Recurrent or residual pelvic bowel cancer: Accuracy of MRI local extent before salvage surgery

PURPOSE: To determine pre-operative MRI accuracy in assessing local disease extent in recurrent/residual pelvic bowel cancer by comparing MRI assessment and staging examination under anaesthesia (EUA), with laparotomy/histopathological findings. MATERIALS AND METHODS: Twenty-seven consecutive patients with recurrent (n = 21) or residual (n = 6) pelvic bowel cancer (13 of the rectum, eleven of the anus and three of the colon) underwent EUA and pelvic MRI (IT) using a phased array pelvic coil. Retrospective analysis of eight specific anatomical regions for tumour involvement on MRI was performed. Findings at EUA and biopsy were recorded. The MRI and EUA findings were correlated with findings at surgery and histopathology. Statistical comparison between MRI and EUA results was performed using the chi-squared test. RESULTS: Overall MRI accuracy in determining tumour invasion for all sites assessed was 452/499 (91%), sensitivity was 95/109 (87%), specificity was 357/390 (92%), positive predictive value (PPV) was 95/128 (74%) and negative predictive value (NPV) was 357/371 (96%). PPV and NPV for specific areas were 21/38 (55%) and 134/136 (99%) for genitourinary tract, 4/6 (67%) and 61/65 (94%) for pelvic side wall, 21/26 (81%) and 40/41 (98%) for pelvic floor, 1/6 (17%) and 40/43 (93%) for the posterior pelvis pre-sacrum/sacrum. For those anatomical sites evaluated by both EUA and MRI, MRI was superior to EUA, with an accuracy of 89% vs 73% (P < 0.05). CONCLUSION: MRI is an accurate technique for assessing disease extent in recurrent/residual pelvic bowel cancer. (C) 2002 The Royal College of Radiologists.