Prostate health index (PHI) and prostate-specific antigen (PSA) predictive models for prostate cancer in the Chinese population and the role of digital rectal examination-estimated prostate volume

• Published in: International urology and nephrology 2016-10, Vol.48 (10), p.1631-1637
• Main Authors: Chiu, Peter K. F., Roobol, Monique J., Teoh, Jeremy Y., Lee, Wai-Man, Yip, Siu-Ying, Hou, See-Ming, Bangma, Chris H., Ng, Chi-Fai
• Format: Article
• Language: English
• Subjects: Aged; China; Digital Rectal Examination - methods; Feasibility Studies; Health Status Indicators; Humans; Male; Medicine; Medicine & Public Health; Middle Aged; Nephrology; Organ Size; Prostate - pathology; Prostate-Specific Antigen - analysis; Prostatic Neoplasms - diagnosis; Risk Assessment - methods; ROC Curve; Urology; Urology - Original Paper
• ISSN: 0301-1623; 1573-2584
• DOI: 10.1007/s11255-016-1350-8

Purpose To investigate PSA- and PHI (prostate health index)-based models for prediction of prostate cancer (PCa) and the feasibility of using DRE-estimated prostate volume (DRE-PV) in the models. Methods This study included 569 Chinese men with PSA 4–10 ng/mL and non-suspicious DRE with transrectal ultrasound (TRUS) 10-core prostate biopsies performed between April 2008 and July 2015. DRE-PV was estimated using 3 pre-defined classes: 25, 40, or 60 ml. The performance of PSA-based and PHI-based predictive models including age, DRE-PV, and TRUS prostate volume (TRUS-PV) was analyzed using logistic regression and area under the receiver operating curves (AUC), in both the whole cohort and the screening age group of 55–75. Results PCa and high-grade PCa (HGPCa) was diagnosed in 10.9 % (62/569) and 2.8 % (16/569) men, respectively. The performance of DRE-PV-based models was similar to TRUS-PV-based models. In the age group 55–75, the AUCs for PCa of PSA alone, PSA with DRE-PV and age, PHI alone, PHI with DRE-PV and age, and PHI with TRUS-PV and age were 0.54, 0.71, 0.76, 0.78, and 0.78, respectively. The corresponding AUCs for HGPCa were higher (0.60, 0.70, 0.85, 0.83, and 0.83). At 10 and 20 % risk threshold for PCa, 38.4 and 55.4 % biopsies could be avoided in the PHI-based model, respectively. Conclusions PHI had better performance over PSA-based models and could reduce unnecessary biopsies. A DRE-assessed PV can replace TRUS-assessed PV in multivariate prediction models to facilitate clinical use.