Genetic mapping of between-population variation using mapping by admixture linkage disequilibrium: Methodology and application to lipoprotein(a)

Most genetic association studies focus on within-population, rather than between-population, differences in disease susceptibility. While between-population variation accounts for only ∼5% of human genetic diversity, consequences for disease risk differences across populations are increasingly apparent. Mapping by Admixture Linkage Disequilibrium (MALD) has been proposed as a method for mapping the genetic basis for between-population disease differences when an admixed population is available with genetic contributions from higher-risk and lower-risk populations. The current restriction of study design considerations for MALD to family-based settings has limited application, since many diseases for which MALD could be appropriate occur later in life, when parental genotypes often are unavailable.; For this dissertation, explicit models of admixture dynamics were used to identify and investigate the relevant parameters for population-based MALD (Chapter 2). The resulting analytic strategy was applied to the heritable coronary heart disease risk factor lipoprotein(a) [Lp(a)] level, which is two-fold higher in African than non-African populations (Chapter 3). The etiology of the difference is unknown and has variously been attributed to environment, cis-acting variation at the apo(a) gene (LPA), and trans-acting factors. Analysis in 249 African-Americans and 534 European-Americans in the CHOICE Cohort strongly implicated LPA in higher African Lp(a). Further investigation of LPA candidate variants identified 3 single-nucleotide polymorphisms (SNPs) that explained 60% of the between-population Lp(a) level difference (Chapter 4). Each was uncommon in one or both populations, demonstrating that multiple rare alleles can generate large differences across populations. This suggests that alternative strategies to those currently being developed for genetic association studies—which assume universally common susceptibility—may be required for identifying the genetic basis for between-population risk variation.