The magnetic resonance imaging-based approach for identification of high-risk patients with upper rectal cancer

Objective: To assess the efficacy of preoperative magnetic resonance imaging (MRI) in identifying upper rectal cancer patients who are at high risk for local recurrence. Methods: 110 upper rectal cancer patients with locally advanced (pT3-4N0 or pTanyN+) tumors treated with tumor-specific mesorectal excision and no adjuvant radiotherapy were identified from an institutional database at a large academic medical center in Korea. Information on the extent of mesorectal invasion, sacral-side involvement was collected from preoperative MRI. Results: At a median follow-up of 47 months, 5 patients (4.5%) experienced local recurrence (LR). LR rates for patients with intermediate risk (T1-2/N1, T3N0), moderately high risk (T1-2/N2, T3N1, T4N0), and high risk (T3N2, T4/N1-2) were 3%, 4.8%, and 8.7%, respectively. Patients who did not have sacral-side involvement or mesorectal invasion of 5 mm or less did not experience LR. The patients with sacral-side involvement and intermediate risk, moderately high risk, and high risk had an LR rate of 4.2%, 5.6%, and 10%, respectively, or 11.1%, 33.3%, and 18.2%, respectively, when combined with those with mesorectal invasion of greater than 5 mm. Multivariate analyses demonstrated the presence of both sacral-side location and mesorectal invasion of greater than 5 mm was significantly associated with adverse disease-free and overall survival (P < 0.05). Conclusions: Patients with mesorectal invasion of greater than 5 mm and sacral-side involvement identified on MRI were at an increased risk of local recurrence. The detection of these features on MRI provides prognostic information that is not available in conventional risk classification systems. Improved identification of a high-risk subset of upper rectal cancer patients may guide indications for preoperative chemoradiotherapy in this subset.