Selective management of intrathoracic anastomotic leak after esophagectomy

Objective: We sought to analyze our experience with management of intrathoracic anastomotic leak after esophagectomy. Methods: All patients who had intrathoracic anastomotic leaks after esophagectomy were reviewed. Management and factors affecting outcome were analyzed. Results: From March 1993 through February 2003, 761 patients had esophagectomy with intrathoracic anastomosis at our institution. Forty-eight (6.3%) patients had an anastomotic leak; one refused authorization to review his medical record and was excluded from further analysis. Twenty-four (51.1%) patients had a contained leak. Twenty-seven (57.4%) patients were managed nonoperatively. Twenty (42.6%) patients required surgical intervention that included primary anastomotic repair in 14 patients, reinforcement of the anastomosis with viable tissue in 6 patients, and esophageal diversion in 2 patients. A single reoperation was done in 15 patients, and 5 patients had 2 reoperations. Median hospitalization in the reoperative group was 31 days (range, 15-97 days) and 20 days (range, 10-42 days) in the nonoperative group (P =.0037). Four (8.5%) patients died. Cause of death was sepsis in 2 patients and multiorgan failure and myocardial infarction in 1 patient each. At follow-up (median, 8 months; range, 1-120 months), 10 (58.8%) patients in the reoperative group were eating a normal diet and 5 (29.4%) patients required at least one dilatation compared with 20 (76.9%) patients in the nonoperative group who were eating a normal diet and 9 (34.6%) who required at least one dilatation. A noncontained leak had an adverse effect on long-term survival (P =.04). Conclusion: Intrathoracic anastomotic leak after esophagectomy is associated with significant morbidity and mortality. Contained leaks often can be managed nonoperatively. When surgical management is required, esophagogastric continuity can often be maintained in the majority of patients. Long-term functional results are satisfactory and similar in both the reoperative and nonoperative groups. However, a noncontained leak adversely affected long-term survival.