308 Machine learning to predict genetic variation and cardiovascular risk in Hispanic patients with Systemic lupus erythematosus

Abstract

OBJECTIVES/GOALS: Cardiovascular disease (CVD) is the most common cause of death in systemic lupus erythematosus (SLE). Genome-wide association studies have identified single nucleotide polymorphisms linked with CVD risk, but the association with SLE is not well established. We aimed to determine associations using machine learning in a multi-ethnic lupus cohort. METHODS/STUDY POPULATION: We will use data from the established SLE cohort study named Genetic Profile Predicting the Phenotype (PROFILE). PROFILE was constituted in 1998 by combining existing cohorts at multiple sites which are also of defined ethnicity (Hispanics of Mexican ancestry and Puerto Rico, African American, and Caucasian). The cohort consists of 3,118 individuals and the database contains socioeconomic–demographic, clinical, laboratory, and genetic variables. Genetic data consist of 196,524 single nucleotide polymorphisms. To detect risk genes and predict an individual’s SLE risk will design a random forest classifier using SNP genotype data. Logistic regression models will be performed with CVD as the outcome, adjusted for age, sex, ethnicity, disease duration, and traditional and nontraditional risk factors for CVD. RESULTS/ANTICIPATED RESULTS: We expect to find several established and new susceptibility genes associated with CVD. DISCUSSION/SIGNIFICANCE: This approach offers an opportunity to characterize distinct genetic risk factors and the relationship of CVD with SLE. These data may be important in the identification of patients at high risk for such events and may allow the design of preventive strategies which may beneficially have an impact on the morbidity and mortality of SLE patients.