|Home | About | Journals | Submit | Contact Us | Français|
The genetic background of atrial fibrillation (AF) in whites and African Americans is largely unknown. Genes in cardiovascular pathways have not been systematically investigated.
We examined a panel of approximately 50,000 common single nucleotide polymorphisms (SNPs) in 2,095 cardiovascular candidate genes and AF in three cohorts with participants of European (n=18,524; 2,260 cases) or African American descent (n=3,662; 263 cases) in the National Heart Lung and Blood Institute's Candidate Gene Association Resource. Results in whites were followed up in the German Competence Network for AF (n=906, 468 cases). The top result was assessed in relation to incident ischemic stroke in the Cohorts for Heart and Aging Research in Genomic Epidemiology Stroke Consortium (n= 19,602 whites, 1544 incident strokes). SNP rs4845625 in the IL6R gene was associated with AF (relative risk (RR) C allele, 0.90; 95% confidence interval (CI), 0.85–0.95; P=0.0005) in whites, but did not reach statistical significance in African Americans (RR, 0.86; 95% CI, 0.72–1.03; P=0.09). The results were comparable in the German AF Network replication, (RR, 0.71; 95% CI, 0.57–0.89; P=0.003). No association between rs4845625 and stroke was observed in whites. The known chromosome 4 locus near PITX2 in whites also was associated with AF in African Americans (rs4611994, hazard ratio, 1.40; 95% CI, 1.16–1.69; P=0.0005).
In a community-based cohort meta-analysis, we identified genetic association in IL6R with AF in whites. Additionally, we demonstrated that the chromosome 4 locus known from recent genome-wide association studies in whites is associated with AF in African Americans.
Genetic variation may affect susceptibility to atrial fibrillation (AF), the most common sustained arrhythmia, beyond clinical variables. Recent AF genome-wide association analyses have identified and replicated genetic loci on chromosomes 4, 16 and 1 with largely unknown function.1–3 Due to the large number of association tests in genome-wide analyses, stringent significance thresholds are used. Whereas a conservative significance threshold limits false positive results, it also leads to false negatives. Biologically interesting associations may therefore fail to reach genome-wide significance. Such concerns provide the motivation for investigating a pathophysiologically focused gene panel to identify additional important genetic variants associated with AF. Significant differences in AF prevalence have been observed between African Americans and whites of European descent.4 The racial variation raises the possibility that the genetic basis of AF in African Americans is different from that in whites.5 It is unknown if any of the single nucleotide polymorphisms (SNPs) associated with AF in European ancestry populations are relevant to AF in African Americans.
We related a gene-centric selection of cardiovascular candidate SNPs to AF data in three community-based cohorts of whites and African Americans participating in the Candidate Gene Association Resource (CARe) study,6 and followed up our results in the German Competence Network for AF.7 Since recent genome-wide genetic investigations indicated a potential relevance of genetic loci related to AF for stroke risk8,9 we further examined our top finding in relation to incident stroke in the Heart and Aging Research in Genomic Epidemiology Stroke Consortium.10
The study design and methods of the CARe study have been described.6 For AF analyses, the NHLBI discovery samples consisted of three prospective, community-based cohorts from the US with genotyping in individuals of European descent, the Atherosclerosis Risk In Communities study (ARIC, n=10,878), the Cardiovascular Health Study (CHS, n=3,952), and the Framingham Heart Study (FHS, n=3,694). Data in African Americans were available in ARIC (n=2,922) and CHS (n=740). Race/ethnicity was by self-report. DNA samples and phenotype information from the CARe cohorts were shipped to the Broad Institute of MIT and Harvard, harmonized across studies (http://www.cdisc.org/models/sds/v3.1/index.html.) and deposited in the central database. Genotype quality control procedures were performed separately for each cohort. Control criteria comprised sex concordance of genotype-inferred sex and reported sex, missing SNP data and genotyping success rate. SNPs with missing data greater than 10% and samples with call rates of less than 90% were discarded.
We followed up our top results in the German Competence Network for AF (AFNET, n=906), a national registry of AF cases (Supplementary Table 1).7 Further information on study design, genotyping and quality control for the discovery and follow-up samples is provided in the Supplement (Methods and Supplementary Table 1).
The outcome was initial, paroxysmal, persistent, or permanent atrial fibrillation or atrial flutter.11 We combined analyses of incident AF and prevalent AF in the meta-analysis. For details on outcome ascertainment in the three cohorts see the Supplement.
We used a cardiovascular gene-centric 50K SNP IBC (ITMAT, Broad, and CARe) array for large-scale genomic association studies comprising 49,320 SNPs in 2,095 genes (http://bmic.upenn.edu/cvdsnp/).
Individuals with missing phenotype or genotype data and age <50 years at baseline were excluded. We chose the age cut-off because AF incidence is low under the age of 50.12 For analyses of incident AF, participants with AF at baseline were excluded. Loss to follow-up was a censoring event. Studies with data available on heart surgery (ARIC, CHS) did not count AF events if coronary artery bypass grafting or valve surgery occurred during the same hospital stay.
We used additive genetic models and all analyses were performed separately by race. Analyses adjusted for age at DNA draw, sex, site (ARIC, CHS) or cohort (FHS). For analyses of prevalent AF, we performed logistic regression. In African Americans, analysis of prevalent AF was not performed due to a very small number of prevalent AF cases (n=38). Framingham applied a generalized estimating equation model to account for correlated family data. For incident AF, we used proportional hazards regression. The outcome was time to AF (or censor) from baseline. For Framingham data we used a robust variance estimate that accounts for familial correlation. We consistently used the allele that comes first in alphabetical order as the effect allele in our analyses.
We used an inverse variance-weighted meta-analysis approach to combine the regression parameter estimates of five analyses for whites across incident and prevalent AF and two analyses for African Americans. The martingale property of Cox models implies that the estimated regression coefficient for CHS (FHS) incident analysis is independent of the estimated regression coefficient for CHS (FHS) prevalent analysis, which justifies combining incident and prevalent results in the meta-analysis. In a previous genome-wide association study of AF,2 associations were fairly consistent across the ARIC, CHS, and FHS cohorts.
An array-wide significance threshold of P<2×10−6 corresponds to an approximate Bonferroni correction of alpha=0.05 by 25,000 independent tests after accounting for the number of SNPs in linkage equilibrium on the IBC gene chip (see Supplement for details). Genes with two or more SNPs meeting the following criteria were candidates to move forward to our replication study: 1) meta-analysis P<0.001; and 2) same direction of effect for at least four out of five analyses that contributed to the meta-analysis. Seven genes met these criteria and we chose one SNP from each of these genes. For six of the genes we selected the SNP with the smallest P value. For one of the genes we chose the SNP with the second-smallest P value because it had a larger minor-allele frequency than the most significant SNP (36% instead of 15%). In AFNET, genome-wide association analyses were performed adjusting for age and sex using logistic regression in relation to prevalent AF. Regression estimates and adjusted standard errors were combined using inverse-variance weighted meta-analysis. Quantile-quantile plots on the distribution of the observed and expected P values are provided Supplementary Figure 1.
In secondary analyses we tried to understand whether the top locus in the IL6R gene located on the same chromosome as the known KCNN33 locus represented an independent signal. We compared the linkage disequilibrium structure for rs13376333 and the newly identified SNP rs4845625 in HapMap and the available 1000 Genomes database (http://www.1000genomes.org). In addition we estimated the proportion of individuals who had risk alleles from both loci by imputing genotypes using additional genome-wide genotyping data in the three cohorts (ARIC, n=9044, CHS, n=3204, FHS, n=4404).2,3 The imputation quality (observed/expected variance) for the two SNPs was good and ranged from 0.83–0.99. Framingham Heart Study data included related individuals. We also provide Pearson correlation coefficients between genotypes.
We further tested the association of the top finding with incident ischemic stroke in silico in the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Stroke Consortium. The consortium incorporates individuals from ARIC, CHS and FHS as well as from the Rotterdam Study with genome-wide data (n=19,602 whites). Study design and methods have been reported earlier.10 We applied an inverse variance fixed effects meta-analysis to combine the study data. Further details are provided in the Supplement.
Study cohorts included a total of 22,186 participants (55.9% women; 16.5% African Americans). In whites analyses were based on 389 prevalent AF cases and 1,871 incident AF events over a total of 218,297 person-years of follow-up and in African Americans on 263 incident AF events in over 56,869 person-years. Details on the characteristics of the discovery and follow-up cohorts are provided in Table 1 and Supplementary Table 2. In whites, only the previously reported region on chromosome 4 neighboring the PITX2 gene reached our pre-defined significance threshold with the SNP rs2200733, relative risk (RR) T allele, 1.40; 95% confidence interval (CI), 1.29–1.51; P=2.5×10−16. rs2200733 showed modest significance, P=0.002 in African Americans (Table 2, results by cohort, Supplementary Table 3). From the three cohorts of white individuals, we carried forward seven SNPs with meta-analysis P<0.001 and effect estimates in the same direction in at least four of the five analyses for additional follow-up in silico testing in AFNET,3 906 individuals of European descent (N=468 cases). All results with P<0.001 are available in Supplementary Table 4.
An intronic SNP, rs4845625 in the IL6R gene, RR, 0.90; 95% CI, 0.85–0.95; P=0.0005, reached the significance threshold of 0.007 (0.05/7 tests) in the replication sample (OR, 0.71; 95% CI, 0.57–0.89; P=0.003). rs4845625 is located close to two non-synonymous SNPs in IL6R, both of which appear associated with AF (rs8192284, r2=0.5; RR, 0.91; 95% CI, 0.86–0.97; P=0.03 and rs4537545, r2=0.49; RR, 0.90; 95% CI. 0.85–0.96; P=0.0006). The association of rs4845625 was of similar magnitude in African Americans but did not reach statistical significance (hazard ratio (HR), 0.86; 95% CI, 0.72–1.03; P=0.09). Regional plots of candidate SNPs genotyped on the IBC chip in the IL6R gene region are presented in Figure 1 for whites and African Americans.
Follow-up for an association with ischemic stroke was performed in the CHARGE stroke consortium with a mean age±standard deviation of 63±8 years and 1544 incident events over an average follow-up time of 11 years. No association between rs4845625 and incident stroke was observed (HR, 0.98; 95% CI, 0.90–1.07; P=0.68).
None of the SNPs in African Americans reached array-wide significance in our analysis. The chromosome 4 locus, an established AF susceptibility locus in whites, was among the strongest associations (rs4611994, HR, 1.40; 95% CI, 1.16–1.69; P=0.0005). Findings with P<0.001 and the same direction of effect in both African Americans samples are tabulated in Supplementary Table 5. Among these SNPs, we identified several variants in potassium channel-related genes that appear associated with AF in the discovery meta-analysis of African Americans, including KCNQ1 (rs463535, P=5.36×10−5), KCNA5 (rs7961013, P=0.0002), KCTD10 (rs1477117, P=0.0004); and MYOCD (rs758185, P=0.0005). Restriction of variants with P<0.001 to those with the same direction of effect in African Americans and whites revealed 4 genes (HPX, hemopexin; SLC2A4, solute carrier family 2 (facilitated glucose transporter), member 4; PCSK6, proprotein convertase subtilisin/kexin type 6; GNAI1, guanine nucleotide binding protein (G protein), alpha inhibiting activity polypeptide 1) of potential interest for follow-up apart from the chromosome 4 locus (Supplementary Table 6).
In secondary analyses we assessed the relation between the previously described KCNN3 and the novel IL6R locus. For participants in the present study, Pearson correlation coefficients between genotypes in the three cohorts ranged from −0.015 to −0.019 in individuals of European descent and from 0.01 to 0.075 in AF cases. The proportion of individuals with at least one risk allele in KCNN3 and at least one risk allele in IL6R was 0.34 in the total sample and 0.39 in individuals with AF.
In our large-scale candidate gene approach assessing approximately 2000 genes across three community-based cohorts in relation to AF in two races, top SNPs were located in the region of recent findings from genome-wide analyses in whites. We further extended the knowledge on the association of the known chromosome 4 locus in whites towards African Americans. In addition, we identified an intronic SNP in the IL6R gene in association with AF risk. We present several additional plausible genes of interest for follow-up in African Americans.
A novel finding of interest is the association with the IL6R gene region in relation to AF. There has been controversy regarding inflammation's role as a causative agent or a risk marker in AF.13–16 Levels of circulating interleukin-6, along with levels of other pro-inflammatory molecules such as C-reactive protein and fibrinogen, have consistently been associated with AF.13,17 Recent investigations of genetic variation of the IL6 gene also provide suggestive evidence (but without replication) for the role of the interleukin-6 system in the pathophysiology of AF.18 A correlation between nearby SNPs, e.g. the exonic SNP rs8192284 in the IL6R gene and interleukin-6 receptor concentrations, interleukin-6, and other inflammatory biomarkers such as C-reactive protein and fibrinogen has been demonstrated,19,20,20,21 as well as metabolic syndrome pathways.22 Diseases also associated with genetic variation at the IL6R locus and closely related to AF are diabetes and coronary artery disease.23,24 As intermediate risk factors they may provide a link between the polymorphisms and AF. If additional research confirms that variation in IL6R is indeed related to development of AF, the interleukin-6 pathway might become an experimental target for modulating the inflammatory milieu to modify AF risk.
A recent investigation identified an association between a KCNN3 locus and lone AF.3KCNN3 is located about 400 kb downstream of IL6R. However, the LD between the top KCNN3 SNP (rs13376333) and the IL6R SNP in HapMap release 22 is very small, with r2=0.008 in CEU and r2=0.001 in YRI. Similarly, in the preliminary 1000 Genomes data (http://www.1000genomes.org) the corresponding r2 values are 0.012 in CEU and 0.005 in YRI. The publicly available results were recapitulated in our data with correlation coefficients from −0.019 to 0.075. Because of the low correlation the IL6R finding appears to represent an independent signal.
Prevalent AF is accompanied by a measurably elevated inflammatory state.13,25 It can be argued that the predictive value of indicators of inflammation may be due to undetected episodes of AF. Recurrent, asymptomatic AF might thus be associated with elevated inflammatory activity prior to the diagnosis of AF. Reverse causation is not possible in the setting of genetic association with disease. Our findings may thus provide evidence for a potential causal role of the interleukin-6 pathway.
Recent observations that genetic loci associated with AF also may be risk indicators for stroke events8,9 led to the hypothesis that rs4845625 may be related to incident stroke. The assumption of an association was supported by consistent data on the interplay of genetic variation at the IL6R locus and circulating interleukin-6 concentrations,20 and observations that interleukin-6 levels were related to cardiovascular events such as stroke.21,26 However, we did not observe an association for the top SNP with ischemic stroke in the large CHARGE stroke consortium. One explanation might be that we did not focus on stroke subtypes since differential associations for atherothrombotic and cardioembolic brain infarction have been observed.27 Unfortunately, our current investigation is underpowered for analysis of ischemic stroke subtypes.
Whereas risk factor associations with AF are similar in whites and African Americans,28,29 AF risk factor burden is higher in African Americans, but prevalence and incidence of AF are lower compared to individuals of European descent.28 These observations suggest risk factors beyond classical risk indicators, and differences in genetic predisposition have been proposed.5,28,29
Though none of the SNPs in African Americans reached array-wide significance in our analysis, the chromosome 4 locus, an established AF susceptibility locus in whites, was among the strongest associations. The comparable effect estimate in African Americans suggests genetic similarities between the two races at this specific locus. Other recently reported SNPs in association with AF in whites at the ZFHX3 and KCNN3 loci2,3 were not represented on the IBC chip. Among SNPs with P<0.001, we identified several potassium channel-related genes including KCNQ1 (rs463535), KCNA5 (rs7961013), KCTD10 (rs1477117), and MYOCD (rs758185), a gene expressed in the myocardium. Mutations in KCNQ130 and KCNA531 have been related to AF in whites, mostly early-onset forms of AF. Little is known about KCTD10,32 a member of the polymerase delta-interacting protein 1 gene family, and MYOCD,33 which codes for a transcriptional co-factor required for cardiomyocyte survival and function.
Apart from the chromosome 4 locus, top results differed by race and associations did not consistently appear in the same direction in African Americans and whites in the current setting indicating a high probability of false positives due to a limited number of AF events and different linkage patterns in African Americans, rather than true differences by race. Overall, our results in African Americans should be interpreted with caution and viewed as hypothesis-generating.
Strengths of the study include the careful AF ascertainment in the cohorts, phenotype harmonization across cohorts, centralized high quality genotyping, uniform quality control and statistical analysis. We present results for a large sample of African American participants with genetic data in relation to AF. Follow-up with additional African American samples is necessary in the future when genetic data and AF events data from other studies become available.
Our a priori selected significance threshold mimics the strategy used in genome-wide association studies. It is based on the assumption of homogenous linkage disequilibrium. Although this assumption is questionable, independent replication data were used to protect against false positive findings. The public deposition of the genetic association data will provide a unique resource for future studies in whites and African Americans. Whereas it is biologically plausible that the IL6R locus provides support for an inflammatory contribution to AF, we acknowledge that the identified SNP may not be causally related to AF or may be in LD with a causal SNP. We also note that our findings may not be generalizable to other ethnicities/races, such as individuals of Asian ancestry. As noted above our data do not provide support for a large contribution of rs4845625 to stroke. However, we cannot exclude that rs4845625 may have a small effect size in relation to stroke, or may be related to specific stroke subtypes.
In conclusion, we identified a novel association between variation in a SNP in the IL6R gene and AF. The association is biologically plausible and may stimulate future research. In addition, we demonstrated that the chromosome 4q25 AF susceptibility locus, originally discovered in whites, is associated with AF in African Americans.
We are only beginning to understand genetic associations with atrial fibrillation (AF) in whites and African Americans. We systematically examined a panel of approximately 50,000 common single nucleotide polymorphisms (SNPs) in 2,095 cardiovascular candidate genes in relation to AF. We studied participants of European or African descent in three cohorts in the National Heart Lung and Blood Institute's Candidate Gene Association Resource, totaling over 20,000 individuals and 2,523 AF cases. SNP rs4845625 in the IL6R gene was associated with AF (relative risk C allele 0.90, 95% confidence interval 0.85–0.95, P=0.0005) in whites and reached borderline statistical significance in African Americans. The results were comparable in follow-up analyses in 906 individuals of European descent in the German AF Network. No association between rs4845625 and incident ischemic stroke was observed in whites in the Cohorts for Heart and Aging Research in Genomics Stroke Consortium with approximately 20,000 white individuals and 1544 incident strokes. Inflammatory activity including circulating interleukin-6 has been related to AF. A correlation between genetic variation at the IL6R locus and interleukin-6 concentrations has been described. If confirmed, our biologically plausible findings render the interleukin-6 system a pathway for further investigation to understand genetic variation and pathophysiology of AF.
In addition we showed that the chromosome 4 locus near PITX2 known from recent genome-wide association studies in whites was associated with AF in African Americans.
Funding sources: We would like to acknowledge the support of the National Heart, Lung, and Blood Institute and the contributions of the all research institutions, study investigators, field staff and study participants in creating the Candidate gene association resource for biomedical research (CARe). The following three parent studies have specifically contributed phenotype data and DNA samples through the Massachusetts Institute of Technology - Broad Institute (N01-HC-65226) to create this genotype/phenotype database that will become publically available:
ARIC (Atherosclerosis Risk in Communities) University of North Carolina at Chapel Hill (N01-HC-55015), Baylor Medical College (N01-HC-55016), University of Mississippi Medical Center (N01-HC-55021), University of Minnesota (N01-HC-55019), Johns Hopkins University (N01-HC-55020), University of Texas, Houston (N01-HC-55022), University of North Carolina, Forsyth County (N01-HC-55018). This study was additionally supported by grants RC1HL099452 (NIH/NHLBI) and 09SDG2280087 (American Heart Association).
CHS (Cardiovascular Health Study) This research was supported by contracts N01-HC-85079 through N01-HC-85086, N01-HC-35129, N01-HC-15103, HHSN268200625226C, and HHSN268200900055C from the National Heart, Lung, and Blood Institute, and by grant R01-HL080295.
FHS (Framingham Heart Study) This research was conducted using data and resources from the Framingham Heart Study of the National Heart Lung and Blood Institute of the National Institutes of Health and Boston University School of Medicine based on analyses by Framingham Heart Study investigators participating in the SNP Health Association Resource (SHARe) project. This work was supported by the National Heart, Lung and Blood Institute's Framingham Heart Study (Contract No. N01-HC-25195) and its contract with Affymetrix, Inc for genotyping services (Contract No.N02-HL-6-4278). A portion of this research utilized the Linux Cluster for Genetic Analysis (LinGA-II) funded by the Robert Dawson Evans Endowment of the Department of Medicine at Boston University School of Medicine and Boston Medical Center. The Framingham Heart Study research was supported by NIH grants 1R01HL092577 (PTE, EJB); RC1HL101056 (EJB, AA); AG028321 (EJB); T32 HL007575 (SAL); DA027021, HL104156 (PTE); The Deutsche Forschungsgemeinschaft (German Research Foundation) Research Fellowship SCHN 1149/1-1 and Emmy Noether Program SCHN 1149/3-1supported RBS' FHS research. JWM was supported by American Heart Association award #09FTF2190028.
AFNET/KORA S4 The study was also supported by the German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung) in the context of the German National Genome Research Network (NGFN), the German National Competence Network on Atrial Fibrillation (AFNET), the Leducq Foundation (07-CVD 03) by grants to Dr Kääb (01GS0499, 01GI0204, and 01GS0838). Dr. Sinner is supported by the German Heart Foundation. KORA S4: The KORA platform is funded by the Helmholtz Zentrum München, the BMBF partly in the context of the German National Genome Research Network, and the State of Bavaria, and as part of LMUinnovativ.
CHARGE Stroke Consortium Supported by grants or contracts from the National Heart, Lung, and Blood Institute (N01-HC-55015, N01-HC-55016, N01-HC-55018, N01-HC-55019, N01-HC-55020, N01-HC-55021, N01-HC-55022 R01HL087641, R01HL59367, and R01HL086694, R01HL093029), the National Human Genome Research Institute (U01HG004402), and NIH (HHSN268200625226C) and the NIH Roadmap for Medical Research (UL1RR025005) to the Atherosclerosis Risk in Communities study; contracts and grants from the National Heart, Lung, and Blood Institute (N01-HC-85079 through N01-HC-85086, N01-HC-35129, N01-HC-15103, N01-HC-55222, N01-HC-75150, N01-HC-45133, U01HL080295, and R01HL087652), the National Institute on Aging (R01AG027002), the National Center for Research Resources (M01RR00069, that partially supported the genotyping), and the National Institute of Diabetes and Digestive and Kidney Diseases (DK063491) to the Cardiovascular Health Study; grants from the National Heart, Lung, and Blood Institute (N01-HC-25195) and its contract with Affymetrix for genotyping services (N02-HL-6-4278), the Robert Dawson Evans Endowment of the Department of Medicine at Boston University School of Medicine and Boston Medical Center for the use of the Linux Cluster for Genetic Analysis; and by grants from the NINDS (NS17950) and the National Institute of Aging (AG08122, AG031287 and AG033193 to the Framingham Heart Study), the Netherlands Organization of Scientific Research (175.010.2005.011), the Netherlands Genomics Initiative (NGI)/Netherlands Organization for Scientific Research (NWO) Netherlands Consortium for Healthy Ageing (050-060-810), the Erasmus Medical Center and Erasmus University, Rotterdam, the Netherlands Organization for Health Research and Development, the Research Institute for Diseases in the Elderly, the Ministry of Education, Culture, and Science, the Ministry for Health, Welfare, and Sports, the European Commission, and the Municipality of Rotterdam to the Rotterdam Study. We thank the staff and participants of the above mentioned studies for their important contributions.
Conflict of Interest Disclosures: None
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.