Integrative multiomic profiling of cfDNA methylation and EV-miRNAs identifies immunotherapy-outcome molecular subtypes in NSCLC.

[BACKGROUND] Patients with non-small cell lung cancer (NSCLC) exhibit heterogeneous responses to immunotherapy (IT) with high resistance rates, highlighting the need for precise biomarkers of treatment outcomes.

[METHODS] In a prospective cohort study, we longitudinally assessed liquid biopsy samples from patients with NSCLC undergoing IT at four distinct time points (T1 pretreatment, T2 post-second cycle, T3 6 months, and T4 1 year). We profiled plasma-derived cell-free DNA methylation and extracellular vesicle-associated microRNAs from 79 patients with metastatic NSCLC treated with immune checkpoint inhibitors (ICIs). High-dimensional omics data were integrated using Multi-Omics Factor Analysis (MOFA2) to uncover latent molecular subtypes, which we termed MOFA-Derived Clusters (MDCs), independently established at baseline (MDC-T1) and post-second cycle (MDC-T2). Differential expression and methylation analyses, pathway enrichment, and immune phenotyping via flow cytometry were used to characterize the molecular and immunological landscape of each MDC. External validation was performed using independent NSCLC cohorts for miRNAs (Genova , 2024, n=54) and methylation (SMC Cohort, GSE119144, n=57).

[RESULTS] MDCs captured divergent survival outcomes and reflected biologically coherent processes including angiogenesis, cytoskeletal remodeling, and immune signaling. Projection of MDCs onto later time points (T3, T4) supported the temporal relevance of early molecular signatures. MDCs also displayed immunological correlates via circulating immune cell subsets. Importantly, MDC classifiers demonstrated consistent survival stratification in external cohorts, particularly MDC-T2.

[CONCLUSION] This study defines a multiomic, liquid biopsy-based framework for molecular subtyping in NSCLC to manage ICI treatment. Our MDC signatures reveal clinically meaningful, treatment-informative biology and offer a path toward minimally invasive patient stratification in immuno-oncology.