To the editor: KATP channels are key to insulin secretion by the beta cell; they couple metabolism to plasma membrane depolarisation and, hence, insulin release. KATP channels consist of two sub-units: Kir6.2 and SUR1, encoded by the genes KCNJ11 and ABCC8, respectively. Multiple large association studies have recently confirmed a Glu23Lys (E23K) polymorphism in KCNJ11 as a Type 2 diabetes susceptibility allele [1, 2]. The K allele confers a Type 2 diabetes odds ratio of about 1.2, and is carried by 58% of the population [1]. Functional studies have shown that KATP channels encoded by the K allele have an increased open probability [3].

Mutations in KCNJ11 that result in constant closure of the KATP channels cause hyperinsulinaemia and, due to increased foetal insulin secretion, higher birth weight [4]. We therefore hypothesised that E23K, as a Type 2 diabetes susceptibility allele, would also affect the quantitative traits of beta-cell function and birth weight. Careful physiological studies, using hyperglycaemic clamps in small cohorts (n=75 and n=159), found no evidence of reduced insulin secretion in carriers of the E23K variant [5, 6]. However, Tschritter et al. did report association of the K allele with impaired glucagon suppression in response to hyperglycaemia [5]. The assessment of beta-cell function using the OGTT is less precise, but has allowed a larger number of subjects to be assessed. Results from such studies have been conflicting: one investigation did not provide any evidence for an effect of the E23K variant on beta-cell function (n=298) [5]; however, a recent larger study (n=519) has shown a reduction in beta-cell function in carriers of the K allele [7]. No assessment of E23K's effect on birth weight has been reported.

We used two large population based cohorts from the Barry-Caerphilly (BCG; n=645 adults) and Exeter Family (EFS; n=430 families) studies, to look for association between E23K genotype and beta-cell function and birth weight. The BCG and EFS cohorts have been described in detail elsewhere [8]. Briefly, BCG is a collection of subjects from Barry-Caerphilly in Wales, UK, studied from birth to young adulthood (average age at OGTT study 30 years). The EFS is an on going consecutive birth cohort of babies, fathers and 28-week pregnant mothers, from a geographically defined region of Exeter, UK (average adult age 30 years). All subjects were healthy UK Caucasians. Birth weight was corrected for sex and gestational age. In the BCG cohort, beta-cell function was assessed using OGTT data, by insulinogenic index (insulin 30 min–insulin 0 min / glucose 30 min–glucose 0 min) and AUC insulin, correcting for sex and current age.

The results from the two cohorts are shown in Table 1. In the BCG OGTT, E23K genotype was not associated with a significant reduction in beta-cell function. Foetal growth, as measured by birth weight and birth length, was not associated with the E23K polymorphism in either cohort.

Table 1. Beta-cell function and birth weight by E23K Kir6.2 genotype (where appropriate adjusted for age, gestational age and sex). All data shown as means (95% CI)
Full size table

It is interesting that, despite strong evidence for a role in Type 2 diabetes, we have not found evidence that the E23K variant affects insulin secretion or alters birth weight; and this is the largest study yet reported. This would suggest the primary effect of E23K on diabetes susceptibility might be through beta-cell dysfunction not detectable in utero or in young adult life. Beta-cell dysfunction has been shown in one study [7], and whilst this could represent stochastic variation for a weak association, there are two other possible reasons why our study and other previous studies have not shown association of E23K genotype with beta-cell function. Firstly, it could be that the use of normoglycaemic subjects limits the power of our study to detect association of diabetes-linked quantitative traits. A second interesting explanation is that E23K is involved in deterioration, rather than persistent dysfunction, of the beta cell. Therefore, detecting an association of E23K genotype with diabetes-linked quantitative traits might not be possible in newborn babies or young adults. Supporting this, KATP channels seem to be involved in cell survival in the endocrine pancreas: pancreatic islets of KCNJ11 knockout mice have reduced numbers of beta cells, apparently due to an increase in apoptosis [9]. In support of this, the mean age of patients in the negative OGTT studies were 30 years (this study) and 38 years [5], compared to 57 years in the positive study [7].

In conclusion, in the largest study so far, we have found no evidence for an association of the Type 2 diabetes susceptibility polymorphism, E23K, with beta-cell dysfunction or foetal growth. This lack of association could reflect a role of the polymorphism in deterioration of beta-cell function, rather than persistent dysfunction throughout life. The lack of evidence for an effect of E23K genotype on beta-cell function might therefore be due to the relatively young age of the subjects studied so far. We suggest further studies are carried out in cohorts of older subjects.