PMC:548130 / 8065-11706 JSONTXT

{"project":"2_test","denotations":[{"id":"15673474-11241407-8410893","span":{"begin":573,"end":574},"obj":"11241407"},{"id":"15673474-10471054-8410894","span":{"begin":575,"end":576},"obj":"10471054"},{"id":"15673474-15350357-8410895","span":{"begin":1637,"end":1638},"obj":"15350357"},{"id":"15673474-15350357-8410896","span":{"begin":2524,"end":2525},"obj":"15350357"},{"id":"15673474-15350357-8410897","span":{"begin":2756,"end":2757},"obj":"15350357"}],"text":"Results and discussion\n\nLDMAS application in identification of LOH markers associated with persistence / progression of cervical intraepithelial neoplasia\nWe divided the CIN groups into disease free indicating cases that become CIN free after treatment, and disease persistence/progression indicating cases that develop show progression or persistence of CIN despite treatment. We used LDMAS to retrospectively examine the prognostic value of LOH at 12 microsatellite markers including 10 from 3p14, 3p22-21, 6p21 and 11q23 which are frequently deleted in cervical cancer [3,4], in 164 cases of CIN lesions using archival cytological/histological specimens. LOH was further correlated with high risk HPV infection.\nInitially MRES was used to automatically parse 4300 patient records and extract clinico-pathological data including age, diagnosis, method of treatment and treatment response during follow up. Out of those, 164 cases with follow up of 3 or more years were chosen for the study and their clinico-pathological information was imported into LDAS. Initially, 71 out of the 164 selected cases were examined for LOH using 12 fluorescent microsatellite markers ran on ABI377 DNA Sequencer. LDAS was then used to identify the microsatellite markers for which LOH was significantly associated with disease persistence/progression of CIN using two tailed student t-test. Figure 2 generated using LDAS shows that microsatellite markers D3S1300 (3p14.2), D3S1260 (3p22.2), D11S35 (11q22.1) and D11S528 (11q23.3) have the highest LOH in CIN lesions displaying persistence/progression than those who were disease free during follow up [5].\n\nValidation of prognostic markers associated with persistence / progression of CIN\nTo validate this finding, LOH at these four markers was investigated in a further series of 93 cases. Compatible results were obtained from these additional cases.\nThe two sets of data were combined and further compared using LDMS. Methodologies included :\n1) comparison using χ2 (chi-squared) test of LOH at each of the four microsatellite markers with age, various methods of treatment, different subtypes of HPV infection and between CINs showing disease free or disease persistence/progression.\n2) correlation of LOH data with histological grade of CIN, treatment response and various HPV subtypes.\nThrough such complex analysis, we showed that concurrent LOH at two of the four microsatellite markers could identify 47% of CINs that showed disease persistence/progression with 100% specificity [5]. Furthermore, LOH at D3S1300 was found to be significantly associated with HPV16 infection. Part of this data analysis is supplied in the LDMAS guide [see Additional file 1]. More detailed analysis of this study is described in [5].\n\nAlgorithm for identifying prognostic disease markers\nBased on the above example, an algorithm can be developed to extract prognostic markers for other diseases. The algorithm can be summarised in the following pseudocode :\n(1) Divide the disease in groups according to the pathology staging\n(2) Parse patient data from clinical records and use the groups defined in part (1)\n(3) FOR each microsatellite marker\ncarry out a two tailed student t-test between the disease groups using LOH data\nIF t-test p ≤ 0.05\nMarker is significant in prognosis of the disease\nELSE\nMarker is not significant in prognosis of the disease\n(4) Validate the prognostic markers using χ2 (chi-squared) test of LOH with clinico-pathological data and correlation of LOH data with histological grade of CIN, treatment response and various HPV subtypes.\nLDMAS has been implemented using the above pseudocode.\n"}