Biomarkers and surrogate end points—the challenge of statistical validation

摘要: Biomarkers and surrogate end points have great potential for use in clinical oncology, but their statistical validation presents major challenges, few biomarkers been robustly confirmed. Provisional supportive data prognostic biomarkers, which predict the likely outcome independently of treatment, is possible through small retrospective studies, it has proved more difficult to achieve robust multi-site validation. Predictive response patients specific treatments, require extensive validation, specifically large randomized trials meta-analysis. Surrogate are even challenging validate, demonstrating both that true point effect treatment on reliably predicts its point. In this Review, we discuss nature predictive points, examine techniques designs required cases where requirements cannot be rigorously achieved, biological plausibility an or might support adoption. No consensus yet exists processes standards pragmatic evaluation adoption absence