Latent-Class TWAS Reveals Opposing Glutamatergic-Plasticity Architectures and Neuroimmune-Pruning Signals in Alcohol Misuse Subtypes

Alcohol misuse is clinically heterogeneous, and genetic studies increasingly show that alcohol consumption, alcohol use disorder, and problematic alcohol use are related but not interchangeable traits. In this study, we used a latent-class framework to ask whether alcohol-misuse subtypes differ in genetically regulated expression profiles across gene sets related to synaptic plasticity, glutamatergic signaling, pruning, microglia, and complement pathways. Using European-ancestry latent-class GWAS summary statistics from six pairwise class comparisons, we performed transcriptome-wide association analyses and compared class-specific gene-set profiles across brain tissues.The strongest pattern was an inversion of synaptic plasticity and glutamatergic TWAS architecture. The Class 2 internalizing-light contrast was strongly anti-correlated with higher-risk contrasts, especially comparisons involving Class 3 and Class 4. Gene-level influence analyses repeatedly nominated CLN3 and ADORA2A. CLN3 points to lysosomal-autophagic synaptic maintenance as a possible resilience pathway, whereas ADORA2A supports adenosine-dependent regulation of dopamine-glutamate plasticity in corticostriatal circuits. Class 4, the broad high-risk class, showed additional evidence of long-term potentiation, dopaminergic signaling, vesicle release, and suggestive pruning/complement involvement, including TREM2, C1Q, C3, CX3CR1, and APOE. These findings suggest that latent alcohol-misuse classes may reflect partly opposing synaptic programs rather than a simple severity gradient. If replicated and supported by colocalization and functional studies, these signals could guide subtype-informed trial enrichment and mechanistic target prioritization.Keywords: alcohol use disorder; latent class; TWAS; glutamate; synaptic plasticity; ADORA2A; CLN3; microglia; complement; precision psychiatry