Abstract 489: A Data Driven Approach to Identify Subtypes of Abdominal Aortic Aneurysm With Distinct Clinical and Genetic Characteristics

Abstract

Background: Abdominal aortic aneurysm (AAA) is a complex and heterogeneous disorder. We hypothesized that dense phenomapping using electronic health records (EHR) can identify subtypes of AAA with distinct clinical and genetic characteristics. Methods & Results: AAA cases (n=1108) and controls (n=4694) were identified from the Mayo vascular disease biorepository and had available high-density genotyping data. 132 Candidate genetic variants (p < 10E-5) associated with atherosclerotic cardiovascular disease (ASCVD) or AAA were selected from GWAS catalog. Using model-based Gaussian-mixture cluster analysis of 46 clinical features from 6 data domains, 24 data elements, 204 variables (including age, sex, vital signs, laboratory data, medication use, family history), spanning over 20 years of observational time in the EHR, we identified two subtypes of AAA: subtype 1(S1, n=421) with faster AAA growth rate (baseline size adjusted mean difference, 95% CI: 0.013, 0.001 to 0.020 cm / year faster, p = 0.03), higher all-cause mortality (age & sex adjusted hazard ratio, 95% CI: 1.36, 1.09 - 1.70, p < 0.01) than subtype 2 (S2, n=687). As compared with the controls, thyroid disorder / rheumatoid or osteoarthritis / infectious conditions were uniquely associated with increased odds ratio for S1; ASCVD / hyperlipidemia / hypertension were uniquely associated with increased odds ratio for S2. After adjustment for all group-specific diseases, genetic variants in CDKN2B-AS1, NOA1-REST, ERG, ZNF335-MMP9 and SMAD3 were associated with S1 and variants in MMP12, DAB2IP, LHFPL2, LPA, FSTL5 and SORT1 were associated with S2 (all FDR p < 0.05). After adjustment for disease comorbidities and genetic variants differently associated with S1 and S2, the increased risk for all-cause death of S1 than S2 and the difference in AAA expansion rate in subgroups attenuated (Both p-values > 0.09) Conclusions: We identified two subtypes of AAA with different rates of aneurysm expansion and all-cause mortality, which were associated with subtype-specific disease comorbidities and genetic markers, suggesting the potential of leveraging EHR to facilitate individualized medicine in patients with AAA.