Biomarker discovery study design consistent with the Receiver-Operator Characteristic

The field of early biomarker discovery is characterized by a lack of consensus on the choice of statistical methodology, which may impede later progress towards clinically useful biomarkers. The Receiver-Operator Characteristic (ROC) is a de facto standard for determining the accuracy of In Vitro Diagnostic (IVD) devices. We demonstrate a biomarker discovery study design that achieves endpoint-consistency through use of ROC analysis from study objectives through sample procurement plan, sample size determination, to data analysis. Through simulations, the investigator can be informed on suitable study size to demonstrate an effect superior to the current best clinically used biomarker for the purpose. The study design is illustrated using proteomic data of newly diagnosed cancer cases and concurrent external controls, and statistically significant composite biomarkers are validated using independent data generated using the same proteomic analysis method. Intriguingly, commonly used feature selection methods do not identify the same composite biomarkers from the same data, and their selections show limited overlap with the ROC-based analysis. The proposed approach can facilitate translation of scientific discoveries into regulatory approved biomarker tests fit for use in clinical medicine.