Accurate staging at the time of prostate cancer diagnosis is critically important because treatment strategy depends directly on whether cancer is confined within the prostate capsule or has spread beyond it. Patients with localized disease are typically candidates for radical prostatectomy with curative intent, while those with extraprostatic extension (EPE) or seminal vesicle invasion (SVI) are more often managed with radiation therapy. Misclassifying an advanced tumor as localized exposes patients to surgery whose benefits may be offset by serious complications including urinary incontinence and erectile dysfunction.
Despite its clinical importance, local staging remains challenging. More than 30% of men with clinically localized prostate cancer are found at surgical pathology to have extraprostatic extension -- meaning staging tools routinely underestimate disease extent. Computed tomography and standard MRI have not proven sufficiently sensitive for routine local staging. Over two decades, endorectal MRI, which uses a coil inserted into the rectum to provide high-resolution images of the prostate and surrounding structures, emerged as the most promising imaging approach for local staging.
This Korean study from 2010 aimed to assess the clinical value of endorectal MRI in a systematic evaluation of 54 patients, examining performance for both extraprostatic extension and seminal vesicle invasion, and investigating whether sensitivity varied by tumor grade (Gleason score) and PSA level.
This retrospective study enrolled 54 patients with clinically localized prostate cancer who underwent radical prostatectomy at a single Korean academic center. Endorectal MRI was performed at least 3 weeks after biopsy for all patients -- a waiting period required to allow post-biopsy hemorrhage to resolve, which can otherwise mimic cancer on MRI. Patients who received neoadjuvant hormonal therapy after MRI were excluded.
Imaging used a 1.5 Tesla GE MRI scanner with a combined endorectal coil and phased-array surface coil configuration. The endorectal coil was inserted rectally and positioned adjacent to the prostate, providing high signal-to-noise ratio for detailed anatomical visualization. Sequences included axial and coronal T2-weighted fast spin echo images, T1-weighted sequences, and gadolinium contrast-enhanced dynamic images. The field of view was 14 cm, optimized for prostate visualization.
Radiologists used established MRI criteria to identify extraprostatic extension: localized bulge of the prostatic contour, thickening or disruption of the prostatic capsule, infiltrative stranding in the periprostatic fat, or asymmetry of the neurovascular bundle. Seminal vesicle invasion was diagnosed based on abnormal low-signal-intensity tissue within the seminal vesicle or seminal vesicle dilatation with asymmetry. All interpretations were made by consensus of two radiologists and compared to the gold standard of surgical pathology.
Of 54 patients, surgical pathology identified 8 (14.8%) with extraprostatic extension. Endorectal MRI suspected extraprostatic extension in 12 patients. Overall performance was sensitivity 50.0%, specificity 82.6%, and accuracy 77.8%. The positive predictive value was 33% and the negative predictive value was 90%, indicating the test is more useful for ruling out extraprostatic spread than for confirming it.
A key finding was that sensitivity improved markedly with increasing Gleason score. Among patients with Gleason score less than 7, sensitivity was 0% (all missed). For Gleason score 7, sensitivity rose to 67%. For Gleason score above 7, all cases were detected (sensitivity 100%). This progressive improvement in detection with higher tumor grade was statistically significant by chi-square trend analysis (p=0.034), confirming a genuine dose-response relationship between tumor grade and MRI detectability.
PSA level also showed a trend toward better sensitivity at higher values (60% in PSA above 10 ng/mL versus 33% below 10 ng/mL), though this difference did not reach statistical significance, likely due to the small sample size. The Partin tables -- a nomogram predicting pathologic stage from clinical variables -- showed comparable but slightly lower accuracy (69%) than endorectal MRI (78%) for extraprostatic extension in this cohort.
Seminal vesicle invasion was detected pathologically in only 4 (7.4%) of 54 patients -- the rarest outcome in this cohort. Despite this small number, endorectal MRI performed notably better for seminal vesicle invasion than for extraprostatic extension. Overall sensitivity was 75.0%, specificity 92.0%, and accuracy 90.7%, with a positive predictive value of 43% and a negative predictive value of 98%.
The excellent negative predictive value of 98% means that when endorectal MRI does not show seminal vesicle invasion, clinicians can be highly confident the seminal vesicles are not involved -- a clinically useful finding for patient counseling and surgical planning. The overall accuracy of 91% suggests seminal vesicle evaluation is one area where endorectal MRI reliably adds clinical value.
Among the three patients with Gleason score above 7, endorectal MRI detected seminal vesicle invasion with 100% sensitivity and 100% specificity, and the negative predictive value across all subgroups was consistently 96-100%. The better performance for seminal vesicle invasion compared to extraprostatic extension is consistent with the literature and likely reflects the anatomically distinct appearance of the seminal vesicles, which are easier to evaluate than the subtle contour changes indicating early capsular penetration.
The study's results place within the range reported in the literature: published sensitivity for extraprostatic extension detection by endorectal MRI spans 13-71%, while specificity ranges from 47-97%. This wide range reflects the fundamental challenge that MRI cannot visualize microscopic extraprostatic extension -- the most common form -- where individual tumor cells have crossed the capsule without creating visible anatomical deformation. Because MRI relies on macroscopic structural changes, it will always miss early microscopic spread.
The finding that sensitivity is higher for high-grade tumors is clinically important and actionable. High Gleason score tumors are more likely to show macroscopic, visible extraprostatic spread rather than microscopic penetration, and their aggressive growth pattern creates the bulging contours and periprostatic stranding that MRI can detect. This suggests endorectal MRI is most valuable as a staging tool specifically in patients with higher-grade disease where it can meaningfully change management.
The study has several important limitations: the small sample size (only 8 patients had extraprostatic extension, limiting statistical power), assessment was restricted to peripheral zone cancers, neoadjuvant therapy patients were excluded, and only the conventional imaging sequences available in 2010 were used. The authors explicitly note that newer MRI technologies -- including diffusion-weighted imaging, dynamic contrast enhancement, MR spectroscopy, and 3 Tesla field strength -- would likely improve performance. These techniques have since become standard practice.
Endorectal MRI demonstrated modest sensitivity but relatively high specificity for predicting extraprostatic extension in clinically localized prostate cancer. Its primary clinical utility lies in its high specificity and negative predictive value -- a negative endorectal MRI result provides reasonable confidence that extraprostatic extension is absent, supporting a decision to proceed with surgical planning assuming organ-confined disease.
The study established that endorectal MRI is more accurate for detecting seminal vesicle invasion than extraprostatic extension, and that its performance improves substantially in higher-grade tumors. These findings support a selective use of endorectal MRI -- particularly in patients with high Gleason scores or elevated PSA, where the pretest probability of advanced disease is higher and where MRI is most likely to detect it.
This 2010 study represents an important historical baseline for prostate MRI staging. Modern multiparametric MRI at 3 Tesla, which combines T2-weighted imaging with diffusion-weighted imaging, dynamic contrast enhancement, and standardized PI-RADS reporting, has since substantially improved upon these results. The evolution from single-coil endorectal imaging to multiparametric protocols reflects the field's recognition that multiple complementary sequences -- rather than any single technique -- are required for reliable prostate cancer staging.