26. MULTI-ANCESTRY STRUCTURAL VARIANT ASSOCIATIONS WITH ALCOHOL USE DISORDER USING SRWGS FROM THE ALL OF US RESEARCH PROGRAM

Background Alcohol use disorder (AUD) genetic risk is only partly captured by common single-nucleotide variants identified through genome-wide association studies (GWAS). Structural variants (SVs), a broad class of genomic alterations including deletions (DELs), duplications (DUPs), insertions, and inversions, represent a major but understudied source of genetic variation that may contribute to its heritability. The All of Us Research Program (AoU) released a well-characterized SV dataset derived from short-read whole genome sequencing (srWGS) data across ancestrally diverse populations, providing an opportunity for an initial genome-wide exploration of SV associations with AUD. Methods Structural Variants were called from 97,061 participants in AoU using GATK-SV pipeline, followed by AoU’s rigorous quality controls (QC). Participants were stratified into three ancestry groups: European (EUR, N=47,162), African (AFR, N=23,625), and Admixed American (AMR, N=15,112). AUD diagnosis was defined using SNOMED codes from electronic health records (Cases: EUR=2,296, AFR=1,717, AMR=662). In this preliminary analysis, we kept DELs and DUPs passing QC and supported by at least two calling algorithms and two types of evidence. Overall and rare SV burdens in AUD were evaluated using a logistic regression model, comparing DEL and DUP burden across multiple SV size classes. GWAS of individual SVs was performed using REGENIE with Firth correction. Single-ancestry analyses were conducted and followed by meta-analysis using METAL. All these analyses were adjusted for age, sex, and first ten genetic principal components. Results A total of 109,002 DELs and 49,058 DUPs were identified across 85,899 participants, with a median of 181 DELs and 117 DUPs per sample. Rare deletion burden was consistently associated with increased AUD risk across multiple size classes (ORs ranging from 1.015 to 1.060, P=1.26E-06 to 2.32E-10), while rare duplication burden showed a modest inverse association (OR=0.995, P=6.24E-03). DELs and DUPs in the 50-500bp range showed divergent associations with AUD risk (DEL: OR=0.944, P=5.57E-12; DUP: OR=1.015, P=2.87E-07). In single-variant analyses, a common duplication at chromosome 19p13.3 (size: 4.2kb, AF=10.1%) was associated with reduced AUD risk in individuals of European ancestry (OR=0.756, P=1.86E-7), with no reported associations at this locus in prior GWAS. In the multi-ancestry meta-analysis, the top signal was a deletion at 8q24.11 (size:1.8kb, OR=1.172, P=5E-05), overlapping EXT1 (Exostosin Glycosyltransferase 1). Discussion This study utilized a large-scale SV callset from srWGS in an ancestrally diverse cohort to explore genetic associations with AUD. A significant rare DEL burden is consistently associated with increased AUD risk across multiple size classes, aligning with patterns observed in other psychiatric disorders. In total burden analysis, DELs and DUPs within the 50-500bp range showed divergent associations with AUD risk – this is a size class previously undetectable by array-based methods. This may reflect gene dosage effects, which could be further examined through burden analyses stratified by functional region. Together with genome-wide SV association findings, these results suggest that the genetic architecture of AUD at the SV level is shaped by variant type, size, frequency, and ancestry. Future work will extend to other SV types and seek expansion and replication in the UK Biobank to further define the role of SVs in AUD heritability.