Skip main navigation

Common Variants in Myocardial Ion Channel Genes Modify the QT Interval in the General Population

Results From the KORA Study
Originally publishedhttps://doi.org/10.1161/01.RES.0000161077.53751.e6Circulation Research. 2005;96:693–701

Abstract

Altered myocardial repolarization is one of the important substrates of ventricular tachycardia and fibrillation. The influence of rare gene variants on repolarization is evident in familial long QT syndrome. To investigate the influence of common gene variants on the QT interval we performed a linkage disequilibrium based SNP association study of four candidate genes. Using a two-step design we analyzed 174 SNPs from the KCNQ1, KCNH2, KCNE1, and KCNE2 genes in 689 individuals from the population-based KORA study and 14 SNPs with results suggestive of association in a confirmatory sample of 3277 individuals from the same survey. We detected association to a gene variant in intron 1 of the KCNQ1 gene (rs757092, +1.7 ms/allele, P=0.0002) and observed weaker association to a variant upstream of the KCNE1 gene (rs727957, +1.2 ms/allele, P=0.0051). In addition we detected association to two SNPs in the KCNH2 gene, the previously described K897T variant (rs1805123, −1.9 ms/allele, P=0.0006) and a gene variant that tags a different haplotype in the same block (rs3815459, +1.7 ms/allele, P=0.0004). The analysis of additive effects by an allelic score explained a 10.5 ms difference in corrected QT interval length between extreme score groups and 0.95% of trait variance (P<0.00005). These results confirm previous heritability studies indicating that repolarization is a complex trait with a significant heritable component and demonstrate that high-resolution SNP-mapping in large population samples can detect and fine map quantitative trait loci even if locus specific heritabilities are small.

Pathological alteration of myocardial ventricular repolarization is a leading cause of ventricular tachycardia and fibrillation.1 It is also suspected to contribute to sudden cardiac death in the context of myocardial hypertrophy or heart failure as well as in drug-induced arrhythmias.2

The cardiac repolarization process is known to be strongly dependent on various parameters, among them heart rate,3 age,4 sex,5,6 plasma levels of electrolytes,7 and medications,8 as well as inherited and acquired pathological conditions.9 The QT interval measured in the surface ECG is the most accessible noninvasive marker of repolarization. After correction for heart rate, its strongest covariate, it is usually referred to as the corrected QT or QTc interval.

Apart from monogenic long QT syndrome (LQT), heritability studies have suggested that genetic factors are also involved in the control of cardiac repolarization at the population level. The heritability of the QTc interval has been estimated between 25% and 52% in three sibpair and in one family-based study.10–13 In a nonparametric linkage analysis, the authors of one of the above studies could demonstrate a significant linkage of the QTc interval to the KCNQ1 (LQT1) and the ANK2 (LQT4) gene loci.12

Several authors have investigated nonsynonymous SNPs in candidate genes for their effect on repolarization. The K897T variant in exon 11 of the KCNH2 gene encoding the α-subunit of the voltage-gated myocardial IKr channel (LQT2) was examined in a study of 226 males and 187 females of Finnish descent. Only in females the 897T-allele had a prolonging influence on the maximum QTc interval measured over all 12 leads, but not in lead V2.14 In 39 LQT patients with the KCNQ1-G589D mutation, the QT interval during exercise was prolonged in those with at least one KCNH2–897T-allele.15 The authors of this and another functional study16 noted that the IKr-897T channel exhibited a decreased current density.

In a study of 1316 Europeans, the 897T-allele shortened the QTc interval at rest in both males and females.17 The effect appeared to be recessive with a shortening of QTc by −10.0 ms in 897TT-homozygotes and was stronger in females than in males. The IKr-897T channel showed a decrease in steady state activation potential predicting a shortening of action potential duration due to an increase in IKr current.

Myocardial repolarization is a fine-tuned process dependent on the delicate coordination of low strength ionic currents at the end of the action potential.18 Gene variants conferring only subtle differences to gene regulation or function, such as intronic or promoter variants, may well influence the repolarization process similar to nonsynonymous variants. We tested the hypothesis that frequent gene variants in the long QT syndrome potassium channel genes KCNQ1, KCNH2, KCNE1, and KCNE2 cause phenotypic variation of myocardial repolarization in the general population and conducted a systematic and high-density linkage disequilibrium (LD)–based SNP association study with a resolution similar to the current HapMap effort19 in search of novel quantitative trait loci (QTL) of the QT interval.

Materials and Methods

Individuals

Between 1999 and 2001, we conducted an epidemiological survey of the general population living in or near the city of Augsburg, Southern Germany (KORA S4). This was the fourth in a series of population-based surveys originating from our participation in the WHO MONICA project. The study population consisted of residents of German nationality born between July 1, 1925 and June 30, 1975 identified through the registration office. A sample of 6640 subjects was drawn with 10 strata of equal size according to gender and age. After a pilot study of 100 individuals, 4261 individuals (66.8%) agreed to participate in the survey, which were ethnic Germans with very few exceptions (>99.5%). During 2002 and 2003, we reinvestigated a subsurvey of 880 persons specifically for cardiovascular diseases. From that subsurvey, 689 individuals were studied to screen for positive genetic associations (screening sample), whereas 3277 different individuals from the total survey were used to confirm positive findings (confirmation sample). A detailed description of samples and the list of exclusion criteria are given in Table 1. Blood samples were drawn after informed consent had been obtained. All studies involving humans were performed according to the declarations of Helsinki and Somerset West and were approved by the local medical ethics committee.

Table 1. Table 1. Population Characteristics

Screening SampleConfirmation SampleP
In the screening sample, the age groups above 55 years were overrepresented, reflecting the age distribution of typical cardiovascular patient populations in an attempt to make the sample also suitable for matched case control designs. Differences in age, RR, and QTc_RAS are due to this overrepresentation.
Survey population, No.8803358
Exclusion criteria
    Atrial fibrillation8300.97
    Pacer/ICD implant8360.67
    Pregnancy2150.36
Randomly excluded to meet requirements of 384-well genotyping1730
Study sample, No.6893277
Male, %342 (49.6%)1617 (49.3%)0.8887
Age, y57.7±12.347.4±13.4<0.00005
Age range, y25–7425–74
RR, ms925.8±147.2942.6±151.10.0076
QT, ms409.3±28.4407.6±28.00.1513
QT range, ms330.0–538.0322.0–550.0
QTc_RAS, ms418.9±18.5417.3±16.90.0278
QTc_RAS range, ms322.8–525.6345.8–541.8

ECG Phenotyping

In the initial survey, we recorded 12-lead resting electrocardiograms (ECGs) using a digital recording system (Bioset 9000, Hörmann Medizinelektronik). QT intervals were determined using the Hannover ECG analysis software (HES-Version 3.22-12) by computerized analysis of an averaged cycle computed from all cycles of the 10-second recording after exclusion of ectopic beats. The QT interval determined by this algorithm represents the earliest begin of depolarization until the latest deflection of repolarization between any two leads. In an international validation study, the HES software was among the best performing digital ECG systems.20 Reproducibility of HES QT-measurements over short- and long-term time intervals has been investigated.21

Covariate Analysis and Phenotype Correction

We adjusted QT for known covariates by a correction formula. Traditional formulas like Bazett’s3 correct only for heart rate in a nonlinear fashion. A linear correction formula for QT has been derived from Framingham Heart Study data.22 We based correction of QT on a multivariate linear regression model including covariates heart rate (RR interval), sex, and age. Correction factors were determined separately for each sex; the QT interval corrected for rate-, age-, and sex was called QTc_RAS. With the correction factors derived from the total sample of 3966 individuals, the formulas for QTc_RAS were determined for males:

and for females:

where RR denotes RR interval in milliseconds.

Genotyping, Determination of Haplotype Blocks, and Haplotypes

We investigated genes encoding the α- and β-subunits of the myocardial delayed rectifier potassium channels IKs (IKs-α: KCNQ1; IKs-β: KCNE1) and IKr (IKr-α: KCNH2; IKr–β: KCNE2). A total of 270 SNPs distributed in and around these genes were chosen from the public dbSNP database, databases on monogenic long QT-syndrome genes,23–25 and diagnostic LQT-patient resequencing. SNPs in exons or intron/exon boundaries were chosen without exception (19 SNPs). Outside those regions SNPs were selected on the criterion of equidistant spacing of ≈1 SNP every 5 kb (251 SNPs). Information about local patterns of LD from HapMap or other sources was not available at the time of SNP selection.

DNA was extracted from EDTA anticoagulated blood using a salting out procedure.26 SNP genotypes were determined using PCR, primer extension, and MALDI-TOF mass spectrometry in a 384-well format (Sequenom). LD measures (D′, r2) and haplotypes were determined with Haploview software.27 Haplotype block boundaries were defined based on the confidence interval of the D′ measure as described in Gabriel et al.28 Haplotype-phenotype association analysis based on sliding window haplotypes was performed using the haplotype trend regression test as described in.29

Of the 270 SNP assays, 33 were not functional, with call rates below 0.8, and 36 were monomorphic. And 174 SNPs had call rates ≥0.8, minor allele frequencies ≥0.02, and Hardy-Weinberg-equilibrium (HWE) P values ≥0.01. The low cut-off value for HWE was accepted because of the relatively large number of SNPs genotyped in the project.

Genotype Phenotype Association Analysis

SNPs were tested for association by linear regression analysis using QTc_RAS as the dependent variable. Significance levels were determined for both the one-degree (1df) and the two-degree of freedom (2df) test. In the 1df test, the independent variable was derived by transforming SNP’s genotypes (AA, Aa, aa) to a relational scale by counting the number of minor alleles (0, 1, 2) assuming a strictly codominant model with identical trait increases between genotypes. This test has a relatively higher power to detect weak effects and was our primary test used during screening and confirmation. In the 2df test, a SNP was decomposed into two variables representing the two genotypic changes and both were included into a bivariate regression. This test accounts for dominance and recessivity by allowing the trait increase of each genotypic change to take an individual value. It was used to specifically quantify each genotype’s effect and significance level in the total sample. The average trait increase per allele was calculated as the mean of both genotypic changes weighted by the genotype frequencies and the variance attributable to a SNP was calculated as the adjusted r2 value from the bivariate regression analysis.

To determine the independence of effects, we performed multivariate linear regression analysis, incorporating the genotypic changes of several SNPs into one model. To determine combined effects, we counted the number of significant genotypic changes in each person to give a QT-prolongation score and performed ANOVA analysis using the score as the independent variable. To investigate if associated SNP-markers had also been identified in a categorical trait analysis, we analyzed groups of individuals with extreme QTc_RAS values in both and individual sexes in a case control–like design using the Cochran-Armitage test for trend.

Association Study Design and Adjustment for Multiple Testing

We designed a two-step association procedure using a small screening and a larger confirmation sample in an attempt to minimize the false-positive error rate. We genotyped the screening sample for all designed SNP assays. Without adjusting for multiple testing, we genotyped all SNPs significantly associated with QTc_RAS (P<0.05 in the 1df test) and nonredundant to each other (pairwise r2<0.6) in the confirmation sample. To adjust for multiple testing in this step, we calculated an adjusted table-wide significance level using 1000 rounds of permutation. As the question if adjustment for multiple testing is necessary in two-step designs is not resolved, we used both the unadjusted and the adjusted significance levels in the discussion of confirmation results. In haplotype blocks of confirmed SNPs, we investigated additional nonredundant markers even if in the screening sample they had not been significantly associated with QTc_RAS.

To determine gender-specific differences of SNP-phenotype associations, we performed sex-specific regression analysis in the total sample. Sample sizes of males (n=1959) and females (n=2007) were similar and therefore comparable for effect strength. To investigate if SNPs with confirmed association to QTc_RAS had also been identified by a categorical trait analysis, we analyzed groups of individuals with extreme QTc_RAS values in both and individual sexes against each other using the Cochran-Armitage test for trend.

Results

Association Analysis of Individual SNPs

In the total sample of 3966 individuals, QTc_RAS corrected to a 60-year-old male with a heart rate of 60 bpm had a mean value of 417.6 ms and a SD of ±17.2 ms.

Of the 174 successfully genotyped SNPs, the average call rate was 0.953 and the average minor allele frequencies were 0.258 (mean) and 0.251 (median). Haplotype blocks are shown in Figure 1 and described in Table 2. In the screening sample, 34 of these SNPs showed association to QTc_RAS in the 1df-test, 18 of these being also significant in the 2df test. We genotyped 13 nonredundant SNPs in the confirmation sample (Table 3a; supplemental Table I, available online at http://circres.ahajournals.org) plus one additional SNP that tags another frequent haplotype in the block of an associated marker. Association was confirmed for four SNPs if the unadjusted significance level of 0.05 was used and for three SNPs if the adjusted table-wide significance level of 0.0041 was applied.

Figure 1. Genomic structure, LD-structure, and genotyped SNPs in the investigated gene regions. 174 SNPs genotyped in the screening sample are denoted as (|), 13 SNPs genotyped in the confirmation sample are marked by •. LD-structure in the regions is marked in form of D′-based haplotype block boundaries28 (▵) and in form of neighboring SNPs exceeding r2-values of 0.5 (▴). For SNPs genotyped in both samples, P values for association with QTc_RAS in the screening (top) and in the confirmation sample (bottom) and for associated SNPs the effect of one minor allele on QTc_RAS in the entire sample are given.

Table 2. Investigated Gene Regions

GeneKCNQ1KCNH2KCNE1KCNE2
KCNE1 and KCNE2 genes are adjacent on Chr. 21 and treated as one region. Start of each gene is given as the position of the first known start of mRNA transcription. All positions are given from human genome assembly hg16. *In the KCNH2 gene, one of five SNPs was significantly associated to QTc_RAS in the two-step design. One additional associated SNP was identified by tagging the third haplotype in the associated block. **This SNP upstream of the KCNE1 gene only showed significant association in the confirmation step when no adjustment for multiple testing was performed.
Genomic region11p15.5-p15.47q3621q22.11-q22.12
Genotyped region
    Start2 406 312149 970 66634 630 685
    End2 856 274150 090 99134 803 884
Length of genotyped region450 kb120 kb173 kb
Length of gene404 kb33 kb13 kb7 kb
Exons in gene model161532
SNP assays setup1319049
Succesfully genotyped SNPs for association in the screening sample815934
Genomic density of succesfully genotyped SNP assays1 SNP/5.6 kbSNP/2.0 kb1 SNP/5.1 kb
Average call rate of successfully genotyped SNPs95.4%94.4%96.6%
SNPs significantly associated with QTc_RAS in the screening sample11167
SNPs genotyped in the confirmation sample55 (6)*3
SNPs significantly associated with QTc_RAS in the confirmation sample11 (2)*1**
No. of haplotype blocks in the entire genotyped region1279
No. of haplotype blocks between Start and Stop codon of the gene11431
No. of SNP-markers not in LD blocks in the entire genotyped region4546

Table 3A. TABLE 3A. Association Results

Gene RegionSNP-MarkerAlleles (A/a)Minor Allele Effect on QTScreening Sample (n=689)Confirmation Sample (n=3277)
(AA)(Aa)(aa)(AA)(Aa)(aa)
nQTc_RASnQTc_RASnQTc_RASP (1df)nQTc_RASnQTc_RASnQTc_RASP (1df)
Results of 13 SNPs in both samples in the two-step approach. Significance levels were determined by the 1df test as described in the Materials and Methods section. All SNPs are located in introns or in flanking genomic regions of the genes except for SNP rs1805123 (KCNH2-K897T). For significance levels of the 2df test, see supplemental Table I.
KCNQ1rs2301700G/A530418.0110420.98428.20.04482590417.5600416.935415.00.2780
KCNQ1rs739677A/G368417.5244421.047422.00.01601707417.81230416.7213417.60.2570
KCNQ1rs757092A/G284417.5305418.8100423.30.01291216416.11541417.9445418.90.0007
KCNQ1rs463924C/T310417.6294419.768422.20.04051557417.41347417.3292417.40.8865
KCNQ1rs2519184G/A535419.9114414.66408.70.00182220417.2460417.129423.20.4870
KCNH2rs885684T/G303420.7302417.982415.60.01081261417.51515417.6421415.70.1665
KCNH2rs956642A/G258416.9322419.7106421.00.02731111417.61413417.2507417.80.9998
KCNH2rs1805123A/C389420.7245416.431419.20.02231665418.01032416.7187414.30.0017
KCNH2rs3800779G/T272417.4317419.182422.00.04601109416.91193417.7327416.70.7194
KCNH2rs1799983G/T320420.7290418.350415.90.03481467417.31330417.4355417.40.8917
KCNE1rs2834456A/G207416.3354419.7122421.60.0079966417.21554417.0611418.00.4451
KCNE1rs2834488T/C598419.576414.02414.90.01712783417.3329417.615422.00.4825
KCNE1rs727957G/T447417.7199420.226427.20.00812046416.9948417.6106420.10.0498

We detected a previously undescribed QTL in intron 1 of the KCNQ1 gene. Although the gene shows remarkably little LD, intron 1 contains a large haplotype block of ≈50-kb size and high LD (both D′ ≥0.94 and r2≥0.79 for 6 of 7 markers) (Figure 1). It contains two major haplotypes with frequencies of 0.570 and 0.379 that can be tagged by rs757092. This SNP showed association in both subsamples (Table 3a), the rare G-allele being associated to a QTc_RAS prolongation of +1.7 ms in heterozygotes and +3.3 ms in homozygotes (Table 3b; 0.38% of variance; P=0.0002).

Table 3B. TABLE 3B. Association Results

Gene RegionSNP-MarkerAlleles (A/a)GenderMinor Allele Effect on QTTotal Sample (n=3966)
(AA)(Aa)(aa)Explained VarianceAverage Δ QTc_RAS Per AlleleP (1df)P (2df)
nQTc_RASnΔ QTc_RASP (Aa)nΔ QTc_RASp(aa)
Effect of the associated SNPs in the total sample and stratified for gender as determined by linear regression over genotypes. Significance levels are given separately for each genotypic change and for the entire effect of the SNP with both the one and two degrees of freedom tests.
KCNQ1rs757092A/GBoth1500416.41846+1.7 ms0.005545+3.3 ms0.0010.38%+1.7 ms<0.000050.0002
KCNQ1rs757092A/GMale747416.8899+1.3 ms0.143275+3.4 ms0.0090.26%+1.5 ms0.00870.0296
KCNQ1rs757092A/GFemale753416.0947+2.0 ms0.011270+3.3 ms0.0040.44%+1.9 ms0.00120.0046
KCNH2rs1805123A/CBoth2054418.51277−1.9 ms0.002218−3.5 ms0.0040.36%−1.9 ms0.00010.0006
KCNH2rs1805123A/CMale1027418.6601−1.2 ms0.181104−3.8 ms0.0430.18%−1.4 ms0.02880.0791
KCNH2rs1805123A/CFemale1027418.5676−2.5 ms0.002114−3.2 ms0.0460.52%−2.3 ms0.00100.0031
KCNH2rs3815459G/ABoth2469416.91251+1.5 ms0.011178+4.5 ms0.0010.35%+1.7 ms0.00010.0004
KCNH2rs3815459G/AMale1226417.0612+2.1 ms0.02292+4.3 ms0.0280.35%+2.1 ms0.00310.0127
KCNH2rs3815459G/AFemale1243416.8639+1.0 ms0.19886+4.7 ms0.0100.28%+1.3 ms0.01340.0230
KCNE1rs727957G/TBoth2493417.01147+1.0 ms0.085132+4.5 ms0.0030.23%+1.2 ms0.00300.0051
KCNE1rs727957G/TMale1239417.0555+1.5 ms0.10066+6.4 ms0.0050.41%+1.5 ms0.00440.0080
KCNE1rs727957G/TFemale1254417.1592+0.6 ms0.44466+2.6 ms0.202<0.01%+0.7 ms0.19460.3682

In the KCNH2 gene, we confirmed the previously published effect of SNP KCNH2-K897T (rs1805123) on the QT interval. The rare 897T-allele was associated with a shortening of QTc_RAS of −1.9 ms in heterozygotes and −3.5 ms in homozygotes (0.36% of variance; P=0.0006). The effect was stronger in females. The K897T variant resides on a large haplotype block extending over 60 kb from exon 3 to 30 kb 3′ of the gene (KCNH2-block 2 in Figure 1), in which four haplotypes with allele frequencies above 0.05 exist (Table 4) among which KCNH2-K897T tags haplotype h2 (Hf=0.205). Typing the confirmation sample with a SNP tagging haplotype h3 (Hf=0.195) revealed a second effect. The rare A-allele of SNP rs3815459 was associated with a prolongation of QTc_RAS of +1.5 ms in heterozygotes and +4.5 ms in homozygotes (0.35% of variance; P=0.0004).

Table 4. Table 4. Haplotypes in LD-Block 2 of theKCNH2 Gene

rs2968864rs2968863rs2907948rs2968853rs1547958rs3815459rs1805123 (K897T)rs1137617 (Y652Y)rs1805121 (L564L)rs1805120 (F513F)rs740952 (I489I)rs3807376rs3778874rs4725386haplotype frequency
Four major haplotypes with frequencies above 0.05 exist in that block of 60 kb size. Haplotype h2 is tagged by SNPs KCNH2-K897T, rs2968864, rs2968863, rs2907948, rs1547958, and rs4725386. Haplotype h3 is tagged by SNPs rs3815459, rs1805120 (KCNH2-F513F), rs740952 (KCNH2-I489I), rs3807376, and rs3778874.
haplotype h1AGCAGGATACCACG0.344
haplotype h2GATGAGCCACCACA0.205
haplotype h3AGCGGAACGTTGTG0.195
haplotype h4AGCGGGACGCCACG0.096

In the KCNE1 gene region, SNP rs727957 showed a positive association to QTc_RAS in both the screening (P=0.0081) and the confirmation sample (P=0.0498), but did not exceed the adjusted significance level. In the total sample, the rare T-allele of the marker was associated with a prolongation of QTc_RAS of +1.0 ms in heterozygotes and +4.5 ms in homozygotes (0.23% of variance P=0.0051).

Among the 174 investigated common SNPs were two further nonsynonymous gene variants, KCNH2-R1047L [Af(min)=0.024], for which functional data indicate no allele differences,16,17 and KCNE1-S38G [Af(min)=0.355], neither of which were associated to QTc_RAS. To clarify the importance of rare nonsynonymous SNPs, we additionally genotyped KCNE2-T8A [Af(min)=0.0073], which had previously been described associated to drug-induced long QT syndrome30 and KCNQ1-G643S [Af(min)=0.00072], for which only one heterozygote was observed. Also, these showed no significant effect (supplemental Table III).

Combined and Categorical Association Analysis

The comparison of multivariate linear regression analysis of both QTc_RAS and QT including covariates demonstrated highly similar significance levels from both methods and independence of SNPs’ effects enabling combined association analysis (Table 5; supplemental Table II). Analysis of the intragenic KCNH2 variants K897T and rs3815459 against all other haplotypes of that block revealed an increase of QTc_RAS mean values among the six genotype groups from 415.0 to 421.1 ms (0.52% of variance; P=0.0002; Table 6a; Figure 2b). The combined intergenic analysis of the KCNH2-K897T and the KCNQ1-rs757092 variants showed an increase of QTc_RAS mean values among the nine genotype groups from 412.9 to 421.2 ms (0.74% of variance; P<0.00005; Table 6b; Figure 2c).

Table 5. Table 5. Multivariate Linear Regression Models of QTc_RAS

Total sample (n=3327)Males (n=1620)Females (n=1707)
Δ QTc_RAS (Aa)Δ QTc_RAS (aa)P (Aa)P (aa)Δ QTc_RAS (Aa)Δ QTc_RAS (aa)P (Aa)P (aa)Δ QTc_RAS (Aa)Δ QTc_RAS (aa)P (Aa)P (aa)
In the multivariate linear regression analysis of QTc_RAS in the total sample, P values represent the additive significance contributed by each genotypic change. Six out of eight genotypic changes were significantly associated; five of them were from SNPs confirmed by our two-step design. None of the other SNPs genotyped in the total sample contributed significantly.
KCNQ1-rs757092+1.9 ms+3.0 ms0.0020.001+1.1 ms+2.7 ms0.2610.053+2.7 ms+3.4 ms0.0010.006
KCNH2-rs1805123−1.8 ms−2.9 ms0.0050.023−1.3 ms−2.6 ms0.1900.175−2.4 ms−3.3 ms0.0050.055
KCNH2-rs3815459+1.4 ms+2.9 ms0.0350.054+2.1 ms+2.9 ms0.0380.189+0.7 ms+2.9 ms0.3970.159
KCNE1-rs727957+1.2 ms+4.2 ms0.0680.011+1.6 ms+6.3 ms0.0950.011+0.7 ms+2.0 ms0.4060.359

Table 6A. Table 6A. Analysis of Combined Effects of SNPs

Combined intragenic effect of the KCNH2-K897T (h2) and rs3815459 (h3) (D′>0.99, r2=0.08) in haplotype block 2 of the KCNH2 gene (Figure 1). hn denotes all haplotypes other than h2 and h3 (Table 4). For QTc_RAS mean, standard deviation and the no. of genotype carriers are given.
KCNH2-rs3815459
KCNH2-K897T (rs1805123)GGGAAA
AA (897KK)417.7±17.4419.2±16.8421.1±19.0
1072 (hn/hn)797 (h3/hn)158 (h3/h3)
AC (897KT)416.4±16.9417.0±18.5
937 (h2/hn)322 (h3/h2)
CC (897TT)415.0±16.7
212 (h2/h2)

Table 6B. Table 6B. Analysis of Combined Effects of SNPs

KCNQ1-rs757092
KCNH2-K897T (rs1805123)AAAGGG
Combined intergenic effect of two SNPs KCNH2-K897T and rs757092 (KCNQ1-intron 1) on QTc_RAS. SNPs are not in LD (D′<0.01, r2<0.01). Rare allele KCNH2–897T shortens QTc_RAS, whereas the rare allele of rs757092 (G) prolongs QTc_RAS.
AA (897KK)417.3±16.9418.9±17.0421.2±18.7
793944297
AC (897KT)415.4±17.0417.6±17.2416.5±18.5
489612161
CC (897TT)412.9±13.1416.3±17.6417.1±21.5
8510427

Figure 2. Normal distribution of QT and QTc_RAS. a, Distribution of QT (dashed line) and QTc_RAS (solid line, with bars) in the total sample (n=3966). QTc values longer than 450 ms are typically suggestive of long QT syndrome. Mean of QTc_RAS is shifted to higher values compared with QT due to correction to a 60-year-old male. b, Combined effect of the two SNPs KCNH2-K897T and rs3815459 on QTc_RAS in the total sample (see Table 6a). Individuals with 897TT/GG genotype (n=212, dashed) have the shortest, and individuals with 897KK/AA genotype (n=158, dotted) have the longest QTc_RAS. c, Combined effect on QTc_RAS of the two SNPs KCNH2-K897T and rs757092 (KCNQ1-intron 1) (see Table 6b). Individuals with 897TT/AA genotype (n=85, dashed) have the shortest, and individuals with 897KK/GG genotype (n=297, dotted) have the longest QTc_RAS.

Five of six genotypic changes of the three confirmed SNPs KCNQ1-rs757092, KCNH2-rs1805123, and rs3815459 were independently significant (P<0.05; Table 5). Individuals harboring the maximum possible number of five QT-prolonging alleles had on average a 10.5 ms longer QTc_RAS than individuals that had no QT-prolonging allele (0.95% of variance; P<0.00005) (Table 6c). When the genotypic change KCNE1-rs727957(aa) was included, a +14.3 ms increase was observed (1.13% of variance; P<0.00005)

Table 6C. Table 6C. Analysis of Combined Effects of SNPs

QT-Prolongation ScoreQTc_RAS±SDFrom Total Sample (n=3966), n
Effect of the five genotypic changes significant in the multivariate regression analysis from the three confirmed SNPs KCNQ1-rs757092 (Aa, aa), KCNH2-K897T (Aa, aa), and KCNH2-rs3815459 (Aa) was determined by a QT-prolongation score (P<0.00005). For each score-class the average QTc_RAS, standard deviation and the no. of individuals are given.
0412.7±13.479
1415.5±16.9462
2416.6±16.91021
3418.3±17.81132
4419.3±16.9641
5423.2±19.4135

Categorical analysis of individuals with extreme QTc_RAS values for all SNPs genotyped in the total sample detected significant effects in 2 of the 4 associated SNPs in 200 individuals from the extremes and in 3 of the 4 associated SNPs in 600 individuals from the extremes (supplemental Table IV). After adjustment for multiple testing, categorical analysis results were only significant for KCNQ1-rs707592 in the analysis of 600 individuals.

Discussion

Covariate Correction of QT

The linear correction factors for heart rate we determined were well in agreement with published ones.22 The comparison between a formula correction and a multivariate linear regression model of QT for detecting SNP association (Table 5; supplemental Table II) supports the view that none of the two methods is superior.

Effects of Individual SNPs

The KCNQ1 gene locus had previously been shown to influence the QT interval in a quantitative trait linkage study.12 We could map this QTL to a 50-kb haplotype block in intron 1 in which only two major haplotypes existed. Of all identified effects, this was the most significant and most robust against testing in both genders. The 50-kb block does not contain any known or predicted exonic or regulatory sequences. Its high LD precludes further fine mapping in our population. The causal variant and its functional nature thus remain elusive at this point. Several association studies have demonstrated the nonsynonymous KCNH2-K897T variant to be significantly associated with repolarization, but results were conflicting. Our data show that in Caucasians of both sexes, the 897T-allele (block 2, h2) shortens the QT interval. The conflicting results may indicate that in other ethnic groups, the LD-relationship of the K897T variant may vary or that the smaller previous association studies were affected by increased type 1 error rates. We have identified another QT-modifying haplotype (h3) in the same block. In Caucasians, the presence of a common nonsynonymous SNP on h3 is unlikely, given the large number of individuals others and we have sequenced to detect mutations in the KCNH2 gene. An effect of the two synonymous SNPs I489I and F513F might be causal as for both amino acids less common codons are present on h3 [I489I: ATC (0.48) > ATT (0.35); F513F: TTC (0.55) >TTT (0.45)].31 In humans, codon usage has been shown to correlate with expression breadth, which covaries with expression levels32 but convincing evidence for codon usage effects in humans and other higher organisms has not been demonstrated.

SNP KCNE1-rs727957 did not fulfill all our significance criteria but showed evidence for association in the combined analysis (P=0.0051). It is located in the 5′ region 50 kb upstream, its haplotype block ending 20 kb upstream of the KCNE1 gene. We argue that an independent replication of this SNP’s effect should be conducted before it is considered a significant QTL.

Sliding window haplotype analysis did not reveal additional associations or improve significance levels (data not shown). For the effect in KCNQ1-intron 1 this result is intuitive, as most of the associated haplotype block’s diversity can be captured by typing only a single SNP. Using the information from the International HapMap Project, currently providing data at an average coverage of 1 SNP per 3.8 kb,19 will aid the capturing of relevant haplotype diversity in future studies.

Combined Effects at Several SNP Loci

We have observed two kinds of combined QTL effects on the QT interval. SNPs in high LD tagging different haplotypes in one block had opposite additive effects on QTc_RAS as seen in the KCNH2 gene. Similarly, SNPs in complete linkage equilibrium also exerted additive effects as seen between the KCNQ1 and KCNH2 genes. The fact that these three gene variants, although together only explaining 0.95% of trait variance, were associated to a monotonous rise in average QTc_RAS of up to 10.5 ms, supports their concerted mode of action irrespective whether they are in LD or not.

Gender Effects

Previous publications of the KCNH2-K897T variant’s association to the QT interval had noted its more pronounced effect in females. We have confirmed and extended this finding, as especially marker KCNE1-rs727957 and to a lesser extent also KCNQ1-rs757092 and KCNH2-rs3815459 showed gender-dependent association. This underscores the importance of considering gender as a potent confounder variable when designing complex trait genotype-phenotype association studies.

Categorical Analysis of the QT Interval

A categorical analysis in the confirmation step using only 200 individuals with extreme QTc_RAS values would only have confirmed KCNQ1-rs757092 (P=0.002, OR=1.92) and KCNH2-K897T (P=0.004, OR=0.44) (supplemental Table IV). Notably, these effects were the most significant ones from the quantitative association analysis. Using a larger sample (n=600) also KCNH2-rs3815459 (P=0.008, OR=1.47) would have been confirmed. Categorical confirmation analysis thus can be considered in similar projects if a focus on the strongest effects at significantly reduced cost is desirable.

Implications for Future Investigations

Although the overall heritability of the QT interval is high, all gene variants identified in this study are only minor quantitative trait loci each explaining less than 1% of trait variance. This finding is in common with the view that important physiological mechanisms are unlikely to tolerate large genetic variance at a single locus. The authors of an early heritability study on electrocardiographic traits already noted that these reflected critical biologic functions, which evolved to an evolutionary optimum and the attainment of this optimum would necessarily tend to eliminate interindividual differences.33

We show that in a large sample of thoroughly phenotyped individuals even minor QTLs can be detected. The population-representative recruiting of individuals from one geographic area with limited recent immigration was helpful to this aim, as complex population genealogies can confound association signals. In two of the known monogenic long QT disease genes KCNQ1 and KCNH2, the common disease or in this case common phenotypes/common variants hypothesis holds true. The confirmation of this hypothesis for cardiac rhythm phenotypes appears a prerequisite to investigate whether common gene variants also influence cardiac patients’ predisposition toward arrhythmias.

Common intronic gene variants may influence repolarization to a similar extent as common nonsynonymous exonic variants. Future fine mapping studies of complex and quantitative trait loci should avoid to focus on exonic effects, but apply SNP coverage based on LD.

Besides studies of monogenic arrhythmogenic diseases and functional studies of recombinant cardiac ion channels, the genome-wide investigation of heritable surface ECG signatures may provide a valuable third route toward the identification of novel genes involved in cardiac electrophysiology that up to now went undetected by other methods.

This manuscript was sent to Harry A. Fozzard, Consulting Editor, for review by expert referees, editorial decision, and final disposition.

This publication contains part of the doctoral thesis of S.J., M.A., and A.S.-W.

Original received November 9, 2004; revision received February 16, 2005; accepted February 21, 2005.

This work was funded by the German Federal Ministry of Education and Research (BMBF) in the context of the German National Genome Research Network (NGFN) and the Bioinformatics for the Functional Analysis of Mammalian Genomes program (BFAM) by grants to Stefan Kääb (01GS0109) and to Thomas Meitinger (01GR0103). The KORA platform is funded by the BMBF and by the State of Bavaria. The KORA group (Cooperative Research in the Region of Augsburg) consists of H.E. Wichmann (speaker), H. Löwel, C. Meisinger, T. Illig, R. Holle, J. John, and their coworkers who are responsible for the design and conduct of the KORA studies. We acknowledge the information on SNPs from diagnostic gene resequencing provided by Hanns Georg Klein (IMGM Martinsried).

Footnotes

Correspondence to Stefan Kääb, MD, LMU-University, Klinikum Grosshadern, Department of Medicine I, Marchioninistr. 15, D-81366 Munich, Germany. E-mail

References

  • 1 Haigney MC, Zareba W, Gentlesk PJ, Goldstein RE, Illovsky M, McNitt S, Andrews ML, Moss AJ, Multicenter Automatic Defibrillator Implantation Trial II investigators. QT interval variability and spontaneous ventricular tachycardia or fibrillation in the Multicenter Automatic Defibrillator Implantation Trial (MADIT) II patients. J Am Coll Cardiol. 2004; 44: 1481–1487.CrossrefMedlineGoogle Scholar
  • 2 Tomaselli GF, Beuckelmann DJ, Calkins HG, Berger RD, Kessler PD, Lawrence JH, Kass D, Feldman AM, Marban E. Sudden cardiac death in heart failure: the role of abnormal repolarization. Circulation. 1994; 90: 2534–2539.CrossrefMedlineGoogle Scholar
  • 3 Bazett HC. An analysis of the time relationship of electrocardiograms. Heart. 1920; 7: 353–370.Google Scholar
  • 4 Reardon M, Malik M. QT interval change with age in an overtly healthy older population. Clin Cardiol. 1996; 19: 949–950.CrossrefMedlineGoogle Scholar
  • 5 Yang H, Elko P, LeCarpentier GL, Baga J, Fromm B, Steinman RT, Lehmann MH. Sex differences in the rate of cardiac repolarization. J Electrocardiol. 1994; 27: 72–73.CrossrefMedlineGoogle Scholar
  • 6 Legato MJ. Gender and the heart: sex-specific differences in normal anatomy and physiology. J Gend Specif Med. 2000; 3: 15–18.MedlineGoogle Scholar
  • 7 Nagasaka M, Yokosuka H, Yamanaka T, Sato T, Nakamura K. QT duration and plasma electrolytes (Ca, Na, and K) in uremic patients. Jpn Heart J. 1972; 13: 187–194.CrossrefMedlineGoogle Scholar
  • 8 Kaab S, Hinterseer M, Nabauer M, Steinbeck G. Sotalol testing unmasks altered repolarization in patients with suspected acquired long-QT-syndrome–a case-control pilot study using i.v. sotalol. Eur Heart J. 2003; 24: 649–657.CrossrefMedlineGoogle Scholar
  • 9 Tomaselli GF, Marban E. Electrophysiological remodeling in hypertrophy and heart failure. Cardiovasc Res. 1999; 42: 270–283.CrossrefMedlineGoogle Scholar
  • 10 Russell MW, Law I, Sholinsky P, Fabsitz RR. Heritability of ECG measurements in adult male twins. J Electrocardiol. 1998; 30: 64–68.CrossrefMedlineGoogle Scholar
  • 11 Carter N, Snieder H, Jeffery S, Saumarez R, Varma C, Antoniades L, Spector TD. QT interval in twins. J Hum Hypertens. 2000; 14: 389–390.CrossrefMedlineGoogle Scholar
  • 12 Busjahn A, Knoblauch H, Faulhaber HD, Boeckel T, Rosenthal M, Uhlmann R, Hoehe M, Schuster H, Luft FC. QT interval is linked to 2 long-QT syndrome loci in normal subjects. Circulation. 1999; 99: 3161–3164.CrossrefMedlineGoogle Scholar
  • 13 Friedlander Y, Lapidos T, Sinnreich R, Kark JD. Genetic and environmental sources of QT interval variability in Israeli families: the kibbutz settlements family study. Clin Genet. 1999; 56: 200–209.CrossrefMedlineGoogle Scholar
  • 14 Pietila E, Fodstad H, Niskasaari E, Laitinen PPJ, Swan H, Savolainen M, Kesaniemi YA, Kontula K, Huikuri HV. Association between HERG K897T polymorphism and QT interval in middle-aged Finnish women. J Am Coll Cardiol. 2002; 40: 511–514.CrossrefMedlineGoogle Scholar
  • 15 Paavonen KJ, Chapman H, Laitinen PJ, Fodstad H, Piippo K, Swan H, Toivonen L, Viitasalo M, Kontula K, Pasternack M. Functional characterization of the common amino acid 897 polymorphism of the cardiac potassium channel KCNH2 (HERG). Cardiovasc Res. 2003; 59: 603–611.CrossrefMedlineGoogle Scholar
  • 16 Anson BD, Ackerman MJ, Tester DJ, Will ML, Delisle BP, Anderson CL, January CT. Molecular and functional characterization of common polymorphisms in HERG (KCNH2) potassium channels. Am J Physiol Heart Circ Physiol. 2004; 286: H2434–H2441.CrossrefMedlineGoogle Scholar
  • 17 Bezzina CR, Verkerk AO, Busjahn A, Jeron A, Erdmann J, Koopmann TT, Bhuiyan ZA, Wilders R, Mannens MM, Tan HL, Luft FC, Schunkert H, Wilde AA. A common polymorphism in KCNH2 (HERG) hastens cardiac repolarization. Cardiovasc Res. 2003; 59: 27–36.CrossrefMedlineGoogle Scholar
  • 18 Näbauer M. Tuning repolarization in the heart: a multitude of potassium channels and regulatory pathways. Circ Res. 2001; 88: 453–455.CrossrefMedlineGoogle Scholar
  • 19 The International HapMap Consortium. The International HapMap Project. Nature. 2003;426:789–796. [Data available at www.hapmap.org, public data release #12.]Google Scholar
  • 20 Willems JL, Abreu-Lima C, Arnaud P, van Bemmel JH, Brohet C, Degani R, Denis B, Gehring J, Graham I, van Herpen G, Machado H, Macfarlane PW, Michaelis J, Moulopoulos SD, Rubel P, Zywietz C. The Diagnostic Performance of Computer Programs for the Interpretation of Electrocardiograms. N Engl J Med. 1991; 325: 1767–1773.CrossrefMedlineGoogle Scholar
  • 21 Perz S, Pfeufer A, Kääb S, Hinterseer M, Holle R, Küfner R, Englmeier K-H, Wichmann H-E, for the KORA Study Group. Does Computerized ECG Analysis Provide Sufficiently Consistent QT Interval Estimates For Genetic Research? In: Jan J, Kozumplik J, Provaznik I, eds. Analysis of Biomedical Signals and Images. Brno, Czech Republic: Vutium Press; 2004: 47–49.Google Scholar
  • 22 Sagie A, Larson MG, Goldberg RJ, Bengtson JR, Levy D. An improved method for adjusting the QT interval for heart rate. Am J Cardiol. 1992; 70: 797–801.CrossrefMedlineGoogle Scholar
  • 23 dbSNP Database. Available at: http://www.ncbi.nlm.nih.gov/SNP/.Google Scholar
  • 24 The Long QT Syndrome Database. Available at: http://www.ssi.dk/graphics/html/lqtsdb/lqtsdb.htm.Google Scholar
  • 25 The Cardiac Arrhythmia Mutation Database. Available at: http://PC4.FSM.it:81/cardmoc/.Google Scholar
  • 26 Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 1988; 16: 1215.CrossrefMedlineGoogle Scholar
  • 27 Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21:263–265. Haploview available at: http://www.broad.mit.edu/personal/jcbarret/haploview/Google Scholar
  • 28 Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, Higgins J, DeFelice M, Lochner A, Faggart M, Liu-Cordero SN, Rotimi C, Adeyemo A, Cooper R, Ward R, Lander ES, Daly MJ, Altshuler D. The structure of haplotype blocks in the human genome. Science. 2002; 296: 2225–2229.CrossrefMedlineGoogle Scholar
  • 29 Zaykin DV, Westfall PH, Young SS, Karnoub MA, Wagner MJ, Ehm MG. Testing association of statistically inferred haplotypes with discrete and continuous traits in samples of unrelated individuals. Hum Hered. 2002; 53: 79–91.CrossrefMedlineGoogle Scholar
  • 30 Sesti F, Abbott GW, Wei J, Murray KT, Saksena S, Schwartz PJ, Priori SG, Roden DM, George AL Jr., Goldstein SA. A common polymorphism associated with antibiotic-induced cardiac arrhythmia. Proc Natl Acad Sci U S A. 2000; 97: 10613–10618.CrossrefMedlineGoogle Scholar
  • 31 Nakamura Y, Gojobori T, Ikemura T. Codon usage tabulated from international DNA sequence databases: status for the year 2000. Nucleic Acids Res. 2000; 28: 292.[Recent data releases available at: http://www.kazusa.or.jp/codon/]CrossrefMedlineGoogle Scholar
  • 32 Duret L, Mouchiroud D. Determinants of substitution rates in mammalian genes: expression pattern affects selection intensity but not mutation rate. Mol Biol Evol. 2000; 17: 68–74.CrossrefMedlineGoogle Scholar
  • 33 Hanson B, Tuna N, Bouchard T, Heston L, Eckert E, Lykken D, Segal N, Rich S. Genetic factors in the electrocardiogram and heart rate of twins reared apart and together. Am J Cardiol. 1989; 63: 606–609.CrossrefMedlineGoogle Scholar

eLetters(0)

eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.