Abstract

We retrospectively evaluated the accuracy of magnetic resonance (MR) imaging in staging colorectal cancer and assessing local tumor extent, nodal involvement, and distant abdominal and pelvic metastases. Forty-eight patients with primary colorectal carcinoma were referred for presurgical abdominal and pelvic MR imaging. MR imaging included T1-weighted, fat-suppressed T2-weighted, and fat-suppressed gadolinium-enhanced spin gradient-echo imaging. The prospective interpretations of the MR examinations were reviewed. MR depiction of local tumor extent, nodal involvement, and distant metastases at 18 anatomic locations was noted and compared with subsequent surgical and histopathologic findings. Overall TNM MR staging agreed with surgical and pathologic staging in 41 (85%) of 48 patients, including 21 (78%) of 27 colon cancers and 20 (95%) of 21 rectal cancers. For depth of tumor penetration, which was evaluable in 44 patients, MR imaging agreed with pathologic results in 38 (86%) of 44 patients, including 22 (88%) of 25 colon cancers and 16 (84%) of 19 rectal cancers. In 42 (95%) of 44 patients, MR images correctly distinguished tumor confined to the bowel wall (T0, T1, and T2) from tumor with transmural tumor extension (T3 and T4). Regional nodal metastases were depicted in 15 of 22 patients (sensitivity, 68%; accuracy, 83%). Nodal metastases were better depicted for rectal cancer in eight of nine patients, compared with colon cancer in seven of 13 patients. Distant metastases were correctly depicted on MR imaging in 13 of 14 patients (sensitivity, 93%; accuracy, 98%). In the site-by-site analysis, MR imaging prospectively depicted 66 of 77 sites of surgically confirmed metastatic tumor in the abdomen and pelvis (sensitivity, 86%; specificity, 99%; accuracy, 98%). MR imaging using currently available techniques can effectively image local tumor extent and distant metastases in patients with colorectal carcinoma. Especially for colon cancer, incomplete depiction of nodal metastases in normal-size lymph nodes remains a limitation of cross-sectional imaging studies.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call