Multiparametric Magnetic Resonance Imaging Is an Independent Predictor of Salvage Radiotherapy Outcomes After Radical Prostatectomy

Background: The Stephenson nomogram is widely used to estimate the success of salvage radiotherapy (sXRT) for prostate cancer (PCa) recurrence after radical prostatectomy (RP). Objective: To determine whether multiparametric pelvic magnetic resonance imaging (mpMRI) performed for biochemical recurrence after RP improves prognostication of sXRT relative to the Stephenson nomogram. Design, setting, and participants: Men undergoing RP at our institution from 2003 to 2012 who had biochemical recurrence evaluated by mpMRI within 12 mo of sXRT were retrospectively reviewed. Exclusion criteria included PCa treatment prior to RP, adjuvant XRT after RP, salvage cryotherapy before sXRT, and hormone refractory disease prior to sXRT. Outcome measurements and statistical analysis: Multivariable Cox regression analyses (adjusting for Stephenson nomogram covariates) associated mpMRI findings with prostate-specific antigen (PSA) recurrence and metastasis after sXRT. The mpMR images were compared in a binary fashion: no lesion versus vesicourethral/seminal vesical bed/prostate fossa lesions. Results and limitations: Among 473 sXRT patients, 57%(204) had lesions on mpMRI: 26%(124) vesicourethral, 28%(135) seminal vesical bed/prostatic fossa, 7%(34) nodal, and 1%(3) bone. Median PSA at mpMRI with lesions was 0.46 versus 0.40 ng/ml without lesions. After excluding nodal/bone lesions, 29% of men developed PSA recurrence and 14% metastasis (median follow-up 45 mo after sXRT). For patients with a pre-sXRT PSA of ≤0.5 ng/ml, negative mpMRI was associated with increased PSA recurrence (39% vs 12%, p < 0.01) and metastasis (16% vs 2%, p < 0.01) at 4 yr after sXRT. For patients with a PSA of ≤0.5 ng/ml, the addition of mpMRI to the propensity score (created using variables from the original Stephenson nomogram) improved the c-statistic from 0.71 to 0.77 for PSA recurrence (hazard ratio [HR] 3.60, p < 0.01) and from 0.66 to 0.77 for metastasis (HR 6.68, p < 0.01). Limitations include evolutions in MRI technique and lack of a cohort of men undergoing mpMRI electing against sXRT. Conclusions: Pre-sXRT mpMRI improves clinicopathologic variables to estimate sXRT success, particularly in the early sXRT setting. Patient summary: Men who have biochemically recurrent prostate cancer after radical prostatectomy often receive salvage radiotherapy. In our study, multiparametric pelvic magnetic resonance imaging prior to salvage radiotherapy was a significant predictor of prostate-specific antigen failure and metastasis after radiotherapy. Multiparametric magnetic resonance imaging (mpMRI) before salvage radiotherapy is a valuable prognostic adjunct to current clinicopathologic variables. Negative mpMRI was an independent risk factor for prostate-specific antigen (PSA) recurrence and metastasis for patients undergoing salvage radiotherapy with a PSA of <0.5 ng/ml.