Coordinated Multicellular Immune Programs and Drug Targets Revealed by Single-Cell Analysis in Driver-Mutated NSCLC

Oncogenic driver mutations in non-small cell lung cancer (NSCLC) activate defined signaling pathways that sustain tumor growth and influence the immune landscape. Yet, how coordinated interactions among diverse cell populations within the tumor immune microenvironment (TIME) contribute to this process remains largely unresolved. To address this, we profiled approximately 200,000 single cells from 45 treatment-naïve NSCLC patients representing seven major driver mutations. This analysis uncovered five multicellular modules (CM1-5) with distinct functional properties, each linked to specific malignant regulatory programs. Among them, CM2 and CM5 exhibited pronounced invasive features and were associated with unfavorable clinical outcomes. CM2 was predominantly observed in EGFR- and MET-driven brain metastases and was defined by strong crosstalk between astrocytes and myofibroblasts. Factors such as SPP1, PTN, and PSAP, together with metabolic alterations, contributed to a microenvironment supportive of metastatic colonization in the brain. By contrast, CM5 was enriched in ROS1-, KRAS-, and EGFR-mutant tumors and consisted of diverse myeloid and endothelial subsets characterized by immunosuppressive and pro-angiogenic signaling, including MIF, GALECTIN, and RETN, collectively facilitating immune escape and vascular remodeling. We further constructed and validated a driver mutation-specific prognostic signature (DMSP.sig) model integrating receptor-ligand interactions and core transcription factors, which effectively stratified patient survival. Leveraging this model, we also identified potential therapeutic candidates linked to these prognostic features, highlighting opportunities for clinical intervention. In summary, our study delineates how oncogenic drivers give rise to distinct TIME architectures, providing a framework for prognostic assessment and precision immunotherapy in high-risk NSCLC.