The association between fibrinogen haplotypes and myocardial infarction in men is partly mediated through pleiotropic effects on the serum IL-6 concentration
Abstract.
Background and objectives. Fibrinogen haplotypes have been associated with risk of myocardial infarction (MI), independently of plasma fibrinogen concentration, and experimental data indicate that fibrinogen exerts pleiotropic effects on interleukin 6 (IL-6) production. Also, the coagulation factor XIII (gene symbol F13A1) Val34Leu haplotype tag single nucleotide polymorphism (htSNP) has been reported to exert pleiotropic effects on serum IL-6 concentration and to be associated with risk of MI. Therefore, in the present case–control study (a substudy to the Stockholm Heart Epidemiology Program), the effects of the fibrinogen gamma (FGG) 9340T>C [rs1049636], fibrinogen alpha (FGA) 2224G>A [rs2070011] and F13A1 Val34Leu [rs5985] htSNPs on concentrations of plasma fibrinogen and serum IL-6 and risk of MI were assessed.
Results. There were no associations between these SNPs and the plasma fibrinogen concentration. In contrast, in male controls the FGA 2224G>A htSNP was significantly associated with serum IL-6 concentration (P < 0.05). Also, in men the FGG-FGA*1 haplotype (containing the major FGG 9340T and FGA 2224G alleles) was associated with increased risk of MI [adjusted odds ratio (OR) 95% confidence interval (CI): 1.29 (1.02, 1.62)] and with higher IL-6 concentrations, whereas the least common FGG-FGA*4 haplotype (containing the minor FGG 9340C and FGA 2224A alleles) conferred lowered risk [adjusted OR (95% CI): 0.70 (0.57, 0.86)] and lowered IL-6 concentrations. In women, fibrinogen haplotypes were not associated with risk of MI after adjusting for cardiovascular risk factors.
Conclusion. In healthy men, fibrinogen haplotypes are associated with serum IL-6 concentrations in a manner consistent with their impact on MI risk.
Introduction
The inflammation and coagulation pathways have been implicated in the aetiology of myocardial infarction (MI) [1, 2]. Interleukin 6 (IL-6), a key component of the inflammation pathway, participates in processes leading to endothelial dysfunction [3] and atherosclerotic plaque development [4]. In sufferers of acute MI, significantly higher serum IL-6 concentrations have been detected at sites of plaque rupture when compared with those from the systemic circulation [5], whereas systemic elevations were reported in individuals with unstable angina pectoris [6] and seem to predict the risk of future MI amongst seemingly healthy men [7]. IL-6 is also known to upregulate the production of acute phase proteins such as C-reactive protein and fibrinogen [8].
Plasma fibrinogen concentration has gained recognition as an independent predictor of MI [9] and several pathophysiological mechanisms (e.g. promotion of smooth muscle cell migration and proliferation, facilitation of low-density lipoprotein accumulation and foam cell formation) may underlie this relationship [10]. Moreover, fibrinogen is the main component of the fibrin clot and may therefore play a key role during acute MI events. Elevated plasma fibrinogen concentrations in young male survivors of a first MI have been associated with lower fibrin clot porosity [11], a property that confers resistance to fibrinolysis [12]. The relationship between plasma fibrinogen concentration and fibrin clot architecture seems to be modulated by the coagulation factor XIII (gene symbol F13A1) Val34Leu haplotype tag single nucleotide polymorphism (htSNP) [13]. Presence of the F13A1 34Leu variant has been associated with decreased risk of MI [14], but conflicting data has also been published [15]. Moreover, a significant association between the F13A1 Val34Leu htSNP and IL-6 concentration has been reported [16]. Interestingly, both in vitro [17–19] and in vivo [20] experiments have demonstrated that fibrin(ogen) stimulates chemokine (monocyte chemoattractant protein-1) and cytokine (IL-6) production. As the source of IL-6 at sites of plaque rupture has been suggested to be either cells trapped in the thrombus or cells in the atherosclerotic arterial wall [5], the ability of fibrin(ogen) to stimulate IL-6 production may be of particular importance in this context. Thus, a dynamic crosstalk between the inflammation and coagulation pathways is likely to contribute to the pathophysiology of atherosclerotic plaque formation and MI.
We have recently reported that fibrinogen haplotypes inferred using genotype data from the fibrinogen gamma (FGG) 9340T>C and fibrinogen alpha (FGA) 2224G>A htSNPs (known as the 1299+79T>C and −58G>A polymorphisms, respectively) are associated with risk of MI, independently of the plasma fibrinogen concentration [21]. Similarly, an association was reported between FGG haplotypes containing the major FGG 9340T allele and deep vein thrombosis which was not accounted for by effects on the plasma fibrinogen concentration [22]. In contrast, the fibrinogen beta (FGB) 1437G>A SNP (known as the −455G/A polymorphism) was found to influence the plasma fibrinogen concentration in male postinfarction participants in the Stockholm Heart Epidemiology Program (SHEEP) [23]. However, this effect did not appear to be clinically translated into an association between the FGB 1437G>A SNP and risk of MI [23], which is consistent with data from a recent meta-analysis [24] and with studies reporting that FGB haplotypes are neither associated with risk of MI [25] nor deep vein thrombosis [22].
As experimental data indicate that fibrinogen is involved in cytokine production, we investigated whether the FGG 9340T>C and FGA 2224G>A htSNPs exert pleiotropic effects (i.e. effects of a gene on more than one phenotype) on the serum IL-6 concentration in the SHEEP study sample comprising survivors of a first MI and population-based controls. Whereas genetic variants in the FGB gene, in particular the FGB 1437G>A SNP, have gained significant attention in previous genotype–phenotype association studies, FGG and FGA SNPs have been studied to a much lesser extent. Interestingly, results from our previous work have indicated that extending the combined FGG and FGA haplotypes with FGB SNPs does not influence the relationship between these haplotypes and MI [21]. Considering that most studies [24], including a report based on the SHEEP study population [23], have indicated lack of association between FGB gene variants and cardiovascular disease, the FGB gene was omitted in the present study. Thus, the FGG and FGA fibrinogen SNPs were selected on the basis of our previous findings in relation to MI and because they are tagging SNPs for haplotypes within each fibrinogen gene, which are linked due to the linkage disequilibrium pattern across the fibrinogen gene cluster [21, 22]. There is evidence to suggest that coagulation factor XIII exerts biological activities beyond haemostasis, which involve gene regulation and are dependent on its transglutaminase activity [26]. Therefore, we also explored whether the F13A1 Val34Leu htSNP exerts pleiotropic effects on the serum IL-6 concentration. A third objective was to study the F13A1 Val34Leu htSNP, fibrinogen SNPs and haplotypes in relation to MI. As there are well-established differences in the molecular and cellular (patho)physiology of the cardiovascular system between genders [27] the data has been evaluated separately in men and women.
Material and methods
Subjects
SHEEP is a population-based case–control study designed to form a basis for studies of genetic, biochemical and environmental factors predisposing to MI. Recruitment strategies and protocol features have been described [28, 29]. Potential study participants (age range 45–70 years) were all Swedish citizens living in Stockholm County without a previous clinical diagnosis of MI. Male cases were recruited between 1992 and 1993 and female cases between 1992 and 1994. The criteria for MI diagnosis were based on guidelines issued by the Swedish Society of Cardiology in 1991 and included: (i) typical symptoms; (ii) marked elevation of the enzymes serum creatine kinase and lactate dehydrogenase and (iii) characteristic electrocardiogram changes. If two of the three criteria were fulfilled, the patient was diagnosed with MI. Five control candidates per case were sampled within 2 days of the case event, in order to enable replacement of potential nonresponders. For each postinfarction patient a randomly selected healthy individual was selected from the study base, after matching for age, sex and catchment area. Due to a late response from some of the initial controls, occasionally both the initial and the alternative controls have been included. The present substudy was based on a database and biobank comprising a total of 1213 cases (852 men and 361 women) and 1561 controls (1054 men and 507 women).
Blood samples were collected approximately 3 months after the index cardiac event in the patients, when the acute-phase reaction had subsided. The investigation of the controls was carried out as close as possible to that of the corresponding case in order to avoid bias due to seasonal variation in blood parameters, and all participants underwent a physical examination. Blood samples were collected in the morning in both cases and controls, following an overnight fast, and the samples were stored in aliquots at −70 °C until analyses were performed.
Biochemical analyses
Plasma fibrinogen concentration was determined as described by Vermylen et al. [30]. Serum concentrations of cholesterol and triglycerides were determined by an enzymatic colometric method (Kodak Echtachem Analyzer; Eastman Kodak, Rochester, NY, USA). Insulin concentration was assayed using a radioimmunoassay (RIA 100; Pharmacia, Uppsala, Sweden) and IL-6 concentration using an enzyme-linked immunosorbent assay (IL-6 Eli-pair, Diaclone, Research, Besancon, France).
Genetic analyses
DNA was extracted using the RapidPrep Macro Genomic DNA Isolation Kit (Pharmacia Biotech, Sweden). Genotyping for the FGG 9340T>C [rs1049636, GenBank accession number AF350254], FGA 2224G>A [rs2070011, GenBank accession number AF361104] and F13A1 Val34Leu [rs5985, GenBank accession number AF418272] htSNPs was performed using the matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry technology (Sequenom Inc., San Diego, CA, USA) [31].
Statistical analyses
Statistical analyses were conducted using the statview package (SAS Institute, Inc. Cary, NC, USA; version 5.0.1 for Windows). Physical inactivity was defined as inactive leisure time, including occasional walks, during the last 5–10 years. Smoking status was considered as former and current smokers versus never smokers. Hypertension was defined as blood pressure ≥140/90 mmHg or antihypertensive medication at the time of inclusion in the study or a history of regular antihypertensive drug therapy during the last 5 years. Hypercholesterolaemia was defined as serum levels of total cholesterol >5.0 mmol L−1 or treatment with lipid-lowering drugs. Triglyceride, insulin, fibrinogen and IL-6 measurements deviate from the normal distribution and are therefore presented as geometric mean values with 95% confidence interval (CI). Differences between groups were estimated by using unpaired t-test, Mann–Whitney U-test and chi-squared test as appropriate. Odds ratios (OR) with 95% CI were calculated using logistic regression analysis.
Hardy–Weinberg equilibrium analyses were performed using the chi-square test. Differences in genotype and allele frequency distribution were assessed using the Metropolitan method [32]. Haplotype analyses were performed using two different approaches, namely a Bayesian haplotype reconstruction algorithm implemented in the phase package, version 2.1 [33] and a maximum likelihood model (the Stochastic-Expectation Maximization algorithm) implemented in the thesias program, version 2 [34].
Ethical considerations
The Ethics Committee of the Karolinska Institutet approved the study. All participants gave their informed consent to participation.
Results
Subject characteristics
The general characteristics of the SHEEP participants have been reported [28]. Differences between the cases and controls included in the present substudy and the frequency distribution of the FGG 9340T>C, FGA 2224G>A and F13A1 Val34Leu genotypes are presented in Table 1. The cardiovascular risk factor burden was higher in cases than in controls. Moreover, there were gender-specific differences in risk factors: the female patients were older (P < 0.001), had more frequently hypercholesterolaemia (P < 0.01), a higher concentration of fibrinogen in plasma (P < 0.001) and were less frequently smokers (P < 0.001).
| Variable | Men | Women | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Cases | n | Controls | n | P-value | Cases | n | Controls | n | P-value | |
| Age (years) | 58.3 (7.1) | 852 | 58.8 (7.1) | 1054 | – | 61.6 (6.8) | 361 | 62.0 (6.7) | 507 | – |
| Physical inactivity (%) | 41.0 | 334 | 32.3 | 331 | <0.001 | 54.8 | 189 | 41.2 | 205 | <0.001 |
| Smoking (%) | ||||||||||
| Never (%) | 20.5 | 169 | 35.5 | 369 | <0.001 | 35.2 | 122 | 51.3 | 257 | <0.001 |
| Former and current (%) | 79.5 | 656 | 64.5 | 669 | 64.8 | 225 | 48.7 | 244 | ||
| Waist-to-hip ratio | 0.96 (0.06) | 847 | 0.94 (0.06) | 1044 | <0.001 | 0.86 (0.09) | 361 | 0.84 (0.09) | 503 | <0.001 |
| Hypertension (%)a | 44.8 | 382 | 50.9 | 537 | 0.008 | 49.6 | 179 | 55.2 | 280 | 0.10 |
| Hypercholesterolemia (%)b | 83.0 | 705 | 78.2 | 823 | 0.008 | 89.1 | 320 | 84.6 | 427 | 0.05 |
| Triglycerides (mmol L−1) | 1.8 (1.7–1.8) | 850 | 1.4 (1.3–1.4) | 1053 | <0.001 | 1.8 (1.7–1.8) | 359 | 1.2 (1.2–1.3) | 505 | <0.001 |
| Insulin (μU mL−1) | 11.0 (10.4–11.6) | 602 | 8.8 (8.5–9.2) | 754 | <0.001 | 11.1 (10.3–12.0) | 297 | 8.3 (7.8–8.7) | 414 | <0.001 |
| Fibrinogen (g L−1) | 3.6 (3.5–3.6) | 789 | 3.4 (3.3–3.4) | 992 | <0.001 | 3.8 (3.7–3.9) | 341 | 3.5 (3.5–3.6) | 467 | <0.001 |
| IL-6 (ng L−1) | 1.6 (1.4–1.8) | 556 | 1.0 (0.9–1.2) | 590 | <0.001 | 1.5 (1.3–1.8) | 275 | 1.5 (1.2–1.8) | 285 | 0.70 |
| Genotype frequency (%) | ||||||||||
| FGG 9340T>C | ||||||||||
| CC | 8.7 | 71 | 9.2 | 94 | 0.66 | 11.2 | 38 | 9.0 | 43 | 0.52 |
| TC | 39.8 | 325 | 41.7 | 424 | 42.2 | 143 | 43.4 | 208 | ||
| TT | 51.5 | 420 | 49.1 | 500 | 46.6 | 158 | 47.6 | 228 | ||
| FGA 2224G>A | ||||||||||
| AA | 14.6 | 119 | 13.6 | 137 | 0.78 | 18.1 | 61 | 15.4 | 73 | 0.59 |
| GA | 45.5 | 371 | 45.0 | 454 | 44.5 | 150 | 46.3 | 219 | ||
| GG | 40.0 | 326 | 41.4 | 417 | 37.4 | 126 | 38.3 | 181 | ||
| F13A1 Val34Leu | ||||||||||
| Leu/Leu | 7.3 | 60 | 7.1 | 72 | 0.95 | 4.7 | 16 | 7.0 | 34 | 0.35 |
| Val/Leu | 37.2 | 305 | 36.7 | 374 | 40.5 | 138 | 38.2 | 185 | ||
| Val/Val | 55.4 | 454 | 56.2 | 573 | 54.8 | 187 | 54.8 | 265 | ||
- Values are mean (SD), geometric mean (95% CI) or percentage (%) of subjects in groups. aHypertension, blood pressure ≥140/90 mmHg and/or treatment with antihypertensive drugs. bHypercholesterolaemia, total cholesterol >5.0 mmol L−1 and/or treatment with lipid-lowering drugs. Group comparisons were made by unpaired t-tests (normally distributed variables), Mann–Whitney U-test (skewed variables), chi-squared test (categorical variables) and the Metropolitan method [32] (genotypes).
The genotype frequency distribution did not deviate from that expected for a population in Hardy–Weinberg equilibrium. No differences in genotype or allele frequency distributions between cases and controls grouped according to gender were detected. Based on genotype data for the FGG 9340T>C and FGA 2224G>A htSNPs, four common haplotypes were detected in this sample: FGG-FGA*1 (TG, prevalence 45%), FGG-FGA*2 (TA, prevalence 25%), FGG-FGA*3 (CG, prevalence 18%) and FGG-FGA*4 (CA, prevalence 12%). It is noteworthy in this context, that by including only two SNPs in the haplotype analyses, a rather large proportion (up to 40%) of the variation in the chromosomes remained unaccounted for in the present study. Amongst postinfarction patients, there was a significant difference in the frequency distribution for the FGG-FGA*1 haplotype between men and women (P = 0.004). Amongst controls, the frequency distribution of the FGG-FGA*2 haplotype was significantly different between men and women (P = 0.01).
Genetic variation in relation to serum IL-6 and plasma fibrinogen concentrations
In male controls, a significant association with serum IL-6 concentration was detected for the FGA 2224G>A htSNP (P = 0.04), homozygotes for the minor A allele having significantly lower levels compared with homozygotes for the major G allele [(ng L−1) geometric mean (95% CI): 0.70 (0.42, 1.17) vs. 1.28 (1.04, 1.58), P = 0.01]. Also, amongst male controls, the FGG-FGA*1 haplotype was associated with a significantly higher serum IL-6 concentration compared with noncarriers of the haplotype [(ng L−1) 1.15 (1.02, 1.29) vs. 0.79 (0.61, 1.02), P = 0.004] (Table 2). Conversely, carriers of the FGG-FGA*4 haplotype had significantly lower serum IL-6 concentrations compared with noncarriers of the haplotype [(ng L−1) 0.82 (0.67, 1.00) vs. 1.19 (1.05, 1.35), P = 0.001]. Moreover, the serum IL-6 concentration was significantly lower in carriers of the FGG-FGA*4 haplotype compared with the FGG-FGA*1 haplotype (P < 0.001). Amongst women controls, carriers of the FGG-FGA*3 haplotype appeared to have significantly higher IL-6 concentrations compared with noncarriers of this haplotype [(ng L−1) 2.23 (1.72, 2.89) vs. 1.28 (1.09, 1.51), P < 0.001]. These specific associations between fibrinogen haplotypes and the serum IL-6 concentration remained statistically significant after correction for multiple testing [35]. The F13A1 Val34Leu htSNP was not associated with serum IL-6 concentration. The FGG 9340T>C, FGA 2224G>A and F13A1 Val34Leu genotypes did not appear to influence the plasma fibrinogen concentration (data not shown).
| Haplotype | Men | Women | ||
|---|---|---|---|---|
| Cases | Controls | Cases | Controls | |
| FGG-FGA*1 | ||||
| Carriera | 1.64 (1.48, 1.81) | 1.15 (1.02, 1.29) | 1.48 (1.28, 1.71) | 1.41 (1.20, 1.65) |
| Noncarrier | 1.53 (1.25, 1.87) | 0.79 (0.61, 1.02) | 1.78 (1.35, 2.34) | 1.78 (1.31, 2.41) |
| P-value | 0.57 | 0.004 | 0.22 | 0.19 |
| FGG-FGA*2 | ||||
| Carrier | 1.63 (1.39, 1.90) | 0.97 (0.79, 1.19) | 1.51 (1.11, 1.70) | 1.38 (1.21, 1.88) |
| Noncarrier | 1.61 (1.44, 1.80) | 1.10 (0.98, 1.24) | 1.58 (1.28, 1.84) | 1.53 (1.35, 1.86) |
| P-value | 0.91 | 0.27 | 0.73 | 0.47 |
| FGG-FGA*3 | ||||
| Carrier | 1.52 (1.28, 1.81) | 1.06 (0.88, 1.27) | 1.55 (1.21, 1.99) | 2.23 (1.72, 2.89) |
| Noncarrier | 1.64 (1.48, 1.83) | 1.06 (0.93, 1.20) | 1.56 (1.34, 1.81) | 1.28 (1.09, 1.51) |
| P-value | 0.50 | 0.99 | 0.98 | <0.001 |
| FGG-FGA*4 | ||||
| Carrier | 1.65 (1.39, 1.96) | 0.82 (0.67, 1.00) | 1.76 (1.43, 2.09) | 1.33 (0.98, 1.79) |
| Noncarrier | 1.60 (1.44, 1.79) | 1.19 (1.05, 1.35) | 1.46 (1.23, 1.72) | 1.55 (1.33, 1.80) |
| P-value | 0.78 | 0.001 | 0.18 | 0.32 |
- Values are presented as geometric mean (95% CI). aCarrier indicates presence of the haplotype on one or both chromosomes.
Genetic variation in relation to MI risk
The FGG 9340T>C, FGA 2224G>A and F13A1 Val34Leu htSNPs did not appear to discriminate between patients and controls when controlling for age and residential area. Haplotype analyses indicated that the FGG-FGA*2 haplotype was associated with increased MI risk in men [OR (95% CI): 1.19 (1.04, 1.37)] (Table 3). The FGG-FGA*1 and FGG-FGA*4 haplotypes appeared to be associated with MI risk after adjustment for cardiovascular risk factors [adjusted OR (95% CI): 1.29 (1.02, 1.62) and 0.70 (0.57, 0.86), respectively], and the discriminative power of the FGG-FGA*2 haplotype was reinforced [adjusted OR (95% CI): 1.29 (1.06, 1.58)]. Complementary analysis using the most common FGG-FGA*1 haplotype as reference group generated similar results for the FGG-FGA*4 haplotype [adjusted OR (95% CI): 0.68 (0.53, 0.87), P = 0.002] whereas the effect of the FGG-FGA*2 haplotype did not differ significantly from that of the most common haplotype. The relationship between the FGG-FGA*4 haplotype and MI remained statistically significant after correction for multiple testing [35]. In contrast, in women, the FGG-FGA*1 haplotype appeared to be protective when adjusting for age, residential area and IL-6 [adjusted OR (95% CI): 0.72 (0.53, 0.97)]. However, this effect was attenuated when controlling for other risk factors in addition to IL-6 [adjusted OR (95% CI): 0.77 (0.55, 1.08)]. Control for the effects of either hormone replacement therapy or menopause did not influence the ORs [adjusted OR (95% CI): 0.75 (0.54, 1.06) and 0.79 (0.56, 1.11), respectively].
| Haplotype | Men | Women | ||||
|---|---|---|---|---|---|---|
| OR (95% CI), P-valuea | OR (95% CI), P-valueb | OR (95% CI), P-valuec | OR (95% CI), P-valuea | OR (95% CI), P-valueb | OR (95% CI), P-valuec | |
| FGG-FGA*1 | 0.94 (0.80, 1.10), 0.45 | 1.12 (0.90, 1.39), 0.31 | 1.29 (1.02, 1.62), 0.03 | 0.76 (0.61, 0.96), 0.02 | 0.72 (0.53, 0.97), 0.03 | 0.77 (0.55, 1.08), 0.12 |
| FGG-FGA*2 | 1.19 (1.04, 1.37), 0.01 | 1.34 (1.11, 1.61), 0.002 | 1.29 (1.06, 1.58), 0.01 | 1.05 (0.85, 1.29), 0.66 | 1.09 (0.84, 1.41), 0.53 | 1.12 (0.83, 1.52), 0.45 |
| FGG-FGA*3 | 0.99 (0.85, 1.16), 0.91 | 0.79 (0.64, 0.97), 0.02 | 0.85 (0.69, 1.06), 0.16 | 1.11 (0.88, 1.40), 0.38 | 1.00 (0.75, 1.34), 0.99 | 0.80 (0.57, 1.12), 0.19 |
| FGG-FGA*4 | 0.89 (0.77, 1.03), 0.12 | 0.81 (0.67, 0.99), 0.04 | 0.70 (0.57, 0.86), <0.001 | 1.08 (0.87, 1.34), 0.46 | 1.21 (0.93, 1.58), 0.16 | 1.17 (0.87, 1.58), 0.31 |
- The reference category is all haplotypes but the one given. aORs adjusted for age and residential area. bORs adjusted for age, residential area and IL-6. cORs adjusted for age, residential area, IL-6, hypercholesterolaemia, triglycerides, insulin, hypertension, waist-to-hip ratio, smoking (never versus former and current) and physical inactivity.
Discussion
We here report that fibrinogen haplotypes inferred using genotype data from the FGG 9340T>C and FGA 2224G>A htSNPs are associated with serum IL-6 concentration, and these haplotypic effects encountered in men are consistent with their relationship to MI risk. To the best of our knowledge, these findings indicating that fibrinogen haplotypes may exert pleiotropic effects on the serum IL-6 concentration are novel. Thus, fibrinogen may contribute to inflammatory processes that are known to be pivotal in developing atherosclerotic plaques. Taking the view that atherosclerosis is an inflammatory disease [1] and that the cross-talk between the inflammation and coagulation pathways may contribute to the development, progression and rupture of atherosclerotic plaques, these findings merit consideration although the mechanisms by which variation in the fibrinogen genes may influence the IL-6 concentration remain unknown. It is possible that the fibrinogen haplotypes studied in the present report quantitatively or qualitatively affect those fibrinogen species that interact with receptors and cells involved in IL-6 expression, i.e. that there are haplotype effects on plasma concentrations or linkage to nonsynonymous variations influencing protein function.
In the present study, gender-specific haplotypic effects were noted, i.e. the FGG-FGA*1 haplotype appeared to be associated with elevated serum IL-6 concentrations and to confer an increased risk of MI in men but not in women. These findings may reflect gender-related differences in gene expression patterns [36] and in the molecular and cellular (patho)physiology of the heart and blood vessels [27] as well as substantial differences in the prevalence of cardiovascular risk factors between men and women in SHEEP [28]. In particular, the serum IL-6 concentration did not differ significantly between female cases and controls, whereas it did between their male counterparts. Accordingly, the serum IL-6 concentration proved to be a better risk indicator in men than in women in SHEEP [37], which is in agreement with recent data from the prospective FINRISK study [38]. Different cardiovascular risk factors are associated with different manifestations of the acute coronary syndrome (ACS) [39] and probably account for a worse prognosis in women [40]. These observations might be reflected by the different haplotype frequency distributions observed in the present study. Age is a well-established risk factor for ACS, and women tend to be older when they suffer their first MI. Interestingly, the OR for the FGG-FGA*1 haplotype changes markedly when evaluated in relation to MI amongst individuals under 60 years, increasing in men [1.45 (1.03, 2.03)], corroborating previous findings [21]. It can be speculated that the potential adverse effect of the FGG-FGA*1 haplotype is underestimated in the group of individuals over 60 years due to a higher incidence of cardiac death in carriers of this haplotype.
The FGG 9340T>C and FGA 2224G>A htSNPs did not appear to influence plasma fibrinogen concentration, in agreement with our previous report [21] and with genome-wide studies [41, 42]. Moreover, we did not find any association between the F13A1 Val34Leu genotype and serum IL-6 concentration, which is in contrast to what has been published [16]. Furthermore, the F13A1 Val34Leu htSNP did not appear to be associated with risk of MI, which is in agreement with previous results [15].
The associations between fibrinogen haplotypes and MI risk detected in the present study are not entirely consistent with recent results from the Stockholm Coronary Artery Risk Factor (SCARF) study [21] as the FGG-FGA*1 and FGG-FGA*4 haplotypes appeared to influence disease risk only after controlling for cardiovascular risk factors. The divergent results could be attributed to differences in age, environmental exposures and study design. As the effect size of the fibrinogen haplotypes is rather modest it could be quite sensitive to confounding. Individuals with type 1 diabetes mellitus, renal failure and chronic inflammatory disease were not included in the SCARF study [43], whereas the same exclusion criteria were not applied in SHEEP. Also, lack of clinical phenotypic homogeneity between the studies could be a potential contributor. The SHEEP patients were recruited during 1992–1994 using diagnostic criteria for MI defined in 1991, whereas the SCARF patients were enrolled during 1996–2000 using more refined criteria. As improved diagnostic tools enable detection of smaller MIs, it cannot be excluded that the SHEEP patients were somewhat sicker. Moreover, both incidence and mortality rates for MI decreased substantially in Sweden during the 1990s [44], which might have affected the haplotype frequency distribution.
One limitation of the present study is the restriction of the analyses to postinfarction patients surviving for at least 28 days after the acute event, as it cannot be excluded that subjects who died had a different genetic predisposition and/or different exposures to the risk factors studied. Also, life-style changes including dietary habits, smoking cessation and increased physical activity might have taken place in the early post-MI period. In addition, medication with lipid-lowering agents and acetylsalicylic acid might have influenced the concentrations of fibrinogen and IL-6. At the time of the examination only 6.2% (n = 75) of the cases and 2.2% (n = 35) of the controls were on lipid-lowering medication, very few of whom were on statins [3.1% (n = 38) of the cases and 1.3% (n = 21) of the controls]. The corresponding numbers for treatment with acetylsalicylic acid were 84.0% (n = 1019) amongst patients and 6.6% (n = 103) amongst controls. However, the concentrations of fibrinogen and IL-6 did not appear to be influenced by treatment with lipid-lowering drugs or acetylsalicylic acid, in neither patients nor in controls (data not shown).
Of note, only few genotype–phenotype association studies have generated consistent results. A possible explanation for this shortcoming is the multidimensionality of complex traits such as IL-6 and MI, i.e. an intricate interplay between genetic and environmental factors is likely to contribute to the phenotypic variation. Nevertheless, it cannot be excluded that the associations noted in the present study may represent false-positive results [45]. However, application of the Benjamini and Hochberg False Discovery Rate correction [35] for multiple testing did not influence the main finding of the present study, namely that the IL-6 concentration might indeed be an intermediate phenotypic link between fibrinogen haplotypes and MI. Importantly, the data herein presented needs to be corroborated in independent study samples and the possible mechanisms underlying the pleiotropic effect of fibrinogen haplotypes on the serum IL-6 concentration should be clarified. In addition, it may be worthwhile to study the concentrations of fibrinogen degradation products in future studies, given their potentially stronger stimulatory effect on IL-6 release as indicated by in vitro studies [17, 18].
In conclusion, the present study indicates that fibrinogen haplotypes, inferred using genotype data from the FGG 9340T>C and FGA 2224G>A htSNPs, exert pleiotropic effects on serum IL-6 concentration in healthy men. These pleiotropic effects are consistent with the impact of fibrinogen haplotypes on risk of MI. The same effects were not observed in women, which is likely to be due to differences in molecular and cellular (patho)physiology of the cardiovascular system between genders.
Conflict of interest statement
None of the authors has any conflict of interest to declare.
Acknowledgements
We thank Gunnel Gråbergs for excellent SAS programming. This study was supported by grants from the Swedish Research Council (8691, 9533), the Swedish Heart-Lung Foundation, the Petrus and Augusta Hedlund Foundation, the Stockholm County Council, the Foundation for Old Servants and Karolinska Institutet.
