Integration of Machine Learning, Bioinformatics, and Experimental Validation to Identify Novel Diagnostic and...

Lung adenocarcinoma (LUAD) is the most prevalent and lethal subtype of non-small cell lung cancer (NSCLC), with its progression closely associated with aberrant succinylation modifications. This study aimed to systematically identify succinylation-related genes in LUAD and evaluate their diagnostic and prognostic significance. By integrating four Gene Expression Omnibus (GEO) datasets, 45 differentially expressed succinylation-related candidate genes were identified. Feature selection using three machine learning methods-Lasso regression, support vector machine recursive feature elimination (SVM-RFE), and Random Forest-yielded seven core genes: TIMP1, SLC2A1, JUP, F12, B3GALNT1, DSP, and MMP1. ROC analysis showed that all core genes achieved AUC values greater than 0.7, indicating strong diagnostic potential. A diagnostic model constructed from these seven genes achieved an AUC of 0.912 in the training cohort, significantly outperforming single-gene models, and was validated in The Cancer Genome Atlas (TCGA) cohort (AUC = 0.893). Prognostic analysis revealed that Kaplan-Meier curves for all seven core genes demonstrated p < 0.05 and HR > 1, indicating that high expression was associated with poor outcomes. A risk prediction nomogram was also developed based on these genes. SHAP analysis clarified each gene's contribution to the model, while drug enrichment and transcriptional regulatory network analyses provided further insights into potential therapeutic targets. Notably, JUP exhibited the highest diagnostic efficacy (AUC = 0.921) and was significantly correlated with immune cell infiltration and tumor microenvironment regulation. Molecular docking suggested stable binding between JUP and potential therapeutic compounds, single-cell analysis confirmed its marked overexpression in tumor and epithelial cells, and experimental validation further established its oncogenic role. In conclusion, this study systematically defines the diagnostic and prognostic value of seven succinylation-related core genes in LUAD, with JUP playing a particularly pivotal role. These findings provide robust evidence supporting its potential as a novel biomarker and therapeutic target.