Abstract 4367317: Leveraging De Novo UK Biobank GWAS to Identify Novel Genetic Loci at the Intersection of Atrial Fibrillation and Cancer

Background: Atrial fibrillation (AF) is a well-established complication of cancer and its treatment. However, the reverse association—cancer arising in individuals with AF—remains poorly understood. This directional relationship may reflect underlying shared genetic architecture, which has not been systematically investigated using individual-level data. Hypothesis: We hypothesized that a de novo genome-wide association study (GWAS) using UK Biobank data could reveal genetic loci uniquely associated with the co-occurrence of AF and specific cancer subtypes—distinct from those implicated in AF or cancer alone—shedding light on biological pathways at the intersection of these conditions. Methods: We conducted a GWAS of AF and cancer co-occurrence in approximately 500,000 UK Biobank participants using individual-level genotype and phenotype data. AF was defined using validated clinical codes. For each of 16 pre-specified cancer types, individuals with both AF and that cancer were compared to participants with neither condition. Genotypes were imputed to the Haplotype Reference Consortium panel. Association testing was performed using mixed-effects logistic regression adjusted for age, sex, genotyping array, and principal components of ancestry. Genome-wide significance was defined as P < 5 x 10^-7. Loci were annotated using OMIM and curated cardiovascular and cancer gene ontology resources. Results: In participants with both AF and non-Hodgkin lymphoma (NHL), we identified a genome-wide significant locus near HK2 (Hexokinase 2) (P = 4.38 x 10^-7), a gene implicated in cardiac metabolism and lymphoma progression. In participants with both AF and colorectal cancer (CRC), we observed a suggestive association near RPGRIP1L (RPGR Interacting Protein 1 Like) (P = 8.67 x 10^-6), a gene involved in ciliary signaling and inflammatory pathways. Conclusions: Our analyses identify HK2 as a novel AF+NHL locus, notable for its divergent roles in cardiac and oncologic biology, suggesting a context-specific mechanism linking these diseases. The signal for RPGRIP1L in AF+CRC, also observed in broader AF-cancer studies, hints at a shared inflammatory or signaling pathway. These subtype-specific findings, which emerged only in co-occurrence models, underscore the utility of intersectional GWAS approaches in uncovering novel genetic contributions to disease interrelationships.