One hundred consecutive pneumonectomies after induction therapy for non-small cell lung cancer: an uncertain balance between risks and benefits.

OBJECTIVE: We sought to assess postoperative outcome after pneumonectomy after neoadjuvant therapy in patients with non-small cell lung cancer. METHODS: This retrospective study included 100 patients treated from January 1989 through December 2003 for a primary lung cancer in whom pneumonectomy had been performed after an induction treatment. Surgical intervention had not been considered initially for the following reasons: N2 disease (stage IIIA, n = 79), doubtful resectability (stage IIIB [T4, N0], n = 19), and M1 disease (stage IV [T2, N0, M1, solitary brain metastasis], n = 2). All patients received a 2-drug platinum-based regimen with a median of 2.5 cycles (range, 2-4 cycles), and 30 had associated radiotherapy (30-45 Gy). RESULTS: There were 55 right and 45 left resections. Overall 30-day and 90-day mortality rates were 12% and 21%, respectively. At multivariate analysis, one independent prognostic factor entered the model to predict 30-day mortality: postoperative cardiovascular event (relative risk, 45.7; 95% confidence interval, 3.7-226.7; P = .001). Four variables predicted 90-day mortality: age of more than 60 years (relative risk, 5.06; 95% confidence interval, 1.47-17.48; P = .01), male sex (relative risk, 8.25; 95% confidence interval, 1.01-67.34; P = .049), postoperative respiratory event (relative risk, 3.64; 95% confidence interval, 1.14-9.37; P = .007), and postoperative cardiovascular event (relative risk, 7.84; 95% confidence interval, 3.12-19.71; P < .001). Estimated overall survivals in 90-day survivors were 35% (range, 29%-41%) and 25% (range, 19.3%-30.7%) at 3 and 5 years, respectively. At multivariate analysis, one independent prognostic factor entered the model: pathologic stage III-IV residual disease (relative risk, 1.89; 95% confidence interval, 1.09-3.26; P = .022). CONCLUSIONS: Pneumonectomy after induction therapy is a high-risk procedure, the survival benefit of which appears uncertain.