Immunity-Related Gene Characterization and Targeted Therapy in Lung Cancer: A Data Mining Approach to Predicting Immunotherapy Response

Non-small cell lung cancer (NSCLC) remains the leading cause of most cancer mortality, while immune-checkpoint inhibitors (ICIs) have transformed treatment in recent years, only a small group of patients show long-term improvement remain insufficiently understood. Herein, study integrated high-throughput platelet transcriptomics with pharmacogenomic and immunogenomic mining to uncover immune-related gene mutations and identify potential therapeutic strategies to enhance patient outcomes. The study curated 228 tumour-educated platelet (TEP) RNA-seq samples from gene expression omnibus (GEO); analyzed NSCLC patients and healthy individuals by comparison of differential expression cancer patients with healthy donors which gave 1,633 significant genes (|log₂FC| > 1.5; FDR < 0.05), prominently including SLC05A1 and RNA28SN family. Machine-learning models, including random forest (RF), support-vector machine (SVM), gradient boosting (GB) and elastic-net logistic regression, trained classifiers that achieved high predictive accuracy (ROC AUC up to 0.96 after SMOTE class balancing), correctly identifying 95% of test samples. Enrichment analyses using Gene Ontology and Reactome highlighted the involvement of olfactory receptor pathways, which is yet to be explored for platelets. Drug repurposing analysis using LINCS L1000 suggested that BET and PI3K–mTOR inhibitors may reverse the platelet-derived transcriptional signature. Immune deconvolution by TIMER, CIBERSORT-ABS, QuanTIseq, xCell, EPIC and MCP-counter linked platelet dysregulation to dendritic-cell and macrophage enrichment, suggesting a systemic platelet-myeloid crosstalk that may suppress anti-tumour immunity. This study provides a data-driven framework for predicting immunotherapy responsiveness and prioritising combination regimens that modulate both tumour-intrinsic and systemic immune pathways.