Accuracy of magnetic resonance imaging in prediction of tumour-free resection margin in rectal cancer surgery

Background Incomplete surgical removal of the circumferential tumour spread is believed to be the main cause of local recurrence after resection of re ctal cancer. This study assessed the accuracy of magnetic resonance imaging (MRI) with a phased-array coil for preoperative staging and prediction of the distance of the tumour from the circumferential resection margin in a t otal mesorectal excision. Methods 76 patients with primary rectal cancer were preoperatively assessed by MRI at 1.5 T, with a phased-array coil. Two observers independently sco red, on two occasions, the tumour stage and measured the distance to the me sorectal fascia. Their findings were compared with the final histological f indings. Findings The MRI tumour stage agreed with the histological stage in 63 (83% ) of 76 patients (weighted kappa =0.77 [95% CI 0.66-0.89]) for observer 1, and in 51 (67%) patients (weighted kappa =0.52 [0.37-0.67]) for observer 2, The intraobserver agreement on the tumour stage was good (kappa =0.80 [0.6 9-0.91]) for observer 1 but moderate (kappa =0.49 [0.34-0.65]) for observer 2, The interobserver agreement was moderate (kappa =0.53 [0.38-0.69]). In 12 patients with an obvious T4 tumour. a margin of 0 mm was correctly predi cted. Of 29 patients for whom the pathologist reported a distance of at lea st 10 mm without specifying the actual distance, a distance of at least 10 mm was predicted in 28 by observer 1 and 27 by observer 2. For the remainin g 35 patients, a regression curve was constructed; from this, a histologica l distance of at least 1 0 mm can be predicted with high confidence when th e measured distance on MRI is at least 5.0 mm. Interpretation MRI with a phased-array coil showed moderate accuracy and re producibility for predicting the tumour stage of rectal cancers. The clinic ally more important circumferential resection margin can, however, be predi cted with high accuracy and consistency. allowing preoperative identificati on of patients at risk of recurrence who will benefit from preoperative rad iotherapy, more extensive surgery, or both.