INTRODUCTION: Accurate clinical staging is critical in guiding treatment for patients with prostate adenocarcinoma. Endorectal magnetic resonance imaging (MRI) has been advocated to improve staging accuracy. In order to assess the learning curve for endorectal MRI interpretation, we compared two cohorts of patients with high-risk prostate who underwent endorectal MRI at a center with limited prior exposure to this imaging modality.
MATERIALS AND METHODS: Data for all patients who received a preoperative endorectal MRI followed by radical prostatectomy were prospectively collected. MRI was performed in patients with a high level of suspicion for extracapsular disease based on biopsy Gleason score, prostate specific antigen level, and digital rectal examination or if the Memorial Sloan-Kettering nomogram predicted a greater than 30% likelihood of extracapsular disease. The MRI results of our first 40 patients (group 1) and our second 40 patients (group 2) were compared to assess for improvement.
RESULTS: Between October 2003 and September 2005, 80 patients underwent an endorectal MRI followed by radical prostatectomy. Mean age and median PSA were 58.4 (range 43 - 74) and 6.4 (range 0.048 -115.0), respectively. MRI findings were compared to the pathological findings from the radical prostatectomy specimen. Sensitivity, specificity, positive predictive value, and negative predictive value for detection of extracapsular disease were 31.3% versus 64.7%, 70.8% versus 78.3%, 41.7% versus 68.8%, and 60.7% versus 75.0%, respectively in group 1 versus group 2. The accuracy of MRI for detecting extracapsular extension was 52.5% in group 1 compared to 72.5% in group 2.
CONCLUSIONS: In our series, endorectal MRI initially did not accurately predict tumor stage in patients with prostatic adenocarcinoma. With further experience, the accuracy of MRI substantially improved and approached the results from centers with significant experience in the interpretation of endorectal prostate MRI.