Surrogate end points for overall survival in trials of PD-(L)1 inhibitors for urinary cancers: a systematic review.

Abstract

Numerous trials for PD-(L)1 inhibitors in the management of advanced urinary cancers (urothelial carcinoma and renal cell carcinoma) were published recently. It is not known exactly what should be the optimal surrogate end point for overall survival (OS) in this context. PubMed database and ASCO meeting library were searched till August 2017. Eligible studies included prospective clinical studies evaluating PD-(L)1 inhibitors for the management of advanced urothelial carcinoma or renal cell carcinoma. The review author extracted relevant data on the characteristics of participants and the outcomes of the different studies. In order to conduct a proper correlation analysis, normality testing using Shapiro-Wilk test was first used. Based on the result of the normality testing, either Pearson's product-moment correlation or Spearman's rank correlation was used. Thirteen trials (nine urothelial carcinoma and four renal cell carcinoma trials) with 2792 participants were included. The correlation of overall response rate with median OS was very weak to weak (for urothelial carcinoma: r= -0.120, n=9, p=0.758; for renal cell carcinoma: r= -0.397, n=6, p=0.436). Likewise, the correlation of progression-free survival with median OS was very weak to weak (for urothelial carcinoma: r= -0.024, n=8, p=0.955; for renal cell carcinoma: r=0.394, n=6, p=0.440). On the other hand, 1-year survival rate may be a better surrogate end point for median OS (for urothelial carcinoma: r=0.806, n=8, p=0.016; for renal cell carcinoma: r=0.941, n=6, p=0.005). RECIST-defined overall response rate and progression-free survival are not reliable surrogate end points for median OS in trials of PD-(L)1 inhibitor therapy for urinary cancers. The use of other surrogate end points (e.g., 1-year survival) in early phase studies may be considered.