Development and Validation of a New Hierarchical Composite End Point for Clinical Trials of Kidney Disease Progression

Abstract

Significance statement: The established composite kidney end point in clinical trials combines clinical events with sustained large changes in GFR but does not weigh the relative clinical importance of the end point components. By contrast, a hierarchical composite end point (HCE) accounts for the clinical importance of the end point components. The authors developed and validated a kidney HCE that combines clinical kidney outcomes with longitudinal GFR changes (GFR slope). They demonstrate that in seven major placebo-controlled kidney outcome trials with different medications, treatment effect estimates on the HCE were consistently in similar directions and of similar magnitudes compared with treatment effects on the established kidney end point. The HCE's prioritization of clinical outcomes and ability to combine dichotomous outcomes with GFR slope make it an attractive alternative to the established kidney end point.

Background: The established composite kidney end point in clinical trials combines clinical events with sustained large changes in GFR. However, the statistical method does not weigh the relative clinical importance of the end point components. A HCE accounts for the clinical importance of the end point components and enables combining dichotomous outcomes with continuous measures.

Methods: We developed and validated a new HCE for kidney disease progression, performing post hoc analyses of seven major Phase 3 placebo-controlled trials that assessed the effects of canagliflozin, dapagliflozin, finerenone, atrasentan, losartan, irbesartan, and aliskiren in patients with CKD. We calculated the win odds (WOs) for treatment effects on a kidney HCE, defined as a hierarchical composite of all-cause mortality; kidney failure; sustained 57%, 50%, and 40% GFR declines from baseline; and GFR slope. The WO describes the odds of a more favorable outcome for receiving the active compared with the control. We compared the WO with the hazard ratio (HR) of the primary kidney outcome of the original trials.

Results: In all trials, treatment effects calculated with the WO reflected a similar direction and magnitude of the treatment effect compared with the HR. Clinical trials incorporating the HCE would achieve increased statistical power compared with the established composite end point at equivalent sample sizes.

Conclusions: In seven major kidney clinical trials, the WO and HR provided similar direction of treatment effect estimates with smaller HRs associated with larger WOs. The prioritization of clinical outcomes and inclusion of broader composite end points makes the HCE an attractive alternative to the established kidney end point.

Trial registration: ClinicalTrials.gov NCT03036150 NCT02065791 NCT02540993 NCT01858532 NCT00308347 NCT00317915 NCT00549757.