Common variants in the regulatory region of the insulin gene are associated with fasting plasma insulin levels in juvenile obesity.

Variable number of tandem repeat polymorphisms (VNTRs) are widely dispersed throughout the genome. They consist of repeats of stereotypic DNA sequences and are generally found in non-coding regions of DNA. For many VNTRs the number of repeat units display large inter-individual variation, which has made them useful tools in forensic medicine and gene mapping studies. In the majority of cases, the function of these VNTR polymorphisms remains a mystery. The insulin VNTR (INS-VNTR) is located 596 base pairs upstream (2596) of the translation initiation codon of the insulin gene on chromosome 11. It was first described by Bell et al. in 1981 (1). The INS-VNTR consists of varying numbers of a 14±15 base pair repeat and is broadly divided into three allelic groups: the short class I alleles consisting of 26±63 repeats, the intermediate class II alleles (approximately 80 repeats) and the long class III alleles consisting of 141±209 repeats (2). The class II alleles have been found predominantly in Black populations and are rarely found in Caucasians and little data are available for this particular class of alleles. In addition to length polymorphisms there are sequence differences within the various repeat units. The polymorphisms of the INS-VNTR are not randomly distributed and have been grouped into lineages based upon sequence and length (3, 4). Moreover, the INS-VNTR polymorphism is in strong linkage disequilibrium with several neighbouring single nucleotide polymorphisms (5). One of these is a T/A polymorphism, which can be identified by the restriction enzyme HphI, and is located 23 nucleotides upstream of the insulin gene translation initiation codon. In Caucasian populations the 223 HphI T or A alleles can be used as surrogate genetic marker for class I and III VNTR alleles respectively. Several reports on associations between particular INS-VNTR alleles and diseases and phenotypic traits have been published, some of which are listed in Table 1. The molecular mechanisms behind the associations with particular INS-VNTR alleles are not clear. Furthermore, due to strong linkage disequilibrium between the various polymorphisms in this region there is, with the exception of type 1 diabetes (6, 7), not formal proof to conclude that it is the INS-VNTR and not one of the neighbouring polymorphisms that are primarily involved. There is, however, evidence from both in vitro (8, 9) and in vivo studies (7, 10) that the INS-VNTR has an effect on mRNA transcription of the insulin gene where the various alleles are associated with different levels of transcription. Recently, Le Stunff et al. (11) demonstrated an association between the INS-VNTR and fasting plasma insulin levels in obese children. Two cohorts of 458 and 157 children of Mediterranean and Central European origin respectively with a mean age of 12 years and body mass index (BMI) above the 85th percentile before the age of 6 years (mean BMI approximately 30 kg/m) were recruited to the study. They used the insulin gene 223 HphI T and A alleles as markers for VNTR class I and III alleles respectively. Fasting insulin levels were 13% and 29% higher among children homozygous for the VNTR class I allele marker compared with the children homozygous for the class III marker or class I/ III heterozygotes. No significant differences in fasting insulin levels were observed between children homozygous for the VNTR class III marker and class I/III heterozygotes. Allele frequencies were the same as in a large cohort of lean controls, which indicated that the INS-VNTR marker was not primarily associated with juvenile obesity, at least not when classifying alleles broadly as VNTR class I and III respectively. The association between fasting insulin levels and VNTR classes was even stronger when the study was limited to the 105 and 44 morbidly obese children from the two cohorts (BMI $ 99.6th percentile). Fasting insulin levels among children with mean BMI 37 kg/ m were approximately 70% higher in homozygotes for the VNTR class I marker compared with the class I/III heterozygotes and class III homozygotes. For the whole cohort of children, Le Stunff et al. (11) demonstrated a stronger influence on the increase in fasting insulin levels secondary to overweight among VNTR class I homozygotes compared with the other genotypes. It was more marked among obese boys than girls, but the relative effect of genotype on the interaction between BMI and insulin levels was not significantly different among the sexes. BMI accounted for approximately 50% of the variance of fasting plasma insulin levels in the group homozygous for the VNTR class I marker, compared with less than 10% among the heterozygotes ISSN 0804-4643 European Journal of Endocrinology (2001) 144 457±459