The validation of predictive and prognostic biomarkers and surrogate end points requires robust statistical analysis of data gathered from multiple, large, independent studies. In this Review, Marc Buyse and coauthors discuss this validation process and the nature of biomarkers and surrogate end points. Furthermore, they consider strategies for the pragmatic evaluation of biomarkers and surrogate end points in the absence of statistical validation.
- Marc Buyse
- Daniel J. Sargent
- Aimery de Gramont