Defining subgroups of patients with intermediate risk prostate cancer for whom active surveillance is safe.

393 Background: A number of guideline groups recommend active surveillance (AS) for patients with low risk prostate cancer as well as selected patients with intermediate-risk prostate cancer (IRPC) – these groups are not well defined. Our group has previously shown that some patients with IRPC have high rates of progression to advanced disease. The goal of this study was to determine a subset of patients with IRPC where AS is safe. Methods: This single-institution, phase II study enrolled men with low-risk and selected IRPC patients since 1995. Patients were followed with PSA every 3-6 months, DRE every 6 months with repeat systematic biopsies at 1 year and every 3 years. Since 2013, MRI staging was done every 2 years with targeted and systematic biopsies done for new or growing lesions. Patients were followed until death or withdrawn consent. Time to treatment, metastasis-free survival (MFS), cause-specific survival (CSS) and overall survival were calculated from date of registration biopsy. Results: 1409 men were analyzed with a median follow-up of 11.0 years (range 1.2 – 26.4 years). Median age and PSA at registration biopsy were 67 years and 5.03 ng/ml. 86.4%, 11.8% and 1.7% had GG1, GG2 and GG3 disease. 72.9%, 5.8%, 7.3% and 13.6% had low, GG1 favourable IRPC (FIR), GG2 FIR and unfavourable risk ds (UIR). 996 men had at least one repeat biopsy, 768 reclassified, 604 underwent treatment, 166 had biochemical failure, 64 developed metastases and 41 died of prostate cancer. 10 year MFS was 97.4%. Absolute percentage pattern 4/5 (APP4) at baseline had a sensitivity of 36%, specificity of 86% and AUC of 64% with an optimal cutpoint of >0.30 for predicting MFS. Multivariable analysis (MVA) revealed APP4 (HR 2.86) at baseline and at repeat biopsy (vs baseline) of high risk vs UIR (HR 5.61) and high risk vs GG2 FIR (HR 10.4) were independent predictors of MFS. 10 year CSS was 98.5%. MVA revealed APP4 (>0.7, AUC 67%; HR 3.13) and high risk vs UIR (HR 34.4) were independent predictors of CSS. Conclusions: Patients with GG1 prostate cancer and those with GG2 FIR and APP4 < 0.30 appear to have very low risk of metastases and prostate cancer death and are suitable for active surveillance.