Genetic epidemiology of cardiometabolic biomarkers : twin studies in the genomic era

Abstract

Following the Human Genome Project, many genomic approaches have been developed in genetic epidemiology to investigate the genetic influences on human complex traits. This thesis aims to answer four genetic epidemiological questions for cardiometabolic biomarkers/ traits, by using classical twin studies and novel genomic methods.

Whether the dominant genetic effects are important for “missing heritability”? Heritability is a population specific estimate reflecting the relative importance of genes (versus environment) for human complex traits. “Missing heritability” is the proportion of heritability that remains unexplained by single nucleotide polymorphisms (SNPs). For 24 cardiometabolic traits, the univariate (study I) and bivariate (study II) heritabilities were estimated by using both twin and SNP models, within the same study base (10,682 twins in TwinGene). Study I supports that the main genetic influences on these traits are additive genetic effects (A), but significant contributions from dominant genetic effects (D) are also identified for certain traits. D effects are often masked by shared environment (C) in twin studies, thus D might have a more prominent role than what the estimates often suggest. It is difficult to distinguish D from A in too small twin studies, so the “missing heritability’’ might be overestimated if all genetic influences (A and D) are erroneously attributed to the narrow-sense heritability.

What’s the pattern of genetic and environmental contributions to the covariation between cardiometabolic traits? Study II demonstrates that the pattern varies by different clusters of cardiometabolic traits. Additive genetic effects (A) and non-shared environment (E) influence the covariation between blood pressure traits. Besides A and E, dominant genetic effects appear to be important for the covariation between obesity traits. However, shared environmental contri-butions seem generally to be weak between cardiometabolic traits in TwinGene samples.

Which genetic variants are associated with the novel cardiometabolic biomarker — immunoglobulin M against phosphorylcholine (IgM anti-PC)? By performing genome-wide association study (GWAS) in four Swedish cohorts (total n=3,648), study III identified a haplotype block at 11q24.1 close to the GRAMD1B gene to be the top locus shared between anti-PC and chronic lymphocytic leukemia (CLL). Prediction from bioinformatics suggests that the SNP rs35923643-G in this locus might be the functional variant by impeding the transcription factor binding. A small nested case-control study indicates a potential reverse causation between anti-PC and CLL.

Whether the associations between blood lipids and amyotrophic lateral sclerosis (ALS) are causal? By using summary GWAS results (~100,000 individuals for blood lipids and ~30,000 for ALS) in the polygenic risk score and Mendelian randomization settings, study IV tested the association and causality between blood lipids and ALS. It supports that high levels of low-density lipoprotein (LDL) and total cholesterol (TC) are risk factors for ALS. Based on current assumptions and evidence, it also suggests potential causal effects of LDL and TC on ALS.

In summary, this thesis quantified the proportion of genetic contributions to the variation (study I) and covariation (study II) for 24 traditional cardiometabolic biomarkers/traits; it identified the genetic variants (common SNPs) associated with novel biomarker IgM anti-PC (study III); it also tested whether polygenic evidence supports the association and causality between blood lipids and ALS (study IV). In general, the thesis suggests that twin studies have continuing important values for genetic epidemiology in the genomic era.