Abstract
Dear Editor,
Escario Gómez and colleagues reported valuable real-world data on first-line nivolumab plus ipilimumab in intermediate- and poor-risk metastatic renal cell carcinoma, a setting in which external validity remains clinically important. Their inclusion of patients often underrepresented in trials, such as those with poorer performance status, brain metastases, and non-clear-cell histology, strengthens the relevance of the cohort to routine practice. At the same time, several methodological features of the study warrant caution when interpreting the survival estimates and exploratory associations they present. 1
The most important concern relates to the reported association between immune-related adverse events and improved overall survival. Because immune-related adverse events arise after treatment initiation, classifying patients simply as having or not having such events can introduce immortal time bias, since patients must survive long enough to enter the exposed group. Recent methodological work has shown that this question should be addressed with landmark methods or time-dependent Cox models rather than conventional fixed-covariate analyses. 2 In the present study, the protective hazard ratio for immune-related adverse events may therefore be overstated and should be interpreted as hypothesis-generating rather than prognostic.
A second issue concerns endpoint definition. The manuscript states that tumor response was evaluated by RECIST 1.1, yet progression-free survival was defined using radiological or clinical progression in a retrospective real-world dataset. This is understandable in practice, but it is methodologically distinct from trial-based RECIST progression and should be explicitly framed as a real-world progression endpoint. Recent oncology methodology literature has emphasized that irregular assessment schedules, heterogeneous documentation, and event misclassification can materially affect the comparability of real-world progression-free survival with trial-based estimates.3,4
The handling of progression-free survival 2 also deserves clarification. Although the endpoint is defined from treatment initiation to progression or death on second-line therapy, only a subset of patients actually received subsequent systemic treatment. Without a clear statement of the analysis population and censoring rules, this estimate is difficult to reproduce and may be vulnerable to selection bias. This is especially relevant because progression-free survival 2 has a specific methodological meaning and is most interpretable when uniformly defined and consistently applied. 4
Finally, the subgroup analyses are clearly exploratory, but the small sample size and reliance on univariable Cox models limit causal interpretation. In a cohort of 43 patients, effect estimates for factors such as ECOG performance status, brain metastases, and histology are likely to be unstable and confounded by baseline severity.
These considerations do not diminish the clinical value of the dataset, but they do suggest that the reported associations should be framed more cautiously and that clearer endpoint definitions would strengthen the manuscript's validity and interpretability.
