Objectives: To evaluate the diagnostic accuracy of CT in postoperative colorectal anastomotic leakage (AL).
Methods: Two independent blinded radiologists reviewed 153 CTs performed for suspected AL within 60 days after surgery in 131 consecutive patients, with (n = 58) or without (n = 95) retrograde contrast enema (RCE). Results were compared to original interpretations. The reference standard was reoperation or consensus (a radiologist and a surgeon) regarding clinical, laboratory, radiological, and follow-up data after medical treatment.
Results: AL was confirmed in 34/131 patients. For the two reviewers and original interpretation, sensitivity of CT was 82 %, 87 %, and 71 %, respectively; specificity was 84 %, 84 %, and 92 %. RCE significantly increased the positive predictive value (from 40 % to 88 %, P = 0.0009; 41 % to 92 %, P = 0.0016; and 40 % to 100 %, P = 0.0006). Contrast extravasation was the most sensitive (reviewers, 83 % and 83 %) and specific (97 % and 97 %) sign and was significantly associated with AL by univariate analysis (P < 0.0001 and P < 0.0001). By multivariate analysis with recursive partitioning, CT with RCE was accurate to confirm or rule out AL with contrast extravasation.
Conclusions: CT with RCE is accurate for diagnosing postoperative colorectal AL. Contrast extravasation is the most reliable sign. RCE should be performed during CT for suspected AL.
Key points: • CT accurately diagnosed clinically suspected colorectal AL and showed good interobserver agreement • Contrast extravasation was the most sensitive and specific CT sign • Retrograde contrast enema during CT improved positive predictive value • Retrograde contrast enema decreased false-negative or indeterminate original CT interpretations.
Keywords: Anastomotic Leak; Colorectal Surgery; Computed Tomography; Enema; Extravasation of Contrast Media.