Objective: We previously reported a high mortality after induction therapy and pneumonectomy for non-small cell lung cancer. Recent reports suggest that operative mortality in these patients is declining. We analyzed our contemporary results to define operative mortality and factors determining surgical risk.
Methods: Eligible patients were identified from our prospective surgical database. Complications were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events 3.0. Uni- and multivariate logistic regression models assessed the association of preoperative tests and clinical characteristics with outcome. Receiver operating characteristic curves and area under the receiver operating characteristic curve (AUC) statistics were calculated in a leave-one-out crossvalidation scheme to evaluate the predictive value of various models.
Results: From January 2000 to December 2006, 549 patients underwent surgery after induction therapy. Median patient age was 64 years (range: 30-86), and 54% were women (298/549). All received chemotherapy, and 17% also had radiation. Lobectomy (388/549, 71%) and pneumonectomy (70/549, 13%) were the most common procedures. Complications occurred in 250 patients (46%), with grade 3 or higher in 23% (126/549). Inhospital mortality was 1.8% (10/549), with only one death after right pneumonectomy (1/30, 3%). Multivariate analysis showed that predicted postoperative (PPO) pulmonary function was associated with postoperative morbidity. By receiver operating characteristic curves, PPO product (AUC = 0.75, p < 0.001), PPO diffusion capacity (AUC = 0.70, p < 0.001), and preoperative % predicted PPO diffusion capacity (AUC = 0.66, p < 0.001) predicted mortality.
Conclusion: Our current experience shows that resection of non-small cell lung cancer after induction therapy, including pneumonectomy, is associated with low mortality. PPO pulmonary function is the strongest predictor of operative risk and should be used to select patients for surgery.