A multi-biomarker panel for non-invasive diagnosis, risk stratification, subtype identification, and therapeutic targeting in NAFLD: bioinformatics and experimental insights

BACKGROUND: Nonalcoholic fatty liver disease (NAFLD) lacks reliable non-invasive diagnostic tools and effective therapeutic targets. This study aimed to identify serum biomarkers for diagnosis, risk stratification, and therapeutic targeting. MATERIALS AND METHODS: Inflammation- and NET-related differentially expressed genes (INRDEGs) were screened from GEO datasets. LASSO/SVM regression constructed a diagnostic model, validated via ROC/calibration/decision curve analyses. Immune infiltration (CIBERSORT/ssGSEA), consensus clustering, and interaction networks (miRNA/TF/drug) were employed. Hub genes were validated in diet-induced NASH mice. RESULTS: TNF, IL6, CCL5, and SPP1 were identified as key biomarkers. The diagnostic model achieved high accuracy (AUC = 0.803) and enabled risk stratification: TNF, SPP1, and CCL5 marked high-risk patients, while IL6 dominated low-risk subgroups. Two NAFLD subtypes with distinct immune profile were classified. Key biomarkers were identified as potential therapeutic targets. Regulatory networks predicted repurposable drugs. Animal experiments confirmed elevated serum levels of these genes in NASH mice (p < 0.001). CONCLUSIONS: This study establishesed TNF, IL6, CCL5, and SPP1 as a multi-functional biomarker panel for NAFLD, enabling non-invasive diagnosis, risk stratification, subtype classification, and target-specific therapy. These findings provide new insights for the management of NAFLD and a feasible theoretical basis for clinical translation.