Stratified medicine: methods for evaluation of predictive biomarkers

Background: Stratified medicine was defined as the use of biomarkers to select patients more likely to respond to a treatment or experience an adverse event. Alms: To investigate the hypothesis that there is a mismatch between the theoretical proposals and practice of predictive biomarker research, focusing on the clinical utility stage. Methods: Methodological research was identified in a systematic review of frameworks for staged evaluation of predictive biomarkers. Actual research supporting 50 real cases identified in European Medicines Agency licensing was analysed. A case study of recent research into ERCC l in non-small cell lung cancer was undertaken. Existing discrepancies between the theory and practice were identified and possible reasons and consequences of these were discussed. Findings: A mismatch between theory and practice was identified. It appeared to be a result of both the practice not following some theoretical requirements, and the underdevelopment of methodology for certain situations. Areas of clinical research with insufficient relevant methodology were identified. Conclusions: The major research priorities identified in this thesis were development of a clear hierarchy of biomarker research designs and development of methodology related to the biomarker threshold