LAUSR

To determine the role of MRI as a predictor of circumferential resection margin (CRM) involvement. To study the impact of CRM status on MRI on recurrence and survival, in correlation with pathology. Analysis of a prospective database was performed over a period of 1 year. All patients with adenocarcinoma of rectum were included in the study. The MRI at presentation for all patients irrespective of stage (MRIT), pre-NACTRT MRI (MRI1) for patients with locally advanced tumours, and post-NACTRT MRI (MRI2) of these patients were analysed separately. The status of CRM on MRI was compared to that on histopathology and as a predictor of recurrence and survival. Two hundred twenty-one patients were included with a median follow-up 30 months. Sensitivity, specificity, positive predictive value, negative predictive value (NPV) and accuracy were 50%, 65.46%, 5.63%, 96.95% and 64.85% for MRIT; 50%, 55.32%, 5.97%, 95.12% and 55.03% for MRI1; and 77.78%, 63.29%, 10.77%, 98.04% and 64.07% for MRI2, respectively. On multivariate analysis, pathological positive margin alone predicted a poor overall survival (OS) whereas involved CRM on pathology and MRIT predicted poorer disease-free survival (DFS) and local recurrence. Pre-treatment and post-treatment MRI scans have a moderate sensitivity, specificity and accuracy and a high negative predictive value to predict CRM status on pathology. Pathological CRM status is the only factor to impact OS, DFS and LR on multivariate analysis. CRM status on MRI at presentation (MRIT) does impact DFS and local recurrence but not OS.