Non–small cell lung cancer and the obesity paradox: A retrospective cohort study using the TriNetX database.

Abstract

e20544 Background: Although obesity is traditionally associated with adverse health outcomes, paradoxical survival benefits have been reported across several malignancies. In non–small cell lung cancer (NSCLC), the prognostic impact of obesity remains uncertain, largely due to small cohorts and limited adjustment for disease and treatment factors. We evaluated the association between obesity and long-term overall survival in NSCLC using a large real-world database. Methods: We conducted a multicenter retrospective cohort study using the TriNetX database, including de-identified records from 94 healthcare organizations. Adults (≥18 years) diagnosed with NSCLC (adenocarcinoma, squamous cell carcinoma, or large cell carcinoma) within the past 15 years were identified. Exclusions included Eastern Cooperative Oncology Group (ECOG) performance status ≥2 and classical or combined small cell histology. Patients were stratified into normal-weight (BMI 20.0-24.9kg/m 2 ) and obese (BMI ≥ 30kg/m 2 ) cohorts based on BMI recorded within 6 months of the index event (diagnosis of NSCLC). Propensity score matching (1:1) controlled for differences in sociodemographics, comorbidities, histologic subtypes, cancer stage, genetic mutations, and cancer-directed treatments. Descriptive statistics summarized baseline characteristics. The primary outcome was 10-year overall survival, analyzed by Cox proportional hazards model and Kaplan-Meier curves. Statistical significance was set at p < 0.05. Results: At baseline, the normal-weight (N = 8,904) and obese (N = 21,748) cohorts were predominantly White (54.2% vs. 65.1%), male (49.5% vs. 48.8%), and non-Hispanic (64.43% vs. 71.92%). The normal-weight cohort was older (66.3 ± 12.2 years) than the obese cohort (61.7 ± 12.0 years, p < 0.0001) and had higher rates of late-stage (III-IV) disease, chemotherapy, and pembrolizumab-based immunotherapy (p < 0.0001). Surgical interventions were equally uncommon, with small standardized differences. Diabetes and hypertensive diseases were more common in the obese cohort (p < 0.0001). Genetic mutations in EGFR, ALK, ERBB3, and ROS1 were slightly more frequent in the normal-weight cohort (p < 0.05). After 1:1 propensity score matching with 7,557 matched pairs, Kaplan-Meier survival analysis showed that normal-weight patients had a median survival of 2,094 days. In contrast, the median survival for obese patients was not reached, with 10-year survival probabilities of 41.7% and 54.7%, respectively (log-rank test, p < 0.0001). A hazard ratio of 1.549 (95% CI: 1.465–1.638) indicated a survival benefit in the obese cohort. Conclusions: This study highlights that future prospective studies should focus on body composition analysis and treatment-related variables to further elucidate the mechanisms underlying the obesity paradox in NSCLC and guide targeted therapies for better outcomes in non-obese patients.