Human genetic insights into lipoproteins and risk of cardiometabolic disease

Abstract

Purpose of review

Human genetic studies have been successfully used to identify genes and pathways relevant to human biology. Using genetic instruments composed of loci associated with human lipid traits, recent studies have begun to clarify the causal role of major lipid fractions in risk of cardiometabolic disease.

Recent findings

The causal relationship between LDL cholesterol and coronary disease has been firmly established. Of the remaining two major traits, recent studies have found that HDL cholesterol is not likely to be a causal particle in atherogenesis and have instead shifted the causal focus to triglyceride-rich lipoproteins. Subsequent results are refining this view to suggest that triglycerides themselves might not be causal but instead may be a surrogate for the causal cholesterol content within triglyceride-rich lipoproteins. Other studies have used a similar approach to address the association between lipids and lipoproteins and risk of type 2 diabetes. Beyond genetic variation in the target of statin medications (HGMCR), reduced LDL cholesterol associated with multiple genes encoding current or prospective drug targets associated with increased diabetic risk. In addition, genetically lower HDL cholesterol and lower triglycerides appear to increase risk of type 2 diabetes.

Summary

Results of these and future human genetic studies are positioned to provide substantive insights into the causal relationship between lipids and human disease and should highlight mechanisms with important implications for our understanding of human biology and future lipid-altering therapeutic development.

Introduction

Several pragmatic motivating factors continue to encourage ongoing human genetic studies focused on resolving the complete genetic architecture of plasma lipid traits. First, every locus identified for these traits has the potential – when resolved to causal variant(s) and gene(s) – to contribute to our understanding of the biology underlying lipoprotein metabolism in humans[1, 2]. Importantly, some loci will emerge from these studies as potential targets for therapeutically manipulating plasma lipids[3, 4]. Finally, resolving the full contribution of rare and common alleles to these traits also has substantial implications for genetic diagnoses in the dawning era of precision medicine.

Considerable progress has already been made in identifying genetic loci that are robustly associated with variability in plasma lipids. In the most recent genome-wide association study (GWAS) by the Global Lipids Genetics Consortium which included nearly 187,000 individuals, investigators identified 157 independent genetic loci which were significantly associated with at least one of four major lipids and lipoproteins: total cholesterol (TC), low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides (TG)[1]. Two observations may lead some to speciously question the utility of performing additional gene mapping studies to identify new loci and resolve existing association signals: current studies only explain a small fraction of the population variance in the trait despite already substantial sample sizes (leading some to wonder if this is the correct approach), and each novel locus is typically associated with increasingly smaller effects on the trait (otherwise the association would likely have been previously identified). If the goal of trait mapping is to understand biology and identify therapeutic targets, however, the variance attributable to either individual or the sum of all loci together is irrelevant. For example, a common variant (rs17238484) in the HMG Co-A Reductase gene (HMGCR) has a very modest effect on LDL cholesterol in the population (1.08 mg/dL lower LDL cholesterol for each copy of the minor allele)[5]. Similarly, gain of function variants in PCSK9 were identified as a cause of autosomal dominant hypercholesterolemia[3]; these specific variants contribute to nearly zero population variance due to their rarity. Despite this, HMGCR and PCSK9 are key elements in the biological pathways of cholesterol biosynthesis and metabolism and substantially effect plasma LDL levels when therapeutically manipulated[6, 7].

Multiple approaches can be used to identify additional loci underlying plasma lipids and to refine the signals within existing loci. In addition to merely increasing sample size (the most obvious means of augmenting statistical power), recent studies have leveraged denser imputation panels[8, 9], studied populations with different or unique genetics[10, 11], and ascertained families with extreme phenotypes[12–14], all of which have enriched our understanding of human lipoprotein metabolism.

In addition to furthering our knowledge of basic human biology, human genetics has the potential to identify causal mechanisms for disease. From a therapeutic perspective it is obviously critical to understand the relationship between genetic loci and risk of disease, as therapeutic manipulation of plasma lipid levels is not a goal unto itself but rather a means to alter disease. Several studies, reviewed below, have contributed to our understanding of lipoprotein metabolism and risk of cardiometabolic disease.

Human genetic insights into the role of plasma lipids and coronary artery disease

Over 50 years ago, investigators at the Framingham study identified plasma cholesterol as a “factor of risk” for the development of coronary heart disease[15]; subsequent basic science investigation[16] and clinical trial results[6] have firmly established elevated LDL cholesterol as a causal risk factor for coronary artery disease. Prior studies identifying genetic variation in PCSK9 that associated with both altered LDL cholesterol and risk of coronary disease[4] have successfully accelerated therapeutic development[17, 18] (e.g. PCSK9 loss of function variants); several recent human genetic studies have contributed to this effort of identifying genetic variation that associates with lipid traits and risk of coronary disease.

Genetic variation modulating LDL cholesterol and risk of coronary artery disease

Nioi et al[19] undertook such a study in which the investigators started with whole genome sequencing (WGS) in 2,636 Icelanders. They imputed the variants discovered by WGS into 119,146 participants with lipid measurements and identified several correlated single nucleotide polymorphisms (SNPs) that were strongly associated with non-HDL cholesterol and which spanned the genes ASGR1 and ASGR2, encoding subunits of the asialoglycoprotein receptor. Nioi et al observed that an intron of ASGR1 had low WGS coverage likely due to local sequence context and performed additional sequencing in this low coverage area. This led to the discovery of a 12-bp intronic deletion in ASGR1 which appeared to be responsible for the statistical association between this locus and non-HDL cholesterol. The 12-bp deletion, which causes aberrant mRNA splicing and the subsequent synthesis of a truncated ASGR1 protein, was significantly associated with reduced levels of non-HDL cholesterol (-15.3 mg/dL for carriers), which appeared to be primarily due to lower LDL cholesterol in carriers (-12.5 mg/dL for carriers). Beyond an association with non-HDL cholesterol, the investigators also demonstrated that carriers had a 34% reduced risk of coronary disease compared with non-carriers. Somewhat intriguingly, the 12-bp ASGR1 deletion was associated with a reduction in coronary disease risk beyond what one might expect based solely on the magnitude of the decrease in non-HDL cholesterol (this comparison was performed using genetic associations in other genes affecting non-HDL cholesterol and coronary disease risk), suggesting there may be non-lipid atheroprotective effects associated with loss of ASGR1.

In addition to identifying potentially novel targets, a human genetic approach can be used to support existing therapeutic targets. We used this type of approach to demonstrate that loss of function mutations in the Niemann-pick C1-like 1 (NPC1L1) gene were associated with a reduced risk of coronary artery disease[20]. Despite being approved for therapeutic use in many countries, there was substantial uncertainty over whether or not ezetimibe (a non-statin medication that reduces plasma LDL cholesterol by inhibiting NPC1L1) would reduce coronary artery disease [21, 22]. To address this question, we sequenced NPC1L1 and found that approximately 1 in every 650 individuals carried a loss of function allele in NPC1L1. These individuals, when compared with those who did not carry NPC1L1 inactivating mutations, had approximately 12 mg/dL lower LDL cholesterol and 53% reduced risk of coronary artery disease[20]. A large randomized control trial later demonstrated that therapeutic inhibition of NPC1L1 with ezetimibe resulted in lower LDL cholesterol and a significant reduction in risk of recurrent coronary disease[23]. Other protein-altering variants associated with LDL cholesterol and risk of coronary artery disease are listed in Table1.

Table 1.

Protein-altering variants associated with plasma lipids and risk of coronary artery disease

Gene	Protein altering variant(s)	Main plasma lipid alteration	Associated risk of coronary artery disease	Reference
LDLR	Multiple loss of function	Increased LDL cholesterol	Increased	[31]
PCSK9	p.R46L, p.Y142*, p.C697*	Reduced LDL cholesterol	Reduced	[4]
NPC1L1	Multiple loss of function	Reduced LDL cholesterol	Reduced	[20]
LPA	p.I1837M	Increased lipoprotein(a)	Increased	[43]
LPA	Multiple loss of function	Reduced lipoprotein(a)	Reduced	[44]
ASGR1	Multiple loss of function	Reduced non-HDL cholesterol	Reduced	[19]
APOA5	Multiple loss of function	Increased TRL	Increased	[31]
ANGPTL4	p.E40K, multiple loss of function	Reduced TRL	Reduced	[45]
APOC3	Multiple loss of function	Reduced TRL	Reduced	[29, 30]
LPL	p.D36N	Increased TRL	Increased	[45]
LPL	p.S447*	Reduced TRL	Reduced	[45]

LDL = low-density lipoprotein; HDL = high-density lipoprotein; TRL = triglyceride-rich lipoproteins

Are HDL cholesterol and Triglycerides causal factors for coronary artery disease?

Beyond LDL cholesterol, human genetics has the potential to identify lipid phenotypes that are causally related to coronary disease and genes and/or pathways that associate with both causal fractions and risk of disease. While LDL cholesterol has been accepted as a causal factor for coronary artery disease, the epidemiological data implicating HDL and TG as causal factors was unclear. HDL and TG are negatively correlated in population studies and both are correlated with coronary artery disease[24]. The relationship between TG and coronary disease typically disappears, however, when HDL cholesterol is accounted for[24], leading some prior studies to conclude that HDL cholesterol was the likely causal fraction of the two. These types of studies are prone to confounding, however, and approaches using the technique of Mendelian randomization have attempted to address this question for HDL and triglyceride rich lipoproteins (TRL)[25–27]. While previous studies have typically focused on SNPs identified in GWAS studies (where most SNPs are of common frequency in the population), Helgadottir et al[28] recently extended this approach to include low-frequency and rare variants that they discovered to associate with lipid phenotypes. First, using the same lipid associations as were used in Nioi et al[19] discussed above, Helgadottir et al created a genetic risk score (GRS) of associated variants for each of the following: non-HDL cholesterol (defined as HDL cholesterol subtracted from total cholesterol), LDL cholesterol, HDL cholesterol, and triglycerides. They then examined the association between these GRSs and risk of coronary artery disease. This analysis demonstrated a clear causal signal for LDL cholesterol with risk of disease and, in keeping with prior studies, also found no evidence to support a causal role of HDL cholesterol in risk of coronary disease.

In contrast to prior studies, however, the results from Helgadottir et al further support a causal role of non-HDL cholesterol (distinct from that of LDL cholesterol) as opposed to triglycerides themselves. Non-HDL cholesterol is, by definition, all of the cholesterol content in the plasma not carried in HDL and includes cholesterol in very low density lipoproteins, intermediate density lipoproteins, chylomicrons, and lipoprotein(a). Thus, the results of Helgadottir et al may be suggesting that the causal role of TRLs is related to the cholesterol content of the fraction and not the triacylglycerols themselves. While perhaps a subtle distinction, clarifying this mechanistically may have important implications for future therapeutic development. For example, approaches focused on modifying the triacylglycerol content in TRLs may not be as effective for treating coronary disease as lowering total TRL concentration (which would reduce both triacylglycerols and cholesterol content in the plasma). Further development in this area is needed to clarify these points.

Current human genetic evidence is coalescing around lipoprotein lipase (LPL) as one potential mechanism for altering TRL concentration in a manner that alters coronary disease risk, beginning with the observation that loss of function mutations in APOC3 were associated with lower concentrations of TRLs and reduced risk of coronary disease[29, 30]. APOC3 is an endogenous inhibitor of LPL, the latter of which functions to hydrolyze triglycerides present in circulating lipoproteins, thereby lowering plasma TRL concentration. As an endogenous inhibitor of LPL, loss of APOC3 function should result in higher LPL activity and loss of an endogenous activator of LPL should result in the opposite effects. Do et al.[31], after sequencing the exomes of nearly 10,000 individuals with and without coronary disease discovered loss of function variants in apolipoprotein A-V (APOA5) – an endogenous LPL activator – were associated in a manner that one would expect, where loss of APOA5 associated with higher triglycerides and increased risk of coronary disease. These observations appear to be somewhat generalizable, as loss of function in an additional endogenous LPL inhibitor (ANGPTL4) was recently found to associate with reductions in plasma TRLs and lower risk of coronary disease[32]. Similarly, these findings are not limited just to regulatory proteins, as loss and gain of function variants in the LPL gene itself are associated with the a priori expected alteration in TRL concentration and risk of coronary disease[32]. This allelic series in LPL and its endogenous regulators, in which loss and gain of gene function modulate both the biomarker and risk of disease in the expected direction of effect, increases the confidence in this pathway as a potential target for therapeutic intervention[33]. A partial listing of other protein-altering variants associated with altered plasma lipids and risk of coronary artery disease is provided in Table 1.

Human genetic insights into the role of plasma lipids and type 2 diabetes

HMG Co-A Reductase, LDL cholesterol, and risk of diabetes

As mentioned above, the relationship between plasma lipid traits and coronary disease has been long recognized. Similarly, prior studies have found a relationship between increased risk of type 2 diabetes with reduced HDL cholesterol and elevated TG concentrations (without any such observation for LDL cholesterol)[34–36]. Despite the epidemiological lack of association between LDL cholesterol and risk of diabetes, therapeutic trials of statin medications (which lower LDL cholesterol) have consistently shown an increased risk of diabetes among statin treated patients compared with those on placebo[37]. Whether or not the increased diabetic risk was due to an off-target effect of statins, an artifact of study design, chance, or numerous other possibilities was debated until a study by Swerdlow et al[5] provided human genetic evidence to suggest it was due to an on-target effect of statins. By studying genetic variation in HMGCR, Swerdlow et al were able to demonstrate that the same genetic variants in HMGCR that associated with reduced LDL cholesterol were also associated with increased risk of type 2 diabetes. This result suggests that the increased risk of type 2 diabetes associated with statin use is not due to an off-target effect from statins but rather from inhibiting HMGCR. Furthermore, their result suggests that further efforts to make statins more specific in an attempt to reduce diabetes risk may be futile[38]. Importantly, these same variants are also associated with lower risk of coronary artery disease, suggesting that even though statin therapy appears to increase the risk of type 2 diabetes by inhibiting HMGCR, the net LDL lowering effect of this mechanism is atheroprotective.

Genetic variation modulating LDL cholesterol and risk of type 2 diabetes

The degree to which this effect extended to other LDL-lowering mechanisms remained unclear until a recent study by Lotta et al.[39] In this study, genetic variants in multiple LDL-associated genes were evaluated for an effect on both type 2 diabetes and coronary artery disease. Similar to Swerdlow, Lotta et al. found that the genetic variants in HMGCR that were associated with lower LDL cholesterol were also associated with lower coronary artery disease risk but higher risk of type 2 diabetes. Beyond HMGCR, Lotta et al. observed similar effects for genetic variants in NPC1L1, ABGC5/ABGC8, and PCSK9 (all current or potential lipid modifying targets) in which the variants associated with lower LDL cholesterol were also associated with lower coronary artery disease risk but increased risk of type 2 diabetes. The finding that LDL lowering not just by HMGCR and statins but also by multiple drug target genetic mechanisms is associated with increased risk of type 2 diabetes has important implications for therapeutic development and future clinical trials.

Are HDL cholesterol and Triglycerides causal factors for type 2 diabetes?

Beyond drugs focused on LDL cholesterol, will therapies focused on TRLs also affect risk of type 2 diabetes? White et al. recently employed a Mendelian randomization technique to estimate the causal role of lipid and lipoproteins in risk of diabetes[40]. Using genetic instruments consisting of SNPs associated with LDL cholesterol, HDL cholesterol, and TGs, the investigators calculated the association between genetically higher lipid levels and both risk of coronary artery disease and type 2 diabetes. Consistent with the results of Lotta et al, White and colleagues found that genetically elevated LDL cholesterol was associated with reduced risk of type 2 diabetes. Beyond LDL cholesterol, they found that both genetically elevated HDL cholesterol and TG were associated with reduced risk of type 2 diabetes (see Table 2). The result that genetically elevated TG concentrations associate with reduced risk of type 2 diabetes is surprising[41] and, due to important implications if valid, deserves further study.

Table 2.

Associations for plasma lipid fractions and risk of cardiometabolic disease

Lipid fraction	Risk of coronary artery disease: genetic estimate	Risk of coronary artery disease: RCT evidence	Risk of type 2 diabetes: genetic estimate	Risk of type 2 diabetes: RCT evidence
Lower LDL cholesterol	Reduced[26–28]	Reduced[6, 23]	Increased[39, 40]	Increased[37]
Higher HDL cholesterol	Null[26–28]	Null[46–49]	Reduced[40]	No evidence
Lower Triglycerides	Reduced[25, 26]	No direct evidence but some indirect evidence for reduced risk[50]	Reduced[40]	No evidence

LDL = low-density lipoproteins; HDL = high-density lipoproteins; Genetic estimate refers to Mendelian randomization studies including multiple genetic variants affecting the trait; RCT evidence = results from randomized control trials of drugs focused primarily on modifying that lipid trait.

It is important to note, however, that the approach of White et al. is designed to study the biomarker generally and does not necessarily apply to specific genetic mechanisms. Unlike LDL cholesterol, for which Lotta et al. demonstrated to be causally related to type 2 diabetes across multiple genetic mechanisms, the HDL cholesterol and TG association with type 2 diabetes may be dependent on specific mechanisms. For example, while genetically elevated HDL cholesterol in general appears to be protective for type 2 diabetes, Clapham et al. did not find an association for diabetes with a null allele in ANGPTL8 that associated with higher HDL cholesterol[42], although it is interesting to note that there was a non-significant trend toward protection. Similarly, while the evidence is clear that HDL cholesterol in general is not causal for coronary disease, specific mechanisms that alter HDL cholesterol may affect risk of disease[13]. Finally, we found that a loss of function variant in ANGPTL4 which associated with reduced TRL and lower risk of coronary artery disease was also nominally associated with a reduction in risk of diabetes[32], raising the possibility that certain mechanisms resulting in reduced TRL (such as modulators of LPL activity, for example) may protect against both coronary artery disease as well as type 2 diabetes.

Conclusion

With advances in the scale and content of human genetic studies, our view of the genetic architecture underlying plasma lipid biology is coming increasingly into focus. Armed with these tools, investigators are expanding the focus of these studies in new and creative ways to investigate the net cardiometabolic effects of lipoprotein metabolism. Results of these and future studies are poised to provide important insights into the causal relationship between these traits and disease and should highlight important mechanisms for novel therapeutic development.

Key points.

• Human genetic studies have identified hundreds of loci associated with plasma lipid traits which have informed our understanding of the biology underlying lipoprotein metabolism in humans

• Several specific genetic mechanisms that alter plasma lipids and risk of coronary artery disease have now been identified, highlighting potential therapeutic targets

• Current studies support a causal role in coronary artery disease for low-density lipoprotein cholesterol in addition to cholesterol found in lipoproteins other than high-density lipoprotein cholesterol

• Lowering low-density lipoprotein cholesterol through multiple mechanisms increases risk for type 2 diabetes

• Genetically elevated high-density lipoprotein cholesterol and triglycerides are associated with lower risk of type 2 diabetes although this relationship may be dependent on specific mechanisms