Abstract
Dear Editor,
We would like to thank the authors for their thoughtful and constructive comments on our real-world study evaluating first-line nivolumab plus ipilimumab in patients with intermediate- or poor-risk metastatic renal cell carcinoma. We appreciate their careful methodological assessment and agree that several aspects deserve clarification to ensure an appropriate interpretation of our findings.
First, we agree that the observed association between immune-related adverse events and improved overall survival should be interpreted with caution. In our study, this analysis was exploratory and based on a conventional univariable Cox model. Because immune-related adverse events occur after treatment initiation, we acknowledge that this association may be affected by immortal time bias. Therefore, our findings should not be interpreted as demonstrating a causal or definitive prognostic relationship, but rather as hypothesis-generating. Future studies with larger cohorts should address this question using landmark analyses or time-dependent Cox models.
Second, regarding progression-free survival, we agree that the endpoint used in our study should be understood as a real-world progression endpoint. Although tumor response was assessed according to RECIST 1.1 when radiological evaluation was available, PFS was defined as time from treatment initiation to radiological or clinical progression or death, reflecting routine clinical practice in a retrospective setting. We acknowledge that this differs from strictly protocol-defined RECIST progression in clinical trials and may limit direct comparability with trial-based PFS estimates.
Third, we appreciate the opportunity to clarify the interpretation of PFS2. In our cohort, 30 patients experienced progression, one was lost to follow-up, and 16 of the remaining patients received subsequent systemic therapy. Accordingly, the PFS2 estimate should be interpreted in the context of this selected subgroup and may be affected by selection bias. We agree that clearer reporting of the analysis population and censoring rules would improve reproducibility and interpretability.
Finally, we agree that the subgroup analyses should be considered exploratory. The limited sample size, the single-center retrospective design, and the use of univariable Cox models preclude causal interpretation. In particular, estimates for variables such as ECOG performance status, brain metastases, histology, and immune-related adverse events may be unstable and influenced by baseline clinical severity. For this reason, these analyses were intended to generate hypotheses and to describe patterns observed in routine practice, rather than to provide definitive prognostic conclusions.
Overall, we agree that the methodological issues raised by the authors are relevant and should temper the interpretation of our results. Nevertheless, we believe that our study contributes useful real-world evidence by describing outcomes in a less selected population than those typically included in pivotal trials, including patients with poorer performance status, brain metastases, and non-clear-cell histology. We thank the authors again for their comments, which help contextualize and strengthen the interpretation of our findings.
