Author Interviews, Cancer Research, JAMA, NIH / 22.06.2015
Oncology Evidence Based On Surrogate Survival May Be Poor
MedicalResearch.com Interview with:
Vinay Prasad, MD, MPH
Medical Oncology Service, National Cancer Institute
National Institutes of Health
Bethesda, Maryland
MedicalResearch: What is the background for this study? What are the main findings?
Dr. Prasad: In medicine, there are two types of endpoints: clinical endpoints and surrogate endpoints. Clinical endpoints, such as survival or quality of life, measure how a patient, feels, functions or lives. In contrast, a surrogate endpoint is not a measure of patient benefit. Instead, it is merely hoped to correlate with one. LDL levels are a surrogate for cardiovascular risk, for instance.
Oncologists use and trust surrogate endpoints, such as response rate, progression free survival and disease free survival. The majority of drug approvals and many guideline recommendations are based on improvements in surrogates. Surrogates are assumed to correlate with overall survival, but we wanted to know if this was true, and under what circumstances.
We reviewed all well done studies of surrogate-survival association. We found that the majority--especially in the setting of metastatic disease--found a poor correlation between a surrogate and survival. In fact, correlations were strong in only a handful of settings, such as adjuvant colorectal cancer. Moreover, we found that correlations were always based on a subset of potentially informative literature, even when authors surveyed unpublished trials. Missing data in these association studies raises the concern that correlations would be different if all data had been considered.
Our overall conclusion was that most surrogate-survival correlations in oncology are based on weak evidence and are poor.





















