Completion pneumonectomy: a multicentre international study on 165 patients.

Objectives: We evaluated factors that influenced morbidity and mortality in patients undergoing completion pneumonectomy (CP).
Methods: A retrospective review of a consecutive series of patients who underwent CP at six international centres.
Results: In total, 165 CP were performed between March 1990 and December 2009: 152 for malignant disease and 13 for benign disease. Forty-two patients (25.4%) underwent neoadjuvant therapy. Right CP was performed in 99 patients (60%) and left in 66 (40%). Thoracotomy was employed in 161 patients and median sternotomy in 4. Stapled closure of the bronchus was performed in 121 patients and hand closure in 44. The overall operative mortality was 10.3% (17 of 165). Operative mortality was 10.5% (16 of 152) in malignant diseases and 7.7% (1 of 13) in benign diseases. Complications occurred in 55.1% (91 of 165) of patients. Mean hospital stay was 16.02 ± 16.8 days (range: 3-151 days). Thirteen patients (7.9%) developed bronchopleural fistulas. No statistically significant relationship was found in mortality or morbidity according to side, gender, induction therapy and surgical approach. Stapled compared with hand closure for the bronchus did not affect the bronchopleural fistula rate (P = 0.4). The overall 5-year survival was 37.6%: 70.1% in benign disease (13 patients), 48.9% in squamous cell carcinoma of the lung (63 patients), 23.9% in primary lung adenocarcinoma (62 patients), 50% in grade 1 and grade 2 neuroendocrine carcinoma of the lung (4 patients), 54.7% in metastatic disease (14 patients) and 0% in primary lung sarcomas. A statistically significant better survival was observed in patients with squamous cell carcinoma versus adenocarcinoma (P = 0.04).
Conclusions: CP shows an acceptable operative mortality with a high morbidity rate. The overall 5-year survival is acceptable in properly selected patients (i.e. squamous cell carcinoma, metastatic disease). Side, gender, induction therapy and surgical approach did not influence mortality and morbidity.