Abstract Body (Do not enter title and authors here): The role of genetic ancestry (GA) in hypertensive pregnancy disorders in Latin-American women is poorly understood.

Using data from a multi-center case-control study (GenPE) of preeclampsia (PE) in young Colombian women (median age = 19) of predominantly low socioeconomic status (2364 controls and 1811 cases), who identify as Afro-Caribbean (AFR-C), White Hispanic (HISP), Amerindian, and Mixed ethnicity, we evaluated associations between 1) reported ethnicity, and 2) empirically estimated GA, with PE. We performed 3-way admixture mapping using European (EUR), African (AFR) and Amerindian (AMR) ancestry references from the Human Genome Diversity Project using the FLARE software to estimate local and global ancestry in GenPE samples. Statistical significance threshold, for three-way local ancestry analyses, was empirically estimated using STEAM (P = 3.45x10^-6).

In multivariable logistic regression models for reported ethnicity, AFR-C were 33% more likely to have PE (OR = 1.33; P = 0.02) than HISP women. In models evaluating empirically estimated global GA, AFR was positively associated (OR per 10% increase in ancestry = 1.05; P = 0.002), while AMR (OR = 0.91; P = 0.035) and EUR (OR = 0.95; P = 0.009) were inversely associated with PE. Additionally, adjusting for reported ethnicity in models evaluating global GA and PE changed estimates only marginally for AFR (OR = 1.04; P = 0.025) and EUR (OR = 0.92; P = 0.009). Evaluation of GA and PE in a subset of women who reported AFR-C ethnicity, showed stronger estimates for all global ancestries: AFR (OR = 1.11; P = 0.013, EUR (OR = 0.82; P = 0.026), and AMR (OR = 0.83; P = 0.01). Association analyses with AFR local GA identified three loci associated with PE. The top locus at chromosome 11, rs2021740 (a smooth muscle enhancer in OTOG1 and near MYOD1), each additional allele of AFR origin associated with 27% increased odds of PE (OR = 1.27; P = 1.13x10^-7). The A-allele for this variant is found in greater frequency in AFR reference populations (22%) than in EUR (5%). Subgroup analyses with HELLP syndrome (279 cases and 2364 controls) shows intriguingly opposite findings with increased risk for global AMR and EUR ancestry and decreased risk for AFR ancestry.

Using a genetically diverse hispanic population, we show genetic ancestry is associated with PE independent of reported ethnicity and further demonstrate the power of admixture mapping to identify a candidate locus for PE.