Background: Chronic inflammation caused by Ulcerative Colitis (UC) causes a pro-neoplastic drive in the inflamed colon, leading to a markedly greater risk of invasive malignancy compared to the general population. Despite colonoscopic surveillance protocols, 50% of cases proceed to invasive cancer before neoplasia is detected. We and others have previously shown an association between epigenetic change, in the form of methylation, specifically in Wnt-signalling associated genes (sFRP1, sFRP2, WIF-1 and others) and neoplasia within colons in patients affected by UC.
The Enhanced Neoplasia Detection and Cancer Prevention in Chronic Colitis (ENDCaP-C) trial is an observational, multi-centre test accuracy study, consisting of two phases. The initial phase aimed to measure the accuracy of a panel of markers on a retrospective cohort of patients with ulcerative colitis in order to ascertain their utility in phase 2, a prospective clinical study.
Methods: Patients were identified from retrospective sample collections with the West Midlands area from patients undergoing colonoscopic surveillance for ulcerative colitis.
These were classified as either having “neoplasia”, defined as any of adenocarcinoma, high-grade or low-grade dysplasia; “non-neoplastic” defined as matched normal mucosal biopsies taken downstream from areas of neoplasia or “control” defined as normal mucosa from patients either with or without chronic inflammation who had been screened for neoplasia without it being found. DNA from biopsy samples then underwent bisulphite pyrosequencing of a 11 marker gene panel (SFRP1, SFRP2, SRP4, SRP5, WIF1, TUBB6, SOX7, APC1A, APC2, MINT1, RUNX3). Percentage methylation was log transformed and the three groups compared by t-testing and multivariate logistic regression to predict accuracy.
Results: In total, 569 blocks from patients undergoing surveillance were retrieved, consisting of 113 neoplastic, 113 non-neoplatic and 343 control blocks. Of the neoplastic samples, 35/113 were adenocarcinoma and 78/113 were dysplastic. All markers had a success rate of >90% apart from SFRP1, MINT1 and RUNX3 which were not taken forward to further analysis. In univariate analysis, five markers accurately detected both dysplasia (p<0.0001) and neoplasia (p<0.0001) – SFRP2, SFRP4, WIF1, APC1A and APC2. A multivariate logistic regression analysis and ROC analysis demonstrated the model using the five marker panel had excellent diagnostic accuracy (AUC=0.87, 95% CI: 0.82–0.92, model p<0.0001.)
Conclusions: A five marker methylation marker panel accurately predicts ulcerative colitis associated dysplasia and neoplasia in this population. This panel is being taken forward to prospective validation and in enabling enhanced surveillance in phase two of the study.