Concordance of preoperative prostate endorectal MRI with subsequent prostatectomy specimen in high-risk prostate cancer patients.

OBJECTIVE

Endorectal MRI (ER-MRI) may identify areas suspicious for prostate cancer. We evaluated the accuracy of ER-MRI compared with subsequent pathology specimen from prostatectomy.

MATERIALS AND

METHODS

We reviewed 309 open radical retropubic prostatectomy cases (RRP) from 2003 to 2008 to identify 94 men with a preoperative ER-MRI, which was obtained in patients with high-risk factors suspicious for local extension (Gleason grade ≥ 4+3, PSA ≥ 10 ng/ml, abnormal rectal exam, or extensive biopsy core involvement). Findings of extracapsular extension (ECE), seminal vesicle invasion (SVI), and lymphadenopathy (LAD) on ER-MRI were compared with subsequent findings on pathology specimens.

RESULTS

Ninety-four men underwent preoperative ER-MRI. No tumor was seen on ER-MRI in 9 men (10%). Of 94 ER-MRIs, 4% showed SVI, and 12% had ECE. At prostatectomy, lymph nodes were pathologically positive in 10 men, none of which were enlarged on ER-MRI. RRP was aborted in 3 of these 10 patients due to positive nodes confirmed on frozen section. Comparing ER-MRI results to subsequent prostatectomy specimen the results for accuracy, positive predictive value, negative predictive value, sensitivity, specificity were 70%, 27%, 76%, 14%, 88% for ECE and 93%, 75%, 94%, 38%, 99% for SVI. The accuracy of ECE prediction was 86% in abnormal rectal exam vs. 66% in normal exam (P < 0.05).

CONCLUSIONS

Endorectal MRI in the evaluation of high-risk prostate cancer was moderately accurate for SV involvement but inaccurate for ECE and insensitive for metastatic lymph node involvement. The predictive accuracy of ER-MRI improved in patients with an abnormal rectal exam.