QUANTIFYING THE BENEFIT OF SALVAGE RADIATION THERAPY FOR BIOCHEMICALLY RECURRENT PROSTATE CANCER AFTER RADICAL PROSTATECTOMY

Abstract

Introduction

The natural history of biochemical recurrence (BCR) after radical prostatectomy (RP) is heterogeneous and may be quite prolonged. As such, determining the impact of salvage radiation therapy (SRT) remains challenging, particularly in the absence of prospective randomized trials with a comparator observation control arm. While “early” SRT at PSA <0.5ng/mL is supported by guidelines based on demonstrated favorable metastasis outcomes, lower PSAs also portend a more favorable prognosis for untreated BCR. Herein, therefore, we employ a time-dependent propensity score-matched analysis to quantify the oncologic benefit of SRT relative to observation for men with BCR after RP.

Methods

We queried our institutional RP registry from 1990-2017 (n=20,688) to identify patients who developed BCR (PSA≥0.2ng/mL). We performed risk-set matching using a time-dependent propensity score. The propensity to receive SRT after BCR was estimated using Cox regression, including covariates at BCR (baseline, time zero) and time-dependent covariates after BCR. Covariates in the Cox regression model were selected a priori based on suspected relationships as potential confounders, including time-independent covariates defined at surgery or at BCR (time zero). These included age at BCR, surgery year, time from surgery to BCR, Gleason score, T-stage, N-stage and margin status at the time of RP. We also included time-dependent covariates ascertained after BCR, including PSA value, log PSA, highest postoperative PSA, and count of PSA values measured since RP. SRT patients were matched to patients with BCR who did not receive SRT to compare the incidence of systemic progression. The number needed to treat (NNT) with SRT to prevent a metastasis at 10 years was calculated as well. Interaction analyses were performed to identify factors that modify the effect size of SRT on metastasis.

Results

A total of 6,881 patients developed BCR, of whom 2109 received SRT. Patients managed with SRT were younger, had higher pathologic Gleason score, and a higher incidence of positive nodes. Median follow-up after BCR was 9.8 years (IQR 5.3, 15.7) during which 947 patients developed systemic progression. After 1:1 propensity score matching (with 2109 patients per cohort), absolute standardized differences in clinicopathologic characteristics between cohorts were negligible (Table). SRT was associated with a lower risk of systemic progression compared to observation at 10 years (22% vs 29%, p<0.001) (Figure). SRT remained independently associated with a decreased risk of systemic progression on regression (HR 0.71, 95%CI 0.60–0.85, p<0.001), translating to an NNT with SRT of 14 (95%CI 11-21) to prevent one case of systemic progression at 10 years after BCR. A significant interaction was identified between PSA at SRT and the association with metastasis (p-interaction=0.002), such that SRT given at PSA<0.4 (HR 1.03, 95%CI 0.76–1.39, p=0.86) was not associated with metastasis reduction, whereas SRT given at PSA>0.4 was associated with a reduced risk of metastasis (HR 0.60, 95%CI 0.49–0.73, p<0.001).

Conclusions

We demonstrate that the majority of patients with BCR after RP will not develop systemic progression. While SRT is associated with a reduction in the risk of systemic progression, the number needed to treat was 14, consistent with the frequently indolent disease course of BCR. Furthermore, the systemic progression benefit of SRT was limited to patients with PSA > 0.4, suggesting that certain subsets of patients with BCR may not benefit from local pelvic radiation. These data support the importance of careful patient selection for SRT, in order to maximize benefit and minimize toxicity and costs from overtreatment. These findings may be utilized to inform patient counseling and shared decision-making regarding the individualized management of BCR after RP, and support testing early SRT versus observation for select patients with BCR after RP in a randomized clinical trial.