Common and rare genetic markers of lipid variation in subjects with type 2 diabetes from the ACCORD clinical trial

Abstract

BackgroundIndividuals with type 2 diabetes are at an increased risk of cardiovascular disease. Alterations in circulating lipid levels, total cholesterol (TC), low-density lipoprotein (LDL), high-density lipoprotein (HDL), and triglycerides (TG) are heritable risk factors for cardiovascular disease. Here we conduct a genome-wide association study (GWAS) of common and rare variants to investigate associations with baseline lipid levels in 7,844 individuals with type 2 diabetes from the ACCORD clinical trial.MethodsDNA extracted from stored blood samples from ACCORD participants were genotyped using the Affymetrix Axiom Biobank 1 Genotyping Array. After quality control and genotype imputation, association of common genetic variants (CV), defined as minor allele frequency (MAF) ≥ 3%, with baseline levels of TC, LDL, HDL, and TG was tested using a linear model. Rare variant (RV) associations (MAF < 3%) were conducted using a suite of methods that collapse multiple RV within individual genes.ResultsMany statistically significant CV (p < 1 × 10−8) replicate findings in large meta-analyses in non-diabetic subjects. RV analyses also confirmed findings in other studies, whereas significant RV associations with CNOT2, HPN-AS1, and SIRPD appear to be novel (q < 0.1).DiscussionHere we present findings for the largest GWAS of lipid levels in people with type 2 diabetes to date. We identified 17 statistically significant (p < 1 × 10−8) associations of CV with lipid levels in 11 genes or chromosomal regions, all of which were previously identified in meta-analyses of mostly non-diabetic cohorts. We also identified 13 associations in 11 genes based on RV, several of which represent novel findings.

Genetic Variants in HSD17B3, SMAD3, and IPO11 Impact Circulating Lipids in Response to Fenofibrate in Individuals With Type 2 Diabetes

“…We implemented a suite of five rare variant tests that can be divided into two classes, burden and nonburden approaches, as previously described. 47 Burden tests collapse a set of rare variants from a gene into a single variable, which is then tested for association with a phenotype. However, simple burden tests do not account for the direction (positive or negative association) of a rare variant effect.…”

“…Subsequently, the combined P value was corrected for multiple comparisons with an false discovery rate (FDR) approach using the R package, qvalue (version 1.36.0) and q < 0.2 was considered to be statistically significant. 49 Additional details regarding the rare variant analysis implemented here can be found in Marvel & Rotroff et al 47 and the Supplementary Material.…”

Population genetic simulation study of power in association testing across genetic architectures and study designs

“…With larger WGS reference panels like the Haplotype Reference Consortium (HRC; McCarthy et al, 2016), large numbers of genotyped samples can be imputed to gain some insight into rare variation. Despite these limitations, imputation has been used to identify rare variant associations in acute macular degeneration (Helgason et al, 2013), lipid levels in type 2 diabetes patients (Marvel et al, 2017), systemic lupus erythematosus (Martínez-Bueno & Alarcón-Riquelme, 2019), among others. As more diverse reference panels become available (for example, TOPMed; Taliun et al, 2019), imputation in non-European and admixed populations will also improve, particularly for rare variants.…”

“…Imputation accuracy falls off at lower minor allele frequencies (MAF), but the use of large WGS reference panels reduces the threshold of acceptable imputation quality (r 2 > 0.3) to~0.004-0.006% (Taliun et al, 2019) in European and African populations. Despite these limitations, imputation has been used to identify rare variant associations in acute macular degeneration (Helgason et al, 2013), lipid levels in type 2 diabetes patients (Marvel et al, 2017), systemic lupus erythematosus (Martínez-Bueno & Alarcón-Riquelme, 2019), among others. It is possible that additional rare variant association signals can be found in imputed data as imputation quality improves, but it is unclear what the statistical properties of RVATs in this setting are.…”

Identifying individual risk rare variants using protein structure-guided local tests (POINT)

“…Here we focus on proprotein convertase subtilisin/kexin 9 (PCSK9 ), as it is the gene 282 reported to be most highly associated with LDL from the baseline study of Marvel and 283 Rotroff et al [46] [46].…”