Can Baseline Histological Score Predict Subsequent Clinical Outcomes in Patients With Ulcerative Colitis?

Abstract

Introduction: Histological inflammation in ulcerative colitis (UC) has been shown to be directly associated with risk of relapse and with neoplasia, both of which ultimately result in higher rates of hospitalization and colectomy. However, histological activity has not yet been studied prospectively as an independent predictor for hospitalization or surgery. This prospective study aimed to utilize a single baseline histologic assessment to predict clinical UC outcomes. Methods: A study cohort of 100 consecutive UC patients prospectively completed colonoscopy or flexible sigmoidoscopy evaluation with biopsies as part of standard of care practice. Histology was assessed by 2 expert GI pathologists per standard of care at our institution using a novel 6-point histologic activity index (the Chicago Scale of Inflammation). Histology scores were analyzed as “continuous” (mean of biopsy scores; maximum score of most inflamed biopsy) or “binary” (active or inactive disease). Patients were followed for up to 3 years to assess UC-related outcomes (neoplasia, hospitalization and surgery). Cox regression analysis was performed to differentiate hazard ratios of clinical outcomes based on initial histological assessments. Results: Eighty-seven of the 100 patients were followed for a median 29 months (0-34) (54% male, 53% pancolitis, median disease duration 17 yrs (7-55)). During follow-up, 5 patients developed new dysplastic lesions, 4 patients were admitted for therapy escalation and 6 patients underwent colectomy for refractory UC (4 were followed and 1 developed pouchitis). Patients with active baseline histologic inflammation had a hazard ratio (HR) of 0.26 (0-20.9) for developing neoplasia, 3.3 (0.4-32.0) for hospitalization, and 1.1 (0.2-5.6) for colectomy. When histological scores were defined as continuous scores, mean and maximum scores respectively, the hazard ratios for the clinical outcomes were similar and did not achieve significance. Patients with hospitalizations were more likely to have prior IMM (p=0.03) or anti-TNF (p=0.01) exposure, while surgery patients were more likely to be exposed to anti-TNFs (p< 0.01) only. Conclusion: In this prospective cohort analysis, increased histological activity at baseline exam was not associated with subsequent hospitalization, surgery or newly diagnosed neoplastic lesions within 3 years. A longer follow-up period and a larger cohort will be required to further assess the clinical predictive value of baseline histology in UC.Table 1: Hazard Ratios of Clinical Outcomes by Baseline Histological Scores