Should Oncology Drug Trials Rely Less on Overall Survival?
Calls for wider adoption of surrogate survival measures as clinical trial endpoints should be considered cautiously.
Overall survival (OS) is the gold standard measure of treatment effectiveness in clinical trials, though some researchers believe that progression-free survival (PFS) should be more widely used as a surrogate endpoint for OS.
Accrual of OS data in a phase 3 clinical trial can take years, noted Dr Paolo Ascierto and Prof Georgina Long, of the Melanoma, Cancer Immunotherapy and Innovative Therapy Unit, Instituto nazionale Tumori Fondazione “G Pascale” in Naples, Italy.
PFS is a “proven surrogate endpoint for overall survival,” they wrote in a commentary recently published in The Lancet Oncology.1 PFS at pre-specified points in time—at 1-, 2- and 3-year “landmarks”—capture information about tumor control and time to disease progression, and such landmarks are easier for patients to understand than statistical measures, such as hazard ratios. They advocate a “shift in emphasis” to landmark PFS.
Late side effects, however, might not be captured by landmark PFS, noted Jason N. Doctor, PhD, director of health informatics at the Leonard D. Schaffer Center for Health Policy and Economics, and associate professor in the School of Pharmacy, University of Southern California in Los Angeles.
“Treatment damage to other organs, such as the heart, might not show up in PFS outcome analysis, but could affect the patient's [overall] survival duration,” Dr Doctor noted. “PFS cannot stand alone as an outcome.”
Dr Ascierto and Prof Long were among the coauthors of a 2014 meta-analysis of data from randomized clinical trials of advanced melanoma treatment, in which dacarbazine served as a control. The analysis showed that PFS is a “strong surrogate for overall survival.”2
The meta-analysis, however, had important limitations, such as addressing only the PFS prediction of OS for a narrow patient population, noted Vinay Prasad, MD, MPH, assistant professor of medicine at the Knight Cancer Institute, Oregon Health & Science University in Portland.
“These results cannot be extrapolated to trials where the control arm does not receive dacarbazine,” Dr Prasad told Cancer Therapy Advisor. “In the era of immunotherapy, for trials that do not use dacarbazine as a comparator, there is no good evidence that PFS is a faithful surrogate for OS.” The meta-analysis did not, furthermore, include unpublished study data.
The initial statistical meta-analysis was “not just industry-sponsored,” but performed by an employee of GlaxoSmithKline, Dr Prasad said. “When it comes to surrogate validation studies, the amount of industry profits from a positive outcome is tremendous, and ways to manipulate the analysis are many, so this is an area where we would want to see non-conflicted studies.”