基于蛋白质组学的肺癌免疫细胞浸润分析

Lung cancer remains a leading cause of cancer-related mortality, accounting for nearly one-quarter of all cancer deaths. Its initiation, progression, and metastasis are tightly influenced by the tumor microenvironment, within which diverse immune cell populations play critical roles. Quantifying tumor-infiltrating immune cells is therefore essential for understanding lung cancer biology and improving therapeutic strategies. While numerous computational methods have been developed for immune infiltration analysis based on transcriptomic data, algorithms optimized for proteomics data are largely unavailable, and the applicability of existing transcriptome-based tools to proteomic datasets remains unclear.In this study, we developed a deconvolution-based algorithm tailored for proteomics data to quantify tumor-infiltrating immune cells in lung cancer. Using a support vector regression model constructed from an immune signature matrix, we decomposed tumor tissue proteomics profiles into proportions of distinct immune cell types. Application of this algorithm to lung cancer proteomics datasets revealed that MO non-classical cells, NK cells, and T4-EMRA cells exhibited the highest infiltration levels. Integrating proteomics and clinical data, we further performed NMF-based subtyping and identified two immune-associated subtypes. Cluster 1, characterized by poorer prognosis, showed significant enrichment in complement cascade signaling, IGF transport and uptake regulation, and post-translational protein phosphorylation pathways. Additionally, NK cells and several CD4⁺ T-cell subsets were more abundant in Cluster 1 than in Cluster 2, indicating stronger immune infiltration and elevated immune activity, potentially contributing to adverse clinical outcomes.These findings provide a proteomics-based framework for immune infiltration analysis and offer new insights into lung cancer molecular subtyping and precision oncology.