Quantitative Trait Loci for Fasting Glucose in Young Europeans Replicate Previous Findings for Type 2 Diabetes in 2q23–24 and Other Locations

Since fat accumulation affects glucose homeostasis, obese patients are at risk for impaired glucose tolerance, impaired fasting glucose, and type 2 diabetes. This risk is, however, widely variable across obese patients. While almost all obese adolescents have normal fasting glycemia (1), many but not all develop type 2 diabetes in later adulthood (1,2). Even a slight elevation of fasting glucose seems a strong long-term predictor of type 2 diabetes in young obese (3) and in nonobese (4) adults. In the general population, heritability for fasting glucose varies between 17 and 51% (5,6). Understanding the genetic predisposition to obesity-related hyperglycemia can be facilitated if genetic and phenotypic heterogeneity are limited by studying relatively homogeneous populations and if reliable phenotypes predicting type 2 diabetes (7) are measured in strictly controlled clinical conditions. To localize quantitative trait loci (QTLs) implicated in fasting glucose variability, we carried out a classic microsatellite-based genome-wide scan in 220 multiplex families (964 individuals and 412 sib-pairs) having a relatively high prevalence of obesity and type 2 diabetes.

Plasma glucose concentration is one of the most exquisitely regulated metabolic traits in humans, and its variations are known to be more dependent on nutritional and physiological conditions than on interindividual variance. Beyond the risk of obesity-related type 2 diabetes, there are thus several reasons for electing to study the quantitative genetics of fasting glucose in young obese patients who have been gaining weight continuously and have stayed in the hospital for 3 days' initial check-up. First, their glucose homeostasis can be studied in stable conditions, including minimization of puncture stress, controlled diet, lack of recent dietary or weight changes, normal physical activity, and precise duration of fasting, a protocol aiming at reducing the ratio of experimental variance to genetic variance of glycemia. Studying the dynamic phase of obesity is important for the quality of the phenotype because later evolution of obesity is exposed to medical or dietary interventions and changing fatness, which can alter the natural history of the trait, as well as to metabolic and hormonal dysfunctions created by the long-term exposure to the obese status itself and aging.

Youth is a period of life favorable to the research of genetic factors implicated in the normal variation of glycemia, since evolutionary selection of many metabolic processes of physiological importance take place during youth; these include energy metabolism during pregnancy and lactation, fetal growth, neonatal survival, adaptation to episodic starvation, and provision of fuel to the large human brain mass. Darwinian views postulate that selection forces for fitness and survival are exerted on young individuals until they reproduce. Therefore, studying individual genotype-phenotype variations during youth matches to our research goals.

In addition, juvenile obesity provides us with the opportunity to maximize the interindividual range of variation of fasting glycemia. We thought that maximizing the variance of the studied phenotype was propitious to genetic association studies. Using the birth places of grandparents, we were also careful at selecting individuals from closely related European geographic areas, which minimizes demographic variations in the genetic background.

The 740 children or adolescents included in the linkage analysis were 13.2 ± 0.2 years old and had a BMI of 20.5 ± 0.4 kg/m². Fifty-one percent were female. Fasting glucose showed a near-normal continuous distribution with a mean value of 4.57 ± 0.62 mmol/l compared with 4.53 ± 0.48 mmol/l in 283 nonobese children from the same population background. There was no significant difference between boys (4.59 ± 0.54 mmol/l) and girls (4.56 ± 0.68 mmol/l). Twelve obese children had impaired fasting glucose with fasting glucose levels between 6.1 and 6.9 mmol/l. Fasting glucose showed a widely dispersed correlation with BMI (r = 0.23, P < 0.001) and with age (r = 0.194, P < 0.001), while pubertal maturation did not change fasting glucose in obese adolescents.

We conducted separate and adjusted genome scans for the two quantitative traits, fasting glucose and BMI. Results of linkage analysis are shown in Fig. 1 and Table 1. A total of 10,000 simulations indicated that six peaks with a logarithm of odds (LOD) score ≥1 would be expected by chance only, one with a LOD score ≥2, and none with a LOD score ≥3, while the current genome scan for fasting glucose found five loci with a LOD score between 1 and 2, two loci with a LOD score between 2 and 3, and one locus with a LOD score of 3.

We detected no loci that could be QTLs for current BMI, since no LOD score for this trait was >1, nor did we replicate reports of regions linked to BMI in Caucasian adults (8–11) or children (12–14).

We found several QTLs for fasting glucose that have no detectable effect on BMI. The strongest evidence for linkage to fasting glucose was on 2q with an unadjusted LOD score of 1.78 on 2q23 (D2S142) near NIDDM1 (D2S125–D2S140). This microsatellite revealed links with fasting glucose since the LOD score increased to 3.00 after adjustment to BMI (1.78 to 3.00). This QTL has previously been identified as a type 2 diabetes locus in French (15) and Australians of European origin (16) (Table 1). This consistency suggests the location of one or several glucose QTLs in this vast region. We also identified two other QTLs specific to fasting glucose on 17p13.3 (LOD = 2.12 on marker D17S831) and 17q12 (LOD = 1.40 on marker D17S798).

Another category of QTLs influence both glycemia and BMI. On chromosome 7q, we found peak evidence for linkage of fasting glucose to D7S684 (7q34, LOD = 2.03) and to D7S661 (7q35–q36, LOD = 1.7), within an ∼10-cM interval (Table 1). On chromosome 8, a LOD score of 1.28 was found in D8S514 (8q24). Linkage to type 2 diabetes was previously found in this region (Table 1) (17). On chromosome 11, we found nominal evidence for linkage of fasting glucose at marker D11S4046 (11p15, LOD = 1.33), another region previously linked to type 2 diabetes (18) (Table 1).

In conclusion, this genome-wide analysis supports the existence of QTLs for fasting glucose on the long arm of chromosomes 2, 7, 8, and 17 and the short arm of chromosomes 11 and 17. In the absence of highly significant scores of linkage—not an unexpected result for a complex and highly regulated trait—these observations call for replication of linkage, including additional studies in Caucasian Europeans or in different populations (Table 1). It should be remembered, also, that whole genome linkage or association is underpowered for identifying common genetic variants that have modest effects on diseases or traits (19), in contrast with locus-specific association (20). The glucose QTLs reported here could be of relevance in only obese juveniles, in whom it may help to predict type 2 diabetes, and of minor if any importance in the nonobese population. That our QTLs replicate several type 2 diabetes localizations in cohorts of European ancestry (Table 1) supports the validity of a linkage analysis of glycemia as a quantitative surrogate for type 2 diabetes. That other QTLs for common type 2 diabetes were not found in our young obese study may be because additional genetic factors are necessary to generate susceptibility to type 2 diabetes later in life. In summary, our study suggests that the complex predisposition of juvenile obesity to type 2 diabetes might benefit from the study of continuous glucose–or insulin-related traits.