Obesity is a multifactorial disease caused by an interaction of genetic, lifestyle, and environmental factors. A sedentary lifestyle, a diet rich in fat and energy, the cultural level, and a genetic predisposition can all contribute to the increased prevalence of obesity (1,2).
Glucokinase is expressed in pancreatic β cells and in hepatocytes. Its expression is controlled by tissue-specific promoters. Pancreatic glucokinase is a glucose sensor for the regulation of insulin secretion. The −30G>A polymorphism of the promoter of the pancreatic glucokinase gene has been associated with a reduced β cell function in Japanese persons with glucose intolerance (3), though other studies in Finnish (4) and Danish (5) persons have failed to find this association. Nonetheless, in obese persons with type 2 diabetes mellitus, the insulin sensitivity index has been associated with the −30G>A polymorphism of the promoter of the pancreatic glucokinase gene (6).
Yamada et al. recently undertook a large study in 3,906 Japanese persons, of whom 1,126 had a BMI ≥ 25. Of the 147 polymorphisms of candidate genes for obesity, only nine polymorphisms showed a significant association with obesity. One of these, the −30G>A polymorphism of the promoter of the glucokinase gene was inversely associated with the risk for obesity (7). The aim of this study was to test the hypothesis of an association between the −30G>A polymorphism of the promoter of the glucokinase gene and obesity in the setting of a population-based cohort study (Pizarra Study) (1).
The frequency of the carriers of the AA genotype was 3.8% (n = 40), 33.4% for the GA genotype (n = 348), and 62.8% for the GG genotype (n = 655); these data are in Hardy-Weinberg equilibrium.
Using the same cutoff point for BMI as Yamada et al, the prevalence of carriers of the A allele in persons with a BMI <25 was 42% whereas its prevalence in persons with a BMI ≥25 was 34% (P = 0.01). After adjusting for age and sex, a logistic regression model (Table 1) showed that the probability of having a BMI ≥25 in carriers of the A allele was significantly lower than expected by chance (odds ratio (OR) = 0.63; P = 0.003). Given the prevalence of the risk genotype, the smallest OR that we could detect with 80% power and α = 0.05 was 0.68. The inclusion of the homeostasis model assessment in the model was, as expected, significantly associated with a BMI ≥ 25 (OR 1.62; 95% confidence interval (CI) = 1.42–1.85; P < 0.001), though it did not significantly modify the strength of association between the A allele and this dependent variable. Six years later, the overall prevalence of obesity (BMI ≥ 30) had risen from 27.5 to 30.5%. At this point, the likelihood of not being obese was significantly associated with the presence of the A allele (AA: OR = 0.33; P = 0.04 and GA: OR = 0.68; P = 0.02) (Table 2). We tested the dominant and additive models using the likelihood ratio test. The additive model did not improve the strength of the association between the polymorphism and the likelihood of being obese.
The incidence of obesity after 6 years' follow-up (i.e., persons who had a BMI < 30 at the first study who had a BMI ≥ 30 six years later) was 14.4%. The incidence and prevalence of obesity after the 6 years was not independent of the presence of the −30G>A polymorphism of the promoter of the glucokinase gene. The increase in weight in the persons who were originally obese (BMI ≥ 30) was less in those who had the A allele (GA/AA) (Table 3). These persons also had a significantly lower probability of remaining obese than the GG homozygotes (OR = 0.22; 95% CI = 0.087–0.576) (Table 4).
We also tested for an association between the presence of the A allele and the risk of increasing waist circumference after the 6 years above the cutoff points of 1022 cm for men or 88 cm for women (the values agreed by the Spanish Society for the Study of Obesity). However, no association was found between the A allele and increase in waist circumference considered as a discrete variable (logistic regression model), nor even when waist circumference was considered a continuous variable (ANOVA model) (data not shown).
The results support the findings of the gene screening study by Yamada et al. (7) and suggest that the −30G>A polymorphism of the promoter of the glucokinase gene is inversely associated with the prevalence of obesity. They also confirm that carriers of the A allele have a lower incidence of obesity and a higher probability of reducing weight.
Phosphorylation of glucose by pancreatic β glucokinase is a key point in the regulation of insulin secretion. Insulin sensitivity and secretion (as measured by the homeostasis model assessment) were very significantly associated with the presence of obesity, as expected. However, they were not associated with the presence of the A allele (data not shown) and neither did their inclusion in logistic regression models modify the strength of the association between the A allele and obesity. The association found with the prevalence and incidence of obesity, as well as with its prognosis, could partly explain the results of Elbein et al. (6), who found that the insulin sensitivity index in persons with type 2 diabetes who were obese was associated with the −30G>A polymorphism of the promoter of the pancreatic glucokinase gene.
A possible explanation for this association with the −30G>A polymorphism of the glucokinase gene promoter is the presence of a linkage disequilibrium with genes associated with obesity, though this has yet to be shown. Accordingly, we used the Hapmap 2 database to scan 20 kbp either side of the −30G>A polymorphism of the glucokinase promoter gene (rs1799884), selecting a set of 48 polymorphisms in Caucasians, of which 38 are polymorphic with a minor allele frequency >0.05. The Haploview 3.32 software tool (http:www.broad.mit.edumpghaploview) to display the r2 showed that eight markers were captured by rs1799884 at an r2 > 0.8. These eight polymorphisms captured by the glucokinase promoter were rs2908282, rs2908289, rs2971668, rs2971669, rs6975024, rs4607517, rs730497, and rs917793. However, these polymorphisms are nonfunctional as they do not interact with any transcription factor (http:www.pupasnp.bioinfo.cipf.es). Moreover, they do not appear among the polymorphisms that Yamada et al. found to be associated with obesity.
Glucokinase is also expressed in the hypothalamus (8), and it has been suggested that certain groups of cells in the anterior hypophysis might function as glucose sensor cells and that direct fuel regulation of such cells may modify the classical indirect neuroendocrine pathways that are known to control hormone secretion from anterior pituitary cells (9).
Yang et al. recently studied mice in which the glucokinase gene, which plays an essential role in neuroendocrine glucose sensing, had been ablated (10). Their results support the hypothesis that glucokinase mediates not only insulin secretion but also neuroendocrine regulation of metabolic economy, including regulation of other hormones and appetite, and that a reduction in glucose sensitivity in the hypothalamus might contribute to the development of obesity in certain persons.
Although other studies are required to confirm this association of the −30G>A polymorphism in the promoter of the glucokinase gene with the BMI, the biological effect is plausible given the possible association of this polymorphism and β cell function (11,12). Furthermore, hepatic glucokinase plays an important role in the pathogenesis of insulin resistance and type 2 diabetes mellitus. This variant may alter hepatic insulin sensitivity, which may also be reflected in the homeostasis model assessment of insulin resistance measurement in the β cell. On the other hand, the low frequency of the AA genotype and the sample variability, together with the low risk of younger persons becoming obese, could argue against finding an association between the polymorphism and the risk of obesity. However, despite these unfavorable factors for the association, our statistical models still suggest its positive association.