Consistent predictive ability of prostate‐specific antigen density prediction model for clinically significant prostate cancer across age strata

Abstract

BackgroundPrebiopsy prostate‐specific antigen density (PSAD) is a well‐known predictor of clinically significant prostate cancer (csPCa). Since prostate‐specific antigen (PSA) and prostate volume (PV) increase normally with aging, PSAD thresholds may vary. The purpose of the study was to determine if PSAD was predictive of csPCa in different age strata.MethodsWe retrospectively reviewed our institutional database for patients who underwent multiparametric magnetic resonance imaging (MRI) between January 2016 and December 2021. We included patients who had post‐MRI prostate biopsies. Based on age, we divided our cohort into four subgroups (groups 1–4): <55, 55–64, 65–74, and ≥75 years old. PSAD accuracy was estimated by the area under the curve (AUC) as a predictive model for differentiating csPCa between the groups. CsPCa was defined as a Gleason Grade Group 2 or higher. Three different PSAD thresholds (0.1, 0.15, and 0.2) were tested across the groups for sensitivity, specificity, and positive predictive value (PPV) and negative predictive value (NPV). Chi‐square and analysis of variance tests were used for bivariate analysis. All analys were completed using R 4.3 (R Core Team, 2023).ResultsAmong 1913 patients, 883 (46.1%) had prostate biopsies. In groups 1, 2, 3, and 4, there were 62 (7%), 321 (36.4%), 404 (45.8%), and 96 (10.9%) patients, respectively. Median PSA was 5.6 (interquartile range 3.4–8.1), 6.2 (4.8–9), 6.8 (5.1–9.7), and 9 (5.6–13), respectively (p < 0.01). Median PV was 42.3 (30–62), 51 (36–77), 55.5 (38–85.9), and 59.3 (42–110) mL, respectively (p < 0.01). No difference was observed in median PSAD between age groups 1–4 (0.1 [0.07–0.16], 0.11 [0.08–0.18], 0.1 [0.07–0.19], and 0.1 [0.07–0.2]), respectively (p = 0.393). CsPCa was diagnosed in 241 (27.3%) patients, of which 10 (16.1%), 65 (20.2%), 121 (30%), and 45 (46.7%) were in groups 1–4, respectively (p < 0.001). For groups 1–4, the PSAD AUC for predicting csPCa was 0.75, 0.68, 0.71, and 0.74. While testing PSAD threshold of 0.15 across the different age groups (1–4), the PPV vs. NPV was 39.1 vs. 93.2, 33.6 vs. 87, 50.9 vs. 80.8, and 66.1 vs. 64.7, respectively.ConclusionsPSAD prediction model was found to be similar among different age groups. In young patients, PSAD had a high NPV but low PPV. With increasing age, the opposite trend was observed, likely due to higher disease prevalence. While PSAD thresholds may be less useful in older patients to rule out higher‐grade prostate cancer, the clinical consequences of these diagnoses require a case‐by‐case evaluation.