Baseline risk assessment of patients with ulcerative colitis: does initial treatment selection influence outcomes?

BACKGROUND AND AIMS: Treatment of ulcerative colitis (UC) typically follows a step-up approach and targets colonic mucosal healing. Although mucosal healing reduces the risk of colectomy, whether or not early treatment of patients with 'high-risk' features using tumor necrosis factor (TNF) antagonists reduces the risk of colectomy is not clear. Accordingly, we aim to evaluate the effect of baseline treatment selection according to the risk profile on 5-year outcomes and identify predictors of poor outcomes. PATIENTS AND METHODS: Adult patients with confirmed UC were retrospectively identified. Baseline clinical and endoscopic data were collected. Patients were assigned a risk profile on the basis of the presence or absence of 'high-risk' features within the first 6 months of diagnosis including moderate to severe endoscopic disease, frequent need for steroids, steroid dependency, and disease involving the entire colon according to endoscopy. Treatment discordance was defined as treating 'high-risk' patients with medications other than anti-TNF therapy during the first 6 months after diagnosis or treating 'low-risk' patients with anti-TNF therapy within 6 months of diagnosis. The associations between discordance and 5-year colectomy and hospitalization rates were statistically calculated through regression analysis, as were predictors of outcomes. RESULTS: A total of 108 patients were identified and studied. The median age was 36 years (interquartile range=27-50) and the average duration of disease was 6.6 (±3.1) years. Females comprised 62% of the cohort and 30% reported cigarette smoking. Seventy three percent of the patients were placed in the 'high-risk' category. The 5-year risk of colectomy was not statistically significantly higher in patients identified as 'high-risk' compared with those who were 'low-risk' (risk ratio=0.86, 95% confidence interval=0.24-3.1, P=0.81), nor was the 5-year risk of hospitalizations (risk ratio=1.63, 95% confidence interval=0.81-3.30, P=0.15). On the basis of stepwise model selection, colectomy was significantly predicted by discordance (P=0.039), arthritis (P=0.007), baseline stool frequency (P=0.019), Adalimumab use within the first 6 months of diagnosis (P=0.006), and pyoderma gangrenosum (P=0.049); hospitalization was predicted by discordance (P=0.018), baseline albumin concentrations (P=0.005), thromboembolism (P<0.005), thiopurine use within the first 6 months of diagnosis (P<0.005), Adalimumab use within the first 6 months of diagnosis (P=0.003), nationality (P=0.016), endoscopic severity (P=0.007), arthritis (P=0.005), and pyoderma gangrenosum (P=0.025). CONCLUSION: Among other clinical parameters, discordance between baseline risk and treatment selection appears to be a significant predictor of outcomes in UC.