Common Variants in KCNQ1 Confer Increased Risk of Type 2 Diabetes and Contribute to the Diabetic Epidemic in East Asians: A Replication and Meta-Analysis

Authors


  • Dr. Haoran Wang and Dr. Kun Miao contributed equally to this work.

Corresponding Author: Hu Ding, MD, PhD, Departments of Internal Medicine, Tongji Hospital, Tongji Medical College, Huazhong University of Science & Technology, 1095# Jiefang Ave., Wuhan 430030, People's Rep. of China. Tel: 86 27 8366 3280; Fax: 86 27 8366 3280; E-mail: huding@tjh.tjmu.edu.cn

Summary

We aimed to evaluate the effect of four common variants (rs2237892, rs2283228, rs2237895, and rs2237897) in KCNQ1 on susceptibility of type 2 diabetes (T2D) by performing a case-control study as well as a comprehensive meta-analysis. We genotyped these four variants in two sets of Chinese Han population, comprising a total of 2533 type 2 diabetic patients and 2643 nondiabetic controls. We also performed a meta-analysis of our results with published studies in East Asians, meanwhile assessing the population attributable risk (PAR) of these variants. By combining our case-control sets, a total of 45,204 T2D cases and 42,832 controls were included in the meta-analyses. The per-allele ORs ranged from 1.24 to 1.33, and the PARs ranged from 15.8% to 31.8%, with SNP rs2237892 being the most widely studied (16 articles containing a total of 38,338 cases and 35,907 controls), showing strongest association (per-allele OR: 1.33, 95% CI: 1.28–1.39) and indicating the highest PAR (31.8%). This study confirmed the strong association between common variants in KCNQ1 and risk of T2D. Variants in KCNQ1 were among the leading genetic factors contributing to the overall burden of T2D in East Asians.

Introduction

It is likely that the disease we call type 2 diabetes (T2D) is heterogeneous and may result from multiple genetic and environmental factors, and their interaction (O'Rahilly et al., 2005). Identification of genetic factors associating with T2D risk has been a hot topic. Recently, a novel gene, potassium voltage-gated channel, KQT-like subfamily, member 1 (KCNQ1), has been reported as a candidate gene for increasing susceptibility to T2D in East Asians (Unoki et al., 2008; Yasuda et al., 2008; Cho et al., 2012). Four SNPs (rs2237892, rs2283228, rs2237895, and rs2237897), which were located in intron 15, showed the strongest association with T2D in the initial reports (Unoki et al., 2008; Yasuda et al., 2008), and were most widely investigated in subsequent studies (Hu et al., 2009; Liu et al., 2009; Qi et al., 2009; Chen et al., 2010). However, inconsistent results were observed regarding both size and direction of the pathogenic effect of KCNQ1 polymorphisms on T2D. For example, Chen et al. could not replicate any association of these four SNPs with T2D; and Tan et al. failed to detect a significant association between SNP rs2237892 and T2D in Chinese living in Singapore (Chen et al., 2010; Tan et al., 2010). Therefore, the credibility of this genetic association should be enhanced by further replications performed by totally independent teams of investigators. Furthermore, the contribution of these variants in KCNQ1 to the epidemic of T2D in East Asians was to be evaluated.

In this report, we first examined the association of four common single nucleotide polymorphisms (SNPs) in KCNQ1 (rs2237892, rs2283228, rs2237895, rs2237897) with susceptibility to T2D in two case-control sets of Chinese Han population mainly coming from Hubei Province. Afterward, we systematically reviewed the association of these four SNPs with risk of T2D in East Asians through a meta-analysis by combining our data with those from previous studies. Finally, we presented the public health relevance by using the population attributable risk (PAR) assessment. We expected to provide a more stable, accurate, and systematic evaluation for the association of these common variants in KCNQ1 with T2D in East Asians and their contribution to the epidemic of this disease.

Materials and Methods

Ethics Statement

Written informed consent was obtained from each participant. The protocol was in accordance with the Helsinki Declaration and was approved by the Institutional Review Committee of Tongji Hospital.

Participants

In total, two independent sets comprising 2533 T2D patients and 2643 nondiabetic healthy subjects were included in our case-control study. Detailed information of the study population has been described in our previous study (Liu et al., 2012). Briefly, T2D cases were confirmed by OGTT or FPG results according to the American Diabetes Association criteria (American Diabetes Association, 2008) or by reports of the use of antihyperglycemia medication or by reviews of medical records. Controls were recruited from geographically matched local communities from Central China (Wuhan, Hubei) by excluding those with a current diagnosis or with a family history of diabetes. In addition, all the cases and controls were unrelated individuals at enrollment.

Clinical Measurements

Height, weight, waist circumference (WC), and hip circumference (HC) were measured in duplicate by trained personnel. Body mass index (BMI) was calculated as weight (kilogram) divided by the square of height (meter2). Waist to hip ratio was calculated as WC divided by HC. Biochemical measurements including levels of total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and triglyceride were performed using standard laboratory assays.

Single Nucleotide Polymorphism Selection and Genotyping

Four SNPs (rs2283228, rs2237892, rs2237895, and rs2237897) were included in our study, and had been explored extensively in previous studies (Yasuda et al., 2008; Hu et al., 2009; Odgerel et al., 2012). Genomic DNA was extracted from peripheral leukocytes isolated from venous blood collected in K2-EDTA tubes, using a DB-S kit (FUJIFILM Corporation, Tokyo, Japan) according to the manufacturer's instructions. Genotyping for the four SNPs was performed using TaqMan allelic discrimination assays commercially provided by Applied Biosystems (Applied Biosystems, Foster City, CA, USA) and Shanghai GeneCore BioTechnologies Co., Ltd. (Shanghai, China) according to the manufacturer's protocol. In order to control genotyping quality, the allelic discrimination plots were visually observed. All genotyping success rates were greater than 98%. In addition, approximately 10% samples including those unsuccessfully genotyped samples were repeated by direct sequencing. All mismatch rates were below 1% in the repeated samples. The primers and probes for genotyping and directly sequencing are listed in Table S1.

Statistical Analysis for the Case-Control Study

The Hardy-Weinberg equilibrium test was performed in the controls before association analysis. The risk of developing T2D was expressed as odds ratio (OR) using logistic regression analysis adjusted for sex, age, and BMI. An additive genetic model was used for the analysis. We also used the conservative Bonferroni correction to control the potential false positive findings because of multiple testing. All reported probabilities (p values) were two-sided, with p < 0.05 considered statistically significant. The statistical analyses were performed with SPSS 13.0 (SPSS Inc., Chicago, USA). Power calculations were carried out using the Quanto software package (Version 1.2.3, http://hydra.usc.edu/gxe/). The power calculations were based on: disease outcome, matched case-control study design, gene only as the hypothesis, an additive inherent model and disease prevalence of 9.7% (Yang et al., 2010). After combining the two sets, considering a minor allele frequency of 30–36% in previous Chinese studies (Hu et al., 2009; Liu et al., 2009), and using an additive model, we had enough power (>95%) to detect a minimal OR of 1.20 at the significance level of α = 0.0125 (adjusted for multiple testing). Pairwise linkage disequilibrium coefficients between SNPs were calculated in all the controls by Haploview software version 4.0 (Daly Lab at the Broad Institute, Cambridge, MA, USA).

Search Strategy

We searched MEDLINE and EMBASE through January 2013 for all publications focused on the association between KCNQ1 polymorphisms and T2D in East Asians (Chinese, Japanese, Korean, and Mongolian) using the key words: “diabetes,” “KCNQ1,” “polymorphism,” and “East Asian.” The reference lists of the retrieved papers were also searched to obtain all the relevant publications. The inclusion criteria were as follows: (1) case-control study or cohort study; (2) only T2D as outcome; (3) available genotype distribution or allele frequencies, or sufficient data for calculation; and (4) published in English. The exclusion criteria were: (1) not focusing on any one of the four SNPs included in this study; (2) the outcome was not T2D; and (3) performed in other populations except East Asians.

Data Extraction and Study Quality Assessment

Two authors (Haoran Wang and Kun Miao) extracted data independently and in duplicate, and agreed upon all items, including: journal and year of publication, name of first author, ethnicity of the study population, representativeness and ascertainment of cases and controls, mean age, male sex percentage in case and control groups, genotype distribution or allele frequencies, Hardy-Weinberg equilibrium status in controls, numbers of cases and controls, and the SNPs included. The results were compared and disagreements were discussed and resolved with consensus. We also quantitatively assessed study quality using quality assessment scores based on traditional epidemiological and genetic considerations (Lichtenstein et al., 1987; Attia et al., 2003). The detailed criteria can be found in Table S2. Total scores ranged from 0 (worst) to 10 (best).

Statistical Meta-Analysis

Data analyses were performed as follows. First, to estimate the pooled risk allele frequency (RAF) of each included SNP, we used data only from control groups and performed this analysis by the inverse variance method. Second, the heterogeneity among study results was assessed using the Cochrane Q-test in combination with the I2 statistic. It was classified as low (<25%), moderate (25–50%) or high (>50%) according to I2 results (Higgins et al., 2003). In the presence of significant heterogeneity (Q test, p < 0.10), the source of heterogeneity was explored by fitting a covariant (sample size, mean age and gender distribution of cases and controls, study quality score) in a meta-regression, in which the dependent variable (the [log] OR) was weighted-regressed against covariates at the study level (Thompson & Higgins, 2002). Third, per-allele ORs from each independent study were pooled together by a fixed-effects model (Q test, p > 0.10) or a random-effects model (Q test, p < 0.10). The significance of the overall ORs were determined by a Z-test. Differences in ORs between Chinese, Japanese, Korean, and Mongolian subjects were compared by the Q test. For a better presentation of how the pooled ORs changed as updated evidence accumulated, we used forest plots from a cumulative meta-analysis method. Fourth, to evaluate the reliability and stability of our results, we assessed publication bias by Begg's funnel plotting and Egger's linear regression (Begg & Mazumdar, 1994; Egger et al., 1997). We also performed sensitivity analyses by recalculation of the pooled ORs after exclusion of studies one by one. Finally, for a better presentation of the public health relevance, we explored the PAR by taking into account both the pooled per-allele ORs and the pooled RAF. PAR was calculated as PAR = (X − 1)/X. Assuming a multiplicative model, X = (1 − f)2 + 2f(1 − f)γ + f2γ2, where γ is the estimated OR and f is the frequency of risk allele. All reported probabilities (p values) were two-sided, with p < 0.05 considered statistically significant. All meta-analyses were performed according to the PRISMA guidelines (Moher et al., 2009) with STATA 10.0 (Stata Corporation, College Station, TX, USA).

Results

Significant Associations between SNPs in KCNQ1 and T2D in Our Case-Control Study

Table 1 presents the clinical characteristics of all subjects from our case-control study. None of the four SNPs in our control samples showed significant deviation from the Hardy-Weinberg equilibrium proportion (p > 0.05). Genotype, allele frequencies, and association study results are presented in Table 2. Under the additive model with adjustment for sex, age and BMI, significant association was detected in all SNPs (rs2237892, per-C-allele OR 1.20, 95% CI 1.10–1.31, p = 3.10 × 10−5; rs2237895, per-C-allele OR 1.27, 95% CI 1.17–1.39, p = 2.08 × 10−8; rs2237897, per-C-allele OR 1.21, 95% CI 1.11–1.32, p = 1.46 × 10−5; rs2283228, per-A-allele OR 1.29, 95% CI 1.17–1.43, p = 2.95 × 10−8). These results remained significant after stringent Bonferroni correction (pcorr = 1.24 × 10−4 for rs2237892, pcorr = 8.32 × 10−8 for rs2237895, pcorr = 5.84 × 10−4 for rs2237897, and pcorr = 1.18 × 10−7 for rs2283228).

Table 1. Baseline characteristics of our study populations
 Population 1Population 2
VariablesControl (n = 1920)Case (n = 1920)p valueControl (n = 723)Case (n = 613)p value
  1. BMI, body mass index; WHR, waist-hip ratio; SBP, systolic blood pressure; DBP, diastolic blood pressure. HDL-C, high-density lipoprotein cholesterol; LDL-C, high-density lipoprotein cholesterol; TC, total cholesterol. TG, triglyceride.

  2. Data are presented as means ± SD or percentages.

  3. 1Age for the case group refers to age at diagnosis; age for the control group refers to age at which the study subject was enrolled.

Gender, male (%)1126 (58.6)1126 (58.6)1.000336 (46.5)386 (63.0)<0.001
Ages (years)156.84 ± 9.6153.23 ± 12.44<0.00154.08 ± 9.0353.34 ± 12.500.365
BMI (kg/m2)24.42 ± 3.2724.72 ± 3.680.00524.34 ± 3.3224.94 ± 3.820.002
WHR0.87 ± 0.070.91 ± 0.06<0.0010.84 ± 0.070.90 ± 0.06<0.001
SBP (mmHg)126.21 ± 19.41134.02 ± 23.15<0.001133.68 ± 21.28136.42 ± 24.510.101
DBP (mmHg)77.86 ± 10.6883.61 ± 13.53<0.00182.20 ± 11.5786.19 ± 14.49<0.001
FPG (mmol/l)4.68 ± 0.599.44 ± 4.35<0.0014.70 ± 0.719.77 ± 3.42<0.001
HDL-C (mmol/l)1.23 ± 0.621.10 ± 0.55<0.0011.45 ± 0.331.09 ± 0.31<0.001
LDL-C (mmol/l)2.43 ± 0.892.51 ± 0.950.0222.85 ± 0.782.52 ± 0.95<0.001
TC (mmol/l)4.33 ± 1.034.78 ± 1.45<0.0014.82 ± 0.944.08 ± 1.48<0.001
TG (mmol/l)1.77 ± 1.922.13 ± 1.890.0071.28 ± 0.931.68 ± 1.20<0.001
Hypertension (%)604 (31.5)777 (40.5)<0.001302 (41.8)297 (48.5)0.014
Hyperlipidemia (%)414 (21.6)613 (31.9)<0.001251 (34.7)175 (35.4)0.016
Smokers (%)745 (38.8)798 (41.6)0.081195 (27.0)240 (39.2)<0.001
Table 2. Results of our case control study
SNP Genotype count2 Genotype count Combined analysis4
Allele1 012 012OR (95% CI)3pOR (95% CI)ppcorr5
  1. 1Alleles are represented as major allele/minor allele while risk alleles are in bold.

  2. 20, 1, 2 represent the number of risk allele.

  3. 3ORs were calculated by logistic regression under an additive model adjusted for sex, age, and BMI.

  4. 4Combined analyses were further adjusted for different set.

  5. 5pcorr were adjusted for multiple testing by using the stringent Bonferroni correction method.

rs2237892Case 11547561010Control 11758608851.19 (1.08–1.32)0.0011.20 (1.10–1.31)3.10 × 10−51.24 × 10−4
C/TCase 249242322Control 2703283251.24 (1.04–1.48)0.015   
rs2237895Case 1790874256Control 19248121841.29 (1.17–1.42)2.66 × 10−71.27 (1.17–1.39)2.08 × 10−88.32 × 10−8
A/CCase 224831649Control 2346321561.24 (1.04–1.49)0.018   
rs2237897Case 1155809956Control 12188328701.21 (1.09–1.34)1.84 × 10−41.21 (1.11–1.32)1.46 × 10−55.84 × 10−5
C/TCase 256245312Control 2843223171.22 (1.03–1.44)0.022   
rs2283228Case 1174851895Control 12449207561.29 (1.17–1.43)3.25 × 10−71.27 (1.17–1.38)2.95 × 10−81.18 × 10−7
A/CCase 266242305Control 2923263051.21 (1.02–1.43)0.026   

Results of the Study Selection and Characteristics of Included Articles

A total of 157 articles were identified from MEDLINE and EMBASE through January 2013, and 121 of these were excluded based on the review of title or abstract, or were not published in English, therefore 36 potentially relevant articles remained and underwent a full review. Two more articles were identified through searching of reference lists, while 17 articles were excluded as they did not focus on any one of the four SNPs included in our study, or contained insufficient data for analysis or the outcomes that were not T2D. Finally, 19 articles were included in the meta-analysis (Lee et al., 2008; Unoki et al., 2008; Yasuda et al., 2008; Hu et al., 2009; Liu et al., 2009; Qi et al., 2009; Takeuchi et al., 2009; Chen et al., 2010; Han et al., 2010; Shu et al., 2010; Tan et al., 2010; Tsai et al., 2010; Xu et al., 2010; Yamauchi et al., 2010; Zhou et al., 2010; Saif-Ali et al., 2011; Tabara et al., 2011; Odgerel et al., 2012; Yu et al., 2012). Figure S2 provides a summary of the literature review, and the characteristics of studies selected into the meta-analysis are listed in Table 3. Using the results provided in these studies, we evaluated the increased risk of the four SNPs in an additive model in the East Asian population.

Table 3. Characteristics of genetic association studies included in this meta-analysis
   CaseControl    
First authorRaceYearNumberMale%AgeNumberMale%Agers2237892rs2237895rs2237897rs2283228
  1. NA, not available. √ represents the SNP was studied.

Unoki et al.Japanese200819464.454.6155836.055.2 
Unoki et al.Japanese2008136761.763.5126653.559.7 
Unoki et al.Japanese2008163060.061.5106460.045.5 
Unoki et al.Japanese2008130459.461.128838.971.6 
Unoki et al.Japanese200865460.067.2    
Unoki et al.Chinese2008149849.163.9188144.035.4 
Yasuda et al.Chinese2008141640.450.0157746.125.1  
Yasuda et al.Japanese2008437855.761.9441246.158.5  
Yasuda et al.Korean200875846.759.263245.464.7  
Lee et al.Korean200890848.458.250253.655.0   
Chen et al.Chinese200957NANA341NANA
Hu et al.Chinese2009176952.161.1173441.457.4 
Liu et al.Chinese2009191241.163.9204131.158.1 
Qi et al.Chinese200942444.358.6190844.358.6 
Tan et al.Chinese20091541NANA2196NANA  
Takeuchi et al.Japanese2009162957.063.9248146.969.0   
Shu et al.Chinese201010190.0NA17100.0NA   
Tsai et al.Chinese2010279851.560.2236750.349.3   
Han et al.Chinese2010102452.756.0100534.158.0   
Xu et al.Chinese2010182543.963.3220038.459.3   
Xu et al.Chinese20106735.861.566736.761.0   
Yamauchi et al.Japanese20104470NA65.83071NA52.5   
Zhou et al.Chinese201053743.057.251036.755.6   
Saif-Ali et al.Chinese201130051.049.823061.352.9 
Odgerel et al.Mongolian2011177NANA216NANA  
Tabara et al.Japanese201150655.360.040254.259.0   

Each C-allele of rs2237892 Was Associated with a 31.8% Increase in Incidence of T2D

Meta-analysis of the relationship between rs2237892 and T2D included 16 articles containing a total of 38,338 cases and 35,907 controls. The pooled risk C-allele frequency (Table S3) was slightly higher in Chinese (0.66) compared to Japanese (0.61), Korean (0.61), and Mongolian subjects (0.64). When comparing effect sizes, high-level between-study heterogeneity was detected (Q = 58.69; I2 = 62.5%; p < 0.001). In the overall estimate including all studies by a random-effect model, the risk C-allele was found to be associated with a 33% increase in the odds of T2D compared with the T-allele (OR, 1.33; 95% CI, 1.28–1.39; p < 10−16). However, no differences in ORs were detected between Chinese (15 studies), Japanese (5 studies), Korean (2 studies), and Mongolian (1 study; ORs, 1.32, 1.33, 1.35, and 1.58, respectively; χ2 = 1.53, p = 0.675; Table S7). In addition, no significant impact on the magnitude of genetic effect was detected from meta-regression analyses for all the factors investigated (sample size, study quality, mean age of cases and controls, and sex distribution in cases and controls). Cumulative meta-analysis indicated that the association remained significant after the initial discovery and the effect size showed a downward trend over the whole time period (Fig. 1). When both the pooled RAF and the pooled per-allele OR were taken into account, the presence of each C-allele would be associated with a 31.8% increase in incidence of T2D according to the PAR estimate.

Figure 1.

Cumulative meta-analysis describing the association between KCNQ1 rs2237892 variant and T2D. The arrow indicates the direction of association and denotes the maximal upper limit of the 95% CI of the pooled OR. Significant heterogeneity was detected (Q = 58.69; I2 = 62.5%; p < 0.001), and a random-effects model was adopted. The pooled OR decreased from 1.38 to 1.33 over the whole time period.

Studies Were Consistent both in Direction and Magnitude for the Association between Rs2237895 and T2D

A total of 16 studies from 10 publications encompassing 56,078 subjects (30,704 cases and 25,374 controls) were combined in the meta-analysis investigating the association of the rs2237895 variant with T2D. The pooled risk C-allele frequency (Table S4) was slightly higher in Japanese (0.34) compared to Chinese (0.32), Koreans (0.30), and Mongolians (0.29). There was no significant heterogeneity in relation to effect size among the included studies (Q = 11.68; I2 = 0.0%; p = 0.703). The pooled estimate from a fixed-effect model revealed that the minor risk C-allele was associated with an augmented T2D risk (OR, 1.28; 95% CI, 1.25–1.32; p < 10−16). The genetic effect sizes were similar between Chinese (12 studies), Japanese (2 studies), Koreans (1 study), and Mongolians (1 study; ORs, 1.26, 1.34, 1.27, and 1.26, respectively; χ2 = 3.40, p = 0.334; Table S7). By performing cumulative meta-analysis according to published year, we detected an upward trend over the whole time period (Fig. 2). When both the per-allele OR and RAF were taken into account, the PAR estimate revealed that each C-allele was associated with a 15.8% increase in incidence of T2D.

Figure 2.

Cumulative meta-analysis describing the association between KCNQ1 rs2237895 variant and T2D. The arrow indicates the direction of association and denotes the maximal upper limit of the 95% CI of the pooled OR. A fixed-effects model was used to pool data due to the absence of heterogeneity (Q = 11.68; I2 = 0.0%; p = 0.703). The pooled OR increased from 1.25 to 1.28 over the whole time period.

ORs Were Significantly Higher in Japanese and Mongolian than in Chinese for the Association between Rs2237897 and T2D

Ten studies extracted from seven articles including a total of 13,374 cases and 14,273 controls were included. The risk C-allele frequency (Table S5) was higher in Chinese (0.65) than in Japanese (0.61) and Mongolians (0.52). High-level heterogeneity in relation to the effect size was observed (Q = 21.36; I2 = 57.9%; p = 0.011). The pooled result from a random-effect meta-analysis demonstrated that the risk C-allele was significantly associated with increased risk of T2D (OR, 1.29; 95% CI, 1.21–1.37; p = 1.33 × 10−15). ORs were significantly higher in Japanese (1 study) and Mongolian (1 study) than in Chinese (8 studies; ORs, 1.40, 1.86, and 1.25, respectively; χ2 = 11.2, p = 0.004; Table S7). In the meta-regression analysis for heterogeneity, neither sample size, study quality, mean age of cases and controls nor sex distribution in cases and controls were significantly correlated with the magnitude of genetic effect (all p > 0.05). In addition, the pooled effect size went down gradually as updated evidence accumulated (Fig. 3). Finally, according to the PAR estimate, the presence of each C-allele would be associated with a 28.7% increase in incidence of T2D.

Figure 3.

Cumulative meta-analysis describing the association between KCNQ1 rs2237897 variant and T2D. The arrow indicates the direction of association and denotes the maximal upper limit of the 95% CI of the pooled OR. Data were pooled by a random-effects model due to high-level heterogeneity (Q = 21.36; I2 = 57.9%; p = 0.011). The pooled OR decreased from 1.40 to 1.29 with evidence accumulating.

Association Studies Performed in Chinese and Japanese Individuals Consistently Confirmed the Association between Rs2283228 Variant and T2D

Eight studies performed in Chinese (7 studies) and Japanese (1 study) individuals extracted from five articles including a total of 13,175 cases and 12,336 controls were combined in the meta-analysis investigating the association between the rs2283228 variant and T2D. The risk A-allele frequency (Table S6) was higher in Chinese (0.63) than in Japanese (0.58). No significant heterogeneity in regard to effect size among the included studies was observed (Q = 5.98; I2 = 0.0%; p = 0.548). The pooled result from a fixed-effect model indicated that the risk A-allele was associated with an increased T2D risk (OR, 1.24; 95% CI, 1.19–1.28; p < 10−16). No difference in ORs was found between Chinese and Japanese (ORs, 1.23, and 1.26, respectively; χ2 = 0.30, p = 0.584; Table S7). The pooled OR changed little with evidence accumulating (Fig. 4). Finally, when both the per-allele OR and RAF were taken into account, the PAR estimate revealed that each A-allele was associated with a 24.1% increase in incidence of T2D.

Figure 4.

Cumulative meta-analysis describing the association between KCNQ1 rs2283228 variant and T2D. The arrow indicates the direction of association and denotes the maximal upper limit of the 95% CI of the pooled OR. ORs were pooled together by a fixed-effects model in that heterogeneity was not detected (Q = 5.98; I2 = 0.0%; p = 0.548). The pooled result changed little (from 1.26 to 1.24) over the whole time period.

Publication Bias and Sensitivity Test

No evidence for publication bias was apparent for any association investigated on the basis of either visual inspection of the funnel plots (Figs S3–S6) or by using the statistical significance test of Egger's linear regression (all p > 0.05). Additionally, results from the sensitivity analyses (Figs S7–S10) indicated that omitting each study in turn did not substantially alter the overall results, further confirming the reliability and stability of these results.

Discussion

In this study, we first investigated the associations of four common variants in KCNQ1 with T2D in two Chinese Han populations originating mainly from Wuhan in central China. All four SNPs conferred increased risk of T2D under an additive model, which was consistent with previous studies. As the study population was totally independent from previous reports, it would provide more optimal evidence of the credibility of this genetic association. Additionally, results from the current large-scale meta-analyses not only confirmed the strong association of KCNQ1 with T2D but also revealed the significant contribution of this gene to the diabetic epidemic in East Asians.

While the genetic effect size was similar among Chinese, Japanese, Korean, and Mongolian individuals for three (rs2237892, rs2237895, and rs2283228) of the four SNPs investigated, our results indicated that ORs were significantly higher in Japanese and Mongolians than in Chinese for the association between rs2237897 and T2D. There were several potential explanations for this heterogeneous association result. First, the risk C-allele frequency differed from 0.52 in Mongolians to 0.61 in Japanese and 0.65 in Chinese. Second, there existed uncertainty in the magnitude of the genetic effect in Japanese (1 report) and Mongolians (1 report) studies, which was due to lack of consistent replication evidence. Third, potential existing gene-gene and gene-environment interactions may also exert their effects on the association results.

In this study, we presented the public health relevance of our genetic association findings by using the PAR estimates. The PAR depends on both the strength of association between exposure to a risk genetic variant and the prevalence of this risk variant within the population. Therefore, it is probably the most useful epidemiological variable for public health administrators. According to our estimates, variants in KCNQ1 explained about 16–32% (PAR estimates, 31.8%, 15.8%, 28.7%, and 24.1% for rs2237892, rs2237895, rs2237897, and rs2283228, respectively) of the T2D incidence in East Asians, which was comparable to that of TCF7L2 in Caucasians (PAR estimates 17–28%; Cauchi et al., 2007; Luo et al., 2009). Therefore, variants in KCNQ1 were among the leading genetic factors contributing to the overall burden of T2D in East Asians.

KCNQ1, located on 11p15.5, encodes the pore-forming subunit of a voltage-gated K+ channel (KvLQT1). Like the case of KCNJ11, ionic mechanisms at KATP and Kv- channels are primarily important in triggering and maintaining glucose-stimulated insulin secretion, therefore it is not surprising that variants in KCNQ1 may play a potential role in the predisposition of T2D by affecting the activity of this ion channel. Previous in vitro studies demonstrated that either knocking out a Kv- channel in rat islets or inhibiting its function by using drugs may enhance glucose-stimulated insulin secretion (Roe et al., 1996; MacDonald et al., 2001; MacDonald et al., 2002; Zhang et al., 2005). However, given that all these SNPs are located in noncoding regions, it is unclear whether these variants may affect the expression level or the function of the protein. In the study by Jonsson et al. (2009), the expression level of KCNQ1 in 10 human islets seems not to be affected by genotype, but this lack of association is underpowered due to the limited sample size. Several studies have already indicated that there may exist some associations between these common variants and diabetes-related quantitative traits; the risk allele carriers exhibit higher fasting plasma glucose (Liu et al., 2009; Qi et al., 2009; Tan et al., 2009) and lower insulin secretion (Yasuda et al., 2008; Holmkvist et al., 2009; Jonsson et al., 2009; Mussig et al., 2009; Qi et al., 2009; Tan et al., 2009; Been et al., 2011). Along with these findings, prospective cohort studies also demonstrated that risk allele carriers exhibit higher risk of future diabetes (Xu et al., 2010). These studies suggest that common variants in KCNQ1 may gradually impair pancreatic beta-cell function over time and eventually lead to the incidence of diabetes.

There were several limitations in this work. First, we were not able to exclude the impact of population substructure on the association results. However, given the homogeneous study population (Central Chinese Han) and the robust association results, we expect this impact to be minimal. Second, the included articles were limited to those published in English, with studies published in other languages systematically excluded. Thus, potential selection bias might exist. Third, the results were based on crude ORs, which do not take into account traditional risk factors, such as age, obesity, and smoking status; more precise estimates could be obtained if individual-level data were available.

Conclusion

In conclusion, the current comprehensive meta-analysis of all association studies derived precise estimates and conclusions on the implication of four common variants in KCNQ1 for risk of T2D in East Asians. Carriers of these risk alleles exhibit significantly increased risk of T2D. We contended that variants in this gene contribute to approximately 16–32% of all T2D in East Asians.

Acknowledgements

HRW, KM, JZZ, LL, GLC, and CC researched data. HRW, KM, HD and DWW contributed to discussion. HRW, KM, JZZ, HD wrote the manuscript. HRW, KM, JZZ, HD, and DWW reviewed and edited the manuscript. Dr. HD is the guarantor of this work, had full access to all the data, and takes full responsibility for the integrity of data and the accuracy of data analysis.

Funding

This study was funded by the research grants from National “973” Project (2012CB518004), National “863” projects (No. 2012AA02A510), and National Nature Science Foundation of China (30930039, 81100066).

Conflicts of Interest

The authors declared that they have no conflict of interest.

Ancillary