The impact of global and local Polynesian genetic ancestry on complex traits in Native Hawaiians

Abstract

Epidemiological studies of obesity, Type-2 diabetes (T2D), cardiovascular diseases and several common cancers have revealed an increased risk in Native Hawaiians compared to European- or Asian-Americans living in the Hawaiian islands. However, there remains a gap in our understanding of the genetic factors that affect the health of Native Hawaiians. To fill this gap, we studied the genetic risk factors at both the chromosomal and sub-chromosomal scales using genome-wide SNP array data on ~4,000 Native Hawaiians from the Multiethnic Cohort. We estimated the genomic proportion of Native Hawaiian ancestry ("global ancestry," which we presumed to be Polynesian in origin), as well as this ancestral component along each chromosome ("local ancestry") and tested their respective association with binary and quantitative cardiometabolic traits. After attempting to adjust for non-genetic covariates evaluated through questionnaires, we found that per 10% increase in global Polynesian genetic ancestry, there is a respective 8.6%, and 11.0% increase in the odds of being diabetic (P = 1.65×10-4) and having heart failure (P = 2.18×10-4), as well as a 0.059 s.d. increase in BMI (P = 1.04×10-10). When testing the association of local Polynesian ancestry with risk of disease or biomarkers, we identified a chr6 region associated with T2D. This association was driven by an uniquely prevalent variant in Polynesian ancestry individuals. However, we could not replicate this finding in an independent Polynesian cohort from Samoa due to the small sample size of the replication cohort. In conclusion, we showed that Polynesian ancestry, which likely capture both genetic and lifestyle risk factors, is associated with an increased risk of obesity, Type-2 diabetes, and heart failure, and that larger cohorts of Polynesian ancestry individuals will be needed to replicate the putative association on chr6 with T2D.

Fig 1. Stratified association testing between global genetic ancestry and BMI.

Individuals were stratified based on T2D disease status. Cases are colored in darker color, controls in lighter color. The association coefficient between ancestry component and BMI within strata are shown (S18 Table). Error bars reflect the estimated standard error from the regression model. P-values for significant association coefficients are labeled. The strongly significant association between PNS ancestry and BMI among T2D controls, but not cases, is suggestive of an interaction between PNS ancestry and T2D.

Fig 2. Manhattan plot of admixture mapping results for T2D.

Dotted line denotes the genome-wide significance threshold for each trait at 2.2x10^-5, determined through permutation.

Fig 3. Association signals with T2D in the broad region of chr 6.

The top panel depicts association signals from admixture mapping on 3,428 unrelated individuals. The middle panel depicts the single variant association result of the same region, using linear mixed model on all 3,940 individuals with genotype dosages imputed from 1000 Genomes Project. The bottom panel shows the corresponding genomic coordinates and nearby genes. Highlighted region at the bottom indicates the signal region as defined in Table 2. LD between rs370140172 and other variants in the region, as shown in the color legend, was r2 calculated in sample using the imputed genotypes.