Magnetic resonance imaging for preoperative staging of rectal cancer in clinical practice: high accuracy in predicting circumferential margin with clinical benefit.

OBJECTIVE: The aims were to determine agreement between staging of rectal cancer made by magnetic resonance imaging (MRI) and histopathological examination and the influence of MRI on choice of radiotherapy (RT) and surgical procedure.
METHOD: In this retrospective audit, preoperative MRI was performed on 91 patients who underwent bowel resection, with 93% having total mesorectal excision. Tumour stage according to mural penetration, nodal status and circumferential resection margin (mCRM) involvement was assessed and compared with histopathology.
RESULTS: Five radiologists interpreted the images. Overall agreement between MRI and histopathology for T stage was 66%. The greatest difficulty was in distinguishing between T1, T2 and minimal T3 tumours. The accuracy for mCRM (MRI) was 86% (78/91),with an interobserver variation between 80% and 100%. In the 13 cases with no agreement between mCRM and pCRM (pathological), seven had long-term RT and nine en bloc resections, indicating that the margins initially were involved with an even higher accuracy for mCRM. Preoperative short-term RT was routine, but based on MRI findings, choice of RT was affected in 29 cases (32%); 17 patients had no RT and 12 long-term RT. The surgical procedure was affected in 17 cases (19%) with planned perirectal en bloc resections in all. CRM was involved (< or = 1 mm) in 14.7% of the 34 cases in which MRI had an effect upon choice of RT and/or surgery compared with 8.8% of the remaining 57 cases where it had no impact.
CONCLUSION: Magnetic resonance imaging predicted CRM with high accuracy in rectal cancer. MRI could be used as a clinical guidance with high reliability as indicated by the low figures of histopathologically involved CRM.