PPARα (peroxisome-proliferator-activated receptor α) regulates the expression of genes that are involved in lipid metabolism, tissue homoeostasis and inflammation. Consistent rodent and human studies suggest a link between PPARα function and cardiovascular disease, qualifying PPARα [PPARA in HUGO (Human Genome Organisation) gene nomenclature] as a candidate gene for coronary artery disease. In the present study, we comprehensively evaluated common genetic variations within the PPARα gene and assessed their association with myocardial infarction. First, we characterized the linkage disequilibrium within the PPARα gene in an initial case-control sample of 806 individuals from the Regensburg Myocardial Infarction Family Study using a panel of densely spaced SNPs (single nucleotide polymorphisms) across the gene. Single SNP analysis showed significant association with the disease phenotype [OR (odds ratio)=0.74, P=0.012, 95% CI (confidence interval)=0.61–0.94 for rs135551]. Moreover, we identified a protective three-marker haplotype with an association trend for myocardial infarction (OR=0.76, P=0.067, 95% CI=0.56–1.02). Subsequently, we were able to confirm the single SNP and haplotype association results in an independent second case-control cohort with 667 cases from the Regensburg Myocardial Infarction Family Study and 862 control individuals from the WHO (World Health Organization) MONICA (Monitoring of Trends and Determinants in Cardiovascular Disease) Augsburg project (OR=0.87, P=0.046, 95% CI=0.72–0.99 for rs135551 and OR=0.80, P=0.034, 95% CI=0.65–0.98 for the three-marker haplotype respectively). From these cross-sectional association results, we provide evidence that common variations in the PPARα gene may influence the risk of myocardial infarction in a European population.

INTRODUCTION

MI (myocardial infarction) is considered a complex disease. Its development and progression underlies interactions between environmental and genetic influences [1]. Genetic association studies of candidate genes focus on the identification of genetic variations that contribute to or protect against the development of the disease [2].

PPARα (peroxisome-proliferator-activated receptor α), a member of the nuclear receptor superfamily, qualifies as an interesting candidate gene for atherosclerotic disease. PPARα is a ligand-activated transcription factor that regulates the transcription of genes involved in lipid metabolism, tissue homoeostasis and proteins that initiate an inflammatory response central to atherosclerotic plaque formation [3]. Ligands for PPARα include fibrates, a substance class of lipid-lowering drugs that have been shown to reduce the risk of cardiovascular events [4,5].

Previously, genetic examinations of the PPARα [PPARA in HUGO (Human Genome Organisation) gene nomenclature] locus have mainly focussed on the L162V polymorphism, which is the only functional variant that has so far been shown to be associated with changes in serum concentrations of lipids and lipoproteins, and with left ventricular hypertrophy [69]. Yet, association studies between the L162V variant and atherosclerosis are limited and provide inconsistent results [10].

However, to date, no comprehensive evaluation of common variations within the PPARα gene and their association with MI has been performed. Therefore the aims of the present study were (i) to assess the LD (linkage disequilibrium) structure of the PPARα gene and to identify possible haplotypes in an initial case-control sample, (ii) to perform single SNP (single nucleotide polymorphism) and haplotype analyses for association with MI, and (iii) to validate available positive results in an independent second case-control sample.

MATERIALS AND METHODS

Study population

All MI patients were participants of the Regensburg MI Family Study. Selection criteria have been described previously [11,12]. In brief, we identified MI families, in which the index patient suffered from MI at the age of 60 years or younger and at least one more family member had been affected with MI and/or severe CAD (coronary artery disease), e.g. percutaneous coronary intervention or coronary bypass surgery. The diagnosis of MI was established by a review of medical records according to the MONICA (Monitoring of Trends and Determinants in Cardiovascular Disease) diagnostic criteria (http://www.ktl.fi/publications/monica/manual/index.htm). All MI patients and all of their available siblings and spouses were contacted and invited to participate in the study. Medical history, risk factor evaluation, clinical examination with anthropometric measurements and blood samples were obtained from all participants.

For the initial case-control study, 433 unrelated MI patients fulfiling our index criteria and 373 unrelated unaffected control subjects were randomly selected to establish genotyping and to analyse gene structure. Control individuals consisted of cardiovascular-disease-free married-in spouses, sisters-in-law and brothers-in-law, from the Regensburg MI Family Study. For the second sample, we identified an additional 667 unrelated MI patients from the Regensburg MI Family Study (age of MI in men <60 years and in women <70 years) and 862 controls from the Third WHO (World Health Organization) MONICA Augsburg survey 1994/1995. Individuals with a history of CAD, cerebrovascular accidents (transient ischaemic attack or stroke) or peripheral artery disease were excluded from the control group. The MONICA Augsburg survey is a large population-based sample of all German residents of the Augsburg area that consists of individuals aged 25–74 years, with approx. 300 subjects per 10-year age increment and gender. The Augsburg project was part of the international collaborative WHO MONICA study [13].

Written informed consent was obtained from all subjects, and the local ethics committee approved the study.

Phenotyping

At recruitment, all patients were studied by comparable design as previously described including a standardized interview, clinical examination and biochemical as well as molecular analyses either at a visit to the study centre (patients from the MONICA Augsburg study) or at a home visit by a physician (patients from rehabilitation centres) [12,14,15]. Validation of cardiovascular events at study entry was performed by a review of medical records. Resting BP (blood pressure) was taken according to MONICA guidelines after participants had been resting in a sitting position [16]. The body weight in kilograms and height in metres were determined with subjects wearing light clothing. Blood samples were taken from non-fasting individuals. Serum levels of LDL (low-density lipoprotein) and HDL (high-density lipoprotein) cholesterol and triacylglycerols (triglycerides) were measured by standard enzymatic methods [17]. Characteristics of both cohorts are shown in Table 1.

Table 1
Anthropometric data of 806 individuals from the initial sample and 1529 individuals from the second sample

Values are means±S.D. (range). Lipid-lowering therapy is defined as a regular intake of statins or fibrates; obesity is defined as BMI >30 kg/m2; hypertension is defined as BP≥140/90 mmHg; antihypertensive therapy is defined as regular intake of established BP-lowering drugs (diuretics, β-blockers, angiotensin-converting enzyme inhibitors, angiotensin 2-blockers, calcium channel blockers, vasodilators and α-blockers), Type 2 diabetes is defined as self-reported diabetes mellitus or regular intake of antidiabetic medication. n.a., not available; n.s. not significant.

Initial sampleSecond sample
CharacteristicMI cases (n=433)Controls (n=373)PMI cases (n=667)Controls (n=862)P
Male gender (%) 73.2 69.2 n.s. 74.2 70.6 n.s. 
Age at first MI (years) 49.7±7.5 (25–60) − − 50.3±8.5 (24–69) − − 
Age at inclusion (years) 59.4±9.4 (35–82) 59.7±9.4 (32–91) n.s. 57.6±9.0 (29–82) 50.2±11.9 (25–69) <0.001 
Systolic BP (mmHg) 134±18 136±18 n.s. 138±19 137±19 n.s. 
Diastolic BP (mmHg) 81±10 82±10 0.043 83±11 83±11 n.s. 
Hypertension (%) 47.7 52.4 n.s. 51.6 43.7 0.003 
Anithypertensive therapy (%) 86.6 42.4 <0.001 87.1 15.8 <0.001 
Total cholesterol (mg/dl) 225±43 239±42 <0.001 228±48 237±45 <0.001 
LDL cholesterol (mg/dl) 143±36 150±33 0.031 156±46 148±43 0.001 
HDL cholesterol (mg/dl) 52±14 57±15 <0.001 49±13 51±16 0.003 
Triacylglycerols (mg/dl) 162±92 182±108 0.006 n.a. n.a. n.a. 
Lipid-lowering therapy (%) 64.2 10.7 <0.001 64.8 3.0 <0.001 
BMI (kg/m227.5±3.9 26.9±3.7 0.023 27.3±3.5 27.2±3.9 n.s. 
Obesity (%) 21.7 17.4 n.s. 22.1 20.1 n.s. 
Type 2 diabetes (%) 16.4 7.4 <0.001 15.1 3.5 <0.001 
Ever smoked (%) 74.1 59.5 <0.001 72.0 59.3 <0.001 
Initial sampleSecond sample
CharacteristicMI cases (n=433)Controls (n=373)PMI cases (n=667)Controls (n=862)P
Male gender (%) 73.2 69.2 n.s. 74.2 70.6 n.s. 
Age at first MI (years) 49.7±7.5 (25–60) − − 50.3±8.5 (24–69) − − 
Age at inclusion (years) 59.4±9.4 (35–82) 59.7±9.4 (32–91) n.s. 57.6±9.0 (29–82) 50.2±11.9 (25–69) <0.001 
Systolic BP (mmHg) 134±18 136±18 n.s. 138±19 137±19 n.s. 
Diastolic BP (mmHg) 81±10 82±10 0.043 83±11 83±11 n.s. 
Hypertension (%) 47.7 52.4 n.s. 51.6 43.7 0.003 
Anithypertensive therapy (%) 86.6 42.4 <0.001 87.1 15.8 <0.001 
Total cholesterol (mg/dl) 225±43 239±42 <0.001 228±48 237±45 <0.001 
LDL cholesterol (mg/dl) 143±36 150±33 0.031 156±46 148±43 0.001 
HDL cholesterol (mg/dl) 52±14 57±15 <0.001 49±13 51±16 0.003 
Triacylglycerols (mg/dl) 162±92 182±108 0.006 n.a. n.a. n.a. 
Lipid-lowering therapy (%) 64.2 10.7 <0.001 64.8 3.0 <0.001 
BMI (kg/m227.5±3.9 26.9±3.7 0.023 27.3±3.5 27.2±3.9 n.s. 
Obesity (%) 21.7 17.4 n.s. 22.1 20.1 n.s. 
Type 2 diabetes (%) 16.4 7.4 <0.001 15.1 3.5 <0.001 
Ever smoked (%) 74.1 59.5 <0.001 72.0 59.3 <0.001 

SNPs and genotyping methods

Seventeen SNPs were selected from public databases (dbSNP, http://www.ncbi.nlm.nih.gov/SNP/) based on the following criteria: (i) MAF (minor allele frequency) of >0.05 in a Caucasian population (according to HapMap Project population data; http://www.hapmap.org/), (ii) average coverage target of one SNP per 5 kb of the PPARα gene and flanking sequences, (iii) evidence of validation status, and (iv) compatibility with our genotyping platform. To allow us to determine the extend of LD beyond the boundaries of the gene, one SNP each in the 5′ intergenic region and in the 3′ UTR (untranslated region) was included.

Genomic DNA was extracted from EDTA-treated blood samples using standard procedures. High-throughput genotyping was performed on the Applied Biosystems ABI 7900 platform according to the manufacturer's protocol using the ABI assay on-demand or by-design service. Genotyping quality was ensured by repetition of 10% (n=233) of all genotypes in independent PCR reactions. Genotypes did not differ between duplicates. For five duplicates, the definite genotype call was obtained in one of the assays with an undetermined call in the other. The overall genotyping call-rate was >98% for all SNPs. In individuals from the second sample, only rs135557, rs135551, rs135543 and rs135539 were genotyped.

Statistical analyses

For each of the 17 SNPs, we tested whether the observed genotype frequencies in MI cases and controls deviated from Hardy–Weinberg equilibrium (P=0.05).

Single SNP association analyses were performed in both case-control samples. Allele and genotype frequencies were compared using the χ2 test and Armitage test for trend reporting OR (odds ratio) and 95% CI (confidence interval). Armitage's trend test, under the assumption of an additive mode of inheritance, allows the reduction of genotype comparisons to one degree of freedom [18].

Additionally, pair-wise LD was determined applying the standard definition of r2 [19]. A haplotype block was defined as a region in which all pair-wise r2 values were >0.4. Haplotypes and their frequencies were inferred from phase-unknown genotype data using PHASE version 2.1 [20,21]. Uncertain haplotypes with a probability of correct phase or allele <90% were excluded from the analyses. The calculated haploblock A as well as the number of copies of hap2 (a three-marker haplotype consisting of the rare alleles of rs135551, rs135543 and rs135539) were tested for disease association by a χ2 test. For interaction analyses, all pair-wise combinations of SNPs were tested in a population-based case/control setting implemented in plink v0.99s [22]. A two-sided P value of <0.05 indicated statistical significance. Multiple logistic regression analyses were used in the multivariate model of the matched cohort adjusting for age, total cholesterol/HDL cholesterol ratio, systolic BP, BMI (body mass index), diabetes mellitus and smoking status with P values from an Effect Wald test. Total cholesterol/HDL cholesterol ratio was chosen because it has been shown to be superior to the use of total cholesterol or LDL cholesterol alone in the prediction of CAD risk and is less affected by ongoing lipid-lowering therapy [2325]. Association results were not corrected for multiple testing.

RESULTS

PPARα LD mapping and haplotype structure

We characterized the LD and haplotype structure of the PPARα gene by genotyping a case-control sample of 806 individuals with (n=433) and without (n=373) MI for 17 SNPs (Figure 1). There were 14 intronic SNPs, one SNP was positioned in exon 6, and one SNP was located in the 5′ intergenic region and the 3′ UTR respectively, yielding an average marker distance of 5.6 kb. All 17 SNPs had MAFs>0.05 and met Hardy–Weinberg expectations in MI cases and controls in both study samples (P=0.05). Haplotypes and their frequencies were constructed from phase-unknown genotype data. A haplotype block was defined as a region in which all pair-wise r2 values were >0.4. A total of four haplotype blocks were identified (blocks A–D; Figure 1). The first haploblock (haploblock A) comprised rs135551, rs135543 and rs135539 and spanned part of intron 2 (6246 bp); haploblocks B, C and D were built up by two (rs4253681 and rs7364220), three (rs4823613, rs5766741, rs4253728) and four (rs4253754, rs4253760, rs4253778, rs11704979) SNPs respectively. Haploblock B covered part of intron 3 and haploblock C spanned most of intron 4, whereas haploblock D commenced within intron 6 and reached across PPARα boundaries into the 3′ UTR (Figure 1). Reconstruction of haploblock A revealed three common haplotypes with a frequency >0.10 which were termed hap1, hap 2 and hap3 with frequencies of 54.3, 28.7 and 12.3% respectively. They comprised 95.3% of the total chromosomes of the screening sample population.

Schematic representation of the PPARα locus

Figure 1
Schematic representation of the PPARα locus

(A) Location of PPARα (GenBank® accession number NM_005036) and neighbouring genes in a region of approx. 147 kb on human chromosome 22q13.3. Exon positions are indicated as vertical lines. (B) Relative positions of genotyped SNPs. Position of the 17 SNPs and corresponding rs numbers from dbSNP are shown with arrows. (C) Relative positions of calculated haplotypes. Haploblocks are indicated as blocks A–D, where block A includes hap1–hap3 focussed on in the present study. Haplotypes were determined from pair-wise LD comparisons with r2>0.4 using PHASE version 2.1 [20,21]. (D) Pair-wise r2 from HapMap release #20 data. Lighter shading indicates low values (low LD) and darker shading indicates higher values (strong LD). Haploblock A does not reach into the neighbouring gene FLJ27365.

Figure 1
Schematic representation of the PPARα locus

(A) Location of PPARα (GenBank® accession number NM_005036) and neighbouring genes in a region of approx. 147 kb on human chromosome 22q13.3. Exon positions are indicated as vertical lines. (B) Relative positions of genotyped SNPs. Position of the 17 SNPs and corresponding rs numbers from dbSNP are shown with arrows. (C) Relative positions of calculated haplotypes. Haploblocks are indicated as blocks A–D, where block A includes hap1–hap3 focussed on in the present study. Haplotypes were determined from pair-wise LD comparisons with r2>0.4 using PHASE version 2.1 [20,21]. (D) Pair-wise r2 from HapMap release #20 data. Lighter shading indicates low values (low LD) and darker shading indicates higher values (strong LD). Haploblock A does not reach into the neighbouring gene FLJ27365.

Association of PPARα variants with MI in the initial case-control cohort

Initially, we carried out single marker association tests with all 17 SNPs. The first cohort was genotyped for calculating ORs from comparisons of allele and genotype frequencies in cases and controls (Table 2). We found a significant association between rs135551 and MI (P=0.012, OR=0.74, 95% CI=0.61–0.94; Armitage's trend test). Additionally, an association trend between rs135543 and MI was observed resulting from an over-representation of the rare alleles in control individuals (P=0.082, OR=0.81, 95% CI=0.68–1.02; Armitage's trend test). The effect was more pronounced in carriers of two copies of the rare alleles as reflected in the homozygous model (22 compared with 11; P=0.005, OR=0.49, 95% CI=0.29–0.81 for rs135551, and P=0.025, OR=0.59, 95% CI=0.38–0.94 for rs135543 respectively; results not shown).

Table 2
Summary of association analyses of the initial case-control sample

For the PPARα locus, 17 genotyped SNPs and their rs numbers from the dbSNP database are shown. Numbers (n) for each genotype (11/12/22) and MAFs are listed separately for MI cases and controls. ORs with 95% CI and P values are given for allelic association and genotypic association based on Armitage's trend test, with significant P values indicated by *.

Genotype MI cases (n)Genotype controls (n)Allele 2 compared with allele 1Armitage's trend test
SNP111222MAF111222MAFOR (95% CI)POR (95% CI)P
rs135557 129 220 80 0.44 100 185 85 0.48 0.86 (0.71–1.05) 0.141 0.86 (0.70–1.05) 0.137 
rs135551 218 183 29 0.28 165 160 45 0.34 0.76 (0.62–0.94) 0.013* 0.74 (0.61–0.94) 0.012* 
rs135543 197 194 40 0.32 158 158 54 0.36 0.83 (0.68–1.02) 0.079 0.81 (0.68–1.02) 0.082 
rs135539 137 208 83 0.44 102 194 75 0.46 0.90 (0.74–1.09) 0.285 0.90 (0.73–1.09) 0.281 
rs881740 330 93 0.12 299 69 0.10 1.26 (0.92–1.73) 0.142 1.30 (0.92–1.70) 0.141 
rs4253662 357 63 0.08 289 56 0.10 0.81 (0.57–1.14) 0.215 0.79 (0.59–1.14) 0.234 
rs4253681 301 121 0.15 279 86 0.13 1.23 (0.92–1.64) 0.160 1.23 (0.93–1.70) 0.139 
rs7364220 260 148 18 0.22 246 114 11 0.18 1.23 (0.96–1.57) 0.104 1.24 (0.96–1.59) 0.100 
rs12330015 337 79 0.10 302 62 0.09 1.08 (0.77–1.51) 0.656 1.06 (0.77–1.53) 0.651 
rs4823613 208 181 35 0.30 196 147 24 0.27 1.16 (0.93–1.50) 0.181 1.17 (0.93–1.46) 0.176 
rs5766741 217 170 36 0.29 196 152 21 0.26 1.12 (0.90–1.40) 0.303 1.17 (0.90–1.41) 0.298 
rs4253728 217 179 34 0.29 198 151 21 0.26 1.14 (0.92–1.42) 0.238 1.17 (0.92–1.44) 0.230 
rs1800206 380 45 0.06 328 42 0.06 0.97 (0.64–1.47) 0.870 0.99 (0.64–1.47) 0.871 
rs4253754 265 147 15 0.21 239 118 12 0.19 1.10 (0.86–1.40) 0.460 1.09 (0.86–1.42) 0.452 
rs4253760 274 133 17 0.20 240 114 14 0.19 1.03 (0.80–1.32) 0.841 1.03 (0.80–1.32) 0.842 
rs4253778 272 137 15 0.20 238 113 15 0.20 1.01 (0.79–1.30) 0.937 0.99 (0.79–1.30) 0.937 
rs11704979 324 96 0.13 294 71 0.11 1.18 (0.87–1.61) 0.287 1.16 (0.87–1.61) 0.287 
Genotype MI cases (n)Genotype controls (n)Allele 2 compared with allele 1Armitage's trend test
SNP111222MAF111222MAFOR (95% CI)POR (95% CI)P
rs135557 129 220 80 0.44 100 185 85 0.48 0.86 (0.71–1.05) 0.141 0.86 (0.70–1.05) 0.137 
rs135551 218 183 29 0.28 165 160 45 0.34 0.76 (0.62–0.94) 0.013* 0.74 (0.61–0.94) 0.012* 
rs135543 197 194 40 0.32 158 158 54 0.36 0.83 (0.68–1.02) 0.079 0.81 (0.68–1.02) 0.082 
rs135539 137 208 83 0.44 102 194 75 0.46 0.90 (0.74–1.09) 0.285 0.90 (0.73–1.09) 0.281 
rs881740 330 93 0.12 299 69 0.10 1.26 (0.92–1.73) 0.142 1.30 (0.92–1.70) 0.141 
rs4253662 357 63 0.08 289 56 0.10 0.81 (0.57–1.14) 0.215 0.79 (0.59–1.14) 0.234 
rs4253681 301 121 0.15 279 86 0.13 1.23 (0.92–1.64) 0.160 1.23 (0.93–1.70) 0.139 
rs7364220 260 148 18 0.22 246 114 11 0.18 1.23 (0.96–1.57) 0.104 1.24 (0.96–1.59) 0.100 
rs12330015 337 79 0.10 302 62 0.09 1.08 (0.77–1.51) 0.656 1.06 (0.77–1.53) 0.651 
rs4823613 208 181 35 0.30 196 147 24 0.27 1.16 (0.93–1.50) 0.181 1.17 (0.93–1.46) 0.176 
rs5766741 217 170 36 0.29 196 152 21 0.26 1.12 (0.90–1.40) 0.303 1.17 (0.90–1.41) 0.298 
rs4253728 217 179 34 0.29 198 151 21 0.26 1.14 (0.92–1.42) 0.238 1.17 (0.92–1.44) 0.230 
rs1800206 380 45 0.06 328 42 0.06 0.97 (0.64–1.47) 0.870 0.99 (0.64–1.47) 0.871 
rs4253754 265 147 15 0.21 239 118 12 0.19 1.10 (0.86–1.40) 0.460 1.09 (0.86–1.42) 0.452 
rs4253760 274 133 17 0.20 240 114 14 0.19 1.03 (0.80–1.32) 0.841 1.03 (0.80–1.32) 0.842 
rs4253778 272 137 15 0.20 238 113 15 0.20 1.01 (0.79–1.30) 0.937 0.99 (0.79–1.30) 0.937 
rs11704979 324 96 0.13 294 71 0.11 1.18 (0.87–1.61) 0.287 1.16 (0.87–1.61) 0.287 

As both SNPs are parts of haploblock A, we next performed SNP haplotype association analysis using the three most common haplotypes hap1–hap3. Haplotype analyses may provide more power to detect association with disease than single marker analyses alone [26]. In the initial sample, we found no statistically significant association between haplotypes hap1–hap3 and MI (Table 3). However, hap2, which is defined by the combination of the rare alleles, showed a trend for association with MI (P=0.067, OR=0.76, 95% CI=0.56–1.02; Table 3). Hap2 was present more frequently in control individuals, thus corresponding to a protective haplotype. Additionally, as for single SNP analyses, we observed a significant association between the number of copies of hap2 and MI. Individuals homozygous for hap2 were at a significantly reduced risk for MI than individuals with no or only one copy of the protective haplotype (P=0.003, OR=0.44, 95% CI=0.25–0.77 and P=0.020, OR=0.52, 95% CI=0.23–0.91 respectively; Figure 2). Importantly, individuals with 0, 1 or 2 copies of hap2 respectively, did not differ with respect to age (54.5±10, 53.8±10.1 and 56.8±8.3 years), BP levels (systolic BP 134±19, 135±17 and 136±16 mmHg; diastolic BP 81±10, 82±9 and 82±9 mmHg), BMI (26.9±3.6, 27.6±4.1 and 27.5 ±4.1 kg/m2) or serum lipid concentrations (LDL cholesterol 146±36, 148±34 and 147±40 mg/dl; HDL cholesterol 54±14, 54±14 and 55±15 mg/dl; and triacylglycerols 165±90, 180±112 and 180±114 mg/dl). The observed gender difference (71.9, 74.6, 58.1% male, P=0.033) did not influence the results after correction in a logistic regression model (P=0.025, OR=0.48, 95% CI=0.25–0.91).

Table 3
Frequencies of hap1–hap3, and association results with MI in the initial and second case-control samples

Padj, P value from multivariate model (adjusted for age, total cholesterol/HDL cholesterol levels, systolic BP, BMI, diabetes mellitus and smoking). From the initial and second sample, n=98 and n=41 individuals were excluded due to improbable haplotype reconstruction. *Significant P value.

Number of haplotypes (frequency) in initial sample (n=708)Number of haplotypes (frequency) in second sample (n=1488)
Haploblock AMI casesControlsOR (95% CI)PadjMI casesControlsOR (95% CI)Padj
hap1 424 (0.55) 345 (0.53) 1.11 (0.78–1.59) 0.549 727 (0.56) 884 (0.53) 1.20 (0.93–1.55) 0.159 
hap2 197 (0.26) 210 (0.32) 0.76 (0.56–1.02) 0.067 312 (0.24) 460 (0.27) 0.80 (0.65–0.98) 0.034* 
hap3 104 (0.14) 70 (0.11) 1.29 (0.90–1.84) 0.167 198 (0.15) 238 (0.14) 1.06 (0.84–1.33) 0.643 
Number of haplotypes (frequency) in initial sample (n=708)Number of haplotypes (frequency) in second sample (n=1488)
Haploblock AMI casesControlsOR (95% CI)PadjMI casesControlsOR (95% CI)Padj
hap1 424 (0.55) 345 (0.53) 1.11 (0.78–1.59) 0.549 727 (0.56) 884 (0.53) 1.20 (0.93–1.55) 0.159 
hap2 197 (0.26) 210 (0.32) 0.76 (0.56–1.02) 0.067 312 (0.24) 460 (0.27) 0.80 (0.65–0.98) 0.034* 
hap3 104 (0.14) 70 (0.11) 1.29 (0.90–1.84) 0.167 198 (0.15) 238 (0.14) 1.06 (0.84–1.33) 0.643 

ORs for MI depending on the number of hap2 copies in the initial sample (with subjects carrying no hap2 copy as a reference group)

Figure 2
ORs for MI depending on the number of hap2 copies in the initial sample (with subjects carrying no hap2 copy as a reference group)

Individuals carrying one or two copies of the protective haplotype hap2 are at significantly reduced risk of MI compared with individuals with no copy of hap2 (overall P value=0.012). Complete haplotypes could be inferred from 708 individuals.

Figure 2
ORs for MI depending on the number of hap2 copies in the initial sample (with subjects carrying no hap2 copy as a reference group)

Individuals carrying one or two copies of the protective haplotype hap2 are at significantly reduced risk of MI compared with individuals with no copy of hap2 (overall P value=0.012). Complete haplotypes could be inferred from 708 individuals.

None of the other SNP markers were significantly associated with MI. In particular, we did not find a relationship between rs1800206, the only SNP resulting in an amino acid exchange (L162V), and MI (Table 2). We further performed comprehensive testing for an epistatic interaction between any of the 17 SNPs and found a statistically significant interaction with an asymptotic P value threshold below 0.01 between rs135539 and rs11704979 (P=0.001, results not shown). To avoid missing an effect by single SNP association analyses, we investigated the other three haploblocks B, C and D for possible haplotype association. These analyses did not reveal significant association results (results not shown). For further association analyses, we therefore focussed on haplotype block A as well as on the single markers contributing to or surrounding it.

Confirmation of the association between MI and PPARα variants in an independent second case-control sample

In order to corroborate the data, we sought to validate our results in an independent second case-control cohort. Therefore an additional 1529 individuals (667 MI cases from the Regensburg MI Family Study and 862 controls from the WHO MONICA Augsburg project) were genotyped for all SNPs of the three-marker haplotype and one 5′ adjacent SNP. In this second cohort, we were able to confirm a single SNP association result for rs135551 (P=0.046, OR=0.87, 95% CI=0.72–0.99, Armitage's trend test; Table 4). Moreover, for rs135543 as well as hap2, which both showed a trend for association in the initial sample, a significant association with MI was observed resulting from an over-representation in control individuals (P=0.014, OR=0.84, 95% CI=0.71–0.96, and P=0.034, OR=0.80, 95% CI=0.65–0.98 respectively; Tables 3 and 4).

Table 4
Summary of association analyses of the second case-control sample

For the PPARα locus, four genotyped SNPs and their rs numbers from dbSNP database are shown. Numbers (n) for each genotype (11/12/22) and MAFs are listed separately for MI cases and controls. ORs with 95% CI and P values are given for allelic association and genotypic association based on Armitage's trend test, with significant P values indicated by *.

Genotype MI cases (n)Genotype controls (n)Allele 2 compared with allele 1Armitage's trend test
SNP111222MAF111222MAFOR (95% CI)POR (95% CI)P
rs135557 192 345 121 0.45 247 432 176 0.46 0.95 (0.82–1.10) 0.496 0.95 (0.82–1.10) 0.488 
rs135551 365 253 49 0.26 407 359 67 0.30 0.85 (0.72–0.99) 0.047* 0.87 (0.72–0.99) 0.046* 
rs135543 342 260 61 0.29 383 384 91 0.33 0.82 (0.70–0.96) 0.014* 0.84 (0.71–0.96) 0.014* 
rs135539 214 334 116 0.43 246 430 175 0.46 0.88 (0.76–1.00) 0.078 0.87 (0.76–1.01) 0.075 
Genotype MI cases (n)Genotype controls (n)Allele 2 compared with allele 1Armitage's trend test
SNP111222MAF111222MAFOR (95% CI)POR (95% CI)P
rs135557 192 345 121 0.45 247 432 176 0.46 0.95 (0.82–1.10) 0.496 0.95 (0.82–1.10) 0.488 
rs135551 365 253 49 0.26 407 359 67 0.30 0.85 (0.72–0.99) 0.047* 0.87 (0.72–0.99) 0.046* 
rs135543 342 260 61 0.29 383 384 91 0.33 0.82 (0.70–0.96) 0.014* 0.84 (0.71–0.96) 0.014* 
rs135539 214 334 116 0.43 246 430 175 0.46 0.88 (0.76–1.00) 0.078 0.87 (0.76–1.01) 0.075 

DISCUSSION

In the present study, we comprehensively evaluated the LD structure of common genetic variations within the PPARα gene and assessed their potential relationship with MI. In a two-step approach, we initially used a case-control sample to investigate the LD structure of the gene and to construct common haplotypes, as well as to explore the association between single SNPs and haplotypes and MI. We found a significant association of a single SNP marker (rs135551) with MI and a trend for association of the adjacent SNP (rs135543). Moreover, we identified a common three-marker haplotype (hap2), consisting of the rare alleles of rs135551, rs135543 and rs135539, that was significantly associated with the disease phenotype. Next, we were able to confirm our results in an independent second case-control sample. In this case, we detected a significant association with MI for the same two single SNP markers, rs135551 and rs135543, and hap2. Hap2 is over-represented in control individuals, thus it suggests protective properties against MI with an OR of 0.80 (P=0.034, 95% CI=0.65–0.98).

The mechanisms by which hap2 or its single marker variants influence the atherosclerotic process remain unclear. PPARα is extensively expressed in vascular cells, such as endothelial and smooth muscle cells or monocytes/macrophages [27]. PPARα activation prompts anti-inflammatory effects at multiple stages of the atherosclerotic pathway [2830]. Thus polymorphisms in the PPARα gene may affect this sequence via a functional mutation resulting, for example, in a defective protein or altered trans-activation or -repression potentials.

However, the genetic variations found in the present study to be associated with the disease phenotype are positioned in intron 2. They neither insert a potential alternative splice site (visual checking for GT-AG consensus sites) nor do they include a consensus sequence for binding of an established transcription factor (analysed by TRANSFAC 7.0 Public) [31]. It is therefore unlikely that one of these SNPs itself represents the causal mutation. We assume that an untested proximate marker in LD with hap2 might be a functional variant. SNP markers of hap2 are part of a weak but distinct LD block covering the promoter region and exon 1 to intron 2 as shown by HapMap data. Marker rs135557, located upstream to the transcription start site, is not associated with MI in the present study. Hence the postulated causal mutation could be located between the core promoter region and intron 2 of the PPARα gene, but is unlikely to be located within the neighbouring gene. A possible effect of such a functional variant could be altered gene expression, as seen with the exonic PPARα L162V missense mutation, where the valine allele possesses higher transcriptional activation in vitro [6,32]. In the present study, the L162V variant did not show any association with MI. Other studies examining the impact of the L162V variant on cardiovascular disease have provided inconsistent results. In the LOCAT (the Lipid Coronary Angiography Trial) study, the Val162 allele was associated with reduced progression of angiographically assessed diffuse atherosclerosis, whereas the prospective Northwick Park Heart Survey found no impact on the L162V variant on the risk for ischaemic heart disease [10]. In a study by Tai et al. [33], 827 men from VA-HIT (the Veterans Affair HDL Intervention Trial) had a reduction in cardiovascular events in Val162 carriers only in the presence of insulin resistance or diabetes mellitus [33].

Some possible limitations of the present study should be considered. MI cases from both the initial and the second sample were recruited from the Regensburg MI Family Study. Our sampling strategy included MI families from all over Germany with at least two living siblings where the index patient was affected with MI under the age of 60 years and the sibling suffered from MI or severe CAD. The use of our index criteria could possibly introduce a bias for MI survival and a familial form of the disease. However, the fact that all MI cases are from unrelated families of Caucasian ethnicity from all parts of Germany excludes a geographical selection bias. Most importantly, replication of our results with an independent control group supports the validity of our findings. However, we cannot support our findings with prospective data from an unselected sample and, hence, our results may not be widely applicable to the general population.

A second possible limitation is the number of female MI patients included in the present study. It is very probable that the lack of statistical power in the female subsample can be attributed to the lower number (approx. 30% of the male subgroup). However, gender-specific effects are unlikely because both allelic- and genotype-association tests point to the same direction and have the same order of magnitude, i.e. relative risk reduction of approx. 20% in male and female MI patients. Furthermore, gender-specific effects could not be observed in any of the previously published studies [10,33,34].

We anticipate the possibility that our findings may represent a ‘false positive’ result: probably the most common problem of many genetic association studies [35]. In particular, our association results did not withstand adjustments for multiple testing. However, we were able to conduct an independent confirmation study that corroborated our initial findings, which suggest that our findings are not a type 1 statistical error (i.e. a false positive result). To place our findings in context, there have been two recently conducted whole-genome association studies on MI and CAD both employing the Affymetrix GeneChip® Human Mapping 500K Array Set [36]. However, this chip provided no coverage of the region between rs135557, rs135551 and rs135543, where we saw the association signal [36]. In these studies, the remaining PPARα gene region was covered by a total of 12 SNP markers, four of which were also included in the present study (rs4253662, rs12330015, rs4253754 and rs4253778). Neither the German ‘Cardiogenics’ MI sample nor the British WTCCC (Wellcome Trust Case Control Consortium) MI/CAD sample revealed an association with any of these SNP markers at a P value <0.05 [36]. Therefore the results from these large whole-genome screens neither support nor impair the findings of the present study.

In conclusion, our results suggest that genetic variations in the PPARα gene, in addition to established genetic and environmental factors, may influence the risk of MI in a European population. The present findings support further investigation to address the role of PPARα in cardiovascular disease and, in the future, detailed analysis of this gene may be warranted.

Abbreviations

     
  • BMI

    body mass index

  •  
  • BP

    blood pressure

  •  
  • CAD

    coronary artery disease

  •  
  • CI

    confidence interval

  •  
  • HDL

    high-density lipoprotein

  •  
  • LD

    linkage disequilibrium

  •  
  • LDL

    low-density lipoprotein

  •  
  • MAF

    minor allele frequency

  •  
  • MI

    myocardial infarction

  •  
  • MONICA

    Monitoring of Trends and Determinants in Cardiovascular Disease

  •  
  • OR

    odds ratio

  •  
  • PPARα

    peroxisome-proliferator-activated receptor α

  •  
  • SNP

    single nucleotide polymorphism

  •  
  • UTR

    untranslated region

  •  
  • WHO

    World Health Organization

The present study was supported by grants from the Deutsche Forschungsgemeinschaft (He1921/9-1), the Vaillant-Stiftung (to C.H.), and the Deutsche Stiftung fuer Herzforschung. The KORA-gen (Cooperative Research in the Region of Augsburg) and the MONICA Augsburg studies were initiated and financed by the GSF-National Research Centre for Environment and Health. The MONICA Augsburg Study was initiated and conducted by Ulrich Keil (Institute of Epidemiology and Social Medicine, University of Münster, Münster, Germany) and co-workers. KORA-gen was supported within the German NGFN (National Genomic Research Network) and by BMBF (Federal Ministry of Education and Research).

References

References
1
Marenberg
 
M. E.
Risch
 
N.
Berkman
 
L. F.
Floderus
 
B.
de Faire
 
U.
 
Genetic susceptibility to death from coronary heart disease in a study of twins
N. Engl. J. Med.
1994
, vol. 
330
 (pg. 
1041
-
1046
)
2
Cardon
 
L. R.
Bell
 
J. I.
 
Association study designs for complex diseases
Nat. Rev. Genet.
2001
, vol. 
2
 (pg. 
91
-
99
)
3
Fruchart
 
J. C.
Duriez
 
P.
Staels
 
B.
 
Peroxisome proliferator-activated receptor-α activators regulate genes governing lipoprotein metabolism, vascular inflammation and atherosclerosis
Curr. Opin. Lipidol.
1999
, vol. 
10
 (pg. 
245
-
257
)
4
Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults
Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP)
JAMA, J. Am. Med. Assoc.
2001
, vol. 
285
 (pg. 
2486
-
2497
)
5
Rubins
 
H. B.
Robins
 
S. J.
Collins
 
D.
, et al 
Gemfibrozil for the secondary prevention of coronary heart disease in men with low levels of high-density lipoprotein cholesterol. Veterans Affairs High-Density Lipoprotein Cholesterol Intervention Trial Study Group
N. Engl. J. Med.
1999
, vol. 
341
 (pg. 
410
-
418
)
6
Flavell
 
D. M.
Pineda Torra
 
I.
Jamshidi
 
Y.
, et al 
Variation in the PPARα gene is associated with altered function in vitro and plasma lipid concentrations in Type II diabetic subjects
Diabetologia
2000
, vol. 
43
 (pg. 
673
-
680
)
7
Jamshidi
 
Y.
Montgomery
 
H. E.
Hense
 
H. W.
, et al 
Peroxisome proliferator-activated receptor α gene regulates left ventricular growth in response to exercise and hypertension
Circulation
2002
, vol. 
105
 (pg. 
950
-
955
)
8
Tai
 
E. S.
Demissie
 
S.
Cupples
 
L. A.
, et al 
Association between the PPARα L162V polymorphism and plasma lipid levels: the Framingham Offspring Study
Arterioscler. Thromb. Vasc. Biol.
2002
, vol. 
22
 (pg. 
805
-
810
)
9
Vohl
 
M. C.
Lepage
 
P.
Gaudet
 
D.
, et al 
Molecular scanning of the human PPARα gene: association of the L162V mutation with hyperapobetalipoproteinemia
J. Lipid Res.
2000
, vol. 
41
 (pg. 
945
-
952
)
10
Flavell
 
D. M.
Jamshidi
 
Y.
Hawe
 
E.
, et al 
Peroxisome proliferator-activated receptor α gene variants influence progression of coronary atherosclerosis and risk of coronary artery disease
Circulation
2002
, vol. 
105
 (pg. 
1440
-
1445
)
11
Broeckel
 
U.
Hengstenberg
 
C.
Mayer
 
B.
, et al 
A comprehensive linkage analysis for myocardial infarction and its related risk factors
Nat. Genet.
2002
, vol. 
30
 (pg. 
210
-
214
)
12
Sedlacek
 
K.
Neureuther
 
K.
Mueller
 
J. C.
, et al 
Lymphotoxin-α and galectin-2 SNPs are not associated with myocardial infarction in two different German populations
J. Mol. Med.
2007
, vol. 
85
 (pg. 
997
-
1004
)
13
WHO MONICA Project Principal Investigators
The World Health Organization MONICA Project (monitoring trends and determinants in cardiovascular disease): a major international collaboration
J. Clin. Epidemiol.
1988
, vol. 
41
 (pg. 
105
-
114
)
14
Lieb
 
W.
Graf
 
J.
Gotz
 
A.
, et al 
Association of angiotensin-converting enzyme 2 (ACE2) gene polymorphisms with parameters of left ventricular hypertrophy in men. Results of the MONICA Augsburg echocardiographic substudy
J. Mol. Med.
2006
, vol. 
84
 (pg. 
88
-
96
)
15
Stark
 
K.
Neureuther
 
K.
Sedlacek
 
K.
, et al 
The common Y402H variant in complement factor H gene is not associated with susceptibility to myocardial infarction and its related risk factors
Clin. Sci.
2007
, vol. 
113
 (pg. 
213
-
218
)
16
Mayer
 
B.
Lieb
 
W.
Radke
 
P. W.
, et al 
Association between arterial pressure and coronary artery calcification
J. Hypertens.
2007
, vol. 
25
 (pg. 
1731
-
1738
)
17
Holmer
 
S. R.
Hengstenberg
 
C.
Mayer
 
B.
, et al 
Lipoprotein lipase gene polymorphism, cholesterol subfractions and myocardial infarction in large samples of the general population
Cardiovasc. Res.
2000
, vol. 
47
 (pg. 
806
-
812
)
18
Armitage
 
P.
 
Test for linear trends in proportions and frequencies
Biometrics
1955
, vol. 
11
 (pg. 
375
-
386
)
19
Ardlie
 
K. G.
Kruglyak
 
L.
Seielstad
 
M.
 
Patterns of linkage disequilibrium in the human genome
Nat. Rev. Genet.
2002
, vol. 
3
 (pg. 
299
-
309
)
20
Stephens
 
M.
Smith
 
N. J.
Donnelly
 
P.
 
A new statistical method for haplotype reconstruction from population data
Am. J. Hum. Genet.
2001
, vol. 
68
 (pg. 
978
-
989
)
21
Stephens
 
M.
Donnelly
 
P.
 
A comparison of bayesian methods for haplotype reconstruction from population genotype data
Am. J. Hum. Genet.
2003
, vol. 
73
 (pg. 
1162
-
1169
)
22
Purcell
 
S.
Neale
 
B.
Todd-Brown
 
K.
, et al 
PLINK: a tool set for whole-genome association and population-based linkage analyses
Am. J. Hum. Genet.
2007
, vol. 
81
 (pg. 
559
-
575
)
23
Ingelsson
 
E.
Schaefer
 
E. J.
Contois
 
J. H.
, et al 
Clinical utility of different lipid measures for prediction of coronary heart disease in men and women
JAMA, J. Am. Med. Assoc.
2007
, vol. 
298
 (pg. 
776
-
785
)
24
Kinosian
 
B.
Glick
 
H.
Garland
 
G.
 
Cholesterol and coronary heart disease: predicting risks by levels and ratios
Ann. Intern. Med.
1994
, vol. 
121
 (pg. 
641
-
647
)
25
Ridker
 
P. M.
Rifai
 
N.
Cook
 
N. R.
Bradwin
 
G.
Buring
 
J. E.
 
Non-HDL cholesterol, apolipoproteins A-I and B100, standard lipid measures, lipid ratios, and CRP as risk factors for cardiovascular disease in women
JAMA, J. Am. Med. Assoc.
2005
, vol. 
294
 (pg. 
326
-
333
)
26
Newton-Cheh
 
C.
Hirschhorn
 
J. N.
 
Genetic association studies of complex traits: design and analysis issues
Mutat. Res.
2005
, vol. 
573
 (pg. 
54
-
69
)
27
Marx
 
N.
Duez
 
H.
Fruchart
 
J. C.
Staels
 
B.
 
Peroxisome proliferator-activated receptors and atherogenesis: regulators of gene expression in vascular cells
Circ. Res.
2004
, vol. 
94
 (pg. 
1168
-
1178
)
28
Marx
 
N.
Sukhova
 
G. K.
Collins
 
T.
Libby
 
P.
Plutzky
 
J.
 
PPARα activators inhibit cytokine-induced vascular cell adhesion molecule-1 expression in human endothelial cells
Circulation
1999
, vol. 
99
 (pg. 
3125
-
3131
)
29
Pasceri
 
V.
Cheng
 
J. S.
Willerson
 
J. T.
Yeh
 
E. T.
 
Modulation of C-reactive protein-mediated monocyte chemoattractant protein-1 induction in human endothelial cells by anti-atherosclerosis drugs
Circulation
2001
, vol. 
103
 (pg. 
2531
-
2534
)
30
Staels
 
B.
Koenig
 
W.
Habib
 
A.
, et al 
Activation of human aortic smooth-muscle cells is inhibited by PPARα but not by PPARγ activators
Nature
1998
, vol. 
393
 (pg. 
790
-
793
)
31
Matys
 
V.
Fricke
 
E.
Geffers
 
R.
, et al 
TRANSFAC: transcriptional regulation, from patterns to profiles
Nucleic Acids Res.
2003
, vol. 
31
 (pg. 
374
-
378
)
32
Sapone
 
A.
Peters
 
J. M.
Sakai
 
S.
, et al 
The human peroxisome proliferator-activated receptor α gene: identification and functional characterization of two natural allelic variants
Pharmacogenetics
2000
, vol. 
10
 (pg. 
321
-
333
)
33
Tai
 
E. S.
Collins
 
D.
Robins
 
S. J.
, et al 
The L162V polymorphism at the peroxisome proliferator activated receptor α locus modulates the risk of cardiovascular events associated with insulin resistance and diabetes mellitus: the Veterans Affairs HDL Intervention Trial (VA-HIT)
Atherosclerosis
2006
, vol. 
187
 (pg. 
153
-
160
)
34
Koch
 
W.
Hoppmann
 
P.
Pfeufer
 
A.
Mueller
 
J. C.
Schomig
 
A.
Kastrati
 
A.
 
No replication of association between estrogen receptor α gene polymorphisms and susceptibility to myocardial infarction in a large sample of patients of European descent
Circulation
2005
, vol. 
112
 (pg. 
2138
-
2142
)
35
Cardon
 
L. R.
Palmer
 
L. J.
 
Population stratification and spurious allelic association
Lancet
2003
, vol. 
361
 (pg. 
598
-
604
)
36
Samani
 
N. J.
Erdmann
 
J.
Hall
 
A. S.
, et al 
Genome wide association analysis of coronary artery disease
N. Engl. J. Med.
2007
, vol. 
357
 (pg. 
443
-
453
)

Author notes

1

These authors contributed equally to the present study.