Stokes WA, Bronsert MR, Meguid RA, Blum MG, Jones BL, Koshy M, Sher DJ, Louie AV, Palma DA, Senan S, Gaspar LE, Kavanagh BD, and Rusthoven CG
Purpose In early-stage non-small cell lung cancer (NSCLC), post-treatment mortality may influence the comparative effectiveness of surgery and stereotactic body radiotherapy (SBRT), with implications for shared decision making among high-risk surgical candidates. We analyzed early mortality after these interventions using the National Cancer Database. Patients and Methods We abstracted patients with cT1-T2a, N0, M0 NSCLC diagnosed between 2004 and 2013 undergoing either surgery or SBRT. Thirty-day and 90-day post-treatment mortality rates were calculated and compared using Cox regression and propensity score-matched analyses. Results We identified 76,623 patients who underwent surgery (78% lobectomy, 20% sublobar resection, 2% pneumonectomy) and 8,216 patients who received SBRT. In the unmatched cohort, mortality rates were moderately increased with surgery versus SBRT (30 days, 2.07% v 0.73% [absolute difference (Δ), 1.34%]; P < .001; 90 days, 3.59% v 2.93% [Δ, 0.66%]; P < .001). Among the 27,200 propensity score-matched patients, these differences increased (30 days, 2.41% v 0.79% [Δ, 1.62%]; P < .001; 90 days, 4.23% v 2.82% [Δ, 1.41%]; P < .001). Differences in mortality between surgery and SBRT increased with age, with interaction P < .001 at both 30 days and 90 days (71 to 75 years old: 30-day Δ, 1.87%; 90-day Δ, 2.02%; 76 to 80 years old: 30-day Δ, 2.80%; 90-day Δ, 2.59%; > 80 years old: 30-day Δ, 3.03%; 90-day Δ, 3.67%; all P ≤ .001). Compared with SBRT, surgical mortality rates were higher with increased extent of resection (30-day and 90-day multivariate hazard ratio for mortality: sublobar resection, 2.85 and 1.37; lobectomy, 3.65 and 1.60; pneumonectomy, 14.5 and 5.66; all P < 0.001). Conclusion Differences in 30- and 90-day post-treatment mortality between surgery and SBRT increased as a function of age, with the largest differences in favor of SBRT observed among patients older than 70 years. These representative mortality data may inform shared decision making among patients with early-stage NSCLC who are eligible for both interventions.