Background: A proportion of men with grade group (GG) 2 intermediate risk (IR) prostate cancer are downgraded to GG1 or harbor favorable pathology (FP, defined as GG1 or GG2 with <5% Gleason pattern 4) at radical prostatectomy (RP). Prediction of downgrading or FP may help identify potential active surveillance candidates within this group that have outcomes similar to biopsy low-risk (LR) disease.
Methods: We performed a comparative cohort study of biopsy LR and IR men who underwent RP at The Johns Hopkins Hospital and Bayview Medical Center between 2005 and 2018. We evaluated pathological outcomes at RP and recurrence-free survival (RFS). Multivariable logistic regression and Cox proportional hazards regression were applied and individual predicted probabilities were calculated.
Results: Among 2943 biopsy GG2 IR patients, 223 (7.6%) were downgraded to GG1, while 525 (17.8%) had FP; 730 of 1325 biopsy LR patients (55.1%) were upgraded (GG >1). Concordance statistics for final predictive regression models were 0.76 for downgrading and 0.70 for upgrading. Biopsy GG2 IR patients downgrading to GG1 or harboring FP had similar RFS to biopsy LR patients. A cutoff of >10% predicted probability of downgrading (24.7% of patients; hazard ratio [HR], 1.55; 95% CI, 0.89-2.68) or >20% predicted probability of FP (37.0% of patients; HR, 1.35; 95% CI, 0.81-2.24) led to similar RFS to biopsy LR patients.
Conclusion: GG2 IR patients who experience downgrading or harbor FP had similar oncologic outcomes as LR patients. The developed models may serve as tools to inform patients about the risks of pathological downgrading/upgrading and help identify a segment of GG2 IR patients who would consider pursuing active surveillance based on predicted probability cutoffs.
Keywords: logistic models; neoplasm grading; prostatic neoplasms; surgical pathology; survival analysis.
© 2020 American Cancer Society.