Routine postoperative CT to detect anastomotic leakage after low anterior resection for rectal cancer has a low sensitivity and specificity and a poor interobserver agreement

Abstract

To compare the accuracy and interobserver variation of routine computed tomography (CT) on postoperative day 6-8 to detect anastomotic leakage (AL) verified by re-operation and/or endoscopy. A secondary objective was to identify the predictive values of different CT findings as an indicator for AL. The material for this study originates from two previous prospective multicentre studies including 277 patients who were scheduled for routine abdominal CT postoperative day 6-8. Inclusion criteria for the present study were routine CT without contrast medium followed by CT with rectal contrast medium. Two independent senior radiologists blinded to the clinical outcome reviewed the CT examinations for specific findings according to a predefined scheme. A total of 52 patients fulfilled the inclusion criteria. AL occurred in 14 patients of which nine were clinical and five subclinical. The two radiologists diagnosed AL at unenhanced CT with sensitivities of 71.4% and 50%, respectively, and of 57.1% and 35.7% with rectal contrast medium. The corresponding specificities were 55.3% and 81.6%, and 94.7% and 92.1%. Peri-anastomotic free air and contrast medium leakage had the highest odds ratios for AL. The diagnostic sensitivity and specificity of routine postoperative CT to detect AL after low anterior resection for rectal cancer is low and with considerable interobserver variation.