Background: Patients with ulcerative colitis (UC) are at risk for colorectal neoplasia. Challenges associated with surveillance colonoscopy with random biopsies for detection of dysplasia/cancer are well-documented. This study extended our findings in UC-associated colorectal cancer to include low-grade dysplasia (LGD) patients, testing whether our biomarker panel detects any UC-associated neoplasm.
Methods: DNA from the LGD area and the corresponding nonadjacent, non-dysplastic section from 171 UC-LGD patients was extracted. TaqMan SNP Genotyping Assays for TNF-α, IL-1β, and IL23R were used to evaluate polymorphisms for each gene. Bisulfite-treated DNA was used for methylation testing of RUNX3, COX2, and MINT1. LGD data were combined with UC-cancer patient data for statistical testing. Logistic regression analyses determined associations between genetic/epigenetic/clinical variables and UC-associated neoplasia. Receiver operating characteristic analyses were performed to determine the final synchronous neoplasm detection panel.
Results: Comparison of nonadjacent, non-dysplastic DNA from UC-neoplasm patients versus UC-controls indicated that TNF-α, IL-1β, and methylation of RUNX3, MINT1, and COX2 were significantly different (P < 0.0001). In multivariable analysis, all remained significant with an area under the curve of 0.85, exceeding the clinical variable panel area under the curve. Combining clinical and experimental variables yielded a neoplasm biomarker panel with an area under the curve of 0.95 (sensitivity and specificity of 82% and 91%, respectively). Analysis of DNA from LGD with known progression compared with LGD without progression indicated a significant difference in RUNX3 methylation.
Conclusions: A combined clinical, genetic, and epigenetic model for detecting synchronous neoplasm by testing of non-neoplastic colonic tissue had favorable operating characteristics and could complement current patient care.