Human metabolome and common complex diseases: A genetic and epidemiological study among African-Americans in the atherosclerosis risk in communities study

A genome-wide association (GWA) study of intermediate traits, such as disease risk factors or functional metabolic intermediates, may increase statistical power and provide biological information into disease mechanism. Metabolomic profiles, the ultimate downstream products of multiple metabolic and cellular activities, are considered potential intermediate phenotypes. A series of studies were conducted to reveal the relationship between metabolomics, genomics and clinical endpoints in the Atherosclerosis Risk in Communities (ARIC) Study. First, I performed a meta-analysis for genome-wide association of cTnT levels among 9,491 European-Americans and 2,053 African-Americans from the ARIC and Cardiovascular Health Study (CHS). A single nucleotide polymorphism, rs10091374, near NCOA2 was identified to be associated with quantitative cTnT levels, and rs12564445 on TNNT2 was identified to be associated with high cTnT category. Second, I performed a GWA study of three incident heart failure related metabolites (pyroglutamine, dihydroxy docosatrienoic acid and X-11787) in 1,260 African-Americans. Significant association was detected for pyroglutamine and X-11787 (rs10463316 and rs13538, respectively), and a genetic risk score created by summing the most significant risk alleles of these three metabolites detected 11% greater risk of heart failure per allele. Third, I assessed the cross-sectional associations of serum metabolites with estimated glomerular filtration rate (eGFR) and longitudinal associations with incident chronic kidney disease (CKD) in 1,921 African Americans. Forty named and 34 unnamed metabolites were found to be associated with eGFR, and two metabolites, 5-oxoproline and 1,5-anhydroglucitol, were further detected to be inversely associated with the risk of developing incident CKD after 19 years of average follow-up. Finally, I assessed the effects of common and rare loss-of-function variants with metabolomics profiles among 1,200 African-Americans. Nineteen significant common variant-metabolite associations were identified, including 15 novel loci. These loci were associated with 7-50% of the difference in metabolite levels per allele. Four novel disease-associated pathways were identified, including two direct longitudinal predictive relationships. Five genes showed a significant burden of rare loss-of-function mutations related to the levels of five metabolites. In summary, these results highlight the value of combining genomics and metabolomics in deeply phenotyped individuals to provide new insights into gene function and disease pathology for African Americans, a population with high disease burden