Document resource
Background Non-small cell lung cancers (NSCLCs) harboring ROS1 rearranged, RET rearranged, or BRAF V600E mutations are rare (<2% each) and traditionally considered ‘oncogene-addicted’ with limited benefit from immune checkpoint inhibitors (ICIs). However, our multi-omics profiling of >200 resected NSCLC specimens revealed that these three subtypes share an ‘inflamed’ tumor microenvironment—characterized by high interferon signaling, intact antigen-presentation machinery, and elevated immunogenic cell-death gene expression—features predictive of ICI responsiveness.Methods We retrospectively identified 177 advanced NSCLC patients (75 ROS1-rearranged, 53 RET-rearranged, 48 BRAF V600E-mutant) treated at the Chinese Academy of Medical Sciences Cancer Hospital and Shanxi Cancer Hospital from 2018–2022 (median follow-up 26.5 months), all of whom received platinum-based chemotherapy (± bevacizumab) in first- or second-line settings; 86 patients also received concurrent PD-1/PD-L1 inhibitors (table 1). To emulate a randomized trial, we applied a target trial emulation framework: within each mutation subtype, we fitted a logistic regression propensity score model (covariates: age, sex, stage, brain metastases, PD-L1 TPS, bevacizumab use, treatment line) and computed overlap weights to balance observed confounders. Weighted Kaplan-Meier curves and Cox proportional hazards models compared progression-free survival (PFS) between chemo+ICI and chemo-alone groups; two-sided p<0.05 indicated significance.Results Overlap weighting achieved excellent covariate balance (all standardized differences <0.05). In weighted analyses, chemo+ICI significantly prolonged PFS in ROS1 rearranged NSCLC (median PFS 10.8 vs 6.2 months; HR 0.43, 95% CI 0.25–0.74; p=0.008) and in RET rearranged NSCLC (median 11.2 vs 7.1 months; HR 0.55, 95% CI 0.33–0.92; p=0.015). BRAF V600E mutant patients showed a favorable trend (median 9.9 vs 7.8 months; HR 0.68, 95% CI 0.42–1.10; p=0.10). Consistent with multi-omics results, ROS1 and RET tumors exhibited durable PFS gains when ICI was added (figure 1).Conclusions Our integrated multi-omics and real-world analysis demonstrates that ROS1- and RET-driven NSCLCs possess immune-activated molecular features and derive significant PFS benefit from platinum-based chemo-immunotherapy. BRAF V600E tumors show a similar trend. Given the low incidence of these subtypes, prospective randomized trials are impractical; instead, target trial emulation with overlap weighting offers a rigorous approach to approximate RCT evidence. These findings support considering chemo-immunotherapy regimens in ROS1/RET NSCLC and highlight advanced causal-inference methods as pivotal for evaluating treatments in rare oncogene-positive populations.Abstract 620 Table 1Baseline characteristics of the original cohort, comparing ICI-containing regimens versus chemotherapy without ICIAbstract 620 Figure 1