Interpreting International Real‐World Survival Outcomes After Avelumab First‐Line Maintenance in Advanced Urothelial Carcinoma: An
RMST
Analysis of Reconstructed Individual Patient Data
Ryo Kasahara, Teiichiro Ueki ABSTRACT
Objectives
Avelumab first‐line maintenance is an established treatment strategy for patients with advanced or metastatic urothelial carcinoma without progression after platinum‐based chemotherapy. Overall survival outcomes reported in international real‐world cohorts are difficult to compare because of differences in patient background, follow‐up duration, and later‐line treatment. We summarized these outcomes within a common time horizon.
Methods
Overall survival data were reconstructed from published Kaplan–Meier curves of five real‐world cohorts comprising 1624 patients: AVENANCE, JAVEMACS, READY, PATRIOT‐II, and a Portuguese multicenter cohort. Reconstruction was validated against published sample sizes, event counts, median overall survival, and numbers at risk. Restricted mean survival time at 24 months was pooled using random‐effects meta‐analysis. Exploratory study‐level meta‐regression described variation with carboplatin‐based first‐line chemotherapy and subsequent enfortumab vedotin use.
Results
The pooled 24‐month restricted mean survival time was 18.06 months (95% confidence interval, 17.00–19.12), with substantial heterogeneity. Ranking cohorts by published median overall survival and by restricted mean survival time gave different orders; the cohort with the longest median overall survival ranked third within the common horizon. Restricted mean survival time was lower in cohorts with a higher carboplatin treatment rate (slope, −0.0945; exact permutation p = 0.0250) and higher with a higher enfortumab vedotin treatment rate, although the latter was not supported by permutation testing ( p = 0.1250). These markers were inversely correlated.
Conclusions
Comparisons of published median overall survival across real‐world cohorts may be misleading. These study‐level findings are descriptive and hypothesis‐generating and do not support causal comparison of avelumab across cohorts.