Assessing Benefit in Patients With Heart Failure and Reduced Ejection Fraction: Analysis of the VICTORIA Trial Using Novel Prognostic Risk Stratification.

Abstract

Background: Randomized trials remain the standard for evaluating novel therapies. Primary endpoint(s) risk heterogeneity and may dilute treatment efficacy. The 5-step Stratified Testing and Amalgamation Routine (5-STAR) methodology helps to address these limitations. We applied this methodology to the original VICTORIA (VerICiguaT Global Study in Subjects With Heart Failure With Reduced Ejection Fraction) database and enriched it by relevant ancillary information.

Methods: Elastic net Cox regression and a conditional inference-tree tool blinded to treatment assignment partitioned the population into risk strata for endpoints based on baseline covariates strongly associated with outcomes. Core laboratory biomarkers and baseline electrocardiographic variables were available covariates. After unblinding, treatments were compared for the primary composite endpoint of cardiovascular death or heart failure hospitalization (HFH) within each risk stratum: stratum-level results were averaged for overall inference.

Results: The 5-STAR analysis revealed greater vericiguat efficacy pooled on the primary composite endpoint than the original VICTORIA analysis (5-STAR HR, 95% CI: 0.85, 0.77-0.94 vs 0.90, 0.82-0.98), and its components (5-STAR HR, 95% CI: cardiovascular death: 0.79, 0.67-0.93 vs 0.93, 0.81-1.06; HFH: 0.89, 0.79-1.00 vs 0.90, 0.81-1.00). Five biomarkers (GDF-15, NT-proBNP, albumin, blood urea nitrogen, and uric acid) determined the risk strata across the 3 endpoints.

Conclusions: The 5-STAR methodology attenuates the dilution of treatment effects inherent in conventional prognostic risk heterogeneity. This retrospective analysis of VICTORIA revealed greater efficacy of vericiguat on the primary endpoint and its components. GDF-15 was consistently the strongest prognostic risk factor across the composite endpoint and its components.