Critical appraisal of combination therapies: the impact of trial design on Benefit-Risk outcomes.

Combination therapies are widely used in oncology due to their potential to improve therapeutic efficacy and overcome drug resistance. However, these potential advantages must be weighed against added toxicity, treatment burden, and costs. In this commentary, we examine some recent examples of divergent opinions by the US Food and Drug Administration and the European Medicines Agency across three indications. In neuroendocrine tumours, NETTER-2 demonstrated that early treatment intensification may improve progression-free survival. However, the higher toxicity, the absence of a demonstrated overall survival or quality of life benefit, and several important design limitations, including the open-label design and a potentially suboptimal control arm, complicate the interpretation of the results. In pleural mesothelioma, CheckMate-743 and KEYNOTE-483 did not prospectively evaluate biologically distinct histological or biomarker-defined subgroups, contributing to uncertainty regarding the populations most likely to benefit. In prostate cancer, the MAGNITUDE trial illustrates how prospectively defined adjacent cohorts, hierarchical testing, and subgroup-specific futility rules can more efficiently identify patients deriving meaningful benefit. Although it is often assumed that adjacent subgroup designs require larger sample sizes, MAGNITUDE challenges this misconception, having enrolled fewer patients than PROpel or TALAPRO-2. These cases illustrate how trial design influences the interpretation of benefit-risk outcomes, particularly when studies rely on broad trial populations, open-label designs, suboptimal control arms, or insufficiently powered subgroup analyses. Future trials of combination therapies should move beyond demonstrating regimen-level efficacy. Efforts must focus on identifying which patients benefit, whether each component is needed, and whether the added toxicity, treatment burden, and costs are justified.