Nomogram for the prediction of high-grade prostate cancer on ultrasound-guided needle biopsy

Abstract

4621 Background: Prostate cancer is characterized by a wide spectrum of biologic heterogeneity that results in divergent clinical outcomes. Thus, not all patients benefit from aggressive therapy. Tumor grade is the most important of the clinicopathologic factors that predict tumor aggressiveness and therefore drive decisions regarding aggressive local therapy vs. observation. The goal of this study was to develop a nomogram for the prediction of high-grade prostate cancer (defined as Gleason ≥ 7) using pre-biopsy factors. Methods: Data were prospectively collected on 1,699 referred men with a serum PSA of ≤ 10 ng/ml who underwent a prostate biopsy (minimum 6 cores). Variables analyzed included: age, race, family history, digital rectal exam (DRE), prostate-specific antigen (PSA), PSA density (PSAD), prostate volume, PSA doubling time (PSADT), and ultrasound (US) findings. Thirty percent of the data were randomly reserved for study validation. Logistic regression analysis was performed to estimate the relative risk (RR) and 95% confidence intervals (CI). Results: The cohort was characterized as follows (values: median ± SD): age = 66.0 ± 7.5 years, PSA = 5.0 ± 2.5 ng/ml/cc, PSAD = 0.12 ± 0.10 ng/ml/cc, volume = 35.0 ± 23.6 cc. High-grade prostate cancer was detected in 157 subjects (9.2%). The results of the multivariate analysis are shown in the table and a user-friendly nomogram was constructed based on these results. Using the independent validation set the area under the ROC was 0.74 for the model. Conclusions: The presence of high-grade prostate cancer was successfully predicted using a multi-factor logistic regression-based model. Models such as this have the potential to significantly reduce the number of biopsy procedures performed when subsequent therapy is contingent upon the detection of high-grade prostate cancer. No significant financial relationships to disclose.