Can Genomic Classifiers Be Used To Identify Those Appropriate For Observation Despite High Risk Pathologic Features Following Radical Prostatectomy?

Three prospective randomized trials have been performed demonstrating improvement in biochemical progression free survival (BPFS) with adjuvant radiation (PORT) following radical prostatectomy (RP) when adverse pathologic features (T3/positive margins) are found. However, defaulting to PORT in all patients would lead to overtreatment as demonstrated in ARO 96-02 where 54% of patients in the observation arm were free of biochemical progression at 5 years. We describe our evaluation of Decipher genomic testing to select patients appropriate for observation despite adverse pathologic features. Beginning in 2014, patients at Thomas Jefferson University with adverse pathologic features and undetectable postoperative PSA underwent Decipher genomic testing. Those with low to intermediate Decipher scores were offered observation with salvage radiation as needed. Biochemical recurrence (BCR) was the primary endpoint with recurrence defined as PSA ≥0.1 ng/ml. A matched analysis was also performed from a retrospective database. Eligible patients underwent RP for localized disease and did not receive adjuvant therapy with radiation or hormonal therapy. Patients were matched by categorical age, categorical pre-operative PSA, Gleason grade groups, pathological stage, and surgical margins. BCR in this cohort was defined as PSA ≥0.2 ng/ml. The prospective cohort included 39 patients after excluding 4 for adjuvant radiation and 2 for missing pathologic information. The median age and preoperative PSA were 62.8 years and 6.4 ng/ml respectively. Grade group 2 and 3 comprised 51.3% and 35.9% of patients. 66.7% of patients were pT3a, 10.3% were pT3b, and 43.6% had positive margins. The median follow up was 16 months. Cumulative incidence of BCR at 36 months was 5.1%. The matched retrospective cohort included 390 patients. The median age and preoperative PSA were 60 years and 6.8 ng/ml respectively. Grade group 2 and 3 comprised 47.2% and 28.5% of patients. 51.8% were pT3a, 22.6% were pT3b, and 47.9% had positive margins. The median follow up was 15 months. The cumulative incidence of BCR at 3 years based on Decipher scoring was 22% for low risk, 20% for intermediate risk, and 36% for high risk. In our prospective cohort of patients we used Decipher genomic testing to stratify men with undetectable PSA and adverse pathologic feature following prostatectomy to observation as opposed to adjuvant radiation. Despite the presence of adverse pathologic features, we achieved low rates of BCR at 3 years, 5.1%, in those with low or intermediate risk decipher scores. In our matched retrospective cohort the 3 year cumulative incidence of BCR was 22% and 20% for those with low and intermediate risk decipher scores respectively. Patients with high risk decipher scores experienced a 3 year cumulative incidence of BCR of 36%. Our findings demonstrate a potential use of genomic classifiers in improving patient selection and reducing overtreatment in the post RP setting.