Introduction: The Cancer of the Prostate Risk Assessment (CAPRA) score is a validated risk assessment tool for biochemical recurrence (BCR) risk after radical prostatectomy (RP) for localized prostate cancer (PCa). The score is calculated from age, PSA at diagnosis, Gleason score pattern, % positive biopsy cores, and clinical stage by digital rectal exam (DRE). Since its publication, transrectal ultrasonography (TRUS) and magnetic resonance imaging (MRI) have increasingly been used in clinical evaluation but have yet to be integrated into PCa risk-stratification models. We used both clinical- and imaging-derived stage as CAPRA score imputations to assess the association between the base and imaging-adjusted CAPRA model and the outcome of BCR.
Methods: We retrospectively analyzed patients who underwent RP for PCa diagnosed between 2000-2017 and staged =T3a on both DRE and imaging (TRUS/MRI). DRE was treated as a binary variable (T1 vs =T2) as was imaging (=T2 vs T3a). BCR was defined as 2 consecutive PSA =0.2 ng/ml starting 8 weeks post-surgery or receipt of salvage treatment. Two CAPRA scores were computed: one with DRE-derived and one with image-based T-stage. Life table estimates were used to evaluate likelihood of BCR-free survival after RP stratified by CAPRA scores. We performed Cox regression to examine the association between clinical CAPRA components and BCR risk and to determine whether stratifying clinical T1, T2, and T3a disease in CAPRA was independently associated with BCR risk on multivariate analysis.
Results: Of 1972 patients studied, 5% had upstaging in CAPRA risk group with imaging-based clinical stage. Life table estimates were comparable between DRE- and imaging-based CAPRA score with regard to likelihood of BCR-free survival. In separate unadjusted Cox regression models, DRE-based (HR 1.57; 95% CI 1.50-1.64) and imaging-based CAPRA (HR 1.56; 95% CI 1.50-1.63) were both independently associated with risk of recurrence after RP and model fit did not differ significantly. On multivariable Cox regression adjusted for CAPRA component variables, clinical stage =T2 on DRE (HR 1.28; 95% CI 1.06-1.54) and T3a on imaging (HR 1.79; 95% CI 1.47-2.19) were independently associated with BCR risk.
Conclusions: Both DRE- and imaging-based CAPRA nomograms are associated with BCR risk after RP for PCa. Each staging modality offers independent information on the risk assessment nomogram, which supports the continued use of DRE-based clinical staging in PCa risk stratification and demonstrates the potential benefit imaging may add to models and staging standards.
Source of Funding: UCSF Goldberg-Benioff Program in Translational Cancer Biology