Usefulness of preoperative MRI diagnosis in prostatic adenocarcinoma

Abstract

Background: We examined the reliability of an MRI diagnosis prior to radical prostatectomy for prostate cancer. Methods: A radical prostatectomy was performed in 24 patients with prostate cancer. Resected specimens were fixed and 5 mm step sections vertical to the urethra were prepared to resemble MRI images. We compared this pathological map with the preoperative MRI diagnosis which included capsular or seminal vesicle invasion and tumor localization in the prostate. We defined a new criterion for the presence of capsular invasion as a chemical shift that occurred on the rectal side on T1-weighted images 5 minutes after gadolinium (Gd) enhancement and the periprostatic venous plexus was not serial. We also examined 4 diagnostic factors of tumor localization including a low-signal intensity area detected in the peripheral zone on T2-weighted images, the presence of an enhanced area on Gd-enhanced T1-weighted images, and a low T2 with either Gd-enhanced or nonenhanced T1-weighted images. Results: The accuracy of a preoperative MRI diagnosis of capsular invasion was 16.7% using the conventional criteria, but 88.9% adding the new criterion. The accuracy of predicting seminal vesicle invasion was 63.2% in a group using a body surface coil compared to 75% in the group using an endorectal surface coil. The accuracy, positive predictive value, sensitivity and specificity of diagnosing tumor localization were 69%, 74.4%, 35.1%, and 91.8%, respectively. Conclusion: This new criterion proved superior for diagnosing capsular invasion in prostate cancer patients. Also, analysis of tumor localization in the peripheral zone demonstrated that cancer detection is increased if the low-signal intensity area is enhanced by Gd.