Statistical Evaluation of Markers and Risk Tools for Prostate Cancer Classification and Prediction.

Adopting novel biomarkers for use in prostate cancer screening requires rigorous scientific evaluation. The predictive accuracy of the markers must be quantified and compared with other potential markers. In this chapter we focus on the statistical approaches commonly used for evaluating biomarkers in the context of early detection for prostate cancer. We cover statistical methods for estimating accuracy summaries for both disease classification and risk prediction, including the true positive fraction (TPF), false positive fraction (FPF), positive predictive value (PPV), negative predictive value (NPV), receiver-operating characteristic (ROC) curve, and predictiveness curve. We also provide methods for combining multiple biomarker tests and comparing biomarkers. An example from the San Antonio Center of Biomarkers Of Risk for Prostate Cancer (SABOR) cohort is used to illustrate these methods. [ABSTRACT FROM AUTHOR]