Evaluating Desirability of Outcome Ranking (DOOR) and Win Ratio for Hierarchical Composite Endpoints in Cryptococcal Meningitis Trials.

Abstract

Traditional antifungal trials have relied on all-cause mortality as the primary endpoint, which may overlook important treatment-related outcomes such as adverse events or tolerability. We evaluated the use of hierarchical composite endpoints to capture clinically meaningful treatment differences in cryptococcal meningitis trials. A Delphi survey determined priorities in constructing the composite endpoint. We analyzed three randomized clinical trials on cryptococcal meningitis with Win Ratio and the Desirability of Outcome Ranking (DOOR). We estimated the Better DOOR Probability (BDP), which is the probability that a participant assigned to the study treatment would have a more desirable outcome. The Delphi survey received 34 respondents and most of them prioritized mortality and serious adverse events. In the AMBITION-cm trial, single-dose liposomal amphotericin B yielded a BDP of 55% (95%CI, 51-59%) and Win Ratio of 1.23 (95%CI, 1.03-1.46), indicating superiority over standard-of-care therapy. In ENACT phase II trial, oral amphotericin B was associated with BDP of 52% (95%CI, 42-63%) and Win Ratio of 1.14 (95%CI, 0.68-1.91), showing no statistically significant difference. The ASTRO-CM trial testing adjunctive sertraline showed no overall benefit with sertraline using either BDP (46%; 95%CI, 41-52%) or Win Ratio 0.87 (95%CI, 0.70-1.08). Hierarchical composite endpoints were feasible to implement in cryptococcal meningitis clinical trials and can enhance clinical relevance, especially in non-inferiority designs where safety and tolerability matter. The two statistical approaches provided complementary perspectives - DOOR offering absolute ranks and Win Ratio providing a more interpretable relative treatment effect estimate.