: 35 to 464). Only 3 (7%) showed 1% to 14% genomic aberration (total CN aberrations 13 to 35). In addition, chromothripsis was observed in 37 of the total 42 cases (88%), predominantly present in cases with high percentage of genome aberrations (>20%). Importantly, chromothripsis involving 17q was observed in 17 (85%) HER2 positive cases by FISH in contrast to 2 (9%) HER2 negative cases by FISH (p<0.001). Conclusions: High-grade breast carcinoma is often associated with high-level complex genome wide copy number alterations. HER2 amplification is associated with chromothripsis involving 17q. Genomic profiling may provide accurate CNAs assessment and prognostic and therapeutic information beyond conventional biomarker analysis. Introduction: Pilocytic astrocytoma (PA) is one of the most frequent brain tumors in childhood. A frequent abnormality documented in PAs is a gain in 7q34 due to duplication and fusion of KIAA1549 and BRAF genes with constitutive activation of BRAF kinase. Methods: Although, there are different methodologies to detect duplication events, we have compared a PCR/pyrosequencing based approach and a Fluorescent In situ Hybridization (FISH) method for the detection of BRAF-KIAA1549 duplication/fusion. Three color interphase FISH analysis was performed using a centromeric (CEP7, Spectrum Aqua) reference probe and 2 custom made locus specific probes on 7q34: RP11-837G3 (covering BRAF up to exon 9; spectrum Orange) and RP1192N1 (covering telomeric end of exon 16 of KIAA1549; spectrum Green). Ploidy of chromosome 7 was evaluated by counting number of aqua signals per nucleus. BRAF-KIAA1549 gene fusion/duplication was considered present in cases with at least 10% of nuclei (based on observations in non-neoplastic tissue) showing closely spaced signal "doublets" for both orange and green probes indicating duplication, as well as an overlap of 1 signal from each doublet resulting in 1 yellow signal indicating fusion. Pyrosequencing analysis approach is based on simultaneous amplification of a DNA fragment containing BRAF 3' UTR (7q34) and its respective pseudogene (BRAFP1) present on Xq13. The sequences of these 2 genes differentially incorporate specific nucleotides, CAG for BRAF gene (at nucleotides 2731, 2737 and 2743) and AGA for the BRAF pseudogene (at nucleotides 3007, 3013 & 3019). The percent allele quantitation of these differentially incorporated nucleotides is done by allele quantification software and a ratio of the sum of nucleotides derived from BRAF versus BRAFP1 is calculated. To obtain "Normal" cutoff value; 5 normal male and 5 normal female tonsil samples were processed multiple times and an average ratio of BRAF: BRAFP1 gene specific nucleotides (CAG:AGA) is obtained. Result: In total, 68 cases were analyzed. Fifty-nine showed concordance with both methods whereas 9 cases showed discordance. Out of these 9 cases, six showed discordance due to aneuploidy that was identified by FISH but not by pyrosequencing, 2 cases showed borderline increased frequency of cells with duplication, whereas discordance in 1 case was never resolved. Conclusions: In this study, we demonstrate 9% discordance between the 2 methods due to aneuploidy, correctly identified by FISH. These were false positives by pyrosequencing analysis, thus, making the FISH method a preferred method of analysis. The use of PCR based quantitative methods may potentially be problematic in the context of aneuploidy as observed in these cases. Cancer gene fusions can be associated with intragenic changes in DNA copy number. Genome-wide array-based DNA copy number analysis can be used to screen for this feature of gene fusions. Spitzoid neoplasms are a distinctive group of melanocytic tumors which lack mutations in common melanoma-associated oncogenes. Recent studies have demonstrated that kinase fusions are important mechanisms of oncogene activation in spitzoid tumors, and ROS1, NTRK1, ALK, BRAF and RET fusions were identified in this tumor type. Here we present the discovery of recurrent NTRK3 gene rearrangements in Spitz tumors using a genome-wide SNP-array platform. Methods: We studied 119 Spitzoid tumor samples, a majority of which were histological diagnosed as atypical Spitz tumors. Genomic DNA was extracted from FFPE tumor material, and 80ng input DNA of each sample was used for genome-wide copy number and allelic imbalance analysis by SNP-array (OncoScan, Affymetrix). Confirmatory FISH study was performed using a dual-color break-apart probe for NTRK3 (Empire Genomics). Results: Genome-wide SNP-array analysis revealed intragenic copy number changes of NTRK3 in 2 Spitz tumor samples, both of which showed copy number gain of 3' NTRK3 DNA segments preserving all coding exons for its kinase domain. In addition, no evidence of other kinase fusions observed in these 2 samples, which is consistent with the mutually exclusive pattern of kinase fusions demonstrated in previous studies in Spitzoid tumors. The intragenic breakpoints of NTRK3 in the 2 samples were mapped to intron 10 and 14 (NM_001012338), respectively. The pattern of copy number changes across the gene locus in these 2 samples was suggestive of the presence of NTRK3 fusions. NTRK3 gene rearrangement FISH study was performed on the former case and confirmed the NTRK3 gene rearrangement. The FISH study could not be performed on the latter case due to insufficient tumor material. Conclusions: NTRK3, similar to NTRK1, encodes a member of the neurotrophic tyrosine receptor kinase (NTRK) family. NTRK1 fusions have been reported in several human cancers including Spitzoid tumors with an alteration frequency of 16%. NTRK3 fusions have been found in several tumor types including congenital fibrosarcoma, leukemias, secretory breast carcinoma, mammary analogue secretory carcinoma of the salivary gland and thyroid cancer. However, to our knowledge, this is the first report of NTRK3 fusion in spitzoid neoplasms. Further analysis is in progress to identify the NTRK3 fusion partner(s) in Spitz tumors.
B.S. Robinson, A.P. Martinez, T. Schnieder, C. Hill Emory University, Atlanta, GA. Introduction: Mutational analysis is increasingly being utilized in the management of colorectal cancer to inform patient prognosis determine therapeutic strategies.
Commonly detected mutations occur in the TP53 (48-58%), APC (43-75%), KRAS (35-51%), PIK3CA (20-31%), SMAD4 (35-51%), and FBXW7 (4-8%) genes, whereas clinically actionable genes include, but are not limited to, KRAS, BRAF, EGFR, PIK3CA, AKT1, MET, KIT, CDKN2A, and PTEN. Recent studies suggest that performing a cancer panel to identify multiple mutations may be cost effective and increase a patient's quality-adjusted life years, compared with single-site mutation testing. We recently implemented a panel to sequence 26 cancer-associated genes in solid tumors using next generation sequencing modalities. The aim of our study was to compare the data from our cancer mutation panel to that of recently published data. Methods: Retrospective review of our laboratory information system identified colorectal cases evaluated with our molecular laboratory's CMP-26 panel. Mutations and mutational frequencies in selected cases were catalogued and compared with publically available databases, including the cancer genome atlas (TGCA) and the catalogue of somatic mutations in cancer (COSMIC). Results: Forty-eight cases of colorectal cancer were identified that had CMP-26 performed between January 2011 and May 2015. Seven cases were excluded due to failure to meet quality parameters. In the 41 remaining cases, 106 mutations were identified in 17 of the 26 cancer-associated genes. 38% (40 of 106) of the mutations identified were found to be clinically actionable. The following high-frequency (i.e., >10%) mutations were identified: TP53-61% (25 of 41), KRAS-51% (21 of 41), APC-49% (20 of 41), PIK3CA-17% (7 of 41), PTEN-15% (5 of 41), and SMAD4-10% (4 of 41). Two of more mutations were identified in 68% (28 of 41) of the cases, with 9 genes showing multiple mutations. Of these 9 genes with recurrent mutations, 7 contained mutations at more than 1 codon. Conclusions: Similar rates of the high-frequency gene mutations, TP53, KRAS, APC, PIK3CA, PTEN, and SMAD4, were identified by our CMP-26 mutation panel as compared to rates previously reported in the literature. A high proportion of these genes are clinically actionable. In addition, more than two-third of the cases analyzed harbored mutations in more than 1 gene, and many of the recurrent mutant genes harbored mutations at multiple codons. These data suggest that using a targeted next-generation sequencing panel to test for multiple mutations will not only help guide clinicians in treatment options, but also remain a more efficient and cost-sparing modality than targeted pyrosequencing and/or allele-specific sequencing panels. . This assay can identify somatic variations in solid tumors including lung, colon, melanoma, gastric and ovarian cancers. However, these tests must be validated in CLIA certified laboratories under CAP guidelines before using them for clinical diagnosis. After validation of this panel, we are now testing patient samples. Methods: The work flow is divided into 2 phases, a wet lab (DNA extraction from formalin-fixed paraffin embedded (FFPE) tissues, library preparation using reagents supplied by illumine and DNA sequencing using MiSeq instrument and a dry lab (software analysis of the sequence to identify the mutations). The sequencing was performed with the Illumina MiSeq instrument whereas sequence data analysis and clinical report was generated by PierianDx (St. Louis, Missouri). Results of clinical report are verified by Variant Studio (Illumina). Results: A total of 390 patient samples were sequenced. Out of those 13% melanoma patient samples, 12.8 % lung cancer patient samples, 10% colorectal cancer patient samples, and the remaining samples include all other types of cancers. Forty-one per cent of melanoma patient samples have BRAF mutations, 9.8 % of samples showed NRAS mutations, and only 3.9% of patents contain KRAS mutations. In lung cancer, we found KRAS mutations in 36% of patients, EGFR mutations in 14% of patients, PIK3CA and NRAS mutations in 6% of patients. In colorectal cancer, KRAS mutations were found in 52% of patient samples, whereas NRAS and BRAF mutations were observed in 7.5% and 2.5% of patients respectively. One hundred and ten mutations were identified in 20 of the 26 genes. Conclusions: After validation of TruSight tumor panel in our clinical laboratory, we are sequencing cancer patient samples by using the reagents provided by Illumina. Analyses of the sequence results and clinical report have been generated by PierianDx. Clinically relevant mutation information along with the FDA approved therapies is provided to the ordering physician. tissues. By comparison of the relative abundance of these peptides between samples, the expression level of proteins can be semi -quantitatively determined. We sought to identify novel therapeutic avenues by analyzing the proteome of cases of FLC compared to paired non-neoplastic liver. Methods: We retrieved 6 paired tumor-normal FFPE specimens with somatic DNAJB1-PRKACA. Ten-micron thick FFPE tissue sections were prepared for laser capture microdissection. Paired tumor and normal liver tissue samples were microdissected in duplicate. An area of 500,000 square microns was microdissected from each case. Tandem mass spectrometry-based proteome analysis was performed on each tumor-normal pair in duplicate using a previously reported method. Differentially expressed proteins in the tumor and normal tissues were identified by calculating a ratio of the level of each protein in the tumor to that in its matched normal pair. The ratios for each protein were averaged across the 6 pairs of samples. The data quality was assessed by confirmation of reproducibility of results in duplicate and identification of known overexpressed proteins in FLC. The data were then examined for biomarkers with known therapeutic agents. Results: We identified 2492 unique proteins per sample within the tumor-normal pairs. From these data, we confirmed consistent overexpression of PRKACA (ratio=2) and CK7 (ratio=6) in the tumor samples relative to normal liver. The most overexpressed protein in the tumors was anterior gradient-2 (ratio=9999) and the most underexpressed proteins were alcohol dehydrogenase 1A, 1B, 1C and 4 (ratios=0.33 to 0.30). No identified tyrosine kinases were differentially expressed. Conclusions: Our data show consistent underexpression (approximately 3 times less than normal) of major alcohol dehydrogenases in tumor tissue of all cases of FLC. These dehydrogenases are responsible for alcohol detoxification within the liver. Taking our data together, we hypothesize that ethanol may be a potential chemotherapeutic agent for FLC. Validation of these findings with in vitro models is needed to confirm the possibility utility of an alcohol-based targeted therapeutic approach. L. Johnson, J. Covino, N. Manoj, M. Bessette, E. Baravik, A. Licon, R. Walters, B. Culver, J. Stahl, J. Haimes, B. Kudlow ArcherDX, Inc., Boulder, CO. Introduction: NGS has enabled multiplexed genotyping of tumor samples for oncogenic drivers. Although focused hotspot panels are effective for identifying mutation-based (SNV/Indel) oncogene activation or tumor suppressor inactivation, many tumors are driven by more complex genomic changes including gene rearrangements and copy number variation (CNVs). We present results from a pair of targeted panels that when combined, permit simultaneous detection of SNV/Indels, fusion transcripts, splice variants and CNVs. We show that the identification of driver mutations by SNV/Indel detection is strongly complemented by ability to detect fusions and CNVs. Methods: SNVs/Indels and CNVs were detected across 67 oncogenes and tumor suppressors with VariantPlex SolidTumor, a targeted NGS assay. Gene rearrangements leading to functional mRNA fusion transcripts were detected in a parallel workflow with the RNA-based FusionPlex Solid Tumor assay. Total nucleic acid was extracted from over 40 archived FFPE nonsmall cell lung carcinomas (NSCLCs). Total nucleic acid input was split between the DNA-and RNA-based workflows. Target-enriched libraries were generated and sequenced on the Illumina NextSeq platform. Variant, fusion, and CNV detection was performed using the Archer Analysis v3.2 pipeline. Results: Known driver mutations were uncovered in many samples; however, the mechanisms of oncogenesis in at least 50% of samples could not be fully explained by the collection of SNV/Indels present in the samples. High-sensitivity detection of CNVs greatly expanded our ability characterize many of these samples. Several samples showed moderate (2X to 4X) to high-level (>4X) amplifications of EGFR. We also identified several samples with moderate or high-level amplifications in MET, MYC, ERBB2, NMYC, CCND1 and CCNE1. FusionPlex Solid Tumor permitted identification of several ALK and ROS1 fusions, despite the poor quality of RNA in the archived FFPE. In general, these amplifications and fusions were found to be mutually exclusive with expected NSCLC SNV/Indels, including those in EGFR, KRAS, and BRAF, suggesting that these CNVs and translocations represented the oncogenic drivers in these patient samples. In addition, we detected several high-confidence, likely single-copy deletions of tumor suppressors including PTEN, STK11, and APC. Conclusions: SNV/Indel detection alone is inadequate in identifying the genetic events driving many NSCLCs. Highly focused NGS-based assays for the detection of CNVs and translocations can significantly complement SNV/Indel detection for tumor genotyping, leading to higher rate of positive identification of oncogenic drivers.
S.S. Talwalkar, T. Lindeman, S. Green, J. Sizemore, J. Barry, L. Eskildsen, K. Edwards CPA Lab / Norton Healthcare, Louisville, KY. Introduction: Thyroid nodules are common in adults, but only a small fraction of them are malignant. Fine-needle aspiration (FNA) cytology provides a diagnosis of benign versus malignant disease in majority of cases. However 25% of nodules are cytologically indeterminate, thereby needing repeat FNA or diagnostic lobectomy for definitive diagnosis. Study aim was to evaluate diagnostic and clinical utility of molecular testing using the newly launched Entrogen thyroid cancer mutation panel on thyroid FNA and formalin-fixed paraffin embedded (FFPE) samples. Methods: Twenty-one cytologically indeterminate samples were used. Sixteen samples were collected in cytolyt preservative, 2 in preservcyt and 3 samples tested had both cytolyt and FFPE cell block. In addition, 2 FFPE blocks of positive controls (RET/PTC1 and ) along with 12 commercially available FFPE standards were also tested. Sixteen most common mutations in BRAF, KRAS, NRAS and HRAS and 3 rearrangements in RET/PTC1, RET/PTC3 and are detected by this assay. Cytolyt and preservcyt samples were spun to discard supernatant and total nucleic acid was extracted from the pellet using Promega total viral nucleic acid kit. DNA and RNA were individually extracted from FFPE blocks. Purity and concentrations were measured using highly sensitive fluorescent detection. Assay input for DNA mutations was between 5 to 10ng per reaction and RNA was 50ng per reaction. Assay sensitivity was evaluated by serial dilutions of the commercial standards using wild-type DNA. Results were compared with cytology diagnoses. Results: Three samples were positive for BRAF V600E mutation, 2 of which were suspicious for papillary carcinoma on cytology. One sample that was diagnosed as atypia of undetermined significance was positive for NRAS codon 61 mutation where as one suspicious for follicular neoplasm was positive for HRAS mutation. The samples that were negative for mutations and translocations were cytologically either atypia of undetermined significance or follicular lesion of undetermined significance. The assay sensitivity was 1% for all mutations except G12R/G12S/G13D (5%) and up to 0.78% for RET/PTC1 and . No cross-reactivity was seen with any of the probes. Total thyroidectomy performed for patients with positive mutation showed papillary thyroid carcinoma (PTC) (n=3), follicular variant of PTC (n=1) and follicular carcinoma (n=1). Conclusions: Thyroid cancer mutation panel is highly sensitive for detection of common genetic abnormalities seen in up to 50% of thyroid carcinomas. This test can be performed on routine cytology as well as FFPE tissue and does not require special collection medium or preservative. It improves the diagnostic yield of cytology and can therefore help in effective clinical management.
H. Rennert, K. Eng, A. Tan, J. Xiang, T. Zhang, R. Kim, W. Tam, H. Beltran, B. Robinson, J. Mosquera, H. Fernandes, A. Sboner, O. Elemento, M. Rubin Weill Cornell Medical College, New York, NY. Introduction: Whole Exome sequencing (WES) has become an effective tool for mutation detection in cancer, providing a comprehensive diagnostic approach for guiding precision cancer treatment. However, the implementation of WES in a clinical setting is challenging and there is little known to date about its performance and applicability for mutation detection in cancer. We have developed a clinical Exome Cancer Test (EXaCT-1) suitable for simultaneous detection of somatic single nucleotide variation (SNV), indels and copy number alterations (CNA), and have examined its utility to detect mutation in cancer patient. Methods: Cancer patients primarily with solid tumor malignancies were prospectively enrolled at a single academic center for paired tumor and normal tissue WES during a 2-year period. gDNA was extracted from macrodissected formalin-fixed paraffin-embedded (FFPE) tumor, or cored frozen, OCT-embedded tumor and peripheral blood lymphocytes. Tumor content was assessed based on sequencing data using CLONET v1.0. EXaCT-1 testing was performed using HaloPlex target enrichment (Agilent) followed by Illumina HiSeq 2500 sequencing (2x100bp). A total of 21,522 genes (37 Mb) were analyzed with an average capture efficiency of 90%. Reads were aligned to GRC37/hg19 DNA reference using BWA and processed accordingly to ExaCT1-pipeline v0.9. A comprehensive computational pipeline was used to detect point mutations, indels and CNAs. Mutations in cancer genes (558 genes), were categorized as category 1 or 2 based on actionability. All other somatic alterations of uncertain significance were classified as Category 3. Clinical reports were generated and discussed in precision tumor board. Results: A total of 200 patients, 313 tumor-germline pairs were successfully sequenced, with an average coverage of 86x and 16 mutations detected per case. Tumor purity ranged from 14-99%. A total of 9,086 mutations were identified in 168 genes. Among these, 72 mutations (~1%) were Category 1 (15 genes), 475 (5%) were Category 2 (153 genes) and 1,474 (94%) were Category 3. In category 1, PIK3CA, KRAS and KIT were the most commonly mutated genes, comprising 21%, 20% and 14% of the genes, respectively. Among Tier 2 genes, TP53 was most frequently mutated accounting for 22% of the mutated genes, followed by APC and KMT2D comprising 5% and 3% of the genes, respectively. Conclusions: The EXaCT-1 accurately detected most common classes of mutations in clinically relevant genes in cancer specimens, greatly expanding its utility for identifying actionable mutations that guide precision cancer treatment.
K. Kalra, S. Thakur BioGenex Labs Inc., Fremont, CA. Introduction: Brain tumors accounts for 2.4% of all cancer-related deaths and have a 5-year survival rate of 33.4 %. Glimas, make up about 80% of malignant brain tumors and Astrocytomas, a type of Glioma, is the most common primary brain
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org tumor among adults and account for roughly 75% of neuroepithelial tumors. microRNAs (miRNAs) are 22 nucleotide-long, small non-coding RNA molecules that modulate various cellular processes. Altered expressions of miRNAs have been associated with various malignancies including brain tumor. In the present study, we have carried out in situ miRNA expression profiling in various types of brain tumors in order to identify miRNA sequences as potential biomarkers for diagnosis of brain cancer. Methods: A total of 35 FFPE cases of different brain tumors (astrocytoma, oligodendroglioma, meningioma and glioblastoma) were chosen for this study. In situ detection of 5 miRNAs was carried out using ISH probes and detection systems (BioGenex, DF400-50KE). Results: All 5 miRNAs, miR-10b, miR-96, miR-146b, miR-155 and miR-200a were down-regulated in the Paired Normal (P N) cerebrum except miR-155 which showed strong staining in one case. miR-146b, miR-200a and miR-155 were up-regulated in 100% (13 of 13), 85% (11 of 13) and 46% (6 of 13) of the cases of astrocytoma, respectively, where moderate to strong staining was seen. miR-10b was down-regulated in astrocytoma. All 5 miRNAs were down-regulated in oligodendroglioma and meningioma; however miR-146b showed moderate and strong staining in 2 cases of oligodendroglioma and miR-10b showed strong staining in 1 case of meningioma. miR-10b was down-regulated in glioblastoma; however miR-155 was up-regulated. Conclusions: Visualization of miRNA expression is an advantage of ISH-based techniques over the PCR and microarray based detection that lacks spatial information. A study encompassing a larger cohort is warranted to establish the up-regulation of miR-146b, miR-155 and miR-200a in astrocytoma and down-regulation in P N Cerebrum. Consistent with the high throughput screening reports, miR-10b was down-regulated in glioblastoma.
K. Kalra, S.S. Sahu, S. Thakur BioGenex Labs Inc., Fremont, CA. Introduction: Squamous Cell Lung Carcinoma (SCC) is a type of non-small cell lung cancer (NSCLC). SCC accounts for 25% to 30% of all lung cancers. Active or passive exposure to smoking is the primary risk factor of SCC and it is more frequent in male than female. SCC is slow growing with poor prognosis in an advanced stage. Unlike lung adenocarcinoma, targeted therapies are not prevalent in SCC due to the lack of molecular markers and genetic characterization of SCC. microRNAs (miRNAs) are 22 nucleotide long, small non-coding RNA molecules known as regulators of gene expression. Dysregulation of miRNA expression has been reported in different cancer types including lung cancer. In the current study, we analyzed differential expression patterns of miR-196b, miR-205 and miR-375 in Lung SCC paired with normal Lung and Lymph node metastasis cases, in order to identify miRNA sequences as potential biomarkers for diagnosis of SCC. Methods: miRNA expression profiling of miR-196b, miR-205 and miR-375 was evaluated in a total 107 FFPE archived cases of paired squamous cell carcinoma (P SCC, n= 34+13), paired lymph node metastasis (P Inv LN, n = 13) and normal lung, adjacent alveoli to tumor (N Lung, n= 34+13). In situ detection of miRNA was carried out using fluorescein labeled miRNA ISH probes and detection system (BioGenex, DF400-50KE). Following chromogenic in situ hybridization, stained slides were scored as negative, weak, moderate or strong. Results: miR-196b and miR-205 were up-regulated (strong staining) in 88% (30 of 34) and 94% (32 of 34) of the cases of SCC respectively, a negative to weak staining pattern was observed in adjacent normal lung 62% (21 of 34) and 65% (22 of 34) cases, respectively. miR-375 was downregulated in 88% (30 of 34) of the cases where weak to moderate staining was observed. Fifty-nine percent normal lung also showed weak to moderate staining with miR-375. miR-196b showed up-regulation in 77% (10 of 13) and 62% (8 of 13) of the cases of P SCC and P Inv LN, respectively. miR-205 was up-regulated in 92% of the cases of both P SCC and P Inv LN, whereas miR-375 was down-regulated in both 77 % (10 of 13) of the cases of P SCC and P Inv LN, where negative to moderate staining was observed. Conclusions: In situ based detection provides visualization of miRNA in a histological context, which is helpful in determining tumor cell-type specific expression. Up-regulation miR-196b, miR-205 and down-regulation of miR-375 in SCC and invasive lymph node metastasis propose a potential diagnostic utility for these miRNAs in SCC, yet this should be further evaluated in light of the treatment response and a panel of miRNA markers. Lung cancer spreads to other organs though the lymphatic system, collaboration of SCC miRNA expression in lymph node may also be potential lead in diagnosis and prognosis of SCC.
K. Murphy, D. Cohen, S. Alsobrook, J. Maybray ProPath, Dallas, TX. Introduction: Fluorescence in situ hybridization (FISH) to the anaplastic lymphoma kinase (ALK) is useful for detecting gene rearrangements associated with non-small cell lung cancer (NSCLC). Rearrangements of the ALK gene can occur by inversion (ALK-ELM4) or translocation of 3' ALK with over 20 partner genes. Detection of these rearrangements is important for predicting patient response to specific therapies such as Crizotinib and Ceritinib. We have compared our experience using 3 different ALK probe sets to detect these rearrangements in a clinical lab setting. Methods: Paraffin embedded tissues were processed using an ALK break apart probe set manufactured by Abbott Molecular (73 cases), Cytocell (640 cases) and Agilent Technologies (361 cases). The FDA approved process was used for the Abbott ALK probe set; whereas the Cytocell and Agilent Technology probe sets were processed using the DAKO Histology FISH Accessory kit. Samples were scored for rearrangement and polysomy. Results: The Agilent probe set needed significantly less hybridization time (1.5 hours) compared to the Abbott and Cytocell probe sets (14 to 20 hours), which resulted in a shorter turn-around-time (3 days compared to 4). The Agilent probe set demonstrated the best probe to background signal ratio, and also had a superior repeat/failure rate (0.28) compared to Abbott (1.37) and Cytocell (0.63). The Agilent probe set also demonstrated superior separation of the 3' and 5' signals in cases with the ALK-EML4 inversion. The rate of polysomy was equivalent using the Agilent and Cytocell probe sets (28.3 and 29.2 respectively), but was significantly lower using the Abbott probe set (2.7). Conclusions: FISH analysis using the Agilent ALK probe set demonstrated superior signal to noise ratios, a shorter turn-around-time, and fewer failures compared to the Abbott and Cytocell ALK probe sets. The Agilent probe set has a smaller gap design (distance between 3' and 5' signals) which improved the ease with which the separation of signals due to the ALK-EML4 inversion was detected. Tumor genotyping experiments are increasingly moving from the characterization of solid primary tumor biopsies to fluid samples like blood and urine and the tracking of dynamic changes of the mutational profile over time. These sample types require the addition of tumor DNA enrichment steps and modifications of the molecular and data analysis protocols be able to detect rare variants. For circulating tumor cell samples, the processing must also function for a low copy input, as sometime only a few cells (<5 cells) are recovered from a blood draw. Here we present validation of a method to detect somatic mutations from a blood draw, where typical tumor cell enrichment above 10% of total cell numbers allows the use of standard amplicon libraries typically employed in the analysis of tissue based biopsies. Methods: Analytical samples were created by spiking a number of different cell lines into 14ml whole blood from normal donors, followed by tumor cell enrichment using the IsoFluxTM System. Further reference samples were created by directly spiking tumor cell line cells into known numbers of donor-derived white blood cells (WBCs). Cells were lysed and DNA was amplified by whole genome amplification (WGA) using the NGS Kit (Fluxion Biosciences), and quantified via qPCR. Targeted libraries were created using a variety of different amplicon sets, including the CHPv2, OncoMine (Ion Torrent) and TruSightTM Tumor sequenced using either the PGM (ThermoFisher) or MiSeq (Illumina) sequencing instruments; data was analyzed using a customized variant calling/ filtering pipeline based on standard alignment tools, variant calling tools (including LoFreq, University of Singapore). Finally, variant filtering and functional interpretation was performed using VarSeqTM. Clinical samples were processed using exactly the same enrichment, DNA isolation and amplification protocols. Example matched clinical samples were enumerated to determine the CTC load, where CTCs were defined as CK+, CD45nucleated cells (DAPI+). All data was analyzed in a blinded manner. Results: Multisite analytical validation data, based on spiking of cells into whole blood, and a matched molecular and bioinformatics approach demonstrates a detection limit down to 10 cells from a blood draw with a very low false positive rate. This translates into a robust detection of variants down to 0.5% mutant allele frequency, without the use of nonstandard NGS techniques like single molecule sequencing. Results are shown for a number of analytical validation experiments, detecting variants down to 2 mutated copies in a background of wild type cells. Exemplary clinical data is presented from a few patients, including biological replicates, and analysis replicates using different amplicon sets and sequencing technologies to validate the results. Conclusions: A sensitive assay for the detection of somatic variants from a blood draw using standard amplicon panels developed for somatic mutation detection from solid cancers has been well validated using analytical samples. Feasibility data has also been generated by applying the same assay parameters to patient samples. Here we report that JAK2 and CALR mutations can co-occur despite prior reports that JAK2, MPL and CALR mutations are mutually exclusive. We also identified a jmd.amjpathol.org ■ The Journal of Molecular Diagnostics novel 9bp deletion in CALR that has not been reported in MPN. Methods: Genomic DNA was isolated from peripheral blood, bone marrow, and FFPE bone marrow clot preparations from 52 MPN specimens with known JAK2 and MPL mutation status. CALR mutation testing was performed using fragment length analysis of a 265bp FAM-labeled amplicon. Sanger sequencing of a 566bp amplicon was also performed to confirm fragment analysis results. Accuracy, precision and sensitivity of the fragment analysis assay were also determined. Results: The assay has correctly identified the mutation status for all 52 specimens when compared with fragment analysis results from a reference laboratory, and by sequencing a subset of specimens with mutant alleles over 20%. Forty specimens (30 JAK2+, 2 JAK2-/MPL+ mutants and 8 JAK2-/MPL-specimens) were negative for CALR mutations. Twelve specimens had CALR mutations, which included 6 5bp insertion, 5 52bp deletion and an in-frame 9bp deletion. This 9 nucleotide (GAGGAGGAC) deletion occurring downstream of mutation hotspot region at 50% allele frequency would result in the deletion of 3 amino acids (EED, (398) (399) (400) . This is a novel mutation, which has not been reported in MPN. One specimen that had a 52bp deletion and another specimen with 9bp deletion also had JAK2 mutations. Overall, 12 of 18 (~67%) specimens negative for JAK2 and MPL mutations had CALR mutations and 6 were triple negative (JAK2-/MPL-/CALR-). The intra-and inter-assay precision had a coefficient of variation <3% for both size of the fragments and mutant peaks. Sensitivity of the assay was determined to be 5% mutant alleles for both the 5bp insertion and the 52bp deletion. Conclusions: Our results show that JAK2, MPL and CALR mutations are not always mutually exclusive. The 9bp deletion that has not been reported in MPN is likely a germline mutation. The fragment analysis is an accurate, sensitive, and precise assay for the detection of CALR exon 9 indels in the diagnosis of MPN.
Introduction: Targeted NGS technologies enable increased depth of coverage, which enhances sensitivity and provides an efficient approach for investigating heterogeneous diseases like cancer. As costs decrease and informatics improve, NGS is becoming an increasingly accessible to clinical laboratories. With growing demand and rising test volumes, the platforms utilized should be evaluated for efficiency and cost-effectiveness in addition to precise and reproducible results. We compared the performance of a 50 gene targeted cancer panel on 2 NGS sequencers from Life Technologies, the Ion Torrent Personal Genome Machine (PGM) with a 1Gb output to the Ion Proton with an output of 10Gb. Methods: DNA from FFPE tissue of 45 tumor specimens was subject to library preparation using the Ion Ampliseq Cancer Hotspot Panel v2, which interrogates hotspots in 50 cancerrelated genes. Sequencing on the PGM was performed using a total of 4 318v2 chips with 12 individually barcoded specimens per chip. The same 45 barcoded specimens were subsequently sequenced on the Proton using a single PIv2 chip. Data obtained from both sequencers was analyzed with the Variant Caller v4.4 software from Life Technologies. Results: A total of 385 variants from 45 specimens were identified in the DNA sequenced on both PGM and Proton. Variants were detected in 20 of the 50 genes in the panel. The average coverage of detected variants by PGM was 997X versus 1450X on Proton. Three hundred and thirty (85.7%) of the variants were concordant between PGM and Proton and their variant allele frequencies (VAFs), ranging from 2.5% to 98% were comparable. There were 55 (14.3%) variants that were discordant, 17 of which were detected only by the PGM and 38 only by the Proton. Ninety percent of discordant variants were single nucleotide variants (SNVs), and 10% were indels. TP53 and PTEN were the most common genes with discordant variants (25 and 7 respectively). All 25 discordant TP53 variants were SNVs, and indels were present in PTEN and APC. The majority Among the 55 discordant SNVs, 40% were G>A substitutions, whereas 30% were C>T substitutions. Sequencing on the Proton generated results 2 times faster and cost two-thirds less than the PGM. Conclusions: The increased sensitivity for detection of SNVs with low VAFs, generated on the Proton could possibly result in nonreproducible C>T/G>A sequencing artifacts in FFPE tissues, caused by the hydrolytic deamination of C to U/T. NGS performed on the Proton is more efficient with cost and turn-around-time from sample to answer. However, caution must be taken to distinguish true variants with low VAF from potential sequencing artifacts present in FFPE specimens. M. Hiemenz, N. Malani, Y. Kemel, M. Moreau, R. Daber Bio Reference Laboratories, Elmwood Park, NJ. Introduction: One major challenge in routine NGS testing of solid tumors is disease indications where specimens are often limiting in size NGS assays have higher input DNA requirements when compared to traditional PCR based methods, and as a result have a higher non reporting, or Quantity not Sufficient (QNS) rate. To address this concern and ensure broader access to NGS based panel testing, we have developed methods to drive input DNA levels to sub nanogram amounts. Methods: Mutational hotspots, surrounding exon regions, and where appropriate, full length genes were captured using a multiplex PCR approach. Relative primer concentrations for each amplicon and primer pooling strategies were optimized using both in silico models and real NGS data. Primers showing non-specific primer-primer interactions were re-designed, removed or re-pooled to reduce the number of NGS reads generated from non specific PCR products. Normal, non-tumor FFPE tissue controls along with mixed tumor samples at different DNA allelic spike in amounts were processed at 20ng, 10ng, 1ng and 500pg input amounts to determine sensitivity, specificity and quantitative precision across a range of DNA input amounts and mutational allelic burdens. Samples were sequenced on an Illumina Miseq with 2 by 175 base pair reads. Sample data was then processed with an in house suite of tools to identify mutation presence or absence as well as allelic burden. Results: NGS libraries were successfully generated for all samples tested and at each DNA input level. Three of the 15 normal FFPE controls failed to generate enough library diversity to meet reportable QC levels. Of the combined 11 variants seen in the exonic regions for both samples, 100% sensitivity was observed across all allelic burdens and input amounts. No new variants were seen across these samples above the 4% limit of detection, demonstrating 100% specificity. As input DNA amounts were decreased, the quantitative accuracy of the allelic burden began to decrease, with deviations observed when 500pg of DNA was used. Conclusions: Through careful assay design and PCR reaction cycling conditions, we have successfully generated an analytical methodology that allows for sub nanogram input DNA levels to be used in NGS library prep generation. Though quantitative accuracy begins to diminish as lower nuclei counts are analyzed, this assay allows for replacement of single gene assays for minute specimens. Introduction: Confirmation of identified variants is critical to establish the accuracy of NGS results and to meet requirements for regulatory bodies. Confirmatory assays in clinical molecular diagnostic laboratories typically include Sanger sequencing and other lab-developed techniques. However, these methods require additional time to perform, incurring delays in reporting final NGS results. To establish a more efficient approach, an orthogonal NGS testing method was developed using a modified single library preparation that can be sequenced concurrently on both Illumina MiSeq and Ion Torrent PGM platforms. Methods: A custom 35 gene SureSelect (Agilent, Santa Clara, CA) panel was designed targeting ASXL1, BCOR, BRAF, CALR, CBL, CEBPA, CSF3R, DNMT3A, ETV6, EZH2, FLT3, GATA1, GATA2, IDH1, IDH2, JAK2, KIT, KRAS, MPL, MYD88, NOTCH1, NPM1, NRAS, PHF6, PTPN11, RUNX1, SETBP1, SF3B1, SRSF2, TERT, TET2, TP53, U2AF1, WT1 and ZRSR2. 200ng of genomic DNA was used with the SureSelectXT Library kit (Agilent) sheared to approximately 150bp. After end-repair and attachment of MiSeq adapter sequences hybrid capture was carried out with SureSelect baits, samples equally divided into two aliquots. Each sample was PCR amplified to incorporate adapter indexes for the Illumina MiSeq and adapters/indexes for Ion Torrent sequencing, respectively. Following NGS, data was processed through our bioinformatics pipeline (CLC Bio Genomics Server for alignment/variant detection and Mayo developed NGS workbench for data visualization/variant annotation) followed by final analysis. Results: Seventy-one samples were analyzed by this cross-platform sequencing strategy (MiSeq representing the primary diagnostic method). Forty-eight samples had a total of 194 variants detected (144 point mutations and 50 insertions or deletions) and 23 samples had no reportable abnormalities. The Ion Torrent confirmed 192 of these 194 (99.0%) mutations. Two samples with CEBPA deletion mutations were not confirmed by the Ion Torrent, and were subsequently verified by a clinical Sanger sequencing assay. Mutant allele frequencies (MAF) obtained by the Ion Torrent were similar to MiSeq results (R2 = 0.936), except for some in/del mutations that presented more difficulty for the Ion Torrent, especially in homopolymer regions. Sequence quality for detected variants was above a PHRED score of 20 for both platforms. Conclusions: Orthogonal sequencing technique using a single capture approach is efficient and can be used reliably in the clinical molecular diagnostic lab for "real time" NGS variant confirmation. GC-rich regions can result in poor coverage depth or difficult interpretation (particularly with Ion Torrent sequencing), necessitating confirmation of primary NGS data by a separate clinical assay. Introduction: Minimally invasive biopsies from solid tumors obtained via fine needle aspiration (FNA) are regularly submitted for downstream mutation profiling via next generation sequencing (NGS). FNA material is often subjected to various cytopreparatory methods for morphologic evaluation yet the cell block preparation is most commonly submitted for downstream NGS. A cell block may not always contain adequate material for NGS testing and this may prompt additional procedures for a patient in an effort to profile the patient's tumor. This is a comprehensive evaluation of the downstream effects of 9 common cytopreparatory methods on DNA derived from FNA samples as part of the validation study of a targeted gene panel via NGS.
jmd.amjpathol.org ■ The Journal of Molecular Diagnostics future expansion. Methods: The panel design includes 2 tiers: Tier 1 with higher clinical oncology relevance (316 genes) and Tier 2 with cancer research content (852 genes). Specific fusion gene territory and cancer-associated virus capture probes are also included. Libraries are prepared from blood, bone marrow or FFPE DNA (Kapa Biosystems), subjected to hybrid capture (Roche Nimblegen) and sequenced via HiSeq 2500 (Illumina). Data analysis was performed on a HIPAAcompliant HPC system (Center for Research Informatics) using in-house developed bioinformatics pipelines, with variant detection performed at a threshold of 10% mutant allelic fraction (MAF). Results: Results from 21 normal FFPE tissues were analyzed to assess basic assay performance and depth profiles. Across the Tier 1 (~1.6 Mb) and Tier 2 (~2.8 Mb) coding areas, average sequencing depth was 780X and 380X, with 98% and 90% of bases showing depth greater than 100, respectively. These depth data were used to refine the final target regions for the Onco1K clinical assay. We then compared Onco1K results against in-house CLIAcertified NGS assays for a group of solid tumors (50 genes) and hematologic malignancies (53 genes). One hundred percent base call concordance was observed for point mutations and insertion/deletion (indel) mutations smaller than 10bp. Two indels larger than 10bp were present in the aligned data, but require de novo alignment or realignment. Conclusions: Across the clinically targeted area, Onco1K shows equivalent sensitivity and sensitivity for point mutations and indels less than 10bp in cancer samples from FFPE tissue or blood/bone marrow compared with our existing CLIA-certified NGS assays, with high correlation of detected MAFs. The tiered design of this assay allows pooling of samples at multiplexing levels appropriate for routine clinical performance volume and economics, while providing adequate depth for reliable detection of scarce variants at or above 10% MAF in the most clinically relevant genes. Introduction: Formalin-fixed, paraffin-embedded (FFPE) tissue and cytology smear slides are routine samples for cancer diagnosis, but these sample types can provide challenges for complex molecular testing such as next-generation sequencing (NGS). FFPE and cytology samples show a wide range of pre-analytical variability that can cause expensive assay failures and re-work for laboratories, in particular reference laboratories receiving specimens from large numbers of unique client pathology workflows. An accurate and reproducible pre-analytical method to evaluate extracted DNA quantity and quality could be used to provide more predictable inputs into downstream molecular assays. This study compares 2 methods of DNA measurement between 2 reference laboratories. Methods: This study was carried out in 2 phases using 120 blinded samples shared between laboratory 1 (L1) and laboratory 2 (L2) to compare the performance of the Quantidex DNA Assay (Asuragen, Inc., Austin, TX), a qPCR-based assay, and Qubit dsDNA Assay Kit (Thermo Fisher Scientific, Waltham, MA), a DNA fluorescence-based assay. In phase 1, 20 FFPE DNA samples from L1 were run in quadruplicate using Quantidex and in singlicate using Qubit at both labs. In Phase 2, 100 DNA samples (50 FFPE and 50 cytology smears) from L2 were evaluated at both labs using the Quantidex assay in duplicate and the Qubit assay in singlicate. Intra-and interlaboratory reproducibility for amplifiable DNA fraction (Quantidex) and DNA concentration (Qubit) were calculated. Results: In phase 1, there was a 2.1-fold average difference between labs for DNA concentrations measured by Qubit (R2 = 0.90), whereas there was only a 1.2-fold average difference between labs in amplifiable DNA copy number measured by Quantidex (R2 = 0.99). Intra-laboratory CV was 16.8% at L2 for Quantidex. Qubit reported a DNA concentration that was on average 5-fold higher than the "functional" Quantidex measurement. In phase 2, fold average and median differences between labs were 3.0 (1.4) for Qubit and 1.2 (1.1) for Quantidex, with inter-laboratory correlation coefficients of 0.65 and 0.88 for Qubit and Quantidex, respectively. Conclusions: In this study, the qPCR-based Quantidex assay showed better inter-laboratory reproducibility than the fluorescence-based Qubit assay. This improved performance may be due to the qPCR measuring the functional amplifiable DNA fraction rather than the raw doublestranded DNA concentration from highly variable FFPE and cytology smear specimens. Consequently, the Quantidex assay may help to standardize DNA inputs across laboratories and better inform the characterization of FFPE or cytology sample DNA prior to downstream molecular analysis.
S. Shen, D. Qin Moffitt Cancer Center, Tampa, FL. Introduction: Short Tandem Repeats (STR) assay has been widely used to monitor stem cell transplantation. To calculate the percentage of donor cells, the representative results from at least 3 markers are usually used. Lab staffs have to spend time to choose the representative markers and the calculation could be time consuming. In case of double donor transplantation, choosing representative markers and calculating are even more challenging. Besides, human errors can occur. To solve these problems, we have developed Excel based software to identify informative markers and calculate the percentage of donor cells. Methods: 1) MicroSoft Excel 2003 or later with macro function. 2) CAP PT samples from 2011 to 2014 10 sets of samples. Results: A) Basic analytic approaches and software features: 1) In STR assay results, some of the donor and recipient alleles could be the same, whereas other alleles are different. There could be more than 150 possible combinations between the donor and recipient alleles. All possible combinations have been coded in our software and a set of formulas to calculate the percentage of donor allele is developed for each combination. 2) The software automatically assigns a code for each marker based on donor and recipient pretransplant genotypes. 3) If the marker is informative, the percentage of donor allele is calculated using the formula corresponding to the particular combination. The results of second and third quartiles of all informative markers are used to calculate average and standard deviation, which are used in report. B) Software features: 1) It only requires a copy of MicroSoft Excel 2003 or after, which are able to run macros; 2) There is no need to manually select informative markers; 3) The software can take data in various formats such as csv; 4) Automatically input of sample information into report; 5) Error prove data input; 6) One press button operation for percentage calculation; 7) Able to analyze the data generated from different test kits, pre-and post-transplant data, recipient and donors' data can be generated by different kits, as long as there are enough informative markers are available. Results of 30 samples from 10 CAP surveys show that the software produced very accurate results. Overall donor percentage calculated by the software show a very strong linear relationship with mean percentage of survey results (correlation = 0.9982, R2 =0.9938). Percentage of donor allele of 359 individual markers from these 30 samples are also evaluated, results also show very strong linear relationship (correlation = 0.9882, R2 =0.9772). Conclusions: A user-friendly software has been developed to analyze STR test for chimerism. It has shown a great accuracy in calculating the percentage of donor allele. Introduction: On February 26, 2015, bioMérieux announced that its molecular affiliate, BioFire Diagnostics, LLC, (Salt Lake City, UT) received FDA clearance for the FilmArray 2.0 system. This system features a single computer device that accommodates 1 to 8 FilmArray 2.0 instruments. It also occupies a small footprint, and is capable of Laboratory Information Systems connectivity. The purpose of this study was to assess the throughput performance of this new system as it compares to the original FilmArray system. Methods: Over a period of 2 months, 1,463 convenience samples were tested using BioFire's FDA-approved assays: the Blood Culture Identification, Gastrointestinal, and Respiratory Panels. In addition to throughput, our observations included ease-of-use, scalability, runtimes, maintenance, hands-on time, invalids, and database management. Results: Up to 168 tests may be performed on the FilmArray 2.0 system within a 24-hour period as opposed to 21 samples on an individual FilmArray instrument. The hands-on time between FilmArray systems remain equivalent (between 70 seconds and 112 seconds). The runtime between FilmArray systems also remains equivalent (between 1 hour, 6 minutes, 25 seconds and 1 hour, 11 minutes, 30 seconds). From the 1,463 samples tested, 11 were invalid (1 pouch vacuum leak and 10 internal control failures). The footprint of the FilmArray 2.0 system occupies 36" x 24" x 36" for 1 to 2 FilmArray instruments and spans up to 72" x 24" x 36", dependent upon the number of instruments added. Comparatively, the original FilmArray system measures between 19" x 24" x 36" and 36" x 24" x 36", depending on its layout. Other improved features include a user-friendly, single database management system and an onboard software control center using a graphical user interface (GUI) configuration. Conclusions: The most significant improvement to the FilmArray 2.0 system was the ability to stack multiple instruments together and connect them to a single computer device for its ease-of-use and footprint. Visual LED lights and an end-of-run audible alert were additionally helpful indicators of run status. Moreover, the software component incorporates sample processing, sample analysis, and a maintenance menu that are easily accessible. We conclude that the BioFire FilmArray 2.0 system is an incremental improvement over the original FilmArray system, which will accommodate future laboratory growth. A.A. Stence, A.D. Bossler, D. Ma University of Iowa Hospitals and Clinics, Iowa City, IA. Introduction: Next generation sequencing (NGS) is widely used in clinical laboratories for detection of point mutations and small deletion/insertions. The limit of detection (LOD) of different NGS platforms varies but usually is around 5%. Mutations in certain genes are clinically important even at a low allele frequency, such as KRAS mutations in colorectal carcinoma. When the mutant allele frequency is below the LOD, the mutations would be filtered out by the pipeline and not reported. Sanger sequencing is not useful in this scenario due to its lack of analytic sensitivity. We developed multiple single nucleotide primer extension (SNPE) assays as confirmatory tests for mutations with allele frequencies around or below the LOD of NGS. Methods: Thirty-four samples which were shown to have mutations at allele frequencies around the 5% LOD of our NGS platforms were tested by SNPE assay. Briefly, genomic DNA extracted for NGS was used for PCR amplification of TT14. Multiplexed Minority Mutation Detection in Gene Panels Using Nuclease-Assisted Minor-Allele Enrichment (NaME) C. Song, R. Fontana, G. Makrigiorgos Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA. Introduction: Identification and sequencing of low-level mutations in tumor specimens or bio-fluids (plasma, urine, saliva) present a challenge, as clinical samples frequently contain far more wild-type than mutant DNA. PCR is almost always used during sequencing library preparation. However, polymerases or sequencing itself may introduce 'noise' which limits detection of rare mutations. Here we describe Nuclease-assisted Minor-allele Enrichment (NaME), a novel method that selectively digests wild type alleles in the starting genomic DNA material and increases the mutation abundance prior to PCR. This increases the detection sensitivity for all downstream detection platforms. Methods: NaME harnesses the preference of a thermostable nuclease DSN for digestion of double stranded over single stranded DNA, in order to render wild type DNA un-amplifiable, while retaining mutated DNA intact. Oligonucleotide probes overlapping mutation positions on top and bottom DNA strands are synthesized. Following denaturation, genomic DNA is incubated with probes and DSN at temperature and buffer conditions that enable probe hybridization and wild-type specific DSN digestion, whereas mutant alleles survive and can be PCR-amplified. This scheme enables multiple mutations along the probes and multiple genes to be enriched simultaneously. We tested NaME in singleplex or multiplexed format, 12 mutations over 10 genes, using serial dilutions of DNA containing low level mutations at multiple genes, as well as in plasmacirculating DNA samples. Droplet digital PCR was used to evaluate mutation abundance before and after NaME. Results: Mutation enrichment of 100 to 1,000fold, depending on conditions, is obtained by applying NaME prior to mutation detection. This major increase in mutation abundance is present on multiple targets addressed by probe-directed selective digestion, whereas non-targeted DNA remained intact. Mutation abundance increases from 0.1% to >80% in KRAS, BRAF, MEK1 and other genes, following NaME applied to mutation-containing DNA and circulating DNA. NaME applied directly on genomic DNA prior to PCR-Sangersequencing, PCR-HRM, COLD-PCR-Sanger, or droplet digital PCR (Biorad), increases the technical detection capability by ~100-fold by applying a digestion step prior to established processes. Conclusions: NaME provides a simple and powerful approach for selective removal of wild-type DNA alleles and enrichment of minor alleles at the genomic DNA level for clinically relevant mutations. Due to its multiplexing capacity, NaME is compatible with next-generation targeted resequencing with minimal deviation from existing protocols. The simplicity, multiplexing ability and low-cost of this new technology provide a useful tool for molecular diagnostics.
TT15. Quality Control in Next-Generation Sequencing Using DNA fingerprinting R. Akabari, Z. Zheng, J. Lal, S. Gandhi, D. Qin Moffitt Cancer Center, Tampa, FL. Introduction: Next-generation sequencing (NGS) is relatively new technology with complex steps. The NGS library preparation usually involves multiple samples.These samples are usually used in different sequencing runs, making it vulnerable to contamination, sample mix-up and pipetting errors. Errors during the bench pipetting are hard to notice. Here we report a QC method using DNA fingerprinting to monitor any possible sample mix-up on the bench. Methods: NGS library preparation: The DNA was isolated using the QIAGEN Blood Mini Kit. The Libraries were prepared using TruSeq Custom Amplicon Kit. Usually, 12 libraries were prepared each time, which are used for 3 sequencing runs. DNA fingerprinting: Promega PowerPlex 16 System is used for DNA fingerprinting. 0.33μl library preparation from each sample, which was theoretically equivalent of 3.3ng of initial DNA sample, was used in the DNA fingerprinting reaction. QC process: 4 cases were randomly selected for QC. DNA fingerprint tests were performed on the DNA samples and the libraries of these 4 cases separately. The results from the DNA samples were matched against that from the prepared libraries to identify any possible sample switch. Results: DNA fingerprinting results from NGS libraries: The DNA survived the NGS library preparation process. The amplicons in the library did not interfere with DNA fingerprinting reaction. We were able to obtain DNA fingerprinting results from all 4 libraries. The library fingerprinting profiles were matched with the samples' profiles. Conclusions: Pipetting errors can occur on the lab bench without an individual being aware of the error and such errors are hard to identify. DNA fingerprinting technology offers a possibility for us to check bench pipetting process. In this abstract, we show that the DNA can survive complex NGS amplicon library preparation and the amplicons generated during the library preparation will not significantly interfere with the DNA fingerprinting reaction. Therefore, DNA fingerprinting technology can be used for NGS library QC to monitor potential sample switches on the lab bench. In fact, we have encountered a scenario where one case of polycythemia vera was negative for JAK2 V617F and a case of Mycosis fungoides was positive for JAK2 V617F. A sample switch was strongly suspected, but DNA fingerprinting QC results indicated that there was no sample switch. It was found later that the "Mycosis fungoides" was a clerical error. The correct diagnosis was Myelofibrosis, which is commonly abbreviated as MF. Myelofibrosis is often positive for JAK2 V617F mutation. Introduction: Manual library preparation for next-generation sequencing (NGS) is a labor-intensive, time-consuming process. As the clinical demand for NGS testing grows, sample throughput is limited by the lengthy library preparation procedure. Library automation offers to streamline NGS workflow and minimize hands-on time.
We evaluated the ability of the Tecan Freedom EVO 100 to efficiently produce AmpliSeq libraries comparable to those made by manual methods for targeted NGS panels. Methods: AmpliSeq (Life Technologies) libraries from FFPE and FNA samples were prepared in duplicate: manually and using the Tecan robot. A limit of detection of a known BRAF p.V600E mutation from an FFPE sample was carried out on the Tecan robot and sequenced to evaluate library quality. Repeatability was tested by preparing libraries from three cell lines in triplicate at different times on the Tecan robot. Library yield, uniformity, percent reads on target and variant frequencies were compared between libraries generated by both methods. Sequencing was performed on Ion Torrent platforms and data was analyzed using Torrent Suite with Variant Caller plugin and an in-house bioinformatics pipeline. Results: Libraries from 19 samples were prepared by both Tecan automated robot and manual set-up and sequenced. Sixty-three percent of samples had higher library yields (7,200pM, mean) when prepared with the Tecan robot as compared to manual preparation (4,100pM, mean). Sequencing of all libraries was successful demonstrating similar alignment to amplicon targets (61% to 97%) and uniformity (97%). All high and low allelic frequency variants were detected in libraries generated by the Tecan robot. High time saving was achieved using Tecan automation: 1 hour hands-on tech time for preparation of 48 libraries by the Tecan versus ~3.5 to 4 hours for manual preparation. Conclusions: We have successfully shown that AmpliSeq library preparation can be automated for use in a clinical laboratory. The automation reduces hands-on tech time without sacrificing library quality. These results show the potential for high-throughput clinical NGS labs to efficiently prepare libraries using Tecan Freedom EVO 100.
Research Using TaqMan Chemistry and High Throughput OpenArray Technology K. Li, S. Patel, I. Pagani, B. Huang, W. Probert, E. Zeringer, N. Fantin, P. Brzoska, E. Diamond, N. Puri, J. Trotta, K. Varma Thermo Fisher Scientific, South San Francisco, CA. Introduction: Studying vaginal microbiota, their dynamics and imbalance provides researchers insights into vaginal health. Various technologies are available in vaginal microbial studies, but some are costly, lack of sensitivity or specificity, and others may have complicated or lengthy workflow. Although 16S rRNA-based sequencing is powerful for bacterial identification or novel bacteria discovery, it may not always give species-level resolution or doesn't work for microbes beyond prokaryotes that are important in vaginal health like Candida, protozoa and virus. PCR-based detection is sensitive and simple, but it may lack of target throughput -only assaying a single or a few targets. Methods: To address its unmet needs, we have developed a large collection of TaqMan assays for vaginal health research by leveraging our TaqMan chemistry and high throughput OpenArray technology. We identify microorganisms that are important for vaginal health research. The selected over 30 species, including normal or pathogenic ones in vaginal flora, cover a wide range of bacteria, fungi, protozoa and even virus. We identify unique gene targets that are species-specific wherever possible for each species. Using the proprietary TaqMan assay design pipeline, we design assays targeting these signature genes for these species. These assays are evaluated with synthetic templates and ATCC genomic DNA samples on 384 well plates and on OpenArray plate, a microscope slide with 3,072 through-holes (48 subarrays/plate and 64 through-hole/subarray). Results: First all the assays are screened with their controls. The accuracy reveals that all the assays display expected on-target Ct values with their corresponding synthetic templates and ATCC gDNA controls. For specificity, each assay is tested against all the rest of plasmids and ATCC controls and no cross-reactivity or off-target effect is observed. Linearity and analytical sensitivity for each assay is determined by a serial dilution of its plasmid and it is demonstrated that 7 log linear dynamic range (with R2 >0.99) with limit of detection (LOD) down to ~50 to 100 copies for all and even lower for some of the analytes. Good PCR efficiency and reproducibility are also showed. Conclusions: We have developed a large collection of TaqMan assays for vaginosis/vaginitis research. Using plasmid and ATCC controls, we have demonstrated excellent assay performance of accuracy, sensitivity, specificity, and reproducibility. In conjunction with OpenArray platform, the application enables researchers to study a large number of vaginal microbial species in a single reaction with a simple workflow, fast turnaround time, and high throughput yet flexible sample/target combinations. For research use only. Not for diagnostic procedures.
S. Kang, D. Moon, S. Jang Chosun University College of Medicine, Gwangju, Korea. Introduction: ApoE genotyping that is a clinically important test, have been analyzed by several techniques for genotyping. However, PCR-RFLP technique for ApoE genotyping is not commonly used to clinical test, because electrophoresis for PCR-RFLP using HhaI restriction enzyme have to discriminate 8bp but conventional agarose gel electrophoresis have lower discriminative power than non-denaturing polyacrylamide and sieving agarose gel electrophoresis. So, PCR-RFLP using conventional agarose mini-gel electrophoresis is not suitable for clinical test for ApoE genotyping. In this study, we developed the rapid single-tube PCR-RFLP for ApoE genotyping using conventional agarose mini-gel electrophoresis: Methods: We used 50 randomly selected samples and 6 controls (e2/e2, e2/e3, e2/e4, e3/e3, e3/e4, and e4/e4) in this study. For rapid PCR-RFLP, we used a rapid amplification reagent (EmeraldAmp GT PCR Master Mix; Takara), directly adding HhaI restriction enzyme (NEB) into a PCR product, a rapid electrophoresis buffer (1mM lithium borate buffer), high voltage electrophoresis equipment with 300V (Solgent) and 3.5% conventional agarose gel (SeaKem LE Agarose; Lonza). To confirm the results of rapid singletube PCR-RFLP with agarose mini-gel electrophoresis for ApoE genotyping, we also performed PCR-direct sequencing. Results: The concordance rate between rapid single-tube PCR-RFLP and PCR-direct sequencing was 100%. The duration of total experiment, PCR, digestion, electrophoresis, and interpretation were 70 minutes, 32 minutes, 5 minutes, 30 minutes, and 3 minutes, respectively. Conclusions: ApoE genotyping using rapid single-tube PCR-RFLP with conventional agarose mini-gel electrophoresis is reliable and can be used as a high-throughput technique in clinical molecular diagnostics.
P. Scott, J.E. Stanchfield SciGene, Sunnyvale, CA. Introduction: All methods specified by FISH probe suppliers require samples be codenatured with probe at a precise temperature for a specific time to achieve optimal results. Despite this, there has been no accurate and reliable method to certify that FISH slide heating equipment actually achieve their critical temperatures and times. We describe here a simple, accurate and affordable method for measuring and logging temperature data from each slide heating position on any FISH hybridizer. Using this method, we collected time and temperature profiles on over 70 instruments in 22 cytogenetics laboratories in North America. Methods: The method used in this study is based on iButton Thermochron Sensors (www.ibuttonlink.com; $89 each) a dime size, wireless sensor with integrated datalogger. Using the manufacture-provided computer adaptor and software, a data collection time interval is selected for each iButton sensor. One or more sensors are placed in slide positions on a FISH hybridizer, such as a ThermoBrite or Hybrite instrument (Abbott Molecular, Dako/Agilent, Leica, etc) and a standardized protocol of 75°C for 2 minutes, then 37°C incubation is initiated. Once the instrument cooled to 37°C, each sensor is removed and inserted again into the adaptor to download time and temperature data into a spreadsheet for analysis. Results: We have used the iButton system to measure slide position temperatures on 74 Thermobrite/HYBrite systems in 22 cytogenetics laboratories. Slide position denaturation temperatures were found to be consistently lower than the set temperature, with deviations between instruments in a single laboratory of 2.0°C to 16.4°C below programmed temperatures. Furthermore, we uncovered that FISH samples are held within 1°C of the peak denaturation temperature for only 50% of the programmed denaturation time. Conclusions: In the absence of routine calibration, slide temperatures and times achieved on FISH slide hybridizers may be much lower and deviate dramatically from those recommended by probe suppliers; contributing unnecessary variability to the quality and reliability of FISH diagnostic results. TT20. Assessment of Fetal Chromosomal Abnormalities Using Cell-Free DNA from Amniotic Fluid R. Duttagupta 1 , S. Venkatapathy 1 , K. Suyenega 1 , J. Cuevas 1 , A. Alka Chaubey 2 , B. DuPont 2 , E. Fung 2 1 Affymetrix, Santa Clara, CA; 2 Greenwood Genetic Center, Greenwood, SC. Introduction: Prenatal testing and assessment for determining risk of fetal chromosomal abnormalities can be undertaken using both invasive and non-invasive methods. The spectrum of non-invasive methods have limitations in diagnostics due to either lower detection rates or indirect sampling of fetal genome through Noninvasive Prenatal Screening (NIPS) of cell-free fetal DNA (cffDNA) from maternal blood. Hence, invasive methods such as amniocentesis and chorionic villus sampling (CVS) continue to be the reference gold standard. Due to the culture limitations required for amniocentesis, there is an increased interest in evaluating the acellular fetal fraction for determining fetal aneuploidy.This left-over material is typically discarded in the clinical setting. The ability to utilize and assess cell-free DNA from amniotic fluid (AF) for copy number assessment presents an alternate molecular strategy for performing confirmatory tests. Methods: A total of 10 AF samples with abnormal NIPS or ultrasound findings were collected from patients ranging between 15 to 29 weeks of gestation. The samples were cultured and cellular DNA extracted using the QIAamp mini kit. Approximately 4ml to 8ml of the de-identified and left-over AF was frozen at -80°C for examination of cff DNA. For purification of cffDNA, samples were centrifuged at 1500 rpm for 10 min at 4°C to pellet residual cells and cffDNA isolated using a commercially available DNA purification kits (Norgen Bioteck Corp). Isolated DNA was analyzed using the OncoScan FFPE kit and compared to karyotyping. Results: The average DNA yield for cff DNA was 9.5+4 ng per mL of AF, with yields typically correlating to gestational age. Karyotyping and OncoScan had 100% concordance for aneuploidy detection, with all 3 aneuploid cases called by both platforms and the remaining 7 karyotypically normal samples also confirmed by OncoScan. Furthermore, with the exception of cases 3 and 4, complete concordance in normal and abnormal aneuploid results were detected between cffDNA and genomic DNA fractions. Results of cffDNA analysis from cases 3 and 4 were inconclusive due to high levels of maternal cell contamination (MCC). Interestingly, NIPS results from both of these cases describe evidence of chromosomal aberrations (Monosomy X for Case 3 and Trisomy 13 for Case 4) which could not be confirmed by either microarray analysis or karyotyping. Conclusions: Taken together, these results indicate that cffDNA isolated from amniotic fluid could be an alternate source of material for the detection of fetal aneuploidies. Also, the turnaround-time for confirmation of NIPS could be greatly reduced with the ability to run this assay on cffDNA rather than waiting for karyotype results. Introduction: Inherited thrombophilia predisposes an individual to thrombotic events such as venous thrombosis, the third most common cardiovascular disease. Genetic analysis has demonstrated that the Factor V Leiden mutation, which has a relatively high prevalence in the general population, accounts for 85% to 95% of APC resistance cases. Patients testing positive for the Factor V Leiden mutation should be considered for molecular genetic testing for the coagulation Factor II (prothrombin) G20210A variant. The Factor II variant is present in 1% to 2% of the general population, and its involvement in venous thromboembolism is well established. Here we report on the validation of assays for both Factor V and Factor II mutations to run concurrently on the same PCR plate at Baystate Health. Methods: A total of 60 de-identified DNA samples previously tested with the FV LightCycler Kit and Factor II (Prothrombin) G20210A LightCycler Kit, on different runs with different cycling parameters on the Roche LightCycler 2.0 were used to assess the accuracy of the same Roche kits run on the Roche cobas z480 instrument on the same PCR plate. Thirty samples analyzed for the Factor II variant consistedof 10 wild type, 14 heterozygous and 6 homozygous mutant. Thirty samples analyzed for the Factor V variant consisted of 10 wild type, 16 heterozygous and 4 homozygous mutant. Polymerase chain reaction was performed utilizing the Roche cobas z480 with the following parameters: 10 minutes at 95°C, 95°C for 1 minute, 40°C for 3 minutes, 72°C for 15 seconds for a total of 45 cycles. The genotypes were identified by melting curve analysis: 95°C for 1 minute, 40°C for 3 minutes and 80°C with 3 acquisitions/°C. Results: Sixty of sixty DNA samples (100%) showed concordance with the expected result. Precision studies showed reproducibility of inter-assay and inter-operator results to be 100% concordant (60 out of 60 replicates). The Factor II wild-type allele exhibited a Tm of approximately 59°C whereas the mutant allele exhibited a Tm of 49°C. The Factor V wild type allele exhibited a Tm of approximately 65°C whereas the mutant allele exhibited a Tm of 57°C. The Tm values for each of the alleles were consistently within +/-2.5ºC. Conclusions: The Roche FV LightCycler Kit and Factor II (Prothrombin) G20210A LightCycler Kit run on theRoche cobas z480 instrument on the same amplification plate provides a fast, easy, and accurate test.
S. Chun 1,3 , T. Kim 2 , Y. Kim 1 , W. Park 2 , S. Jang 1,3 1 Asan Medical Center, Seoul, Korea; 2 Asan Institute for Life Sciences, Seoul, Korea; 3 College of Medicine, Ulsan University, Seoul, Korea. Introduction: Identification of specific somatic mutation is essential step for the personalized cancer medicine. Various molecular methods including Sanger sequencing, real-time PCR, microarray, pyrosequencing and next-generation sequencing have been used for detection of mutations in specific genes or gene panels. However, there are still clinical unmet needs for mutation detection with high levels of sensitivity because of frequent low mutant allele fraction in clinical samples such as plasma and tissues with tumor heterogeneity and low tumor cellularity. Methods: In this study, novel method named as ultra-high sensitive (UHS) was developed for enrichment of mutant allele specific amplicon, thereby increase the detection sensitivity. Amplification of target region and enrichment of mutant allele
The Journal of Molecular Diagnostics ■ jmd.amjpathol.org could be performed simultaneously in single PCR reaction by using three primes, pair of universal sense and anti-sense primer and mutation specific sense primer. Initial amplification of target region including mutation site was started using universal sense and antisense primers with gDNA as template, and followed by exponential enrichment of mutant allele specific amplicon with pair of mutation specific primer and universal antisense primer using previous amplified mutant allele amplicon as template. Enriched product was analyzed using MassArray system (from Sequenom) composed of iPLEX chemistry and MALDI-ToF analysis. Results: Detection sensitivity of this method was evaluated as 0.1%~0.5% using serial diluted gDNA isolated from H1975 cell line has L858R mutation of EGFR gene. This method was applied to detect mutation of EGFR_L858R, KRAS_G12D, KRAS_G12V and BRAF_V600E with 45 cell free circulating DNAs extracted from lung cancer clinical samples, which are known as EGFR mutation positive in resected tumor tissues, L858R (18 cases) and exon 19 indel (27 cases), respectively. L858R mutation was detected in 6 (33.3%) out of 18 samples known as positive, thereby sensitivity of mutation detection was much higher compared to cast-PCR (3, 16.6%) and conventional iPLEX chemistry (1, 5.5%). Conclusions: This unique OncoUHS method could potentially become a sensitive, specific, simple, rapid and safe approach for detection of somatic mutation.
K.R. Carney, C.T. Wittwer University of Utah, Salt Lake City, UT. Introduction: DNA melting is an inexpensive and quick method of "finger printing" pieces of DNA for laboratory and medical use. Conventional industry DNA melting instruments create the melted states of a uniform DNA as a function of time. We present a high resolution method of spatial or steady-state melting, which creates all of the melted states of a DNA as a function of position. This new technique shows promise for increased resolution of melted DNA. Methods: A new prototype instrument was constructed to melt DNA on a spatial scale as opposed to a temporal scale. The instrument consists of a narrow 17 cm long copper bar that is heated at one end creating a temperature gradient along its length. A 15cm long glass capillary (OD: 1.0mm, ID: 0.5mm), filled with a DNA solution and LCGreen dye, is sealed and placed on the copper bar. The DNA in the capillary quickly feels the temperature gradient of the copper bar and soon forms all its melted states along the length of the capillary. A 450nm excitation laser and imaging optics, mounted on a single axis ball screw drive, are used to create multiple precise scans of the capillary's fluorescence profile. During data collection the temperature profile of the bar is slowly increased so that each scan is slightly different from the one before. This slow heating allows the use of a simple mathematical algorithm to eliminate systematic background noise and create multiple standard derivative plots which are averaged together to produce a single high resolution derivative plot. Results: Using the HR-1 as a comparison instrument, our steady-state melter was able to detect an equivalent or increased number of DNA melting transitions for several test samples. Conclusions: High resolution steady-state DNA melting is a new technology that is in the process of being explored and optimized. Given our results with our phase 1 instrument, we are encouraged to keep developing this technology to set the new standard in high resolution DNA melting.
A. Ieamniramit 1 , S. Li 1 , J. Jin 2 , X. Wu 2 , H. Wu 2 , D. Ma 1 , W. Li 1 1 National Medical Services Labs (NMS Labs), Willow Grove, PA; 2 City of Hope National Medical Center and Beckman Research Institute, Duarte, CA. Introduction The 4-microRNA Signature for Clear Cell Renal Cell Carcinoma (ccRCC) Metastasis and Prognosis was previously identified by microarray and qPCR assay at a research laboratory of City of Hope (COH). Using an independent 265-case test cohort of primary ccRCC formalin-fixed, paraffin-embedded (FFPE) tissue, the COH researchers have shown the qPCR-based signature with overall sensitivity and specificity greater than 70% and highly correlated with patients' cancer specific survival rates. In collaboration with COH, NMS Labs has conducted studies on the analytical and clinical performance of the qPCR assays to further develop and validate the assay in a CLIA setting. Methods: Normal Human Kidney RNA and synthetic miRNAs were used to define assay specifications including the reportable range, precision, accuracy, analytical sensitivity and specificity in the analytical validation. RNAs used in the clinical validation was isolated from FFPE tissue blocks followed by cDNA synthesis and qPCR using ABI 7900. Results: 1) Analytical Validation: The PCR efficiencies of four miRNAs were shown from 86% to 92% with the standard deviations of 1% to 2%. The reportable range is 0.3 ng to 300 ng of human kidney RNA with R2 value at 0.99 to 1.0 for all miRNAs including the internal control miRNA. The Low Limit of Quantitation (LLOQ) for all five miRNAs were evaluated and established whereas the analytical specificity was shown 100%. Three assay calibrators were created for the purpose of quality assurance and quality control, 1) Low Risk High Concentration Control (LRHC) for Low Risk Cut-off Control and High Concentration Control of five analyte miRNAs, 2) High Risk Medium Concentration Control (HRMC) for High Risk Cut-off Control and Medium Concentration, and 3) Low Concentration Control (LCC) for Low Concentration. Thus the assay acceptance criteria were established based on the statistical analysis of the five analyte miRNAs and the calculated risk scores generated from 54 replicates in multiple runs over the period of two weeks. The results also suggested the acceptable assay precision, with CV<3.3% for all five analyte miRNAs. 2) Clinical Validation: The preliminary data obtained from 26 training cohort FFPE tissues has shown the assay accuracy as 100% for High Risk Samples (14/14) and 91.7% for non-High Risk Samples (11/12) in comparison to COH. We are currently expanding the validation study to include over 200 patient samples for the clinical sensitivity and specificity. Conclusions: We have created the quality control calibrators for the 4-miRNA Signature qPCR assay and successfully completed the analytical validation and the preliminarily clinical validation at NMS Labs. R. Kanchi Ravi, M. Panjikaran, T.K. Wang, I.K. Mitchell, Y.K. Xia, R.K. Tseng, A.K. Papoutsis, S.K. Gadient, J.K. Kines, S.K. Nahas, C.K. Scott, G. Sims, J. Falk, B. Dabbas, A. Graber Genoptix, a Novartis Company, Carlsbad, CA. Introduction : DNA mutations at the gene level have been used for patient screening to assess the level of risk for developing cancer and to identify drugs that provide the most benefit to a patient. However, methods such as Sanger sequencing focus on small groups of genes and are unable to screen for numerous genes from multiple patients simultaneously whereas amplicon-based next-generation sequencing (NGS) assays frequently only cover hotspot mutations and do not provide a range of information on genomic alteration types such as copy number variants (CNVs) and translocations. The goal of the current study was to validate a 173-gene panel for accuracy, sensitivity, specificity to detect single nucleotide variant (SNV), indels, CNV, and translocations in different clinical tissue types using NGS techniques for diagnosis, prognostic staging, and treatment decisions. Methods: A total of 83 samples including reference cell lines, formalin-fixed, paraffin-embedded (FFPE) tissue, peripheral blood, and bone marrow were used to assess the accuracy of the target capture-based NGS assay. The 173-gene panel was validated with a target capture-based parallel NGS method developed at Genoptix, which used DNA/RNA isolated from characterized reference cell lines, FFPE tissue, peripheral blood and bone marrow clinical samples and variant data analysis through custom built pipeline tools. Variants detected by the multi-gene NGS assay were confirmed using the Ion AmpliSeq CHPV2 50 gene panel for SNV, Nanostring nCounter method for CNV, and Oncomine Fusion Gene Panel from Life Technologies for the structural variants/translocations. Results: Eight hundred twenty (820) variant calls were compared across the target capture-based NGS assay and Ion AmpliSeq Panels resulting in a 99.76% concordance. All indels detected by the NGS assay were confirmed using the AmpliSeq Panels. The validation results of CNV using NGS assay showed 99.7% concordance with comparison to Nanostring CNV method or published data references. All the translocations detected by the NGS assay were confirmed by Oncomine Fusion Gene Panel, RT-qPCR, and data references. The lower limit of detection of the multi-gene NGS assay was observed to be 5% for SNVs and translocations, and 10% for indels. Conclusions: Overall, this study provides thorough validation of the 173-gene NGS panel and the analysis tools that can be used in a clinical laboratory for routine testing with the ability to sequence multiple patient samples including those with only small amounts of FFPE DNA. We also observed the assay to be specific and sensitive for SNV mutation analysis, indels, CNV, and translocations. Keywords: next-generation sequencing, cancer, FFPE, translocation, target capture.
K. Culp, J. Houghton, B. Caughron, E. Bram, G.J. Latham, A. Hadd Asuragen, Inc., Austin, TX. Introduction: Accurate library quantification is a critical component in nextgeneration sequencing (NGS) workflows to balance sample-to-sample read coverage and optimize available sequencing bandwidth. Quantitative PCR (qPCR) is the method of choice for functional library assessments since it directly measures on-target amplifiable DNA, requires very low sample input, and provides high specificity and sensitivity. Current qPCR assays necessitate a standard curve, which increases both the costs and the hands-on time for the assay. Here, we describe the development and performance of a streamlined, calibration-curve-free, competitive qPCR (cqPCR) assay for the quantification and normalization of targeted NGS libraries. Methods: An exogenous DNA standard was designed for use as an internal single-point reference for the quantification of targeted NGS libraries using qPCR. Differences in cycle threshold between the exogenous standard and the input library were used to determine the concentration for each sample library and subsequently guide the normalization of sample pooling. cqPCR assay performance was compared to a direct quantification method (Quant-iT DNA Assay Kit (high sensitivity), Life Technologies) and to a common standard curve-based qPCR assay that relies on absolute quantification (KAPA Library Quantification Kit, KAPA Biosystems) with NGS outputs as comparison metrics. Source gDNA was derived from FFPE, fresh frozen, and whole blood residual clinical samples as well as cellline extracts. NGS studies were performed using the Quantidex Pan Cancer Kit (Asuragen) and the MiSeq (Illumina). Results: The cqPCR library quantification method eliminated the need for a calibration curve and reduced the number of pipetting steps up to 79% compared to absolute quantification, thereby decreasing hands-on time and simplifying the workflow. Using a set of 28 samples, cqPCR measurements were within 2-fold of those using a common calibration-curve-based qPCR method and within 3-fold of the concentration measured using a fluorescent intercalating dye assay. Results across all three assays were highly correlated with one another (R2 > 0.93). Repeatability studies demonstrated assay precision with less than 12% CV. Incorporation of the cqPCR assay into the NGS workflow guided optimal flow-cell seeding on the MiSeq and allowed uniform sequencing coverage. Conclusions: Results from this study demonstrate the utility of a cqPCR approach to accurately quantify and pool targeted NGS libraries. This streamlined approach offers quantitative accuracy and consistency, simplifies library analysis, and provides a new tool for integration into comprehensive workflows for clinical NGS.
Quality and Impacts Variant Calling by Targeted NGS R. Blidner 1 , J. Plyler 1 , J. Kemppainen 1 , B.C. Haynes 1 , R. Zeigler 1 , A. Hiddessen 2 , K. Rose 2 , G.J. Latham 1 1 Asuragen, Inc., Austin, TX; 2 Purigen Biosystems, Pleasanton, CA. Introduction: Advances in precision medicine in oncology mirror innovations in molecular profiling techniques and the utility of new biomarkers. A common denominator for workflows that enable these advances is the up-front sample preparation of tumor biopsies. The choice of DNA isolation technology, particularly for low-quality or low-quantity FFPE samples, can impact downstream assay results. To quantify this impact, we compared and contrasted multiple isolation methods, including a novel technology in isotachophoresis (ITP), for pre-analytical functional DNA yield and quality and analytical results following targeted NGS. Methods: Three commercial FFPE isolation kits from Qiagen, Norgen, and Thermo Fisher were used to extract DNA from 15 low-quality melanoma FFPE specimens. In addition, two commercial kits (Qiagen and Norgen) and a custom ITP method (Purigen) were employed for DNA isolation from a mixed tissue FFPE cohort. DNA QC was performed using spectrophotometry (NanoDrop), fluorimetric assay (Qubit dsDNA HS Assay), and qPCR (Quantidex DNA Assay). Targeted NGS of each sample was performed using variable amplifiable copy number inputs into the Quantidex Pan Cancer kit (Asuragen) with sequencing on the MiSeq (Illumina). Results: Recovered FFPE DNA was evaluated by yield, functional quality, qPCR inhibition, and variant calls from targeted NGS. A set of 15 melanoma samples revealed distinct DNA yields and quality from the same tissue input; the fraction of amplifiable DNA varied by >4-fold across three commercial isolation kits. Inhibition in qPCR was observed in 3 of 15 samples and was dependent on the choice of extraction method, indicating differences in the removal of interfering substances such as melanin. An increase in NGS variant calls was also observed between different isolation methods, particularly with reduced template inputs. Evaluation of a separate FFPE cohort confirmed these findings, and extended them by including an innovative method for isolating DNA using ITP. Compared to conventional extraction chemistries, ITP produced 2-to 3-fold higher yields of both total DNA and amplifiable DNA. All three methods enabled accurate NGS variant calls at relatively high DNA inputs, but one commercial kit manifested a categorically elevated "noise" profile when challenged with low-input DNA. Conclusions: The results underscore differences in FFPE DNA yield, and functional quality and purity, using commercial and emerging sample preparation technologies. Moreover, these differences can impact results using downstream molecular analyses such as qPCR and NGS. Our findings highlight the need to improve and standardize DNA extraction methods from challenging clinical specimens, and illustrate the potential of a new approach using ITP.
TT28. Choosing an Accurate, Economical Platform for Typing Self-identifying Antigens Encoded by HLA Genes J. Sims 1 , J. Kaur 2 , E. Earley 1 , K. Robasky 1 , K. Robasky 1 , E. Lai 2 1 EA|Quintiles, Durham, NC; 2 Takeda Pharmaceuticals International, Inc, Deerfield, IL. Introduction: The human leukocyte antigen (HLA) system is the genomic region that regulates the human immune system by encoding proteins for the cell wall that are responsible for self-recognition. This genomic region plays a role in immunological disease, oncological immunotherapy, and drug toxicity, among other things. Accurate, cost-effective typing of this region enables biomarker identification and prediction of adverse drug reactions (ADR), as well as many other clinical applications. However, HLA genes are the most polymorphic region in the genome making characterization elusive by traditional practices and can cost over $1000 per sample. More sensitive and specific assays being developed using NGS have nearly halved that cost. Meanwhile, microarray-based techniques cost even less. Here, we quantify the accuracy trade-offs for these variable cost technologies. Methods: Classic HLA Class I and Class II molecular typing were performed using a standard method (Luminex kit). HLA-typing were additionally conducted via microarray using generic computational phasing tools and NGS using platform-specific tools. Principal component analysis was performed on the genotypes to ascertain population stratification for each sample. DNA samples were purchased from rheumatoid arthritis patients from BioServe to be used for the comparison of these methodologies. Classic HLA genotype calls were analyzed for concordance to the Luminex platform and cost comparisons were performed. Results: HLA-typing performed with NGS technologies achieved 97% to 100% concordance with Luminex for all Class I alleles and for HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQB1 and HLA-DRB3 Class II alleles. Due to limitations in the microarray HLA-typing platform, only same-population samples and 8 of the classical HLA genes were comparable (HLA-A, HLA-B, HLA-C, HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQB1, HLA-DRB1). The HLA-C gene was found to have the highest concordance on every platform (up to 92% on microarray platforms), and the HLA-DRB1 gene had the lowest concordance on any platform. Recall was within 1% on all platforms, but the microarrays had lower concordance than NGS calls (by 8-44%) for every gene with the exception of HLA-DRB1, which ranged from 59% to 60% concordance across all platforms. Cost analyses show NGS is largely less expensive than traditional methods. Conclusions: Newer NGS technologies can be applied at nearly half the cost of traditional methods for HLA-typing with comparable accuracy. Microarrays can again halve the cost of NGS-based HLA-typing assays, at the cost of accuracy, and may be applicable to large legacy datasets or in circumstances where sample size and cost are sufficient to achieve the statistical power necessary for associations.
TT29. Evaluation of a New, Accurate, Automated, and Cost-Effective Genotyping Approach for the Analysis of Pharmacologically Important Single Nucleotide Polymorphisms W.R. Higgins 1 , C. Keppel 2 , L. Linz 2 , A. Rodriguez 2 1 Douglas Scientific, Alexandria, MN; 2 Douglas Scientific, LLC, Alexandria, MN. Introduction: The use of genetic information by medical practitioners to guide treatment and drug dosage decisions has increased in prevalence over the last decade. Recent pushes towards personalized medicine have resulted in the need for accurate, automated, and low-cost methods for performing relevant pharmacogenomic SNP analysis in clinical and research settings. Three commonly tested SNP alleles, CYP2C9*2, CYP2C9*3, and VKORC1, have been widely demonstrated to impact the metabolism of warfarin, a commonly prescribed anticoagulant. Using probe-based SNP genotyping assays, this study analyzes human samples for these SNPs using the IntelliQube, a new integrated liquid handling and qPCR instrument. Through comparison with data previously published by Pratt, et al and results generated with standard qPCR instrumentation, we assess the accuracy and reproducibility of this platform for pharmacogenomic SNP analysis. Methods: Purified gDNA samples from 34 cell lines were obtained from the NIGMS Human Genetic Cell Repository at the Coriell Institute for Medical Research. Custom BHQplus probe-based genotyping assays were purchased from Biosearch Technologies and Genotyping ToughMix PCR mastermix was purchased from Quanta BioSciences. These materials were prepared and used according to manufacturer instructions. The IntelliQube instrument from Douglas Scientific was used for the automated assembly of 1.6μL reactions in Array Tape consisting of 800nL of gDNA and 800nL of primers, probes, and Genotyping ToughMix. Following setup, qPCR was performed on the IntelliQube. For comparison, 5μL reactions were manually assembled in a standard 384-well plate and a Life TechnologiesTM ViiATM 7 instrument was subsequently used for qPCR. Results from the plate-based method were then compared to those generated by IntelliQube. Results: All sample replicates tested were successfully genotyped and the genotype calls for 1.6μL reactions on the IntelliQube showed consistency with those generated in 5μL reactions on the ViiA7. Results from both methods were 100% concordant with those previously published by Pratt, et al. and the Coriell consensus genotypes. Conclusions: The IntelliQube, when used in conjunction with BHQplus assays, provides an accurate and streamlined qPCR-based method for genotype analysis of human DNA. All sample replicates tested in this study were successfully genotyped and concordant with published allele calls, demonstrating that the IntelliQube produces accurate and consistent genotype results. Combined with the automated workflow and economic benefits of miniaturized reactions in Array Tape, the IntelliQube may prove to be a useful and powerful platform for high-throughput pharmacogenomic testing. Introduction: Next-generation sequencing (NGS) technologies have rapidly made their way into the clinical laboratory setting. Performance characteristics that span pre-analytical, analytical and post-analytical variables are important factors that must be taken into account when developing a custom assay for clinical use. Here, we evaluated the impact of implementing the KAPA BioSystems HyperPlus Library Preparation Kit (KHPLPK) into our current workflow for a laboratory-developed NGS assay, the JAX Cancer Treatment Profile (JAX-CTP), with the goal of lowering DNA input concentration, turn-around time and cost while increasing data quality. Methods: JAX-CTP is designed to sequence coding exons of 358 cancer related genes. Using 1 HAPMAP and 23 FFPE tumor samples, which were also processed with Agilent SureSelect XT (ASXT) in the current version of the assay, we evaluated library preparation with the KHPLPK. All samples were hybridized and sequenced on the Illumina HiSeq 2500 or MiSeq. Data analysis and concordance were performed using an internal Clinical Genome Analytical pipeline (JAX-CGA). Results: Using the enzymatic fragmentation of the KHPLPK, we were able to reduce our DNA input concentration from 1ug to 200ng. A one-tube preparation method allows for better Hybridization Buffer it ranged from 17 to 20 hours for cytological and FFPE specimens, respectively. Conclusions: The new fast working Vysis Hybridization Buffer demonstrated the ability to complete FISH hybridizations within a single work shift (on average less than 6 hours) by using a 2 hour hybridization for all specimen types, reducing overall assay time to 3 to 6 hours from the standard 17 to 20 hours. Further, the fast working Vysis Hybridization Buffer 2 hour hybridizations demonstrated equivalent performance to Vysis LSI/WCP Hybridization Buffer overnight hybridizations.
A. Lal 1 , C.A. Kunder 1 , M.D. Ewalt 1 , I. Schrijver 1 , J. Zehnder 1 , C.J. Suarez 2 1 Stanford University, Stanford, CA; 2 Stanford University, Palo Alto, CA. Introduction: The TruSeqAmplicon-Cancer Panel (TSACP) is a target enrichment assay manufactured by Illumina to analyze hotspots in 48 cancer-related genes. It uses sequence-specific oligonucleotides to generate 212 amplicons in a highly multiplexed reaction. Although the assay is generally robust, here we illustrate potential pitfalls we have observed with this approach, which are under-recognized and under-reported. Methods: DNA from peripheral blood (PB), bone marrow or fresh frozen paraffin-embedded neoplastic specimens (N=152) was processed using the TSACP. Sequencing was performed on the IlluminaMiSeq (2x150). Alignment and variant calling was accomplished using NextGENev2.3.4 (SoftGenetics). Selected variant calls were reviewed for accuracy on NextGENe's viewer. Results: Case 1: A lung adenocarcinoma biopsy specimen was sequenced and KRAS c.34G>A was detected. Two months later, the tumor was resected, and this specimen was sequenced as well, using an in-house developed hybrid capturebased next-generation sequencing (NGS) assay. In addition to the KRAS variant originally detected by TSACP, STK11 c.644_645delGC was also identified. Sanger sequencing of the STK11 region confirmed the deletion in the biopsy and the resection specimen. Further analysis of the biopsy's read alignments showed that the deletion was located within a sequence targeted by one of the oligos, which resulted in allele drop-out. Case 2: A rectal adenocarcinoma biopsy specimen was sequenced and KIT c.2439A>T was identified. Careful analysis of the read alignments demonstrated two sets of forward reads. A predominant set accounted for the expected 183-bp amplicon, and a minor set containing the variant and derived from a 170-bp amplicon. Further evaluation of the minor set showed that its 5'sequence, which included the KIT variant, completely matched one of the MET oligos. Apparent mispriming of the MET oligo to the KIT sequence generated the 170-bp amplicon and artifactually introduced the KIT variant. Case 3: Sequencing of an acute myeloid leukemia PB specimen revealed NOTCH1 c.4817T>C. Careful review of the read alignments demonstrated that the variant was present only in the forward reads and was located within the 26-nucleotide stretch targeted by one of the oligos. Therefore, the variant was considered an artifact likely introduced by the oligo through a yet undefined mechanism. Conclusions: Just as with other NGS assays, awareness of TSACP's limitations along with cautious interpretation of assay results can minimize false variant calls. In addition, trimming of the reads to remove the oligo sequences or modification of the BED files to preclude calls on those sequences would prevent the introduction of spurious variants at those locations. G. Durin, B. Packard, E. Rudd, A. Purdy, J. Han, H. Khoja, J. Laugharn Covaris Inc, Woburn, MA. Introduction: Formalin-fixed, paraffin-embedded (FFPE) tissues are a valuable source of DNA and RNA for targeted and whole genome sequencing or transcriptome analysis. However, both the formaldehyde fixation and the tissue dehydration required for paraffin embedding complicate reproducible extraction of DNA and RNA in sufficient quantity and sufficient quality for next-generation sequencing (NGS) analysis. In this poster, we present the truXTRAC Reagent Kit, a robust and efficient method for the extraction of NGS-grade DNA and RNA from FFPE tissue based on Covaris Adaptive Focused Acoustics (AFA) technology. The process combines AFA-based extraction from FFPE tissues with a bead-based purification for an automated nucleic-acid preparation pipeline. The combined workflow enables extraction of DNA or RNA from up to 96 FFPE samples in 4 hours in a 96-well plate format, with only 5 minutes of hands-on-time. Methods: truXTRAC utilizes a highly controlled focused acoustic energy field for the effective removal of paraffin from FFPE tissues. Importantly, the AFA process also promotes concomitant rapid tissue rehydration, which facilitates subsequent tissue digestion and crosslink reversal. The entire process is fully automated and carried out without the use of any organic solvents or heat. After extraction, DNA is purified using an automated magnetic bead-based system. DNA and RNA were extracted from four different FFPE tissues with the truXTRAC extraction and purification pipeline. In parallel, DNA was also extracted from the same FFPE tissues with a traditional, organic solvent and heat based method and from matched fresh-frozen tissues using standard commercial kit. DNA quality was assessed by qPCR assays as well as wholegenome sequencing using Illumina NGS technology. RNA quality was assessed by qRT-PCR with a 67bp amplicon. Results: All truXTRAC-derived DNAs passed the delta Cq score threshold for the Illumina FFPE QC kit (designed for the TruSeq Amplicon -Cancer Panel), whereas half of the samples extracted with a traditional extraction method failed the same QC threshold. Whole-genome sequencing showed similar coverage depth and evenness for truXTRAC-derived DNA and control DNA extracted from fresh-frozen tissue, whereas coverage depth and evenness were lower and less consistent with DNA extracted from FFPE tissues with a traditional extraction method. For the tissues tested, up to several fold more amplifiable RNA was recovered using truXTRAC. Conclusions: Covaris truXTRAC FFPE yields high-quality, NGS-grade DNA and RNA from FFPE tissues. The automated process requires minimal hands-on-time and supports processing of 8 to 96 FFPE samples in parallel.
D. Grölz 1 , T. Voss 1 , A. Ullius 1 , N. Dettman 1 , T. Buirkle 2 , E. Provencher 2 , S. Grömminger 3 , W. Hofmann 3 , R. Wyrich 1 1 Qiagen GmbH, Hilden, Germany; 2 Becton, Dickinson and Company, Franklin Lakes, NJ; 3 LifeCodexx, Konstanz, Germany. Introduction: Non-invasive prenatal testing (NIPT) and circulating tumor DNA (ctDNA) testing based on circulating cell-free DNA (ccfDNA) are rapidly emerging fields. Sensitivity of ccfDNA testing is, however, compromised by the release of genomic DNA (gDNA) from lymphocytes due to mechanical lysis or apotosis during blood collection, storage and transport. PreAnalytiX has developed the PAXgene Blood ccfDNA System, consisting of the PAXgene Blood ccfDNA Tube, a plastic BD Vacutainer tube with non-crosslinking chemistry for stabilization of blood cells, and the QIAsymphony PAXgene Blood ccfDNA Kit for automated ccfDNA extraction from up to 5 ml of plasma. Here we present results of ccfDNA testing of the new system compared to EDTA whole blood and the Streck Cell-Free DNA BCT. Methods: Study 1: Duplicate blood samples were collected from apparently healthy subjects into PAXgene and EDTA tubes, stored for 7 days at 25°C or one day at 35°C. Transport was simulated by continual inversion of blood filled tubes for 5 hours during storage. Plasma was processed by double centrifugation and ccfDNA was extracted on the QIAsymphony with the dedicated kit and protocol. ccfDNA was quanitfied and size distribution was determined by copy number quantification of 66, 180 and 500 bp fragments of the 18s rDNA gene. In addition, gDNA was purified manually from PAXgene and EDTA tube blood with the QIAamp Mini Kit and analyzed by agarose gel electrophoresis. Study 2: Paired blood samples were collected from 20 pregnant women into PAXgene and Streck tubes. ccfDNA was extracted manually with the QIAamp cNA Kit. Fetal ccfDNA was quantified by qPCR using the QuantYfeX assay and analyzed for trisomies by NGS (PrenaTest, LifeCodexx). Results: Apoptotic bands appeared in agarose gels of gDNA extracted from blood collected and stored in EDTA tubes, and copy numbers of ccf 18s rDNA gene fragments increased by up to 100-fold after 7 days of storage. No apototic bands were visible in genomic DNA purified from blood collected and stored in PAXgene tubes for up to 7 days, and plasma 18s rDNA fragment copy numbers did not significantly increase during storage and were comparable to those from EDTA plasma prepared immediately after blood draw. ccfDNA fetal fractions from pregnant women were comparable between PAXgene and Streck tubes (5% to 20%). All DNA samples met quality standards and produced identical results with the PrenaTest.
The PAXgene Blood ccfDNA Tube stabilizes blood cells during transport and storage, and when used with the QIAsymphony PAXgene Blood ccfDNA Kit, provides for accurate detection and quantitation of ccfDNA. For research use only. Not for use in diagnostic procedures.
E. Golomb 1 , R. Feingersh 2 , I. Bejarano-Achache 1 , A. Lagziel 2 , S. Carmi 3 , N. Moskovits 4 , S. Stemmer 4 , I. Haviv 3 1 Shaare Zedek Medical Center, Jerusalem, Israel; 2 Genesort, Zafed, Israel; 3 Faculty of Medicine in the Galilee, Bar Ilan University, Zafed, Israel; 4 Rabin Medical Center, Petah Tikva, Israel. Introduction: Somatic mutations emerge as superior biomarkers for rationalized drug selection in combating cancer. To trace the full cancer heterogeneity and detect mutations in cancer cells within DNA preparation that includes neighboring normal stromal cells, multiple target enrichment tools, developed to allow sequencing of the most relevant area of the genome, at the deepest possible, exhibit limited sensitivity, typically requiring 50ng gDNA. Methods: We assessed the sensitivity and specificity over a common genomic area, encompassing all frequently mutated exons of over 150 cancer causing genes, using a unique combination of target enrichment tools on Illumina HiSeq2500. This combination improves the sensitivity to less than 1 ng gDNA, from selected defined tumor cells dissected from H&E stained slides. Results: The technology we describe exhibit comparable sensitivity, reproducibility and specificity, without failing cases, even in small biopsies, with limited material. Furthermore, the use of Agilent Sureselect has the advantage of detecting gene fusions and CNVs. Conclusions: These results suggest that clinically driven tumor sequencing should read the samples at relatively high depth, even when the source material is very limited. cytotrophoblast, sheared DNA, blended with DNA from immortalized B cells, accurately replicates a cfDNA isolated from a pregnant individual when interrogated by MPS. Further, incorporation of the small fragment DNA at low concentration into a nucleosome mimetic not only confers an enhanced stability of the analyte but also facilitates its utility as a whole process control and as reference material for validation of various test methods.
C.P. Van Loy, D. Topacio, M. Carvallo, K. Rhodes, M. Andersen Thermo Fisher Scientific, Carlsbad , CA Introduction: Standard library preparation methods for next-generation sequencing research applications require purified DNA as input. However, purification methods are time consuming, costly, and result in variable sample quality, which lengthen the time required to progress from sample to answer. In addition, much of the DNA is lost during the purification process, meaning that extremely small samples cannot be used. Methods: Herein, we describe the use of biological samples as direct input in a highly multiplexed target amplification to generate libraries for subsequent analysis with the Ion Torrent NGS platform. We have developed a Direct Reagent for simple processing of retrospective (or archived) samples of formalin-fixed, paraffinembedded (FFPE) samples, saliva, buccal swab, and blood prior to library generation. This method is resistant to inhibitors that are present in biological samples therefore allowing their direct use in the amplification reaction. The Direct Reagent is compatible with Ion AmpliSeq panels, no purification is required. We evaluated each type of sample using Ion AmpliSeq DNA research panels with 207, 1267 or 3900 assays designed to known oncogenes. Results: Libraries generated directly from FFPE tissue slices, and retrospective samples from whole blood, saliva, and buccal swab samples all resulted in similar sequencing performance as compared to DNA purified from FFPE using commercial nucleic acid purification kits. For the 207 assay DNA panel, uniformity of base coverage was >95%, base reads on target was >95%, and target bases with no strand bias was >97%. Additionally, greater than 98% of the assays were covered at >500X base read depth with a minimum of 2500X average base coverage depth when 6 libraries were included per sequencing run. For the 1267 and 3900 assay DNA panels, libraries made directly from biological samples also resulted in similar metrics as those observed for purified DNA controls. We were able to generate and successfully sequence DNA libraries from single FFPE slices as small as 2 mm X 2 mm X 5 microns and from a single microliter of archived blood or saliva samples.
We developed a research method to directly use biological samples and low input samples for generating libraries to reduce the time from sample to analysis on an NGS platform. Extremely small samples were successfully processed with this method, which reduces the total time from sample to sequence to 24 hours. For research use only. Not for use in diagnostic procedures. Introduction: All molecular tests have an inherent limit of detection (LOD). Therefore, the ability of a molecular oncology assay to reliably detect a particular genetic alteration requires a sample with a sufficiently high percentage of neoplastic cells. Many laboratories employ dissection techniques to enrich for neoplastic cells prior to molecular testing. However, optimal procedures for guiding dissection, ensuring the adequacy of dissection and monitoring relevant QA measures have not been established. Methods: Several modifications to the tumor dissection procedures and workflow were sequentially introduced over a period of one year. These included transition of tissue review to pathologists board-certified in anatomic and molecular and genetic pathology, post-review of unstained slides following dissection, the direct marking of unstained slides in select cases and the use of a dissection microscope. During these changes, key QC measures were monitored including mutation allele frequencies (KRAS Sanger sequencing), the percentage of cases deemed to be QNS (quantity not sufficient), the percentage of inadequate results (EGFR fragment analysis) and the percentage of cases requiring re-extraction. Results: Each modification of the tumor dissection procedure resulted in a significant improvement to the dissection process based on one or more QC measures. The performance of tissue review by an experienced molecular pathologist resulted in a 55% reduction in QNS rate (p < 0.001), with a slight, nonstatistically significant increase in EGFR inadequate results and without a decrease in KRAS mutation allele frequency. The review of unstained slides following dissection resulted in an average of 2% of cases requiring re-extraction and a 90% reduction in low-level KRAS mutations (<15% allele frequency; p = 0.004). The direct marking of unstained slides and the use of a dissection microscope resulted in a trend toward a decreasing re-extraction rate and a further decrease in the QNS rate. Conclusions: The review of unstained slides following dissection and the monitoring of mutation allele frequencies is an important part of tumor dissection QA that should be considered by all laboratories performing dissection in order to allow this process to be optimized and to avoid potential false negative results. The optimization of tumor dissection procedures results in a measurable improvement in the quality and reliability of this process.
G. Kang 1 , K. Rhodes 2 , C. Van Loy 2 , C. Beadling 3 , M. Andersen 2 , C. Corless 3 1 Sanggye Paik Hospital, Inje University, Seoul Gyeonggi-do, Korea; 2 Thermo Fisher Scientific, Carlsbad, CA; 3 Oregon Health and Science University, Portland, OR. Introduction: The quality of next-generation sequencing results is critically dependent on the input material. Unfortunately, many biopsy specimens submitted for sequencing contain too little tissue to support DNA extraction, quantification and NGS library preparation. Yields from commercially available DNA extraction kits vary and DNA losses are inevitable during protease digestion and subsequent purification. We tested a novel approach that allows amplicon-based NGS libraries to be prepared directly from pieces of formalin-fixed, paraffin-embedded (FFPE) tissue. Methods: Tumor material was scraped with a sterile scalpel blade from 5 micron unstained sections that were not deparaffinized, and was melted in TE buffer at 99°C twice for 2 min with a brief intervening spin. Direct Reagent was added at room temp for 15 min, followed by the addition of Switch Solution 2, then 5X Ion AmpliSeq HiFi master mix and primers for the Ion AmpliSeq Cancer Hotspot Panel v2 were added. Libraries were prepared using Ion AmpliSeq v3 reagents. Emulsion PCR was performed on an Ion OneTouch 2 instrument, and sequencing was completed on Ion 318 Chip Kit v2 chips run on an Ion PGM system. Results: We tested a variety of archived FFPE samples that had previously been analyzed, including melanoma (in skin, brain, lung), lung adenocarcinoma (in lung, lymph node, brain), astrocytoma, urothelial carcinoma, adenocarcinomas of the small and large bowel, GI stromal tumor (GIST), and B-cell lymphoma in bone marrow. Between 32 mm2 and 64 mm2 of tumor-rich material was collected from a single slide of each tumor. In addition, from the GIST sample and from a metastatic melanoma sample we collected areas measuring 4 mm2 (2700-4300 nuclei) and 16 mm2, again from single slides. Libraries from all 21 samples yielded excellent sequence data, with 100% concordance for all previously reported mutations (18 SNV, 1 in/del). An additional 40 FFPE samples from different commercial sources, including a reference standard, also yielded high quality libraries. All variants with mutant allele frequencies greater than 1% were detected in the reference standard. Conclusions: By eliminating tissue deparaffinization, protease digestion and DNA purification, the Ion AmpliSeq Direct approach supports highly efficient NGS library preparation from very small quantities of FFPE tumor tissue. Work is ongoing to define the lower limit of input tissue, and to further simplify the workflow. Direct extraction/library preparation holds the promise of salvaging biopsy samples that are otherwise too small for standard NGS analysis.
C.E. Huang, R. Vemula, J. Dickens, B. Anekella SeraCare Life Sciences, Gaithersburg, MD. Introduction: Non-infectious, whole process controls are needed by laboratories and test developers as they design, manufacture and validate new diagnostic assays. New assays are needed to prepare for emerging viruses, such as the recent outbreak of Ebola 2014 Virus, to combat drug resistance in highly adaptive viruses such as HIV-1, and to advance diagnostics for non-culturable viruses such as HBV. SeraCare's AccuPlex technology uses recombinant viruses containing target sequences from the pathogen of interest and has many advantages as a NAT quality control material. First, it mimics clinical samples because it undergoes the entire extraction procedure. Second, it is non-infectious and ensures safety for lab personnel. Finally, AccuPlex recombinant viruses can be highly multiplexed and have extended stability at 2 Methods: Recombinant RNA virus was produced using portions of Ebola 2014 virus nucleoprotein (NP), envelope glycoprotein (GP), and VP24 genes. Similarly, recombinant virus was produced bearing HIV-1 sequence with 48 mutations in Protease, Reverse Transcriptase and Integrase genes, which represent the current, clinically relevant mutations for drug resistance. This "multi-mutant" virus was mixed at low levels with WT recombinant virus to form whole process controls for minor variant detection. SeraCare also developed HBV controls of the genotype desired and at titers >1.0E+08 copies/mL. The reference materials are formulated in defibrinated plasma and gene sequences are severely truncated, stop codons are introduced, and are heat treated to ensure the product safety. Results: The AccuPlex recombinant Ebola GP/NP Reference material was tested on Cepheid Xpert Ebola assay, Primerdesign Ltd Ebola genesig Kit and in-house developed TaqMan assay. Performance studies conducted over ten (10) days using multiple operators showed the %CV was <2%. AccuPlex HIV-1 Multi-drug resistant material was functionally tested by NGS. Stress stability was analyzed at 37 multiple rounds of freeze-thaw to establish a shelf life of >24 months at
Ongoing studies also indicate that the materials are stable in both liquid and dry state at room temperature for at least eight (8) weeks. Real-time studies were performed at -20 detected across the nine (9) months of storage, even for samples stored at ambient temperatures. Conclusions: SeraCare has developed non-infectious, whole process controls for both RNA and DNA virus pathogens that perform consistently and are highly stable. These reference materials will enable laboratories to develop, validate and train personnel to ensure preparedness for emerging and reemerging pathogens.
|
