Diagnosis and management of a mediastinal leak following radical oesophagectomy

Abstract

Background The aim of this study was to evaluate the diagnosis, management and outcome of mediastinal leaks following radical oesophagectomy with a stapled intrathoracic anastomosis. Methods Some 291 consecutive patients underwent two-phase subtotal oesophagectomy with gastric interposition for malignancy. Patients with clinical suspicion of a leak were investigated with contrast radiology and flexible upper gastrointestinal endoscopy. Results Nineteen patients (6·5 per cent) developed a proven mediastinal leak at a median of 8 (range 3–30) days following surgery. Contrast radiology and flexible upper gastrointestinal endoscopy identified that 13 patients had an isolated leak from the oesophagogastric anastomosis and two had widespread leakage secondary to gastrotomy-line dehiscence. Endoscopy revealed a further four patients with gastric necrosis in whom contrast radiology was normal. In six patients the diagnosis of leakage followed an apparently normal routine contrast examination on day 5–8. All 13 isolated anastomotic leaks were managed non-operatively with targeted mediastinal drainage, intravenous antibiotics and antifungal therapy, nasogastric decompression and enteral nutrition; the mortality rate was 15 per cent (two of 13). Patients with gastrotomy dehiscence or gastric necrosis had a more severe clinical picture; they were managed with repeat thoracotomy and either revision of the conduit or resection and exclusion. Despite early intervention four of the six patients died. Conclusion Routine postoperative contrast radiology cannot be recommended. On clinical suspicion of a leak patients require both contrast radiology and endoscopic evaluation. Isolated anastomotic leaks can be managed successfully with non-operative treatment, whereas more extensive leaks from the gastric conduit require revisional surgery which carries a high mortality rate.