Genetic analysis of haptoglobin polymorphisms with cardiovascular disease and type 2 diabetes in the diabetes heart study

Cardiovascular disease (CVD) is one of the major complications associated with type 2 diabetes mellitus (T2DM). As of 2011, 25.8 million Americans had diagnosed T2DM [1]. More than 50% of individuals with T2DM had coronary heart disease, stroke, or cardiac disease [2]. T2DM is an independent risk factor for development of CVD with the relative risk of CVD mortality of 2.1 in men and 4.9 in women, relative to non-T2DM affected individuals [3, 4]. There is increasing evidence that genetic and environmental factors contribute to this risk.

Haptoglobin (HP) is a 54 kDa protein, found abundantly in the serum [5, 6]. The HP gene has two major alleles: HP1, (containing five exons) and HP2, (containing seven exons) which likely arose from a duplication event involving exons 3 and 4, producing a 61 kDa protein [6]. In its ancestral form, HP is a dimer, however, the HP 1 2 encoded protein exists as linear polymers containing 2–8 monomers, while the HP 2 2 encoded protein exists as circular polymers of 3–10 Hp monomers [6]. The expanded polymerization in the HP 1 2 and HP 2 2 genotypes is due to the duplication of the multimerization domain in exon 3 [6]. Genotype frequencies vary in different ethnicities. In European Americans (EA) they have been reported as 16% HP 1 1, 48% HP 1 2, and 36% HP 2 2[5].

HP may prevent oxidative damage through mechanisms including stabilization of the heme iron within hemoglobin (Hb) [7]. The HP-Hb complex is rapidly removed from circulation via CD163 mediated endocytosis by hepatic Kupfer cells [8]. The HP 1–1 protein is both more efficient than HP 2–2 at preventing oxidation caused by the heme iron [9] and is internalized and cleared from circulation more rapidly; with half lives of approximately 20 minutes and 50 minutes for HP 1-1-Hb and HP 2-2-Hb respectively [10, 11]. Binding of the HP-Hb complex to CD163 induces the production of several cytokines and anti-inflammatory mediators [9, 12] with a much larger production of anti-inflammatory mediators induced by HP 1–1 compared to HP 2–2 [13, 14].

HP has been implicated in both T2DM and T2DM-associated CVD [15, 16]. In the latter context the binding of HP to apolipoprotein A1 (ApoA1) has also been reported [17]. HP binds to ApoA1 in the same location as lecithin-cholesterol acyltransferase (LCAT), subsequently decreasing LCAT activity and therefore limiting high density lipoprotein (HDL) maturation. This inhibits reverse cholesterol transport causing HDL to become proatherogenic [17]. In addition, the tethering of Hb to HDL via the HP-ApoA1 allows the oxidation of HDL and its acquisition of proatherogenic and proinflammatory properties [18]. Due to the multimerization of the Hp 2 protein, individuals with the HP 2–2 genotype have significantly more HP attached to HDL via ApoA1 increasing these properties [11].

Due to the striking differences in properties of the HP 1 and HP 2 proteins, several studies have investigated the impact of the HP phenotype on CVD risk. There have been differing results when examining different populations and different outcomes. Studies investigating incident CVD in individuals affected by T2DM show an increased risk with the HP 2–2. One study [19] found that individuals with T2DM and the HP2 2 genotype had increased risk for CVD events. In addition, Suleiman et al. in 2005 [20] found that individuals with T2DM and the HP 1–1 phenotype had decreased 30-day mortality and heart failure after acute myocardial infarction compared to individuals with the HP 2–2 phenotype, again suggesting the HP 2-2 phenotype as the risk phenotype. This association was not seen in individuals without T2DM. Indeed similar observations have been made in type 1 diabetes; Simpson et al. [21] found that in individuals with type 1 diabetes the HP 2 2 genotype predicted coronary artery calcification progression, a measure of subclinical CVD. In contrast, in cohorts where rates of T2DM are low or individuals with T2DM excluded, the HP1 1 genotype has been shown to be associated with an increased risk for mortality due to coronary heart disease [22]. Similarly, in the Framingham offspring study the HP 1–2 or HP 2–2 phenotypes were associated with decreased rates of prevalent CHD [23].

The HP duplication has also been examined for association with T2DM. The role of HP in regulation of inflammation suggests a potential role in T2DM pathogenesis. There are several studies showing that the HP duplication was associated with T2DM risk in different populations [24, 25].

Thus, the relationship between HP polymorphism and CVD in T2DM-affected individuals is likely complex and association with T2D risk has been documented to a limited degree. Based on these prior studies we hypothesized that if the HP 2 2 genotype is associated with CVD events in people with T2DM, then a similar association would likely be observed with measures of subclinical CVD in predominately T2DM-affected populations. We have taken advantage of the richly phenotyped Diabetes Heart Study [26] (DHS) sample with measures of coronary artery calcification (CAC; or calcified plaque), carotid wall intima-medial thickness (IMT), and blood lipid traits to investigate this hypothesis. Further, the DHS provides a base from which to investigate whether the HP locus is directly associated with T2DM risk.

HP duplication genotypes were determined for 1208 individuals (Table 1). The genotype frequencies were 13.9% HP 1–1, 41.8% HP 1–2, and 44.3% HP 2–2 and were consistent with Hardy-Weinberg proportions.

This study evaluated association of HP gene polymorphisms with subclinical CVD, mortality, and T2DM in 1208 EA individuals from the DHS. The HP 2–2 genotype was associated with increased carotid IMT (p = 0.001, Table 1) in this T2DM enriched population. However, we did not observe significant evidence of association between HP genotype and calcified plaque as a different measure of subclinical CVD. An association with triglyceride concentrations was also observed; the HP 2–2 genotype was associated with lower concentrations. The biological mechanism for this latter association is as of yet, unknown. We also observed suggestive evidence for association of the HP duplication polymorphism with CVD related mortality in the DHS. In addition, we found that the HP 2–2 genotype was associated with T2DM status (OR: 1.49; 95% CI: 1.18-1.86; p = 6.59x10^-4).

Several prior studies have investigated HP polymorphisms and CVD risk in T2DM. In 2002 Levy et al. [19] reported an OR of CVD events in diabetes five times greater with the HP 2–2 phenotype, than with HP 1–1 in a study that included 206 CVD patients and 206 CVD controls (146 and 93 were affected by T2DM, respectively, as part of the Strong Heart Study). In 2004, a subsequent study by Levy et al. [23] included 3273 individuals in the Framingham Heart Study, however only a subset of 433 individuals were affected with T2DM, and of these, only 86 had a history of prevalent CVD. Finally, a 2003 study in individuals with acute myocardial infarction (AMI) reported individuals with T2DM and the HP2 allele had increased mortality following AMI compared to individuals with T2DM and the HP 1 1 genotype (included only 224 T2DM-affected individuals) [20]. In the present study we detected modest evidence of association with carotid IMT, but did not strongly replicate association with history of prior CVD and only nominally with CVD mortality. Parenthetically, IMT and measures of vascular calcification are not highly correlated [26]. The DHS is predominately comprised of T2DM-affected subjects (1013 of 1208 participants). Our primary measures were the subclinical measures of CVD, CAC and IMT which may not be as strongly influenced by HP polymorphism. Of the DHS subjects, 435 were T2DM-affected participants with a history of prevalent CVD, based upon self-reported history and prior intervention which was not associated with HP genotype. The analysis with CVD mortality, a firm endpoint, suggests a possible contribution to risk. Given the association of the HP 2 2 genotype with risk for mortality, it is possible that a survival bias may be present. However, genotype frequencies were consistent with Hardy-Weinberg equilibrium. In addition, the genotype frequencies of the HP duplication in this study were similar to those reported previously [5].

In prior reports, two promoter SNPs, rs5470 and rs5471 were associated with altered levels of HP expression [34, 35] with rs5471 reported to be associated with the Haptoglobin 1–2 modified (HP1-2mod) phenotype. In individuals with the rs5471 “C” allele and the HP 1 2 genotype, normal expression levels of the HP 1 protein, but decreased levels of HP 2 have been reported [34]. It has been suggested that the decreased levels of HP 2 lead to greater oxidative stress [36]. We did not observe evidence of association for these three genotyped HP SNPs genotyped (rs5467, rs5470, and rs5471) with measures of subclinical CVD, history of CVD, or mortality. One possible explanation is the low MAF for both rs5470, and rs5471 (0.0004 and 0.0008 respectively). The combination of the HP1 2 and the rs5471 SNP has been reported in approximately 10% of African Americans [34], but we are unaware of reports of the frequency of the HP1-2mod phenotype in other populations. In this study there were no minor allele homozygotes for rs5471, nor rs5471 heterozygotes with the HP1 2 genotype; as such, the Haptoglobin 1–2 modified phenotype is unlikely to have confounded the HP associations described here. In addition, in LD analysis (see Additional file 1) these two SNPs were in low LD with the HP duplication. Thus these two promoter SNPs along with rs5467 probably do not have an impact on CVD or T2DM status. However, there are other SNPs that are known to impact circulating HP concentrations (e.g. rs2000999) [37] that we did not genotype in the current study which may also contribute to the variance in HP and its role in CVD risk. The lack of measured HP concentrations in the DHS and the inability to further control for these additional genetic variants is one limitation of this work.

Several previous studies have investigated the effect of the HP polymorphism on T2DM risk. A previous study by Stern et al. in 1986 [24] found that the HP1 allele was associated diabetes risk in Mexican Americans. They found that a single copy of the HP1 allele increased T2DM risk by 50% and a second copy increased risk by 100%. A second report from 2006 by Quaye et al. [25] found that the HP 2–2 phenotype was a risk factor for T2DM in a population in Ghana. The current study had a larger sample size than either of the two previous studies, albeit in a different ethnicity. In this study it was found that EA individuals with the HP 2–2 genotype are more likely to have T2DM with an OR of 1.49. These studies, when combined suggest that HP is a risk gene for diabetes or is in LD with a risk gene. Several studies have shown that the different alleles lead to different levels of circulating HP protein [22, 37]. Higher circulating HP has been suggested to be associated with metabolic syndrome, high blood pressure, and elevated glucose [38]. This could possibly explain the association of the HP polymorphism with T2DM. Different risk alleles across populations are problematic as it could be difficult to assess risk across different populations.

Importantly, the T2DM association may be difficult to further investigate without subsequent data generation since the duplication is not included in the major, publically available, databases. Furthermore, any subsequent studies will require a targeted phenotyping approach either through analysis of HP in serum or genotyping through fragment size analysis as performed in this study, since the duplication is not captured by the current commercially available genome-wide genotyping platforms and does not appear to be tagged by other common polymorphisms [37]. A genome wide association study (GWAS) analysis has been performed in the DHS and LD was analyzed for the HP duplication and the two SNPs on the GWAS chip that are closest to the duplication (rs16973636 and rs2287998). The SNPs were found to have low LD with the duplication with an r² of ≤0.01 (data not shown).

Overall, we detected limited association of haptoglobin polymorphisms with CVD. The HP2 2 genotype was found to be positively associated with carotid IMT and the HP2 allele was associated with decreased serum triglyceride concentrations. We identified an association with T2DM in EAs that has not been reported previously [39–41]. Further studies are needed to extend and replicate these relationships.