Diagnosis and management of a mediastinal leak following radical oesophagectomy

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.