Genetic Variation in the Leukotriene Pathway and Carotid Intima-Media Thickness
A 2-Stage Replication Study
Background and Purpose— The recent finding that genetic variation in the leukotriene biosynthesis pathway may confer an increased risk of ischemic stroke and atherosclerosis has implicated the leukotriene family as potential mediators of cardiovascular disease. Using a 2-stage replication design, we have examined whether polymorphisms in 8 genes related to this pathway are associated with early atherosclerosis and remodeling as measured by carotid artery intima-media thickness.
Methods— We assessed 969 individuals from the Carotid Atherosclerosis Progression Study (CAPS), a community based study of normal subjects, for 39 variants in the leukotriene pathway. Significant associations and gene–environment interactions were found for 21 variants in the initial cohort and were examined in the next 1905 consecutive cases from the same CAPS population.
Results— No replicable association between any individual polymorphism and carotid intima-media thickness itself was present after correction for multiple testing. A single gene–environment interaction was replicated between rs17222814 on bifurcation intima-media thickness and alcohol consumption exceeding 30 grams per day.
Conclusion— The genetic variants we examined in the leukotriene biosynthesis pathway have little effect on early atherosclerosis and remodeling risk as determined by carotid intima-media thickness. Our study cannot exclude them as being risk factors for more advanced stages in the atherosclerotic process.
Cardiovascular disease and stroke remain one of the leading causes of death in the developed world,1 although the genetic basis of the condition is largely uncharacterized. Considerable evidence suggests inflammation may play an important role in cardiovascular disease, and recently genetic variants in inflammatory response genes have been shown to interact with both conventional proinflammatory risk factors such as smoking, as well as with chronic infection.2,3 One such pathway is the leukotriene biosynthesis pathway; initially, variants in 5-lipoxygenase activating protein (FLAP) and leukotriene A4 hydrolase (LTA4H) were associated with increased stroke risk.4–6 More recently variants associated with the leukotriene B4 receptors BLT1 and BLT2 receptor region were also associated with ischemic stroke risk, implicating the full leukotriene synthesis pathway.7 The leukotriene pathway has also been implicated in atherosclerosis, with leukotriene B4 (LTB4) shown to be expressed in carotid plaques8 and coronary artery smooth muscle cells.9
It is unclear at what stages of the atherosclerotic process genetic alterations in leukotriene activity act to increase disease. The enhanced inflammatory processes could promote a number of stages, including early endothelial dysfunction, established plaque progression, and plaque instability. Carotid intima-media thickness (IMT) is widely used as a measure of arterial remodeling and early atherosclerosis, and has been extensively used to examine the contribution of genetic variants to early atherosclerosis. It is an independent predictor of both myocardial infarction and stroke.10
We performed a 2-stage replication study in a large community population to determine whether genetic variation in the leukotriene pathway was associated with early atherosclerosis and remodeling as indicated by increased IMT. IMT is a continuous variable and can be measured in all individuals. Its use therefore offers considerably more power than case control studies of symptomatic disease. This, and our large sample size, allowed us to also evaluate gene–environment interactions with proinflammatory risk factors in the same 2-stage replication design.
We studied subjects from the Carotid Atherosclerosis Progression Study.11 All members of a German primary health care service population (n=32 708) who lived within a radius of 50 kilometers from 5 study sites in Western Germany were invited to participate. Within a predefined time limit, 6962 (21.3%) agreed to participate. In this study we included the first 969 consecutively recruited subjects as the initial study population, and the next 1905 consecutively recruited subjects as the replication population.
Vascular risk factors were assessed using a standardized computer-assisted interview performed by a physician experienced in vascular medicine. Risk factors determined included pack-years of cigarette smoking and smoking category (never/ex/current smoker), history of arterial hypertension, history of diabetes mellitus, and body mass index. The mean value of 3 supine blood pressure measurements was taken as the actual arterial blood pressure. Fasting blood samples were taken for estimation of serum cholesterol and glycosylated hemoglobin A1 (HbA1c). Total serum cholesterol was determined enzymatically using a commercial kit (Boehringer). Baseline high-sensitivity C-reactive protein circulating levels were measured using an IMMAGE automatic immunoassay system (Beckmann-Coulter). Informed written consent was obtained from all participants, and the study protocol was approved by the ethical review committee of the Hospital of J.W. Goethe-University Frankfurt am Main.
For ultrasonic examinations, a 7.5- to 10.0-MHz linear array transducer was used (P700SE; Phillips Medical System). Preprocessing configurations (log gain compensation [60 dB], and image persistence) were held constant during all examinations. The gain was adjusted so that the least dense arterial wall interface was just visible. With the use of antero-oblique insonation, far-wall carotid IMT was visualized within the common carotid artery (CCA) 20 to 60 mm proximally from the flow divider, the carotid bifurcation (bifurcation IMT) 0 to 20 mm proximally from the flow divider, and the interior carotid artery (ICA) bulb (ICA bulb IMT) 0 to 20 mm distally from the flow divider, bilaterally. The images were digitally captured during the systole of a single heartbeat for off-line measurements. Vertical and horizontal calibration measurements were performed every 100th measurement using an ultrasound assurance phantom. Automated image analysis software (Matlab; The Mathworks, Inc) was used to determine both the blood/intimal and the medial/adventitial borderline automatically with a gray value-based edge detection algorithm combined with higher-degree polynomial fitting along these borderlines. The mean distance between the blood/intimal and the medial/adventitial interfaces was calculated and defined as the IMT of the corresponding arterial segment. The mean length of the arterial segment in which IMT was determined was 14.35 mm for the left CCA IMT, 12.85 mm for the right CCA IMT, 5.75 mm for the left bifurcation IMT, 5.8 mm for the right bifurcation IMT, and 3.45 mm for the ICA bulb IMT on both sides.
Choice of Genetic Polymorphisms and Overall Study Design
In the initial association study in the first 969 consecutive subjects, we studied 39 variants in the 8 genes regulating the leukotriene synthesis pathway (Table 1, Figure). The genes were: FLAP, 5-LO, LTA4H, BLT1/BLT2, leukotriene C4 synthase (LTC4S), and cysteinyl leukotriene receptors 1 and 2 (CYSLTR1/CYSLTR2). In addition, variants in ALOX15 were also examined on the basis of familial similarity and previous findings in a stroke case control analysis.7 Eighteen of the variants were in FLAP and 21 were in the remaining 7 genes.
This was followed by a replication study in the next 1905 consecutive subjects of variants for which associations with either carotid IMT itself, or via gene–environment interactions, had been found. On this basis we tested for replication of 21 variants.
We tested for associations between gene variants and IMT at 3 sites in the carotid artery, namely the CCA ICA and the bifurcation of the carotid artery (BIF). All analyses were conducted using SPSS v15. After univariate analysis, adjustment was made for age and sex, and then for common cardiovascular risk factors. The following risk factors were included: body mass index, hypertension, smoking (pack-years), low-density lipoprotein cholesterol, alcohol intake >30 grams/day, and a measure of impaired glucose intolerance (defined as diabetes mellitus or HbA1C >6%). Our prespecified hypothesis was that interactions would exist with proinflammatory environmental stimuli; therefore, we determined interactions with the following risk factors: smoking, body mass index, alcohol, and impaired glucose tolerance.
Correction for multiple testing in the regression model was performed based on the number of genes examined, with a nominal threshold of P=0.01 for the initial screen. A Bonferroni correction based on the number of genes would represent a threshold of P=0.006. However, to ensure associations in the initial cohort were not missed, a less conservative threshold of P=0.01 was chosen. A full Bonferroni correction based on the number of genes was made for the replication cohort, setting a significance threshold as P=0.006 because variants in 8 genes were replicated. For gene–environment interactions, only a single gene was tested at a time. Bonferroni correction was therefore applied to the number of environmental risk factors tested for, namely 4. A significance level of P=0.0125 was therefore considered as significant for gene–environment interactions.
Power calculations suggest that assuming a significance level of 1×10−5 allows the detection of a genetic variant contributing between 1% and 1.5% variation of IMT thickness with >80% power in a sample size of 1000 individuals (Quanto; http://hydra.usc.edu/gxe/). Thus our initial population is adequate in size to identify changes <0.05±0.01 mm in IMT, a figure previously observed with single polymorphisms in previous genetic association studies in this population.3 The use of a larger replication population ensures that such a finding should be replicated if it is a true association in the initial population of 969 individuals.
Demographic characteristics of the study population are given in Table 2. The replication cohort was younger than the primary cohort and had a lower frequency of individuals with hypertension. All other variables were not significantly different between the 2 populations. Genotypes were assessed for Hardy-Weinberg equilibrium and a full listing of variants investigated are available in Table 1, whereas the leukotriene pathway is shown in the Figure. Data in the manuscript refer only to positive associations after correction for multiple testing because of the large number of variants investigated.
Initial Study in 969 Individuals
Demographic details including prevalence of risk factors and mean IMT values are shown in Table 2. Significant findings for age-/sex-adjusted and multivariable models after correction for multiple testing are shown in Table 3. There was no significant association between CCA IMT and any variant. A single variant in FLAP (rs17216466) showed a significant association with ICA IMT after correction in a multivariable model. Associations with BIF IMT were shown for 3 variants in 3 genes (rs17222814 in FLAP; rs2072510 in LTA4H; and rs2664593 in ALOX15) after similar correction. No individual variant was associated with IMT at >1 carotid site.
Gene–environment interactions were determined with the potentially proinflammatory stimuli: body mass index, alcohol, smoking, and impaired glucose tolerance After correction for multiple testing, numerous interactions in CCA, ICA, and BIF IMT were identified in the initial cohort (Table 4). Interactions were shown with smoking (rs1108372 in LTA4H) and impaired glucose tolerance (rs320995 in CYSLTR1 and rs2664593 in ALOX15) on ICA IMT alone, in addition to multiple interactions with alcohol in excess of 30 grams/day in CCA and BIF IMT independently. Of these, 2 variants in FLAP showed an interaction with alcohol excess at >1 carotid site (rs17216349 and rs17222926 in CCA and BIF). Both variants showed a similar direction of effect in the 2 vessels, with the rare allele being associated with higher IMT at both sites for rs17216349 and lower IMT at both sites for rs17222926 (Table 4).
Replication Study in 1905 Subjects
Any variant that showed an association with IMT on multivariable analysis, or that showed gene–environment interactions in the initial sample of 969 individuals, was then replicated in a further sample of 1905 individuals from the same Carotid Atherosclerosis Progression Study population. A total of 21 variants were replicated.
After correction for multiple testing, no variant was identified as significant in the replication sample (Table 3). When examining gene–environment interactions, only a single association was replicated in the second sample set, namely rs17222814 in FLAP and alcohol consumption on BIF IMT. A second variant was identified in a different IMT segment, rs1108372 in LTA4H, with BIF IMT in the replication cohort rather than ICA in the initial cohort (Table 5).
Under the initial replication hypothesis, after correction for multiple testing, no variant studied was shown to be significantly associated with carotid IMT. When considering gene–environment interactions under the same model and after correction for multiple testing, only a single gene–environment interaction has been identified and confirmed in the same carotid segment.
In this primary replication study, in a total of 3000 subjects we found no evidence of a major role of variants in genes encoding components of the leukotriene pathway in increasing carotid IMT risk. Previous studies have implicated a number of these variants in ischemic stroke risk, but the stage of the pathogenic process at which this occurs is unclear. Our study suggest that these associations are not accounted for by accelerated early atherosclerosis or vessel remodeling, as evidenced by increased IMT.
Given the large number of variants investigated, we chose a 2-stage replication design of sufficient size to allow associations to be detected in the first sample, with positive associations being confirmed in a larger replication sample from the same population. This methodology represented a balance between cost and power, an issue of increasing importance with the advent of large-scale genetic association studies. Although not part of our initial study design, for completeness we did perform a global analysis of all data combined (ie, combining the initial and replication samples and analyzing jointly), but did not identify any significant finding for the 21 variants when this was possible (data not shown).
The initial analysis in 969 individuals revealed few associations with IMT, with no single variant associated with increased IMT at >1 carotid site. In contrast, multiple gene–environment interactions, most notably with excess alcohol consumption, were identified. However, despite the initial large sample size, the chance of detecting false-positive gene–environment interactions is much higher than that for primary association with IMT itself. This is a consequence of the requirement to split a sample by both a genetic and an environmental component, particularly if the genetic variant is rare. Therefore, it is particularly important that these interactions are replicated.
In the replication study of a further 1905 individuals from the same Carotid Atherosclerosis Progression Study, no association with IMT itself was replicated. Only 1 of the gene–environment interactions was replicated, and this was 1 out of many and may therefore be a chance finding. Consequently, we uncovered little evidence for these genetic variants playing a significant role in IMT risk. Whereas a difference in the 2 populations cannot be excluded given the age and sex demographics, both cohorts were consecutively collected series of patients without selection bias. The younger age and consequently slightly lower frequency of hypertension in the replication cohort could be masking a replication effect, although analysis before correction for risk factors also revealed no replication for any variant investigated (data not shown). Thus, although a replication cannot be totally excluded, it remains extremely unlikely.
Correction for multiple testing can be approached in a number of ways, from the most conservative full Bonferroni on all variants, genes, and phenotypes examined, through evidence-based correction on linkage disequilibrium between single nucleotide polymorphisms such as the spectral decomposition found by Nyholt,12 to no correction but cautious interpretation. We have adopted a modified Bonferroni correction based on the number of genes investigated or a full Bonferroni correction based on the number of gene–environment interactions investigated as appropriate. Such a process reduces the risk of false-positive associations, particularly in gene–environment interactions in which stratification by both a genetic factor and an environmental factor is likely to lead to groups with small sample sizes.
Genetic variation in the 5-lipoxygenase promoter has previously been associated with increased atherosclerosis.13 We have failed to replicate this finding in our cohort, although we have looked at single nucleotide polymorphisms rather than the tandem Sp1 motif previously studied. The effects observed by Dwyer et al13 were also related to dietary intake, particularly levels of n-6 polyunsaturated fatty acid and marine n-3 fatty acids, which were found to promote and inhibit, respectively, leukotriene-mediated inflammation leading to atherosclerosis. We do not have such dietary data for our cohort, and it may be that this gene–environment interaction is more significant than the risk factors examined as part of this study.
Similarly, the Carotid Atherosclerosis Progression Study population represents a community population with predominantly early atherosclerosis, and there were insufficient advanced plaque and stenoses to exclude associations with advanced atherosclerosis. As such, we cannot exclude the possibility that the risk previously identified with ischemic stroke is manifest through later pathogenic mechanisms such as plaque instability and plaque rupture rather than the early atherosclerotic mechanisms of vessel remodeling or changes in IMT. Further specific studies will be required to assess the role of leukotrienes in such processes.
In summary, no variant was identified that associated with carotid IMT in both samples. A single gene–environment interaction was detectable in both samples independently at the same carotid site and remains significant after correction for multiple testing. The lack of association at >1 single carotid site, however, suggests this finding is likely to be of limited importance. As such, we would conclude that the genetic variants investigated in this study have little influence on early atherosclerosis and remodeling as measured by changes in IMT. Further, these results also emphasize that when examining gene–environment interactions, sample sizes should be as large as possible to ensure false-positive associations are not detected as a consequence of population substratification.
- Received May 13, 2008.
- Revision received July 28, 2008.
- Accepted July 30, 2008.
Yusuf S, Reddy S, Ounpuu S, Anand S. Global burden of cardiovascular diseases: part I: general considerations, the epidemiologic transition, risk factors and impact of urbanisation. Circulation. 2001; 104: 2746–2753.
Jerrard-Dunne P, Sitzer M, Risley P, Steckel DA, Buehler A, Kegler S, Markus HS. Inetrleukin-6 promoter polymorphism modulates the effects of heavy alcohol consumption on early carotid artery atherosclerosis. Stroke. 2003; 34: 402–407.
Jerrard-Dunne P, Sitzer M, Risley P, Buehler A, Kegler S, Markus HS. Inflammatory gene load is associated with enhanced inflammation and early carotid atherosclerosis in smokers. Stroke. 2004; 35: 2438–2444.
Helgadottir A, Manolescu A, Thorleifsson G, Gretarsdottir S, Jonsdottir H, Thorsteinsdottir U, Samani N, Gudmundsson G, Grant SFA, Thorgeirsson G, Sveinbjornsdottir S, Valdimarrson E, Matthiasson SE, Johannsson H, Gudmundsdottir O, Gurney ME, Sainz J, Thorhallsdottir M, Andresdottir M, Frigge ML, Topol EJ, Kong A, Gudnason V, Hakonaeson H, Gulcher JR, Stefansson K. The gene encoding 5-lipoxygenase activating protein confers risk of myocardial infarction and stroke. Nat Gen. 2004; 36: 233–239.
Helgadottir A, Manolescu A, Helgason A, Thorleifsson G, Thorsteinsdottir U, Gudbjartsson D, Gretarsdottir S, Magnusson KP, Gudmundsson G, Hicks A, Jonsson T, Grant SFA, Sainz J, O'Brien SJ, Sveinbjornsdottir S, Valdimarrson E, Matthiasson SE, Levey AI, Abramson JL, Reilly MP, Vaccarino V, Wolfe ML, Gudnason V, Quyyumi AA, Topol E, Rader DJ, Thorgeirsson G, Gulcher J, Hakonaeson H, Kong A, Stefansson K. A variant of the gene encoding leukotriene A4 hydrolase confers ethnicity-specific risk of myocardial infarction. Nat Gen. 2006; 38: 68–74.
Helgadottir A, Gretarsdottir S, St. Clair D, Manolescu A, Cheung J, Thorleifsson G, Pasdar A, Grant SFA, Whalley LJ, Hakonarson H, Thorsteinsdottir U, Kong A, Gulcher J, Stefansson K, MacLeod MJ. Association between the gene encoding 5-lipoxygenase activating protein and stroke replicated in a Scottish population. Am J Hum Genet. 2005; 76: 505–509.
Bevan S, Dichgans M, Wiechmann E, Gschwendtner A, Meitinger T, Markus HS. Genetic variation in members of the leukotriene biosynthesis pathway confer an increased risk of ischaemic stroke—a replication study in two independent populations. Stroke. 2008; 39: 1109–1114.
Back M, Bu D, Branstrom R, Sheikine Y, Yan Z-Q, Hansson GK. Leukotriene B4 signalling through NF-KB-dependent BLT1 receptors on vascular smooth muscle cells in atherosclerosis and intimal hyperplasa. Proc Natl Acad Sci U S A. 2005; 102: 17501–17506.
Lorenz MW, Markus HS, Bots ML, Rosvall M, Sitzer M. Prediction of clinical cardiovascular events with carotid intima-media thickness: a systematic review and meta-analysis. Circulation. 2007; 115: 459–467.