Performance of existing predictive scores of response in patients with moderately active ulcerative colitis treated with oral corticosteroids.

Abstract

Background: the prediction of response to corticosteroids in ulcerative colitis (UC) might allow early introduction of advanced therapies. Predictive scores have been proposed for severe UC, but their performance in moderate flares is unknown.

Aim: to evaluate the performance of established indexes to predict early corticosteroid response in moderately active UC patients treated with oral prednisone and identify baseline predictors of response.

Methods: data from two prospective clinical studies in patients with moderate UC treated with oral prednisone were analyzed. All patients underwent a colonoscopy before starting corticosteroids and clinical and biological assessments at baseline and after seven days. Non-response was defined as the persistence of rectal bleeding in > 50 % of bowel movements and > 3 additional bowel movements from baseline after seven days of therapy or if rescue therapy was prescribed. Albumin, C-Reactive Protein, and Endoscopy Score (ACE), Degree of Ulcerative Colitis Burden of Luminal Inflammation (DUBLIN), and C-reactive protein/albumin ratio (CAR) scores were calculated.

Results: among 94 patients (51 % extensive colitis), 44 achieved clinical remission and 16 were classified as non-responders at day 7. Most patients met high-risk scores for the DUBLIN score, 17 % for CAR index, whereas only one patient met the high-risk ACE score. No index reliably predicted non-response. In the univariate analysis, younger age at baseline (p = 0.04) and shorter disease duration at admission (p = 0.012) were the only variables associated with lack of response.

Conclusions: the existing clinical-biological scores developed for therapeutic response in severe UC lack predictive accuracy for moderate UC flares treated with oral corticosteroids. Novel, tailored prediction tools are needed in this clinical setting.