Marc Buyse1, Stefan Michiels. 1. International Drug Development Institute, Louvain-la-Neuve, Belgium. marc.buyse@iddi.com
Abstract
PURPOSE OF REVIEW: The derivation of signatures using -omics technologies is increasingly integrated in the design of clinical trials in oncology. In this review, we investigate the clinical trial designs for the validation of prognostic and predictive signatures. RECENT FINDINGS: Using real-life breast cancer trial examples, we highlight the pros and cons of clinical utility designs for prognostic signatures. For predictive signatures, we first review alternative procedures to test the effect of treatment in the overall population as well as in the signature-positive or signature-negative subgroup. We proceed to show why the recent literature on signature-based strategy designs discourages the use of this design. We conclude by discussing adaptive signature designs to identify and validate a signature in a single trial using cross-validation techniques. SUMMARY: Use of -omics technologies should not be an add-on to clinical trials, it must become an integral part of their design.
PURPOSE OF REVIEW: The derivation of signatures using -omics technologies is increasingly integrated in the design of clinical trials in oncology. In this review, we investigate the clinical trial designs for the validation of prognostic and predictive signatures. RECENT FINDINGS: Using real-life breast cancer trial examples, we highlight the pros and cons of clinical utility designs for prognostic signatures. For predictive signatures, we first review alternative procedures to test the effect of treatment in the overall population as well as in the signature-positive or signature-negative subgroup. We proceed to show why the recent literature on signature-based strategy designs discourages the use of this design. We conclude by discussing adaptive signature designs to identify and validate a signature in a single trial using cross-validation techniques. SUMMARY: Use of -omics technologies should not be an add-on to clinical trials, it must become an integral part of their design.