Progression-free survival as a surrogate for overall survival: a new paradigm?
PURPOSE: In cancer, it is vital to develop and validate surrogate endpoints so that long-term and costly Phase III trials for comparing treatments can be avoided.
METHODS: We propose development of a new framework for assessing surrogacy in which the surrogate endpoint is considered to be validated if two criteria are satisfied: (1) the surrogate endpoint has to have positive probability of occurring before the true endpoint, conditional on treatment group; (2) the treatment effect on the surrogate endpoint, as part of a composite outcome with the true endpoint, must be associated with the treatment effect on the true endpoint more than expected by chance.
RESULTS: Applying this to two cancer clinical trial datasets, we show that complete remission is a valid surrogate for survival in acute myelogenous leukemia, and time to progression is a valid surrogate for survival in advanced colorectal cancer.
CONCLUSION: This new surrogacy paradigm will lead to surrogate endpoints that are composite outcomes that will occur more quickly than observing the true endpoint only. At the same time, treatment effects on these combined outcomes will be informative about those on the true endpoint.
Debashis Ghosh. "Progression-free survival as a surrogate for overall survival: a new paradigm?" Technical Report, Department of Statistics, Penn State University (2008).
Available at: http://works.bepress.com/debashis_ghosh/28