Diagnostic accuracy of patient-reported outcomes in predicting endoscopic subscore in patients with ulcerative colitis

Abstract

Background

Tofacitinib is an oral, small molecule Janus kinase inhibitor for the treatment of ulcerative colitis (UC). In patients with UC, associations between endoscopic findings and UC symptoms are not well described.

Aims

Post hoc analysis of data from two randomised, placebo-controlled, 8-week, phase 3 studies of tofacitinib for the treatment of patients with UC.

Methods

Associations of stool frequency and rectal bleeding subscores with endoscopic improvement (Mayo endoscopic subscore ≤1) were assessed and relationships studied using regression analyses.

Results

Analysis of two-by-two contingency tables showed that dichotomised stool frequency and rectal bleeding were each or both not good predictors of endoscopic improvement. Using stool frequency and/or rectal bleeding as predictors of endoscopic subscore, regression modelling analyses demonstrated a weak relationship between variables. However, a robust relationship was observed with endoscopic subscore as a predictor of stool frequency and rectal bleeding. In OCTAVE Induction 1, normal/inactive disease (endoscopic subscore 0) corresponded to a least-squares mean value of 0.05 for rectal bleeding (no blood), and severe disease (endoscopic subscore 3) corresponded to a value of 1.5 (interpreted as streaks of blood with stool <50% of the time [score of 1] or obvious blood with stool most of the time [score of 2]). OCTAVE Induction 2 results were similar.

Conclusions

Results suggest that the likelihood of endoscopic improvement or normalisation is higher in patients with normal stool frequency and without rectal bleeding, but that these symptoms alone are not predictive of endoscopic improvement or normalisation, and endoscopy is needed for disease assessment.

ClinicalTrials.gov: NCT01465763; NCT01458951.