Fate of Pneumonectomy Patients Variably Captured by Non-Small Cell Lung Cancer Staging System.

Abstract

Lung cancer patients rely on survival estimates to weigh risks and benefits of treatment. However, pneumonectomy-requiring lung cancer may have inherent oncologic or physiologic survival implications not captured by the current stage classification. Stage-specific survival was evaluated to refine survival expectations for patients with pneumonectomy-requiring disease. The National Cancer Database was queried for treatment-naive patients who underwent lobectomy or pneumonectomy for stage I to III non-small cell lung cancer between 2004 and 2013. Patients who died within 90 days after resection were excluded. Three-way propensity score weighted multivariable Cox models were built and incorporated into adjusted 5-year overall survival (OS) curves. A total of 79,953 patients met inclusion criteria: 75,708 lobectomies (95%) and 4,245 pneumonectomies (5%). Stage I and II patients undergoing right pneumonectomy had worse adjusted 5-year OS than patients undergoing left pneumonectomy, which was worse than lobectomy (stage I: 55%, 58%, 67%; stage II: 37%, 44%, 48%; indicating right pneumonectomy, left pneumonectomy, lobectomy). Stage III right pneumonectomy patients had worse adjusted 5-year OS; however, left pneumonectomy and lobectomy patients were similar (33%, 39%, 40%). A doubly robust Cox model identified a similar pattern for mortality risk for stage I and II (right pneumonectomy > left > lobectomy); however, stage III right pneumonectomy patients had higher mortality risk than lobectomy patients (hazard ratio [HR] 1.23, 95% confidence interval [CI]: 1.17 to 1.28, p < 0.001), whereas left pneumonectomy was similar to lobectomy (HR 1.02, 95% CI: 0.97 to 1.06, p = 0.47). Pneumonectomy-requiring lung cancer embodies a 5-year mortality risk not completely captured by the lung cancer staging system. Refined survival estimates for pneumonectomy patients may enhance shared decision making in this population.