One of the topics at EADO 2015 was "Which endpoint for which trial?" regarding whether progression free survival (PFS) or overall survival (OS) should be the primary endpoint for a clinical trial. OS is regarded as the golden standard primary endpoint in oncology clinical trials but comes at a cost with longer trials, larger patient population (harder to recruit) and that the OS may be affected by crossover design trials (which lets patients to continue with another treatment/study arm if the first one didn't work). Clinicians want their numbers right, but but who wants to stay on an inferior arm waiting to die just to prove that the other stuff is better? Not me.
The speaker Prof. Dr. Axel Hauschild showed in his talk a retrospect study that the correlation between PFS and OS is very high, with an R-value of 0.95 (1.0 = absolute correlation) so why don't we use PFS as the new "golden standard"?
The retrospect study was done on melanoma trials and according to http://www.cancerworld.org/pdf/8995_pagina_39_43_eGranround.pdf (posted on the MPNE forum) PFS as primary endpoint cannot be used for every single cancer.