A recent analysis published in JAMA Network Open addresses the ongoing debate regarding the selection of endpoints in heart failure (HF) clinical trials, specifically focusing on the importance of heart failure-specific versus all-cause endpoints. The study, a meta-analysis of 113 HF trials, highlights the complexities in endpoint selection and their implications for trial design and interpretation.
Understanding Hospitalization Endpoints in HF Trials
The meta-analysis, led by Sayed et al., examined HF trials published over the last three decades, evaluating the association of heart failure hospitalization (HHF) with all-cause hospitalization (ACH). The findings indicate that ACH was reported in approximately half of the evaluated trials, with no significant improvement across time. Notably, less than half of ACHs were attributed to HHF, reflecting the multi-morbidity often seen in HF patients. This is particularly relevant in patients with HF with preserved ejection fraction (HFpEF), who often experience a higher proportion of non-cardiovascular hospitalizations.
The study revealed a strong correlation between treatment outcomes on HHF and ACH, with a trial-level R² of 90.1%. However, the treatment effect on ACH was, on average, half that observed for HHF. This discrepancy has significant implications for sample size estimation in clinical trials. For instance, a trial designed to detect a 25% reduction in the odds of HHF might require a sample size of 2424 participants. In contrast, to detect a corresponding 12.5% reduction in the odds of ACH, the sample size would need to increase by 62%, to 3936 participants.
Implications for Trial Design and Interpretation
The choice between cause-specific and all-cause outcomes in HF trials is critical. According to Sanjay Kaul, MD, Department of Cardiology, Cedars-Sinai Medical Center, cause-specific endpoints should be chosen as the primary outcome for HF trials and utilized for sample size estimation. However, all-cause endpoints should be pre-specified in trial protocols as secondary or exploratory outcomes. Case report forms should clearly capture detailed information necessary to adjudicate whether cause-specific outcome is primary, contributory, or noncontributory. Even when a cause-specific primary end point is significantly reduced by an intervention, all-cause end points should be assessed to rule out potential adverse effects. A reduction in an all-cause end point is even more reassuring.
The analysis also points out that the proportion of HHF to ACH can vary significantly based on the HF subtype. For example, in the EMPEROR-Preserved trial with empagliflozin, the treatment effect on hospitalization diminished when the analysis broadened beyond HF events, decreasing from a 29% reduction in first HHF to an 8% reduction in first ACH. This is contrasted by the EMPEROR-Reduced trial, where the reduction in ACH was greater (18%) due to a larger proportion of hospital admissions being related to worsening HF.
Recommendations for Future Trials
Based on these findings, the authors recommend a balanced approach to endpoint selection in HF trials. Cause-specific endpoints, such as HHF, are more efficient for evaluating treatment benefits, while all-cause endpoints provide a broader perspective on the net clinical benefit, capturing both favorable and adverse effects. Regulators should require collection and reporting of both outcomes, physicians, and guideline writers should prioritize, and payers should reward therapies that favorably impact both outcomes. The study emphasizes the importance of pre-specifying all-cause endpoints in trial protocols and thoroughly capturing detailed information to adjudicate cause-specific outcomes.
Ultimately, the management of HF requires a multidisciplinary approach that addresses both HF and non-HF comorbidities, rather than focusing exclusively on HF outcomes. This holistic approach ensures a comprehensive evaluation of efficacy, safety, and health economics in HF clinical trials.
