A comparison of prostate cancer prediction models in men undergoing both magnetic resonance imaging and transperineal biopsy: Are the models still relevant?

Abstract

To externally validate and compare the performance of the European Randomized Study of Screening for Prostate Cancer risk calculator 3/4 (ERSPC-RC3/4), the Prostate Biopsy Collaborative Group risk calculator (PBCG-RC) and the van Leeuwen model to determine which prediction model would perform the best in a contemporary Australian cohort undergoing transperineal (TP) biopsy. Materials and MethodsA retrospective review identified all patients undergoing TP biopsy across two centres. Of the 797 patients identified, 373 had the data required to test all three risk calculators. The probability of high-grade prostate cancer, defined as International Society of Urological Pathology Grade Group >1, was calculated for each patient. For each prediction model discrimination was assessed using area under the receiver-operating characteristic curve (AUC), calibration using numerical and graphical summaries, and net benefit using decision curve analysis. ResultsAssessment of model discrimination for detecting high-grade prostate cancer showed AUCs of 0.79 (95% confidence interval [CI] 0.74-0.84) for the ERSPC-RC3/4, 0.81 (95% CI 0.77-0.86) for the van Leeuwen model, and 0.68 (95% CI 0.63-0.74) for the PBCG-RC, compared to 0.58 (95% CI 0.52-0.65) for prostate-specific antigen alone. The ERSPC-RC3/4 was the best calibrated in the moderate-risk range of 10-40%, whilst the van Leeuwen model was the best calibrated in the lowrisk range of 0-10%. The van Leeuwen model demonstrated the greatest net benefit from 10% risk onwards, followed closely by the ERSPC-RC3/4 and then the PBCG-RC. ConclusionThe ERPSC-RC3/4 demonstrated good performance and was comparable to the van Leeuwen model with regard to discrimination, calibration and net benefit for an Australian population undergoing TP prostate biopsy. It is one of the most accessible risk calculators with an easy-to-use online platform, therefore, we recommend that Australian urologists use the ERSPC-RC3/4 to predict risk in the clinical setting.

External validation of two mpMRI-risk calculators predicting risk of prostate cancer before biopsy

“…This manuscript is the first external validation study testing RC-R and RC-A in two centers within a "real-life" scenario. Contrary to previously published validation studies [11][12][13][14][15][16][17][18], our validation cohort had significant heterogeneity. We included patients with mpMRIs not performed by dedicated uro-radiologists but multiple radiologists with differing degrees of expertise.…”

“…Most recently, various research groups implemented mpMRI findings into the ERSPC-RC. Among these, the RCs of Radtke et al [9] and Alberts et al [10] showed the most promising results in internal and external validation studies [11][12][13][14][15][16][17][18]. As mpMRI quality is highly dependent on the radiologist's experience, among other factors, it is unclear whether these RCs accurately predict the risk of PCa and csPCa when including mpMRI reports not performed by dedicated uro-radiologists.…”

Accommodating heterogeneous missing data patterns for prostate cancer risk prediction

“…We compared six methods that have been proposed for handling missing data with the objective of finding the method most likely to perform well in multiple external validation studies of a globally accessible online risk tool. As with all online risk tools, online publication of the original PBCG continues to result in published external validation studies providing evidence for or against its generalizability to other populations, particularly in comparison to other published tools [4,[21][22][23][24][25][26]. To date, by exclusion of prostate volume, the original PBCG tool has competed less favorably with the other tools incorporating this information.…”

Globally accessible end-user-friendly prostate cancer risk prediction tools based on contemporary cohorts with heterogeneous missing risk factors

“…We compared six methods that have been proposed for handling missing data with the objective of nding the method most likely to perform well in multiple external validation studies of a globally accessible online risk tool. As with all online risk tools, online publication of the original PBCG continues to result in published external validation studies providing evidence for or against its generalizability to other populations, particularly in comparison to other published tools [21,22,23,24,25,26,27]. To date, by exclusion of prostate volume, the original PBCG tool has competed less favorably with the other tools incorporating this information.…”