Matching-adjusted indirect comparisons of efficacy outcomes between etrasimod and ozanimod for moderately to severely active ulcerative colitis.

Abstract

Aim: Etrasimod and ozanimod are selective sphingosine 1-phosphate receptor modulators targeting the S1P_1,4,5, and S1P_1,5 receptors, respectively, for the treatment of patients with moderately to severely active ulcerative colitis (UC). No head-to-head trial data exist between the two treatments. We compared these treatments indirectly using key efficacy outcomes from pivotal trials with induction and maintenance phase data adjusting for differences in clinical trial design and populations. Materials & methods: Individual patient data for etrasimod were matched to published aggregate data of ozanimod by key baseline characteristics. An anchored matching-adjusted indirect comparison (MAIC) was conducted for the induction period. An unanchored MAIC was utilized during the maintenance period due to differences in placebo arms between trials as a result of differing trial designs. Matching characteristics measured at baseline were age, sex, corticosteroid use, duration of UC, biologic exposure, modified Mayo score, and presence of left-sided colitis. Outcomes were clinical response and clinical remission for the induction period, and clinical response and clinical remission among induction phase responders for the maintenance period. Two sensitivity analyses were conducted. The first matched on prior TNFi exposure rather than biologic exposure, the second sensitivity analysis included an induction only etrasimod trial (ELEVATE UC 12). Results: There were no significant differences between etrasimod and ozanimod at the end of the induction period for clinical response and clinical remission, respectively (relative risk [RR] 0.98 [95% confidence interval (CI): 0.76-1.33], RR: 1.25 [95% CI: 0.71-2.92]). At the end of maintenance, etrasimod demonstrated improved outcomes compared with ozanimod for both clinical response (RR: 1.18 [95% CI: 1.05-1.30]) and clinical remission among induction phase responders (RR: 1.33 [95% CI: 1.12-1.55]). In the sensitivity analysis that matched on prior TNFi exposure rather than biologic exposure, there were no notable differences compared with the primary analyses. In the sensitivity analysis pooling ELEVATE UC 12 and ELEVATE UC 52 data, results were similar for clinical response (RR: 0.90 [95% CI: 0.75-1.10]) but etrasimod showed reduced efficacy for clinical remission (RR: 0.72 [95% CI: 0.50-1.12]) compared with the primary analysis, though overall remained not significantly different from ozanimod. Conclusion: MAIC results suggest that patients receiving etrasimod have similar induction results but are more likely to have clinical response and clinical remission at the end of the maintenance phase compared with patients receiving ozanimod. Despite the approach to ensure similarity between the trials by weighting, residual imbalance is possible, and results should be interpreted in the context of the assumptions.