Giovanni Lughezzani1, Massimo Lazzeri2, Alexander Haese3, Thomas McNicholas4, Alexandre de la Taille5, Nicolò Maria Buffi2, Nicola Fossati2, Giuliana Lista2, Alessandro Larcher2, Alberto Abrate2, Alessandro Mistretta2, Vittorio Bini6, Joan Palou Redorta7, Markus Graefen3, Giorgio Guazzoni2. 1. Department of Urology, San Raffaele Turro, Vita-Salute San Raffaele University, Milan, Italy. Electronic address: lughezzani.giovanni@hsr.it. 2. Department of Urology, San Raffaele Turro, Vita-Salute San Raffaele University, Milan, Italy. 3. Martini-Clinic Prostate Cancer Center, University Clinic Hamburg, Eppendorf Hamburg, Germany. 4. South Bedfordshire & Hertfordshire Urological Cancer Centre, Lister Hospital, Stevenage, UK. 5. Department of Urology, INSERM U955Eq07 APHP Mondor Hospital, Créteil, France. 6. Department of Internal Medicine, University of Perugia, Perugia, Italy. 7. Urologic Oncology Section of the Department of Urology and Radiology Department, Fundació Puigvert, Cartagena, Barcelona, Spain.
Abstract
BACKGROUND: External validation of a prediction tool is mandatory to assess the tool's accuracy and generalizability within different patient cohorts. OBJECTIVE: To externally validate a previously developed Prostate Health Index (PHI)-based nomogram for predicting the presence of prostate cancer (PCa) at biopsy. DESIGN, SETTING, AND PARTICIPANTS: The study population consisted of 883 patients who were scheduled for a prostate biopsy at one of five European tertiary care centers. Total prostate-specific antigen (tPSA), free prostate-specific antigen (fPSA), and [-2]pro-prostate-specific antigen (p2PSA) levels were determined. The fPSA-to-tPSA ratio (%fPSA), p2PSA, and PHI ([p2PSA / fPSA] × √tPSA) were calculated. INTERVENTION: Extended initial and repeat prostate biopsy. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Logistic regression models were fitted to test the predictors of PCa and to determine their predictive accuracy. A calibration plot was used to evaluate the extent of overestimation or underestimation between nomogram predictions and observed PCa rate. Decision curve analysis (DCA) provided an estimate of the net benefit obtained by using the PHI-based nomogram. RESULTS AND LIMITATIONS: Of 833 patients, 365 (41.3%) were diagnosed with PCa at extended prostate biopsy. In accuracy analyses, PHI was the most informative predictor of PCa (0.68), outperforming tPSA (0.51) and %fPSA (0.64). The predictive accuracy of the previously developed nomogram was 75.2% (95% confidence interval, 71.4-78.1). Calibration of the nomogram was good in patients at a low to intermediate predicted probability of PCa, while calibration was suboptimal, with a tendency to overestimate the presence of PCa, in high-risk patients. Finally, DCA demonstrated that the use of the PHI-based nomogram resulted in the highest net benefit. The main limitation of the study is the fact that only Caucasian patients were included. CONCLUSIONS: At external validation, the previously developed PHI-based nomogram confirmed its ability to determine the presence of PCa at biopsy. These findings provide further evidence supporting the potential role of the nomogram in the biopsy decision pathway for European men with suspected PCa. PATIENT SUMMARY: In the current study, we externally validated a Prostate Health Index-based nomogram to predict the presence of prostate cancer (PCa) at biopsy. This tool may help clinicians determine the need for a prostate biopsy in European patients with suspected PCa.
BACKGROUND: External validation of a prediction tool is mandatory to assess the tool's accuracy and generalizability within different patient cohorts. OBJECTIVE: To externally validate a previously developed Prostate Health Index (PHI)-based nomogram for predicting the presence of prostate cancer (PCa) at biopsy. DESIGN, SETTING, AND PARTICIPANTS: The study population consisted of 883 patients who were scheduled for a prostate biopsy at one of five European tertiary care centers. Total prostate-specific antigen (tPSA), free prostate-specific antigen (fPSA), and [-2]pro-prostate-specific antigen (p2PSA) levels were determined. The fPSA-to-tPSA ratio (%fPSA), p2PSA, and PHI ([p2PSA / fPSA] × √tPSA) were calculated. INTERVENTION: Extended initial and repeat prostate biopsy. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Logistic regression models were fitted to test the predictors of PCa and to determine their predictive accuracy. A calibration plot was used to evaluate the extent of overestimation or underestimation between nomogram predictions and observed PCa rate. Decision curve analysis (DCA) provided an estimate of the net benefit obtained by using the PHI-based nomogram. RESULTS AND LIMITATIONS: Of 833 patients, 365 (41.3%) were diagnosed with PCa at extended prostate biopsy. In accuracy analyses, PHI was the most informative predictor of PCa (0.68), outperforming tPSA (0.51) and %fPSA (0.64). The predictive accuracy of the previously developed nomogram was 75.2% (95% confidence interval, 71.4-78.1). Calibration of the nomogram was good in patients at a low to intermediate predicted probability of PCa, while calibration was suboptimal, with a tendency to overestimate the presence of PCa, in high-risk patients. Finally, DCA demonstrated that the use of the PHI-based nomogram resulted in the highest net benefit. The main limitation of the study is the fact that only Caucasian patients were included. CONCLUSIONS: At external validation, the previously developed PHI-based nomogram confirmed its ability to determine the presence of PCa at biopsy. These findings provide further evidence supporting the potential role of the nomogram in the biopsy decision pathway for European men with suspected PCa. PATIENT SUMMARY: In the current study, we externally validated a Prostate Health Index-based nomogram to predict the presence of prostate cancer (PCa) at biopsy. This tool may help clinicians determine the need for a prostate biopsy in European patients with suspected PCa.
Authors: Stacy Loeb; Sanghyuk S Shin; Dennis L Broyles; John T Wei; Martin Sanda; George Klee; Alan W Partin; Lori Sokoll; Daniel W Chan; Chris H Bangma; Ron H N van Schaik; Kevin M Slawin; Leonard S Marks; William J Catalona Journal: BJU Int Date: 2016-11-22 Impact factor: 5.588