GENOME-WIDE ASSOCIATION STUDY OF OPIOID USE DISORDER AND LONG-TERM OPIOID EXPOSURE REVEALS 62 LOCI AND GENETIC OVERLAP WITH PAIN

In 2024, over 7.6 million Americans misused prescription opioids and 4.8 million had opioid use disorder (OUD). Electronic health records (EHRs) linked with biobanks provide an opportunity to increase sample sizes for genome-wide association studies (GWAS). Case identification in EHRs typically relies on International Classification of Diseases (ICD) codes. However, OUD ICD codes are often underutilized. Prior work across Vanderbilt, Geisinger, and Mass General Brigham showed that individuals with OUD ICD codes and those with long-term opioid prescriptions share similar phenotypic profiles, suggesting overlap. However, the impact of combining these groups for genetic studies is unclear. We conducted two OUD GWAS across five EHR-linked biobanks (N=1,067,513 individuals of European-, African-, and Admixed American-like genetic ancestry). Two case definitions were applied: (a) individuals with at least one OUD ICD code, and (b) individuals with at least 10 opioid prescriptions within 12 months. Based on these definitions, we performed GWAS of (1) OUD ICD code only (OUD ICD) and (2) OUD ICD code and/or long-term opioid prescription (OUD+longTerm Rx). We meta-analyzed European-like summary statistics (N=860,493), and calculated SNP-based heritability and genetic correlations with 12 other complex traits. To examine enrichment of pain loci, we accounted for pain-mediated signals in the OUD+longTerm Rx analysis by conditioning on a previously published pain intensity GWAS. The meta-analysis of OUD ICD GWAS identified 14 loci, 7 of which were novel, and the meta-analysis of the OUD+longTerm Rx GWAS identified 62 loci, including 41 novel loci relative to prior OUD and pain GWAS. The top locus in both analyses mapped to OPRM1 (OUD ICD: rs1799971, P=2.73e-16; OUD+longTerm Rx: rs3778147, P=2.11e-21). Both GWAS showed strong genetic correlation with OUD (rg=0.93±0.02 for OUD ICD; rg=0.73±0.03 for OUD+longTerm Rx). The OUD+longTerm Rx GWAS had a higher genetic correlation with chronic pain (rg=0.44±0.03 vs 0.67±0.02). After conditioning on pain intensity, 14 loci remained significant in the OUD+longTerm Rx analysis, 12 of which were not identified in the OUD ICD GWAS and included loci associated with risk-taking, smoking initiation, and alcohol use disorder measurement. The conditioned results showed a strong genetic correlation with OUD (rg=0.74±0.04) and a modest correlation with chronic pain (rg=0.09±0.04). Our findings show that including individuals with long-term opioid prescriptions alongside those with OUD ICD codes can improve discovery power in EHR-based biobanks, identifying a broader phenotype with substantial shared genetic liability with OUD, retaining canonical OUD signal, and yielding new loci that are OUD-relevant even after accounting for pain. Cross-ancestry and African-like meta-analyses are ongoing.