Purpose: We compared the accuracy of endorectal magnetic resonance imaging (erMRI), transrectal ultrasound (TRUS) guided biopsy and digital rectal examination (DRE) for detecting the location of cancer in the prostate gland and seminal vesicles.
Materials and methods: This is a retrospective study of 106 consecutive patients with prostate cancer who were referred for erMRI before radical prostatectomy. Step-section pathological data and erMRI were available in 90 patients, DRE data were available on 86 and individually labeled sextant core biopsies were available in 45. T1 and T2-weighted erMRI was interpreted by a single reader, who scored the likelihood of tumor on a 5-point scale in each seminal vesicle and in 12 locations in the prostate gland. MR spectroscopy data were not used for erMRI interpretation. One pathologist reviewed whole mount serial sections of radical prostatectomy specimens. The area under ROC curves was used to evaluate accuracy.
Results: The area under ROC curves for tumor localization was higher for erMRI than for DRE at the prostatic apex (0.72 vs 0.66), mid gland (0.80 vs 0.69) and base (0.83 vs 0.69). It was likewise higher for erMRI than for TRUS biopsy in the mid gland (0.75 vs 0.68) and base (0.81 vs 0.61) but not in the apex (0.67 vs 0.70). On mixed model analysis erMRI significantly increased the accuracy of prostate cancer localization by DRE or TRUS biopsy (each p <0.0001).
Conclusions: For prostate cancer localization erMRI contributes significant incremental value to DRE or TRUS biopsy findings (each p <0.0001).