Genetic Variation in KCNQ1 Associates With Fasting Glucose andβ-Cell Function: A Study of 3,734 Subjects Comprising Three Ethnicities Living in Singapore

The study used data from the cross-sectional population-based 1998 Singapore National Health Survey (NHS98; n = 4,723). Genotype data were available for 3,734 subjects comprising 2,520 Chinese, 693 Malay, and 521 Asian Indians. A total of 1,881 Chinese subjects with normal glucose tolerance (NGT) had previously served as the control subjects for the Singapore replication arm in our original report (6), whereas type 2 diabetes case subjects were from a separate study (the Singapore Diabetes Cohort Study). In this present study, all subjects (including subjects with NGT, impaired fasting glucose, impaired glucose tolerance [IGT], and type 2 diabetes) were derived from the NHS98 study. Details of the NHS98 survey have been previously described in greater detail (9). Briefly, this was a population-based, cross-sectional survey conducted between September and November in 1998. The reference population was 2.16 million Chinese, Malay, and Asian Indian Singapore residents aged between 18 and 69 years. The survey was based on the World Health Organization (WHO)-recommended model for field surveys of diabetes and other noncommunicable diseases and the WHO MONICA protocol for population surveys. The research protocol for NHS98 was approved by the Singapore General Hospital Institutional Review Board (#54/2001).

Fasting blood samples were collected for glucose and insulin after an overnight fast of 10 h. All subjects underwent a 75-g oral glucose tolerance test except those taking oral hypoglycemic agents or insulin. Subjects were classified as having diabetes if they gave a history of diabetes or if their fasting glucose was >7 mmol/l or 2-h post-challenge glucose >11.1 mmol/l. Impaired fasting glycemia was defined as fasting glucose of 6–7 mmol/l and impaired glucose tolerance as 2-h post-challenge glucose of 7.8–11.1 mmol/l. Other measurements included the following: BMI and waist and hip circumference. Insulin resistance was estimated using the homeostasis model assessment method (HOMA-IR) (10). β-Cell function was assessed using the corrected insulin response at 120 min (CIR₁₂₀) (11).

Genotyping of four SNPs in intron 15 of KCNQ1 rs2237897 (c>T), rs2237895 (A>C), rs2237892 (C>T), and rs2283228 (A>C) was performed using the TaqMan SNP genotyping assay (Applied Biosystems, Foster City, CA). Genotyping success rate for rs2237897, rs2237895, rs2237892, and rs2283228 was 92, 99, 92, and 87%, respectively. To assess reproducibility, 1% of samples were analyzed in duplicate; genotyping was 100% concordant for these samples.

Minor allele frequency, Hardy-Weinberg equilibrium, and linkage disequilibrium (LD; reported using r²) were estimated using Haploview (12). The distribution of glucose and insulin measures were skewed and therefore normalized by natural logarithmic transformation. Means were subsequently back transformed for presentation. Quantitative traits are presented as means and SDs. Linear regression analyses were performed to study the associations between diabetes-related traits with genotypic groups. Individuals were as assigned as 0/1/2 according to their number of minor alleles under an additive model of inheritance. There was no significant heterogeneity between the sexes (P> 0.05), and subsequent analyses were performed with the sexes combined and adjusted for sex. Linear regression with adjustment for ethnicity was used to estimate the summary effect size of the SNPs in the combined sample from the three ethnic groups. Logistic regression was used to estimate the association between KCNQ1 SNPs and type 2 diabetes. All analyses were stratified by ethnic group and adjusted for age, sex, and BMI (where appropriate). Analysis of association with β-cell function was further adjusted for insulin resistance (13). Analyses were performed using STATA (version 9.1 for Windows) (14).

The anthropometric and biochemical characteristics of the participants are detailed in Table 1. Among these individuals, there was a higher prevalence of type 2 diabetes among the Malay and Asian Indian populations and a lower mean BMI and fasting glucose level among the Chinese population.

Allele frequency for rs2237897, rs2237895, rs2237892, and rs2283228 were similar between the Chinese and Malays but different in Asian Indians (Table 2). rs2237895 was in weak LD with rs2237897, rs2237892, and rs2283228 (r² < 0.25), whereas moderate LD was observed between rs2237897, rs2237892, and rs2283228 (Chinese:r² = 0.56–0.79 Malay:r² = 0.74–0.86, Asian Indian:r² = 0.39–0.62) (Supplemental Fig. 1, found in an online-only appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db08-1138/DC1). All SNPs in the Chinese and Malays were in Hardy-Weinberg equilibrium (P> 0.05). In the Asian Indians, three of the SNPs that had low minor allele frequency (rs2237897, rs2237892, and rs2283228) were not in the Hardy-Weinberg equilibrium. Consequently, the results for the Asian Indians should be interpreted with caution.

Table 2 shows the association between KCNQ1 SNPs with type 2 diabetes in the Chinese, Malay, and Asian Indian population in Singapore. Significant associations were observed in the Chinese, with rs2237897 showing the strongest effect: odds ratio (OR) 1.50 (1.15–1.96), P = 0.003. In the combined sample of the three ethnic groups, the risk alleles of rs2237897 (C), rs2237892 (C), and rs2283228 (A) were also associated with type 2 diabetes in our study, consistent with previous reports (6,7). Of these, rs2237897 showed the strongest association (OR 1.48 [1.20–1.83], P = 0.0003). In the combined sample of only Chinese and Malays subjects, the association with rs2237897 remained significant (OR 1.45 [1.16–1.81], P = 0.001).

Table 3 shows the association between KCNQ1 SNPs with diabetes-related traits. The risk alleles for rs2237897, rs2237892, and rs2283228 showed statistically significant association with higher fasting glucose levels (P = 0.014, 0.011, and 0.034, respectively) and reduced CIR₁₂₀(P = 0.007, 0.013, and 0.014, respectively) in the Chinese population. After restricting to subjects without diabetes, the association remained significant with CIR₁₂₀(P = 0.011, 0.020, and 0.015, respectively; online appendix Table 1). A similar trend was observed among the Malays and Asian Indians, although the associations did not reach statistical significance, possibly because of the limited sample sizes for these minority ethnic groups. In the combined analysis, rs2237897 and rs2237892 were significantly associated with fasting glucose levels (P = 0.029 and P = 0.021, respectively), whereas rs2237897, rs2237892, and rs2283228 were significantly associated with lower β-cell function (CIR₁₂₀) (P = 0.013, 0.021, and 0.020, respectively). These SNPs also showed associations with BMI in the Malay population and waist-to-hip ratio in Asian Indians. However, these associations were no longer statistically significant in the combined analysis. Based on a risk allele frequency of between 0.3 and 0.6, power calculations estimate that the combined sample provided 90% power to detect a 10% change in the examined traits.

Polymorphisms within KCNQ1 have recently been shown to be strongly associated with an increased risk of type 2 diabetes in the East Asian and European populations (6,7). Of the three polymorphisms within KCNQ1 previously described, rs2237897 appeared to have the strongest association with type 2 diabetes (6), whereas another study reported the strongest association with rs2237892 (7). Consistent with these findings, we found that both these SNPs were significantly associated with type 2 diabetes in the combined analysis of three ethnic groups. In addition, the presence of the risk allele for KCNQ1 variants rs2237897, rs2237892, and rs2283228 were associated with increased fasting glucose level and decreased β-cell function in the Chinese population and combined sample. This is in agreement with the study by Yasuda et al. (7), which reported an association of these KCNQ1 variants with β-cell function. Together, these findings corroborate the hypothesis that the role of this protein in the pathogenesis of type 2 diabetes is likely mediated through its effects on the pancreatic β-cell, although there is still a possibility that these polymorphisms may increase the risk of type 2 diabetes through regulation of nearby genes. In contrast, there was no association between rs2237895 with any diabetes-related traits, specifically highlighting the importance of the polymorphisms rs2237897, rs2237892, and rs2283228 (which are in moderate LD) in increasing the risk of developing type 2 diabetes. Further fine mapping of SNPs in KCNQ1, especially within the LD block containing rs2237897 and rs2237892, may allow the identification of the causal variant.

KCNQ1 is located on 11p15.5, which encodes the pore-forming α-subunit of the I_KS K⁺channel, which is expressed mainly in the heart and, to a lesser extent, the inner ear, stomach, small and large intestine, liver, and kidney. KCNQ1 is also expressed in the pancreas, where it is coexpressed with products of other regulators such as KCNE1, which may alter its biophysical characteristics and role (15). As such, it is plausible that polymorphisms within the KCNQ1 gene alter the properties and role of the I_KS K⁺channel, causing decreased pancreatic β-cellfunction and insulin production, leading eventually to hyperglycemia.

Whereas we do appreciate that the associations observed in the Chinese do not represent an independent replication, since the controls used are largely the same as those in the study by Unoki et al. (6), we have provided data on Malays and Indians and, in our opinion, it was reassuring that similar trends for an effect of these three polymorphisms on association with type 2 diabetes, fasting glucose, and β-cell function were also observed in the Malay and Asian Indian subjects, even though their limited numbers certainly resulted in a reduction of study power. Although associations with BMI in Malays and waist-to-hip ratio in Asian Indians were found, the smaller sample sizes for these ethnic groups, together with the fact that these observations were not found in other ethnic groups, inevitably led us to be cautious in interpreting these findings. One caveat of this study is that we only examined the SNPs, which showed the strongest association, identified by Unoki et al. (6) and Yasuda et al. (7). Although the Chinese show strong and consistent associations with these SNPs, the association among Malays and Asian Indians is less clear. Further fine mapping of the KCNQ1 locus, for instance, through deep DNA resequencing may allow the identification of population-specific causative variants.

The use of CIR₁₂₀represented a limitation of our study, since it only served as an approximate measure of β-cell function, which may be better assessed using the CIR₃₀. Thus, if CIR₃₀measurements were available, the positive associations that had been observed in our study may be further strengthened. It has also been suggested that CIR₁₂₀could inadvertently capture information on additional aspects of glucose intolerance instead of insulin secretion alone. The previous finding of the association between CIR₃₀with KCNQ1 SNPs, in nondiabetic European subjects of the Botnia prospective study during their follow-up visit (P = 0.024) (7), however, appeared to corroborate our hypothesis that these SNPs may exert their effect on type 2 diabetes through an effect on insulin secretion.

In conclusion, the risk alleles of rs2237897, rs2237892, and rs2283228 within the KCNQ1 gene are associated with decreased pancreatic β-cell function and fasting glucose levels, suggesting that the impact of these polymorphisms on the risk of type 2 diabetes may be mediated through an impact on β-cell function rather than insulin resistance. Further studies will be useful to replicate these promising findings and to fully delineate the role of KCNQ1 and its related pathways in the pathogenesis of type 2 diabetes.

This work was funded by the Singapore National Medical Research Council (NMRC/1115/2007).

No potential conflicts of interest relevant to this article were reported.