Skip to main content
Premium Trial:

Request an Annual Quote

Combining Electronic Health Records, GWAS Data Uncovers Variants Linked to Blood Lipid Levels

NEW YORK (GenomeWeb) – Researchers have linked more than 120 SNPs to blood lipid levels using a large cohort and associated electronic health records.

Researchers from the University of California, San Francisco, and the Kaiser Permanente health system conducted a genome-wide association study of nearly 95,000 people. As reported today in Nature Genetics, they combined about 479,000 measurements captured in the subjects' electronic health records (EHRs) with genotyping data to uncover a suite of serum lipid level-linked SNPs.

Because their cohort was so large, the researchers said they were also able to examine differences by ancestry and gender. Additionally, with pharmacy data captured by the EHRs, they developed a genetic risk score to predict when patients would begin lipid-lowering treatments. 

"[O]ur results demonstrate numerous strengths of leveraging large, single cohorts in which longitudinal EHRs with independent measurements are linked to genome-wide data and provide insights into the underlying genetic architecture of plasma lipids to guide future research and clinical care," UCSF's Neil Risch and his colleagues wrote in their paper.

The researchers conducted a GWAS using 94,674 people from the Genetic Epidemiology Resource on Adult Health and Aging (GERA) study, analyzing each ancestry group separately and then in combination in a meta-analysis. Blood lipid measurements had been taken on the participants.

Overall, they noted that African Americans had the highest HDL and LDL cholesterol levels, though the lowest total cholesterol levels. At the same time, South Asians had the highest triglyceride levels, while East Asians had the highest total cholesterol levels.

The researchers uncovered 171 loci that reached genome-wide significance for association with at least one lipid trait, including 46 novel loci. For each of those novel loci, they tested their association with lipid traits in 94,595 individuals of European ancestry from the Global Lipids Genetics Consortium (GLGC) dataset and 460,088 individuals from the UK Biobank. Of those loci, the researchers replicated 17, and a further 14 had nominal significance.

In a combined GERA and GLGC meta-analysis, they teased out another 60 new lipid-associated loci, 32 of which they replicated in the UK Biobank cohort.

Additionally, they used the GERA cohort to conduct conditional analyses through which they found 33 loci with 74 genome-wide significant conditional variants, 15 of which remained significant after further testing.

All in all, Risch and his colleagues reported 121 new lipid-linked loci.They noted that there was biological support for their new SNPs. For instance, one of the SNPs associated with triglyceride levels was a nonsynonymous variant in SLCO1B1, which encodes OATP1B1, a transporter of drugs and ligands into the liver. It was previously linked to statin-induced myopathy and blood metabolites.

The effects of some of these SNPs, though, varied by sex or ancestry, the researchers noted. About 64 percent of the HDL or LDL cholesterol-linked SNPs had stronger effects in women, as did 56 percent of the total cholesterol-linked SNPs.

Using the known, novel, and conditional SNPs linked to lipid levels, the researchers constructed genetic risk scores, which they found to largely vary between the sexes as well as between ancestry groups. African Americans, for example, had the highest mean HDL genetic risk score, but the lowest mean LDL, triglyceride, and total cholesterol scores, and East Asians had the lowest mean HDL genetic risk scores, but the highest mean LDL, triglyceride, and total cholesterol scores.

Risch and his colleagues also combined their genetic risk scores with pharmacy data from the participants' EHRs. Both the LDL and triglyceride genetic risk scores predicted the age at which patients began taking lipid-lowering drugs — typically statins — they reported.

"These findings highlight the value of longitudinal EHRs for identifying new genetic features of cholesterol and lipoprotein metabolism with implications for lipid treatment and risk of coronary heart disease," the authors wrote.