Abstract
BACKGROUND: Trigeminal neuralgia (TN) is a neurological disorder affecting the maxillofacial region. Current treatment relies on pharmacotherapy and invasive interventions; however, the growing prevalence of drug-resistant cases highlights the need for a mechanistic understanding and novel biomarkers. In the present study, transcriptome-wide association study (TWAS) and Mendelian randomization (MR) approaches were adopted to identify priority causative genes and therapeutic targets, thereby bridging the gap in our understanding of the polygenic etiology and tissue-specific pathophysiology of TN. METHODS: We employed two TWAS frameworks to integrate multi-tissue expression quantitative trait loci (eQTL) data from GTEx v8 with a TN genome-wide association study (GWAS). Candidate loci were systematically validated through a multi-tiered analytical cascade, encompassing MR, summary-based MR (SMR), Bayesian co-localization, and replication across four independent cohorts (BrainMeta v2, PsychENCODE, eQTLGen, and Westra blood). Cell-type specificity was mapped using gsMAP analysis. Besides, a rat model of TN was established to validate the expression of the target. In addition, to elucidate shared risk factors, two-sample analysis quantified the role of Risk-Gene pathways. Finally, phenome-wide association studies (PheWAS) was performed to interrogate potential off-target effects of prioritized therapeutic targets using FinnGen R10 data. Analytical robustness was ensured via sensitivity analyses and Cochran’s Q heterogeneity testing. RESULTS: TWAS prioritized L3MBTL2 as a Multi-tissue regulator, showing consistent protective effects across tissues (OR range 0.56–0.82; smallest P = 9.13 × 10(− 14)) validated by MR, SMR and co-localization (posterior probability > 0.8), leading to its designation as Tier 1. In contrast, CIDEA (Tier 2) exhibited tissue-specific risk in esophageal mucosa (β = 0.42, FDR P = 0.0046) but lacked cross-tissue reproducibility. gsMAP analysis revealed spatial enrichment of TN-associated cells in tissues, including the epidermis. TN animal experiments showed lower expression of L3MBTL2 in the hypothalamic region, confirming our genomic findings. Finally, risk factor analysis identified basal metabolic rate as a shared risk factor for both reduced L3MBTL2 expression and increased TN risk. PheWAS analysis indicated that therapeutic targeting of L3MBTL2 may elicit adverse effects, specifically obesity-related phenotypes and benign uterine fibroids. CONCLUSIONS: This study identifies L3MBTL2 as a high-confidence candidate gene for TN, with experimental evidence supporting its role in the hypothalamus. These findings provide novel insights into TN pathophysiology and suggest that L3MBTL2 warrants further investigation as a potential therapeutic target. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s10194-025-02218-6.