INTRODUCTION: Cigarette smoking negatively affects lung cancer prognosis. Incorporating smoking history into stage-stratified survival analyses may improve prognostication.
METHODS: Using the International Association for the Study of Lung Cancer ninth edition NSCLC database, we evaluated the association between smoking status at diagnosis and overall survival (OS) using Kaplan-Meier plots and multivariate Cox proportional hazard regression models adjusted for age, region, sex, histologic type, performance status, and TNM stage. Subgroup analyses evaluated within- and between-stage survival according to smoking status. Recursive partitioning analysis evaluated the effect of smoking relative to anatomical staging factors.
RESULTS: Among 48,531 patients, 26%, 41%, and 34% currently, formerly, or never smoked, respectively; 75% had adenocarcinoma and 21%, squamous cell carcinoma. Compared with those who never smoked, those who currently (adjusted hazard ratio [aHR] = 1.39, 95% confidence interval [CI] [1.34-1.46], p < 0.0001) or formerly (aHR = 1.32, 95% CI: 1.27-1.36, p < 0.0001) smoked had worse OS. Current smoking was associated with worse OS (aHR = 1.05, 95% CI: 1.02-1.06] p = 0.005) compared with former smoking. These associations were consistent within each stage. In pair-wise comparisons, those who formerly or currently smoked had similar OS to patients in the next higher stage who never smoked. These associations remained consistent among patients from Asia, patients with adenocarcinoma, and both sexes. In recursive partitioning analysis, among patients with early stage (up to T3N1M0), smoking had greater prognostic impact than N0 versus N1 categories.
CONCLUSIONS: Smoking is an important prognostic factor and may be more impactful on prognosis than anatomical staging. Further exploration of combining smoking status with TNM classification for lung cancer is planned in the tenth edition International Association for the Study of Lung Cancer staging project.