Genetics of Coronary Artery Disease: An Update

Abstract

In 2007, the first genetic risk variant, 9p21, was simultaneously discovered by two independent groups. 9p21 increases the risk of coronary artery disease in individuals with premature heart disease by twofold, and in the overall population the heterozygote is associated with a 25% increased risk and the homozygote with a 50% increased risk. It is of note that the risk mediated by 9p21 is independent of known risk factors. Since then, with the development of new technologies and the international consortium of CARDIoGRAM, there is now a total of 50 genetic risk variants confirmed and replicated for CAD. Of these 50, 35 mediate their risk by unknown mechanisms, indicating that the pathogenesis of atherosclerosis and myocardial infarction is due to additional factors as yet unknown. The role of genetic risk factors in the management of CAD is yet to be determined. Since many of them are independent of known risk factors, the genetic risk will in the future have to be incorporated into the guidelines, which recommend the target level of plasma LDL-C to be achieved based on the number of risk factors.

Introduction

Coronary artery disease (CAD), the number-one killer in the world, is largely preventable. Modification of conventional risk factors such as cholesterol has consistently shown a 30% to 40% reduction in mortality and morbidity.1, 2 Genetic risk factors for CAD, well documented by epidemiological studies, have until recently been elusive.3 The current explosion in the discovery of genetic risk variants for CAD will, in the future, provide more comprehensive primary prevention and the hope of further reduction in the incidence of CAD and its sequelae. It remains to be determined whether it will meet the challenge to eliminate or markedly attenuate CAD in the 21st century as claimed by several investigators.4, 5

21st Century: a Genetic Landfall for Coronary Artery Disease

The challenge to prevent CAD in the 21st century has had a great start. In 2007, we reported in this journal that the technology had arrived to pursue genes predisposing to polygenic disorders such as CAD.3 The technology referred to is a chip containing 500,000 DNA markers selected to genotype the entire human genome, making possible the first genome-wide association studies (GWAS). These 500,000 DNA markers are single nucleotide polymorphisms (SNPs) occurring at a frequency greater than 1% that had been mapped to their chromosomal location in the human genome. For the human genome of 3 billion nucleotides, these 500,000 SNPs provided, on average, a marker every 6,000 nucleotides. Using the case-control association approach, one could genotype cases and controls and compare the frequencies of each DNA marker in cases to that of controls. Any DNA marker occurring statistically more frequently in cases than controls would reflect a DNA region that was associated with increased risk for that disease. The chip has since been updated to contain approximately one million SNPs. The analysis of multiple SNPs requires a statistical correction, which by convention is a Bonferroni correction whereby a P value of 0.05 is divided by one million, giving a P value < 0.00000005 (i.e., P value < 5 x 10^-8); this is referred to as genome-wide significant.6 Furthermore, the results have to be replicated in an appropriate independent population. Our discussion in this review focuses solely on the results of GWAS, in which the cases had documented CAD and the genetic risk variants discovered are genome-wide significant and have been replicated in an independent population.

To enrich for genetic predisposition, cases in the Ottawa Heart Genomics Study (OHGS) were required to be < 55 years for males and 65 years for females and to have obstruction ≥ 50% in one or more coronary vessels on a coronary angiogram or documented myocardial infarction (MI). Controls were required to be asymptomatic and ≥ 65 years for males and 70 years for females; in addition, those having had a coronary angiogram were required to have < 30% obstruction in either vessel. We phenotyped, genotyped, and performed a GWAS on individuals in the OHGS,7 with replication in multiple independent populations from Texas (Houston and Dallas) and Denmark. The total sample size was greater than 23,000, enabling us to discover the first genetic risk variant for CAD, in 2007, located on the short arm (p) of chromosome 9, now commonly referred to as 9p21.8 Simultaneously and independently, the deCODE group also discovered 9p21.9 Within months, 9p21 was confirmed by multiple investigators around the world.10-13 Subsequent technological advances markedly facilitated the pursuit of genetic risk for CAD, including the mapping of more than 16 million SNPs to their chromosomal location for use as DNA markers. Several GWAS were performed for CAD as well as other diseases, and by 2009, 12 genetic risk variants had been discovered.14 It was realized from this data that multiple genetic risk variants contribute to CAD, each associated with only mild to moderate genetic influence. This would require much larger sample sizes than initially expected to discover genetic risk variants for CAD. Many of the centers already pursuing GWAS for genetic predisposition for CAD agreed to collaborate and leverage their patients, expertise, and other resources. Together we formed an international consortium dedicated to the pursuit of discovering genes associated with CAD15 that is the largest collaboration in the history of cardiology. The initial international consortium was referred to as the Coronary ARtery DIsease Genome-Wide Replication And Meta-Analysis (CARDIoGRAM) study, which involved 14 GWAS, each of which had been previously successful, and included investigators from the United Kingdom, Germany, United States, and Canada. This provided a sample size of 86,995 individuals (22,223 cases vs. 64,762 controls) of European ancestry for the discovery genotyping followed by replication of results in an independent population sample size of 56,682. The study led to the discovery of 13 new genetic risk variants for CAD and confirmation of 10 previously identified risk variants.16 This was followed by the results from the Coronary Artery Disease C4D Genetics Consortium, which identified four additional genetic risk variants for CAD.17 The IBC 50K CAD Consortium, using a 50K SNP array, identified three additional risk variants for CAD.18 Subsequently, CARDIoGRAM joined with the C4D group to become CARDIoGRAMplusC4D with a total sample size of more than 190,000 individuals. Meta-analysis of this sample size led to the discovery and confirmation of 46 genetic risk variants associated with CAD.19 There are currently a total of 50 genetic risk variants predisposing to CAD of genome-wide significance with confirmation in independent populations (Table 1).

Table 1.

Chronological list of 50 genetic variants (genome-wide significant) associated with coronary artery disease or myocardial infarction.

Chromosomal Location	SNP	Nearby Genes	Risk Allele Frequency (allele)	Odds Ratio	Delivery Route
Risk Variant Associated with LDL Cholesterol
6q25.3	rs3798220	LPA	0.02 (C)	1.92 (1.48-2.49)	2009
2p24.1	rs515135	APOB	0.83 (G)	1.03	2012
1p13.3	rs599839	SORT1	0.78 (A)	1.29 (1.18-1.40)	2007
19p13.2	rs1122608	LDLR	0.77 (G)	1.14 (1.09-1.19)	2009
19q13.32	rs2075650	APOE	0.14 (G)	1.14 (1.09-1.19)	2011
2p21	rs6544713	ABCG5-ABCG8	0.29 (G)	1.07 (1.04-1.11)	2011
1p32.3	rs11206510	PCSK9	0.82 (T)	1.15 (1.10-1.21)	2009
Risk Variant Associated with HDL Cholesterol
6p21.31	rs12205331	ANKS1A	0.81 (C)	1.04	2012
Risk Variant Associated with Triglycerides
8q24.13	rs10808546	TRIB1	0.65 (A)	1.08 (1.04-1.12)	2011
11q23.3	rs964184	ZNF259, APOA5-A4-C3-A1	0.13 (G)	1.13 (1.10-1.16)	2011
Risk Variant Associated with Hypertension
12q24.12	rs3184504	SH2B3	0.44 (T)	1.13 (1.08-1.18)	2009
10q24.32	rs12413409	CYP17A1, CNNM2, NT5C2	0.89 (G)	1.12 (1.08-1.16)	2011
4q31.1	rs7692387	GUCYA3	0.81 (G)	1.13	2012
15q26.1	rs17514846	FURIN-FES	0.44 (A)	1.04	2012
Risk Variant Associated with Myocardial Infarction
9q34.2‡	rs579459	ABO	0.21 (C)	1.10 (1.07-1.13)	2011
Risk Variant Mechanism of Risk Unknown
9p21.3	rs4977574	CDKN2A,CDKN2B	0.46 (G)	1.25 (1.18-1.31) to 1.37 (1.26-1.48)	2007
1q41	rs17465637	MIA3	0.74 (C)	1.20 (1.12-1.30)	2007
10q11.21	rs1746048	CXCL12	0.87 (C)	1.33 (1.20-1.48)	2007
2q33.1	rs6725887	WDR12	0.15 (C)	1.16 (1.10-1.22)	2009
6p24.1	rs12526453	PHACTR1	0.67 (C)	1.13 (1.09-1.17)	2009
21q22.11	rs9982601	MRPS6	0.15 (T)	1.19 (1.13-1.27)	2009
3q22.3	rs2306374	MRAS	0.18 (C)	1.15 (1.11-1.19)	2009
10p11.23	rs2505083	KIAA1462	0.42 (C)	1.07 (1.04-1.09)	2010
1p32.2	rs17114036	PPAP2B	0.91 (A)	1.17 (1.13-1.22)	2011
5q31.1	rs2706399	IL5	0.48 (A)	1.02 (1.01-1.03)	2011
6q23.2	rs12190287	TCF21	0.62 (C)	1.08 (1.06-1.10)	2011
7q22.3	rs10953541	BCAP29	0.75 (C)	1.08 (1.05-1.11)	2011
7q32.2	rs11556924	ZC3HC1	0.62 (C)	1.09 (1.07-1.12)	2011
10q23.31	rs1412444	LIPA	0.34 (T)	1.09 (1.07-1.12)	2011
11q22.3	rs974819	PDGF	0.29 (T)	1.07 (1.04-1.09)	2011
13q34	rs4773144	COL4A1, COL4A2	0.44 (G)	1.07 (1.05-1.09)	2011
14q32.2	rs2895811	HHIPL1	0.43 (C)	1.07 (1.05-1.10)	2011
15q25.1	rs3825807	ADAMTS7	0.57 (A)	1.08 (1.06-1.10)	2011
17p13.3	rs216172	SMG6, SRR	0.37 (C)	1.07 (1.05-1.09)	2011
17p11.2	rs12936587	RASD1, SMCR3, PEMT	0.56 (G)	1.07 (1.05-1.09)	2011
17q21.32	rs46522	UBE2Z, GIP, ATP5G1, SNF8	0.53 (T)	1.06 (1.04-1.08)	2011
5p13.3*	rs11748327	IRX1, ADAMTS16	0.76 (C)	1.25 [1.18-1.33]	2011
6p22.1*	rs6929846	BTN2A1	0.06 (T)	1.51 (1.28-1.77)	2011
6p24.1**	rs6903956	C6orf105	0.07 (A)	1.65 (1.44-1.90)	2011
6p21.3	rs3869109	HCG27 and HLA-C	0.60 (C)	1.15	2012
1q21	rs4845625	IL6R	0.47 (T)	1.09	2012
Chr4	rs1878406	EDNRA	0.15 (T)	1.09	2012
7p21.1	rs2023938	HDAC9	0.10 (G)	1.13	2012
2p11.2	rs1561198	VAMP5-VAMP8	0.45 (A)	1.07	2012
Chr2	rs2252641	ZEB2-AC074093.1	0.45 (A)	1	2012
Chr5	rs273909	SLC22A4-SLC22A5	0.14 (C)	1.11	2012
6p21	rs10947789	KCNK5	0.76 (T)	1.01	2012
6q26	rs4252120	PLG	0.73 (T)	1.07	2012
8p22	rs264	LPL	0.86 (G)	1.06	2012
13q12	rs9319428	FLT1	0.32 (A)	1.1	2012

*Variant identified only in Japanese; **Variant identified only in Han Chinese; ‡ The risk variant at 9q34.2 is associated with myocardial infarction but not with coronary atherosclerosis. A: adenine; C: cytosine; CI: confidence interval; G: guanine; OR: odds ratio; SNP: single nucleotide polymorphism; T: thymine.

Common Features in CAD Genetic Risk Variants

The widespread use of GWAS to discover genetic risk variants for common polygenic diseases has met with remarkable success. In just over 5 years, more than 2,800 genetic variants have been discovered as risk factors for more than 300 diseases.20 The genetic risk variants for CAD have many features that are similar to genetic variants for other polygenic disorders:

1. The genetic risk variants for CAD are very common, occurring on average in 50% of the population with a frequency varying from 2% to 91% (Table 1).

2. The relative increased risk of each genetic variant is small, averaging 18% with an odds ratio varying from 2% to 90%.

3. For CAD as well as other common polygenic disorders, multiple genetic risk variants are inherited by everyone. Those at high genetic risk for CAD have a greater genetic risk burden due to inheritance of a greater number of common risk variants, as opposed to inheriting one or more genetic variant of high risk. In a CARDIoGRAM analysis of 23 genetic risk variants for CAD, the average number inherited per individual (case or control) was 17, varying from a minimum of 7 to a maximum of 37.

4. Most of the genetic risk variants for CAD are located in DNA sequences that do not code for protein. This means the risk variant mediates its increased risk for CAD directly or indirectly through regulation of DNA sequences that do code for protein.

5. All DNA genetic risk variants need only be assessed once, since one’s DNA does not change over one’s lifetime nor do genetic risk variants vary with time, meals, drugs, or gender.

Pathological and Therapeutic Implications of Genetic Risk Variants for CAD

A brief analysis of Table 1 indicates that only 15 of the 50 genetic risk variants are associated with conventional risk factors for CAD: seven associated with low density lipoprotein-cholesterol (LDL-C); one with high density lipoprotein (HDL); two with triglycerides; four with hypertension; and one with coronary thrombosis. The remaining 35 risk variants operate through mechanisms yet to be determined. While the hope is to ultimately use these genetic risk variants for more effective primary and secondary prevention, the immediate surprise is that many other factors contribute to the pathogenesis of atherosclerosis and CAD that are yet unknown. Research can now be directed towards these new genetic risk factors with the hope of identifying new pathways that lead to CAD. This implies a great opportunity to develop new biomarkers for detecting early CAD as well as unique targets for novel therapy. Just as 10 of these genetic risk variants mediate their risk through lipids, it is expected that the 35 genetic risk variants of unknown function will mediate their risk through only a few pathways. Identification of these molecular pathways will provide for early detection and more effective prevention of this disease. It is self-evident that until we identify these pathways, we are unlikely to be comprehensive in our prevention of CAD. The identification of PCSK9 has already led to the development of new therapies for CAD as described below.

Genetics Leads to New Therapy for CAD: PCSK9 Inhibition

Evidence that cholesterol plays a major role in the pathogenesis of atherosclerosis has been known for more than five decades. However, one of the major observations confirming the link between cholesterol and heart disease was from human genetics. In the 1970s, members of a family with hypercholesterolemia, due to a mutation in the LDL-receptor, experienced heart attacks in their 2nd and 3rd decade of life.21 This observation catalyzed efforts to find a drug to lower plasma levels of LDL-C. Two decades later, a drug that inhibits cholesterol synthesis was introduced; all drugs with this mechanism are referred to as statins. Statins are essentially the only drug for primary and secondary prevention of hypercholesterolemia. The worldwide budget for statins alone is more than $70 billion.

In 2003, Seidah et al. discovered PCSK9, an enzyme that increases the degradation of LDL receptors.22 Since LDL receptors are a major mechanism for the removal of LDL-C, PCSK9 is associated with hypercholesterolemia and increased mortality from heart disease. Subsequently, other mutations in the gene encoding for PCSK9 have been identified. Those associated with increased function are associated with higher cholesterol levels and increased cardiac morbidity and mortality. This is in contrast to mutations inducing loss of function of PCSK9, which are associated with hypocholesterolemia and a decreased incidence of MI and death. It was well recognized and recently confirmed in a U.K. study that only 28% of individuals receiving a statin reached the recommended target for plasma LDL-C.23 There are several reasons for not obtaining this target, but one is intolerance associated with high doses of statins. Inhibition of PCSK9 provides a complementary therapy to statins since it can lower the plasma levels of LDL-C without affecting the synthesis of cholesterol. African Americans that inherited hypocholesterolemia due to loss of function mutations in PCSK9 showed a mean reduction of 28% in plasma LDL-C levels and a mean reduction of 88% in the risk of CAD. Despite these families being exposed to hypocholesterolemia throughout their lives, there were no adverse side effects.24 Several therapies have been developed to inhibit PCSK9 and are now undergoing clinical trials.25-28 The one appearing most promising is a monthly injection of a monoclonal antibody.27, 28 Results of phase I trials showed no significant side effects and LDL-C reductions of 41% to 58%.29 Phase II trials were in individuals with hypercholesterolemia receiving atorvastatin treatment. Those receiving 80 mg of atorvastatin alone had a mean reduction of 17% in their LDL-C versus a 72% reduction in LDL-C for those receiving 80 mg atorvastatin plus the PCSK9 antibody.29 Phase III clinical trials are currently ongoing. In just a few years, since this genetic discovery, a new and potent therapy is emerging for the treatment of hypercholesterolemia. Thus, genetic observations have again provided us new insight and novel therapy for CAD.

Blood Groups A and B are Risk Variants for CAD with Therapeutic Implications

In a CARDIoGRAM study, a GWAS was performed in 4,372 patients with documented CAD by angiography and confirmed MI and in 2,739 patients with documented CAD without MI.30 There was a strong association between the ABO blood group locus at 9q34.2 and MI but no association with CAD. This was replicated in an independent population. Epidemiologists have claimed for decades that blood group O offers protection from MI. Blood groups A, B, and O are different forms of the same gene at 9q34.2. The A and B genes encode for a protein (alpha 1, 3N-acetylgalactosaminyltransferase) that transfers a carbohydrate moiety onto von Wille-brand Factor (vWF). This prolongs the life of vWF and predisposes to coronary thrombosis and MI. The blood group O gene codes for a protein that has been mutated and lacks any biochemical activity and thus does not transfer the carbohydrate moiety onto vWF. As a result, individuals with blood group O show no increased risk for MI.

The frequency of the gene that encodes for A or B blood group occurs in about 57% of Caucasians. The average relative increased risk for MI is about 20% depending on the genotype. In the recent Nurses’ Health Study and Health Professionals Follow-up Study of more than 90,000 individuals, 4,070 developed heart disease. In this 20-year follow-up study, having blood group A or B alone was associated with an increased risk of MI of about 10%; however, the combination of A and B blood groups increased the risk to 20%.31 It also has been shown that plasma levels of vWF complex are approximately 25% higher in individuals with A, B, or AB blood groups as opposed to blood group O.32

These results have important implications for people undergoing angioplasty, bypass surgery, and other such procedures. For example, should individuals of blood group A or B receive some form of antiplatelet therapy such as aspirin?

9p21 Predisposes to Coronary Atherosclerosis and not Myocardial Infarction

The 9p21 risk variant for CAD is perhaps the most robust genetic variant and the most studied of those risk variants with unknown function. This risk variant is contained in a long non-protein coding RNA (LncRNA) of 126,000 bps referred to as Anril, which remains of unknown function. The 9p21 risk variant was not introduced into the genome until the arrival of higher primates and is highly conserved in the human genome. The 9p21 risk allele occurs in 75% of humans except for Africans (50% heterozygous, 25% homozygous). Each risk variant is associated with an increased relative risk for CAD of about 25%. The risk of 9p21 is consistently observed by investigators throughout the world to be independent of conventional risk factors such as cholesterol, diabetes, or hypertension. In individuals with premature CAD, 9p21 homozygosity is associated with a 2-fold increased risk for CAD. The 9p21 risk variant also contributes to increased risk for intracranial and abdominal aortic aneurysms33 and Alzheimer’s disease34 and has recently been associated with periodontitis35 and gout,36 diseases with a marked inflammatory component. It is of note that the 9p21 risk variant does not associate with C-reactive protein.37, 38

Determining the function of 9p21 is further complicated by the observation that the risk variant is not present in the mouse genome, the favorite animal model for assessing gene function. Harismendy et al. had suggested that interferon-gamma may mediate the risk of 9p21 for CAD.39 However, we have recently shown that interferon-gamma acts independently of the 9p21 risk variant.40

All studies have consistently shown that the 9p21 risk variant is associated with atherosclerosis and not with MI.30, 41-43 Several studies have also indicated that the 9p21 risk variant is associated with progression of coronary atherosclerosis as suggested by the correlation between the number of vessels involved and the number of copies of the 9p21 risk variant.41, 43 However, there are other studies that have not confirmed the correlation between 9p21 and progression of CAD.42, 44, 45

Genetic Risk Variants and Management of CAD

Where do these genetic risk variants fit in the management of CAD? Currently, the answer would be that they do not. One might argue that until there is some therapy to alter their risk, why would one screen for these genetic risk variants? If one has to await the development of drug therapy, it could certainly be 10 years away other than what has already been identified for PCSK9 or antiplatelet therapy for blood groups A and B. One approach to incorporating independent genetic risk variants such as 9p21 into the management of CAD is on the basis of increased burden of risk as outlined by the Adult Treatment Panel III (ATP III). Currently, the ATP III recommends that LDL-C ≥ 190 mg/dL be reduced in individuals with one other risk factor and that LDL-C ≤ 160 mg/dL be reduced in those with two other risk factors. One of these other risk factors could be an independent genetic risk factor such as 9p21, since there is universal agreement that 9p21, like the 34 other genetic risk factors, is independent of conventional risk factors. The ATP panel could then assess whether individuals positive for one or more of these genetic risk variants should have LDL-C treated since it provides for an independent risk factor. It is important to note that in individuals with premature CAD, 9p21 is associated with a 2-fold increased risk—greater than the risk from smoking or that associated with a moderate increase in blood pressure or plasma LDL-C.

The Hope for the Future

The challenge for the next decade will be to identify the molecular mechanisms mediating the risk of those genetic risk variants that do not act through known conventional risk factors. There is good evidence that several of these genetic risk variants predispose to CAD through inflammatory pathways.19, 46 This would appear to be a major pathway in keeping with previous epidemiological suggested evidence. Genetic observations have already contributed to the mainstay therapy of prevention, namely, statin drugs to prevent CAD, and PCSK9 inhibition that will likely enhance and complement statin therapy based on its effects in ongoing clinical trials. It is reasonable to assume that genetic risk variants will lead to markers for earlier detection of CAD as well as drug therapies to interrupt or attenuate the risk. This is occurring along with the overall trend of personalized medicine, in which the disease and the individual will be treated with more specific therapies to match their genome susceptibilities.

Funding Statement

Funding/Support: Dr. Roberts receives grant support from CIHR#MOP82810 (RR)/Canada and CFI#11966 (RR)/Canada.