Predicting the outcome of salvage radiotherapy for recurrent prostate cancer after radical prostatectomy

Abstract

4514 Background: Salvage radiotherapy may potentially cure patients with recurrent prostate cancer after radical prostatectomy (RP). However, the outcome is highly dependent on patient selection factors and current diagnostic modalities for locally recurrent prostate lack acceptable sensitivity and specificity. We developed a model to predict the likelihood of a durable response to salvage radiotherapy for patients with post-RP biochemical recurrence (BCR). Methods: Using multivariable Cox proportional hazards regression analysis, we modeled the clinical data for 1540 patients who received salvage radiotherapy at 1 of 17 North American tertiary referral centers for post-RP BCR. The primary endpoint was disease progression after salvage radiotherapy, defined as a serum prostate-specific antigen (PSA) value ≥ 0.2 ng/mL followed by another increase, initiation of systemic therapy, or clinical recurrence. The median follow-up was 53 months. Results: Disease progression was observed in 866 patients and the 5- and 10-year progression-free probability was 38% (95% CI, 35–40) and 19% (95% CI, 15–23), respectively. Significant variables in the nomogram were: pre-radiotherapy PSA (P < .001), RP Gleason score (P < .001), neoadjuvant androgen-deprivation therapy (P< .001), negative surgical margins (P < .004), PSA at BCR (P = .011), PSADT (P = .029), and regional lymph node metastasis (P = .034). The nomogram was accurate and discriminating with a concordance index of 0.68. Published models predicting the probability of metastasis progression or PCSM for BCR patients based on disease-free interval, prostatectomy Gleason score, and/or PSA doubling time discriminated poorly for a durable response to salvage radiotherapy among patients in our cohort (concordance index, 0.56–0.61). Conclusions: A nomogram has been developed that predicts the 5-year progression-free probability after salvage radiotherapy for men with post-RP BCR. This represents the first model to predict the outcome of salvage therapy for BCR and is useful for quantifying the anticipated benefit of salvage radiotherapy for prostate cancer recurrence. Published predictive models for metastasis progression or PCSM are of limited clinical utility in selecting patients for salvage radiotherapy. No significant financial relationships to disclose.