Optimal threshold of the prostate health index in predicting aggressive prostate cancer using predefined cost–benefit ratios and prevalence

Abstract

© 2019, Springer Nature B.V. Purposes: The aim of the study was to determine optimal threshold of the Prostate Health Index (Phi) for predicting aggressive prostate cancer (PCa), taking into account misclassification costs, prevalence, and plausible risk factors. Methods: This prospective cohort study analyzed patients undergoing prostate biopsy and Phi testing. The primary endpoint was aggressive PCa, defined as biopsy Gleason score ≥ 7. The data about age, total prostate-specific antigen (PSA), percentage of free PSA (%fPSA), and digital rectal examination (DRE) were extracted from the patient files. We divided the patients to the low- and high-risk group. The clinical usefulness of the Phi was assessed by the decision curve analysis. The predictive performance was assessed using the area under the receiver operating characteristic curve (AUC), per-class metrics, and the potential reduction in unnecessary biopsies. The uncertain interval of Phi values was also determined. Results: There were 200 men included in the study, 35 (17.5%) of them having aggressive PCa. Important predictors of aggressive PCa were %fPSA, DRE, Phi, and belonging to the high-risk group. With optimal threshold of 30.7, about 32% unnecessary biopsies would be avoided. The optimal threshold of Phi was lower in the high-risk group than in the low-risk group. The AUC for detection of aggressive PCa was 0.791. Per-class metrics showed that the Phi has insufficient diagnostic accuracy. The lower and upper limits of the uncertain interval were 41.8 and 51.4, respectively. Conclusion: Different thresholds of the Phi could be optimal, depending on prevalence, patient characteristics, and misclassification costs. Further studies with a larger patient sample are necessary to confirm our conclusions.