Abstract
BACKGROUND: Temporomandibular disorders (TMD) show substantial clinical and genetic overlap with anxiety, yet it remains unclear whether TMD risk reflects shared anxiety-related liability or distinct anxiety-independent genetic mechanisms. Disentangling these components is essential for understanding TMD heterogeneity beyond symptom-based classifications. METHODS: We applied GWAS-by-subtraction using genome-wide summary statistics for TMD (20,799 cases and 479,549 controls; FinnGen Release 12) and anxiety disorders (74,973 cases and 400,243 controls), partitioning TMD heritability into two orthogonal latent components: an anxiety-dependent factor (F(Anxiety)) and an anxiety-independent factor (F(Non-Anxiety)). To delineate the mechanisms underlying each component, we integrated fine-mapping, transcriptome- and proteome-wide association analyses, genetic colocalization, brain imaging–genetics, and single-cell RNA sequencing from human embryonic temporomandibular joint tissue. RESULTS: Anxiety showed significant genetic correlation with TMD (rg = 0.4417, p = 1.98 × 10(− 1 9)) and accounted for 19.50% of TMD heritable variance. F(Anxiety) yielded multiple genome-wide significant loci (CNTNAP5, PCLO, PRSS16, BTN1A1, RAB27B), whereas F(Non-Anxiety) produced a single independent signal near GPNMB, demonstrating sharply divergent genetic architectures. Multi-omic integration identified RAB27B as a driver of the anxiety-related pathway, implicating synaptic vesicle trafficking and neuroimmune regulation, while GPNMB and KLHL7 supported anxiety-independent pathways involving musculoskeletal remodeling and peripheral inflammation. BrainXcan analyses showed that F(Anxiety) predominantly affected limbic and external capsule microstructure, whereas F(Non-Anxiety) mapped to thalamic–sensorimotor white matter networks. Single-cell mapping further revealed distinct enrichment patterns of RAB27B, KLHL7, and GPNMB across TMJ cell types. CONCLUSION: These findings demonstrate that TMD genetic liability comprises separable anxiety-related and anxiety-independent dimensions with distinct molecular, neurostructural, and cellular signatures. Rather than defining clinical subtypes, these latent components represent associative dimensions of genetic risk at the population level. This integrative framework clarifies the genetic architecture underlying TMD heterogeneity and provides a foundation for future studies integrating individual-level phenotyping to assess clinical relevance and causal mechanisms. GRAPHICAL ABSTRACT: [Image: see text] SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s10194-026-02304-3.