Poor correlation between progression-free and overall survival in modern clinical trials: are composite endpoints the answer?

It can be difficult to identify endpoints that accurately reflect patient benefit in metastatic solid tumors. Overall survival (OS) is the gold standard although progression-free survival (PFS) is sometimes used as a surrogate for OS. Statistical modelling has suggested that the association between OS and PFS becomes weaker in diseases with longer survival post-progression (SPP). To evaluate these statistical hypotheses we determined the relationship between PFS and OS in control and experimental arms of randomised trials conducted in the last 10years, which have led to drug approval. Our data confirm that PFS is a poor surrogate for OS when SPP is long, but it is a better surrogate where SPP is short. In cancers with short SPP designing trials to show OS benefit is feasible and, therefore, remains the preferred approach. In tumours with long SPP, PFS is not clinically meaningful unless it is also associated with improvement in patient reported outcomes such as quality of life. The oncology community should consider the further development and validation of composite endpoints including patient reported outcomes and PFS across different disease sites. Such endpoints have been successfully used in cancer trials in the past. With improvements in therapy and prolonged survival of patients with many cancers, and with increasing pressure from healthcare payers to prove that treatment leads to patient benefit, the choice of optimal endpoints for clinical trials is increasingly important. Composite measures comprising patient reported outcomes and intermediate endpoints such as PFS may be the solution and should be investigated further.