Local staging of rectal cancer using combined pelvic phased-array and endorectal coil MRI.

PURPOSE: To assess the accuracy of MRI, using a pelvic phased-array coil and an endorectal coil, for preoperative local staging of rectal cancer. MATERIALS AND METHODS: Fifty-one patients (26 males and 25 females) with adenocarcinoma of the rectum underwent preoperative MRI and surgical resection of their tumors. Surgical pathology staging was compared to MRI staging (using the TNM classification) obtained both retrospectively by a reader blinded to surgical findings and prospectively (radiological reports). In addition, patients were stratified according to surgical treatment groups (stage I = T1-2/N0, stage II = T3/N0, stage III = Tx/N1-2). RESULTS: At pathology, 36 of 51 (68%) tumors were classified as T0-T2, and 15 (32%) were classified as T3. Overall, the sensitivity and specificity of MRI readings for T3 staging were 93% and 86%, respectively (positive predictive value (PPV) = 74%, negative predictive value (NPV) = 97%, accuracy = 88%). MRI correctly predicted lymph node metastases in 11 of 13 patients with a sensitivity of 85% and specificity of 69% (PPV = 58%, NPV = 90%, accuracy = 74%). MRI correctly predicted surgical treatment groups in 33 of 39 (85%) patients. Interobserver agreement between the retrospective and prospective readings was excellent (kappa = 0.85) for prediction of T3 tumor and good (kappa = 0.80) for prediction of nodal involvement. CONCLUSION: Combined endorectal and pelvic phased-array coil MRI can be used reliably to select which patients should receive preoperative chemoradiotherapy. It is highly predictive in terms of excluding T3 tumors, but still has limitations in predicting lymph node metastasis