We recently reported an association between common allelic variation (class I or class III) at the variable number of tandem repeat (VNTR) locus in the promoter region of the insulin gene (INS) and birthweight1. In a contemporary, well-nourished group of 758 children, the Children in Focus sub-cohort of the Avon Longitudinal Study of Pregnancy and Childhood (ALSPAC), the frequency of the III/III genotype was 9% and associated with an increase of 200 g in birth weight. The class III allele and III/III genotype have also been associated with insulin resistance, diabetic hypertriglyceridaemia, atherosclerosis, cardiovascular disease and central obesity2, anovulatory polycystic ovary syndrome3 and, in some studies, type 2 diabetes2. These results, however, conflict with data showing strong associations between low birth weight and a raised prevalence of impaired glucose tolerance (IGT) and type 2 diabetes in adult life4,5.

To address this paradox, we analysed the INS VNTR in 218 men born in Hertfordshire between 1920 and 1930 in whom the link between low birth weight and IGT/type 2 diabetes was first described5. IGT or type 2 diabetes was found in 52 of 218 men, as determined by blood glucose levels 2 hours after a 75-g glucose drink (IGT, glucose 7.8-11.0 mmol/l; type 2 diabetes, glucose>11.1). We were not able to confirm the association between the INS VNTR and birth weight in this study. This was not unexpected, as the sample size was small, birth weight was often only measured to the nearest pound and size at birth in these men may have been more greatly influenced by maternal-uterine restraint than in contemporary cohorts. We did, however, detect a significant association between the III/III genotype and IGT/type 2 diabetes (OR=4.6, P=0.01; Table 1), but only in those subjects whose growth in utero would have been relatively unaffected by maternal-uterine factors such as nutrition, and who subsequently showed little change in weight standard deviation (s.d.) score from birth to one year of age ('non-changers', Table 1). Similarly in the ALSPAC cohort, it was in this 'non-changers' group, who followed their genetic growth trajectory, that the association between the III/III genotype and birth weight was most apparent1. Hence, by incorporating data on birth weight and early postnatal growth, the association between INS VNTR III/III genotype and IGT/type 2 diabetes was stronger (OR=4.6) than that obtained by meta-analysis of all previously reported conventional case-control studies of the association of the III/III genotype with type 2 diabetes, which were carried out without regard for birth weight and early growth (OR=1.4, P=0.037; Table 2).

Table 1 IGT/type 2 diabetes, genotype and birth weight
Full size table
Table 2 Insulin gene VNTR class III/III genotype and type 2 diabetes
Full size table

Low birth weight was strongly related to IGT/type 2 diabetes (Table 1). For the first time, we show that this association was more evident in 'changers' than in 'non-changers' (Table 1), indicating the important role of restraint of fetal growth according to intrauterine environment and nutrition. It has been proposed that fetal metabolic and endocrine adaptations to maternal nutritional restraint in utero persist throughout life and lead to insulin resistance and increased susceptibility to type 2 diabetes6. The relationship between birth weight and IGT/type 2 diabetes was observed regardless of INS VNTR genotype (Table 1). Likewise, the association between INS VNTR and IGT/type 2 diabetes persisted after adjustment for birth weight (Table 1), suggesting factors that restrain fetal growth and increase susceptibility to subsequent disease act independently of, and add to, the effect of the INS VNTR (or a closely linked polymorphism). Therefore, we have resolved the paradox1,4 by showing independent effects of INS VNTR and birth weight on risk for IGT/type 2 diabetes.

The INS VNTR III allele has been associated with reduced transcription of INS in fetal pancreas7,8 and of the gene encoding insulin-like growth factor 2 (IGF2) in placental tissue9. We propose that reduced insulin secretion or altered placental nutrient transfer may lead to a muscle-specific insulin resistance10 in the fetus. As long as nutrient supply is adequate, selective peripheral insulin resistance may enhance the anabolic actions of insulin and thereby promote growth, as occurs during puberty11. In the Hertfordshire men, the III/III genotype was weakly associated with higher insulin levels two hours after oral glucose (P=0.09, data not shown; more direct measures of insulin resistance were not available). The hypothesis that the III/III genotype may confer a selective insulin resistance is also supported by reported associations of the INS VNTR class III allele with a number of conditions in which insulin resistance is a major feature2,3, such as polycystic ovary syndrome and central obesity. Thus, selective peripheral insulin resistance could explain the III/III genotype associations with both larger size at birth and insulin resistance or type 2 diabetes risk. Similarly, in conditions of fetal growth restraint, such as poor nutrition in pregnancy, peripheral insulin resistance may also represent a fetal metabolic adaptation that diverts nutrients to protect brain and skeletal growth in utero12, but may lead to disease in adulthood.