Comparison of clinical outcomes after conservative and surgical treatment of isolated anastomotic leaks after esophagectomy for esophageal cancer.

The clinical course and outcome of isolated anastomotic leaks (IALs) after esophagectomy are significantly different from those of necrotic leaks. The purpose of this study was to investigate the clinical features, diagnosis, treatment, and long-term outcome in patients with IALs after esophagectomy with reconstruction for esophageal cancer. A total of 663 patients underwent esophagectomy with esophageal reconstruction because of esophageal cancer between 2000 and 2010 at the Seoul Asan Medical Center. IALs occurred in 23 patients (3.5%). All patients with IAL were male, with a median age of 61 years. Patients with IAL were divided into three groups based on their clinical course. group A comprised patients who had definite clinical symptoms and/or signs indicating mediastinal contamination or leak before routine contrast esophagography was performed. Groups B and C comprised patients who had no definite clinical symptoms and/or signs of leaks before the routine contrast examination. Furthermore, group B contained those patients who resumed oral intake because no leak was found in the routine contrast examination and was diagnosed some days after resuming oral intake. Group C contained those patients who kept fasting because the leak was found in the routine contrast examination. The median follow-up period was 30 months. The mean time to closure of the IAL was 70.1 ± 96.0 days (range 4-364). There was a 72.7% overall closure rate within 60 days. By univariate analysis, the mean time to closure of the IAL was found to be significantly longer for group A patients or in cases where the patients had an uncontained leak, leukocytosis, or empyema. However, there was no statistically significant differences in age, neoadjuvant treatment, site of anastomosis (cervical vs. thoracic), fever, or treatment of the leak. By multivariate analysis, group A was found to be an independent predictive factor for the time to closure of the IAL. Repeat contrast studies revealed no anastomotic leaks in 18 patients and the formation of contained fistula in four cases (excluding one patient who died in hospital). The four patients with a contained fistula showed no clinical symptoms or signs, and tolerated resumed oral intake. IALs were resolved in most cases with low leak-related mortality, and resolution of the leaks occurred within 2 months in the majority of patients.