Prefrontal cortex-extracellular vesicle RNAs coupling in alcohol use disorder: towards non-invasive biomarkers and...

Alcohol use disorder (AUD) poses a substantial global health burden, with its molecular mechanism remaining elusive. This study integrated datasets (GSE253155, GSE180722, GSE182173 and GSE176122) and clinical specimens to identify non-invasive biomarkers and therapeutic candidates for AUD. Transcriptomic analysis of post-mortem prefrontal cortex identified 34 differentially expressed genes (DEGs), significantly enriched in TNF/IL-17 signalling, ribosome, cytokine-cytokine receptor interaction and endoplasmic reticulum protein processing. Among multiple machine learning algorithms, the support vector machine demonstrated superior classification performance, prioritizing MDK, FAM225B and SERPINA3 as critical biomarkers. These were incorporated into a highly predictive diagnostic nomogram. Intersection with plasma extracellular vesicle (EV) RNAs showed that all three biomarkers were detectable and upregulated in AUD patients versus controls, which was confirmed by quantitative real-time PCR (qRT-PCR) in a clinical validation cohort. SERPINA3 exhibited a particularly robust association with AUD, maintaining significance after adjustment for confounders and suggesting direct effects via mediation analysis. Cross-species validation confirmed a conserved dysregulation of murine Serpina3n. Computational drug repurposing identified loratadine, doxepin, citalopram and imipramine as promising therapeutic candidates targeting SERPINA3. Collectively, this work delineated PFC signature genes, proposed a EVs-supported brain-periphery molecular crosstalk, and provided translatable biomarkers and repurposable candidates for AUD precision medicine.