“Whether and to what degree surrogates predict survival is an important question in oncology,” Dr Prasad said. “In some settings the correlation is strong, and in other settings it is weak.”

Avastin (bevacizumab) was approved for breast cancer based on PFS data, but was later found not to improve OS in 3 different randomized clinical trials, and its approval was revoked, Dr Prasad said. “Many of us remember this painful episode.”

The US Food and Drug Administration (FDA) allows provisional, accelerated regulatory approvals based on surrogate endpoints like PFS, which are deemed “reasonably likely to predict” patient survival or quality of life, Dr Prasad added.

Dr Prasad and Chul Kim, MD, MPH, of the National Cancer Institute (NCI) in Bethesda, Maryland, recently reviewed FDA cancer drug approvals. They found that many drugs were approved on the basis of PFS or tumor response as surrogate endpoints, even though there existed no published studies of how well those surrogates predict OS for the indications in question.3 Only 10 (18%) of the 55 approved drugs reviewed later proved to improve OS, Dr Prasad and Dr Kim found.

“The issue is what you are trying to accomplish with a randomized controlled trial,” said Edward L. Korn, PhD, a statistician at the NCI’s Division of Cancer Diagnosis and Treatment in Rockville, Maryland. “In some situations, you want to see if the agent has enough activity to move it forward to use in an earlier setting. In that case, PFS would be sufficient.”

RELATED: PFS, Response Rate Suitable Surrogates for Overall Survival in RCTs of Soft Tissue Sarcoma

Where things get controversial, Dr Korn said, is in definitive phase 3 trials. In metastatic settings, it’s “definitely safer” to use OS as the definitive endpoint, he believes.

“Some people argue that PFS is a surrogate for OS, so that even though they’re not directly measuring patient benefit, if you had a larger trial and followed patients for longer, you would see an OS benefit,” Dr. Korn said. “That’s difficult, because that might happen or it might not happen.”

Meta-analyses can be problematic because “there tends to be a lot of variability in those trials,” he cautioned.

Dr Korn is skeptical of the argument that OS is confounded by post-progression treatments. “That’s the real world,” he said. “That’s what people are getting! I don’t really see that showing that OS isn’t useful.”


1. Ascierto PA, Long GV. Progression-free survival landmark analysis: a critical endpoint in melanoma clinical trials. Lancet Oncol. 17 Jun 2016. doi: 10.1016/S1470-2045(16)30017-1 [Epub ahead of print]

2. Flaherty KT, Hennig M, Lee SJ, et al. Surrogate endpoints for overall survival in metastatic melanoma: a meta-analysis of randomised controlled trials. Lancet Oncol. 2014;15:297-304.

3. Kim C, Prasad V. Strength of validation for surrogate end points used in the US Food and Drug Administration’s approval of oncology drugs. Mayo Clin Proc. 2016;91(6):713-725.