Abstract
OBJECTIVE: To determine the diagnostic value of 3 Tesla MR imaging in detection of mucosal (Tis), submucosal (T(1)) and muscularis propria (T(2)) invasion in patients with early rectal cancer. METHODS: A total of 50 consecutive patients who underwent 3 Tesla MR imaging and curative-intent intervention for MRI-staged Tis/T(1)/T(2) rectal cancer from March 2012 to December 2016 were included. The radiological T category of each rectal tumour was compared retrospectively with histopathological results assessed according to the tumor, node, metastasis (TNM) classification. The sensitivities, specificities, and overall accuracy rates of 3 Tesla MR imaging for Tis, T(1), and T(2) cases were calculated using MedCalc statistical software v. 16. RESULTS: The sensitivity, specificity, PPV, NPV of 3 Tesla MR imaging in T categorization for T(2) were: 93.7% [95% CI (0.79-0.99)], 77.7% [95% CI (0.52-0.93)], 88.2% [95% CI (0.75-0.94)] and 87.5% [95% CI (0.64-0.96)]; for T(1) were 92% [95% CI (0.63-0.99)], 91.8% [95% CI (0.78-0.98)], 80% [95% CI (0.57-0.92)] and 97.1% [95% CI (0.83-0.99)]; for Tis were: 20% [95% CI (0.51-0.71)], 100% [95% CI (0.92-1)], 100%, 91.8% [95% CI (0.87-0.94)], respectively. MR categorization accuracy rates for T(2), T(1) and Tis were calculated as 88, 92 and 92%, respectively. CONCLUSION: 3 Tesla MR imaging seems to be useful for accurate categorization of T-stage in early rectal cancer, especially for T(1) cancers. The method is not a reliable tool to detect Tis cases. The potential for overstaging and understaging of the technique should be realized and taken into consideration when tailoring the treatment protocol for each patient. Advances in knowledge: High-resolution MR with phased-array coil is being increasingly used in the pre-operative assessment of rectal cancer. 3 Tesla high-resolution MR imaging allows improved definition of bowel wall and tumour infiltration.