Abstract PD-1/PD-L1 immune checkpoint inhibitor (ICI)-based combination regimens have transformed metastatic urothelial cancer treatment. However, phase 3 trials of these regimens have shown varied results. We sought to understand these differences by assessing whether these combinations are synergistic, additive, or less-than-additive, which is crucial for translational research and clinical development. We systematically identified five randomized phase 3 trials of ICI-based combinations for metastatic urothelial carcinoma and tested whether progression-free survival (PFS) distributions of combinations were as predicted by a null hypothesis of additivity, defined as the sum of PFS durations of individual drugs. For KEYNOTE-361 and IMvigor130, which allowed cisplatin or carboplatin, we evaluated platinum-specific predictions. We found that PFS distributions in trials adding PD-1/PD-L1 ICI to enfortumab vedotin (EV-302), gemcitabine plus cisplatin (CheckMate 901), or CTLA-4 blockade (DANUBE) were statistically indistinguishable from additivity. In trials using choice of cisplatin or carboplatin (KEYNOTE-361, IMvigor130), ICI combinations with cisplatin-gemcitabine yielded additive PFS, whereas ICI combinations with carboplatin-gemcitabine yielded less-than-additive PFS, indicating that carboplatin blunts the potential benefits of PD-1/PD-L1 inhibitors. The predictable success of EV-302 and CheckMate 901 demonstrate that substantial improvements in clinical outcomes can be achieved by additive combinations of individually effective agents. Overall, modeling PFS from single-agent data could have anticipated most contemporary phase 3 trial outcomes, potentially informing rational trial design and prioritization.
Schlachter et al. (Fri,) studied this question.