Abstract
IMPORTANCE: The optimal sequencing of radiotherapy (RT) combined with immunotherapy (iRT) and the value of chemotherapy remain undefined for advanced non-small cell lung cancer (NSCLC), where randomized data are limited. OBJECTIVE: To compare real-world overall survival (OS) between sequential and concurrent iRT in newly diagnosed advanced NSCLC, assess the effect of immune checkpoint inhibitor (ICI) maintenance after RT in refractory disease, and evaluate the association of chemotherapy with survival. DESIGN, SETTING, AND PARTICIPANTS: This is a territory-wide study (OCEANUS) based on the Hong Kong Hospital Authority Clinical Data Analysis and Reporting System (more than 90% population coverage). Patients with NSCLC diagnosed from January 1, 2010, to December 31, 2021, who subsequently received iRT for advanced or refractory disease were included. Overlap weighting was the primary propensity score-weighted method, with inverse probability of treatment weighting used for sensitivity analysis. Data were analyzed from December 2024 to April 2025. EXPOSURES: Sequential vs concurrent iRT for newly diagnosed advanced NSCLC; RT with vs without ICI maintenance for refractory NSCLC; receipt of chemotherapy. MAIN OUTCOMES AND MEASURES: The primary outcome was real-world OS after landmark, estimated with weighted Kaplan-Meier and Cox models. When proportional hazards were violated (per Schoenfeld residuals), treatment effects were summarized using restricted mean survival time. RESULTS: Of 3522 patients who received ICIs, 335 received RT, including 155 with newly diagnosed advanced and 180 with refractory NSCLC. Of these, 247 (73.7%) were male, and the median (range) age was 64 (34-90) years. In newly diagnosed NSCLC, patients treated with sequential iRT had significant longer real-world OS than those treated with concurrent iRT (median, 20.3 months [95% CI, 13.3 to not reached] vs 16.0 months [95% CI, 8.3-30.0]; adjusted hazard ratio, 0.68; 95% CI, 0.47-0.99; P = .045). Chemotherapy was also associated with longer survival in patients with newly diagnosed advanced NSCLC. In refractory NSCLC, RT with ICI maintenance was associated with a numerically longer median real-world OS (11.2 months [95% CI, 7.9-20.6] vs 6.7 months [95% CI, 4.4-17.4]; P = .20). Addition of chemotherapy was not significant for real-world OS. Inverse probability of treatment weighting analyses produced similar estimates. CONCLUSIONS AND RELEVANCE: In this cohort study, sequential iRT was associated with longer survival than concurrent iRT in patients with newly diagnosed advanced NSCLC, and chemotherapy was associated with longer survival. In patients with refractory NSCLC who survived at least 90 days, RT with ICI maintenance resulted in nonsignificantly longer survival and an unclear association with chemotherapy. These findings are hypothesis generating and support prospective randomized studies to define optimal sequencing of iRT and use of systemic treatment partners.