Local staging of prostate cancer with endorectal MR imaging: correlation with histopathology

• Published in: American journal of roentgenology (1976) Vol. 166; no. 4; pp. 845 - 852
• Main Authors: Jager, GJ, Ruijter, ET, van de Kaa, CA, de la Rosette, JJ, Oosterhof, GO, Thornbury, JR, Barentsz, JO
• Format: Journal Article
• Language: English
• Published: Leesburg, VA Am Roentgen Ray Soc 01-04-1996; American Roentgen Ray Society
• Subjects: Adenocarcinoma - diagnosis; Adenocarcinoma - pathology; Adenocarcinoma - surgery; Aged; Biological and medical sciences; Genital system. Mammary gland; Humans; Investigative techniques, diagnostic techniques (general aspects); Magnetic Resonance Imaging - methods; Male; Medical sciences; Middle Aged; Neoplasm Staging; Predictive Value of Tests; Prostate - pathology; Prostatectomy; Prostatic Neoplasms - diagnosis; Prostatic Neoplasms - pathology; Prostatic Neoplasms - surgery; Radiodiagnosis. Nmr imagery. Nmr spectrometry; Retrospective Studies; Sensitivity and Specificity; Human; Correlation; Urinary system disease; Carcinoma; Prostate disease; Exploration; Histology; Malignant tumor; Nuclear magnetic resonance imaging; Pathology; Male genital diseases; Prostate; Semiology
• ISSN: 0361-803X; 1546-3141
• DOI: 10.2214/ajr.166.4.8610561

To evaluate the accuracy of MR imaging of the prostate with an endorectal surface coil in determining presence, localization, volume, and local stage of prostate carcinoma.
MR images of 34 patients with biopsy-proven cancer were correlated retrospectively with the histologic mappings of radical prostatectomy specimens. The volume and number of tumor lesions of MR images were calculated and compared with the surgical specimens used as the gold standard. Tumor stage based on MR imaging was compared with the pathologic stage according to the TNM classification. Predictive values were calculated separately for all lesions and for the lesions correctly localized with MR imaging.
MR imaging correctly depicted the location of 67% of the tumors. Twenty percent of the lesions depicted by MR imaging appeared to be false-positive errors. The tumors that were missed were located centrally and ventrally in the prostate. Tumor volume as shown by MR imaging was within a 25% range of the actual tumor volume in 10 cases, overestimated in 16 cases, and underestimated in eight cases. Histopathology showed capsular penetration in 12 of 34 patients (35%) and in 14 of 52 lesions (27%). Sensitivity, specificity, and positive predictive values were 43%, 84%, and 55%, respectively. Histologically, capsular penetration extended less than 1 mm into the periprostatic adipose tissue in seven patients. Sensitivity for capsular penetration less than 1 mm was 14%. Sensitivity for capsular penetration more than 1 mm was 71%. Accuracy for differentiating a pT2 from a pT3 tumor was 68%.
Results from this study indicate that the accuracy of the technique was not satisfactory for predicting actual tumor volume. Tumor detection and localization was more accurate in the peripheral zone than in the central zone. Accuracy was poor for detecting capsular penetration of less than 1 mm, but accuracy was much better for penetration of more than 1 mm. Because recent reports suggest that capsular penetration of less than 1 mm does not adversely affect surgical cure, MR imaging still may be practical in the selection of patients for radical prostatectomy.