Clinical Burden of Anastomotic Leak After Elective Colorectal Cancer Surgery: A Multicenter Cohort Study.

Background

Anastomotic leak remains a major complication after colorectal cancer surgery. While most studies focus on prediction or prevention, less is known about the clinical course after leak diagnosis. We aimed to evaluate the occurrence, severity, management, and early postoperative consequences of anastomotic leak after elective colorectal cancer surgery.

Methods

This retrospective multicenter cohort study included consecutive patients undergoing elective colorectal cancer resection with restoration of bowel continuity in five tertiary referral centers. Patients without an evaluable anastomosis were excluded. The primary endpoint was 30-day anastomotic leak incidence. Secondary endpoints included Clavien-Dindo severity, therapeutic management, major morbidity, surgical reintervention, length of hospital stay, intensive care unit admission, readmission, and 30-day mortality. Firth penalized logistic regression was used for adjusted analyses.

Results

Among 943 patients with an evaluable anastomosis, 55 developed an anastomotic leak (5.8%). Leak was more frequent after rectal than colon cancer surgery (8.8% vs. 4.5%; crude OR 2.04, 95% CI 1.18-3.53; p = 0.016). After adjustment, rectal cancer remained independently associated with leak occurrence (adjusted OR 3.61, 95% CI 1.74-7.48; p < 0.001), whereas intraoperative indocyanine green fluorescence angiography was not (adjusted OR 1.02, 95% CI 0.48-2.17; p = 0.962). Once leak occurred, 92.7% of patients developed major morbidity, 76.4% required surgical reintervention, and, among patients with available length-of-stay data, 65.4% had a postoperative hospital stay exceeding 14 days; 30-day mortality was 10.9%.

Conclusions

Anastomotic leak occurred in fewer than 6% of elective colorectal cancer resections but was associated with severe postoperative deterioration. Its impact is better captured by evaluating severity, management, and downstream consequences, rather than incidence alone.