OP012 Risk factors for colorectal neoplasia in ulcerative colitis: results from the largest and longest-running colonoscopic surveillance programme
C. H. R. Choi*1, 2, I. Al Bakir1, 2, N. S. J. Ding1, M. Moorghen1, S. Thomas-Gibson1, J. Warusavitarne1, B. Saunders1, M. Rutter3, T. Graham2, A. Hart1
1St. Mark’s Hospital, London, United Kingdom, 2Barts Cancer Institute, Tumour Biology, London, United Kingdom, 3University Hospital of North Tees, Department of Gastroenterology, Stockton-on-Tees, Teesside, United Kingdom
While patients with ulcerative colitis (UC) have an increased risk of developing colorectal cancer (CRC), its absolute risk remains low. As most CRCs develop from pre-existing dysplasia, identifying those patients at high risk of developing dysplasia is important. The aim of this study was to identify risk factors associated with development of colorectal neoplasia (CRN) in patients with UC.
Patients with extensive UC who were under surveillance from 2003 to 2012 were identified and followed up to July 2013. Each surveillance episode was scored for endoscopic and histological severity of inflammation based on the segment worst affected by colitis: (0, normal/quiescent; 1, mild; 2, moderate; 3, severe active). Potential predictors (n = 18) were correlated against colorectal neoplasia (CRN) outcome, which was defined as non-polypoid low-grade dysplasia, high-grade dysplasia, or CRC. To take into account the variables that may change over time, we employed Cox regression analysis with time-dependent covariates.
In the study, 987 UC patients underwent 6 985 colonoscopies (median, 6; interquartile range (IQR), 4–9 per patient) over 12 305 patient-years (median follow-up, 11; IQR, 7–17) since starting their surveillance. Of these, 97 (9.8%) developed CRN (41 cancers). After multivariate analysis, following variables remained a significant contributory factor for the CRN outcome: increased average histological inflammation score (average-HIS; average score of all surveillance colonoscopies for each patient), endoscopic signs of disease chronicity (ie, tubular, featureless, or shortened colon), colonic stricture, and primary sclerosing cholangitis (see Table 1 and Figure 1).
Final multivariate model showing predictors associated with development of CRN
Kaplan-Meier plots showing cumulative risk of colorectal neoplasia by
a) histological inflammation score from immediately preceding colonoscopy, b) PSC, c) signs of disease chronicity, and d) colonic stricture.
Further, although average-HIS was the most influential predictor, HIS of the immediately preceding colonoscopy showed comparable predictive value (Table 1). Postinflammatory polyps and scarring were not significant at the multivariate level.
This is the largest detailed cohort study to date looking at individual risk factors for CRC in UC. Patients with PSC, pertinent macroscopic features (tubular, featureless, shortened colon, or stricture), or persistent microscopic inflammation require intensive colonoscopic surveillance. In addition, the histological assessment of inflammation severity at the time of colonoscopy may be used to predict future CRN risk.