Sterol regulatory element–binding protein (SREBP)-1 transcription factors play a central role in energy homeostasis by promoting glycolysis, lipogenesis, and adipogenesis. The sterol regulatory element–binding protein gene (SREBF)-1 is a good candidate gene for obesity and obesity-related metabolic traits such as type 2 diabetes and dyslipidemia. The SREBF-1 molecular screening of 40 unrelated obese patients by PCR/single-strand conformation polymorphism revealed 19 single nucleotide polymorphisms (SNPs). Six SNPs were genotyped for an association study in large French obese and nonobese cohorts. Case-control studies using two independent nonobese cohorts indicated that SNP17 (54G/C, exon 18c) is associated with morbid obesity (odds ratio 1.5, P = 0.006 and P = 0.02, respectively). SNP3 (−150G/A, exon 1a), SNP5 (−36delG, exon 1a), and SNP17 are found in high linkage disequilibrium (D′ > 0.8). The haplotype including wild-type alleles of these SNPs (C/G/G/T/C/G, HAP2) is identified as a risk factor for morbid obesity (P = 0.003). In the obese group, SNP3, SNP5, and SNP17 are associated with male-specific hypertriglyceridemia (P = 0.07, P = 0.01, and P = 0.05, respectively). SNP17 is also associated with type 2 diabetes (P = 0.03) and increased prevalence of nephropathy (P = 0.028) in a diabetic cohort. Our results indicate a role of the SREBF-1 gene in genetic predisposition of metabolic diseases such as obesity, type 2 diabetes, and dyslipidemia.
Sterol regulatory element–binding protein (SREBP) transcription factors are major regulators of carbohydrate and lipid metabolism (1,2). Three isoforms have been identified: SREBP-1a and SREBP-1c, produced from a single gene named sterol regulatory element–binding protein gene (SREBF)-1 (by specific promoters and alternative splicing), and SREBP-2, produced from a separate gene SREBF-2 (3,4). Several studies have demonstrated specific isoform functions in transactivating target genes (1). SREBP-1a activates genes of fatty acid and cholesterol synthesis pathways and is predominant in cell lines, spleen, and intestine. SREBP-1c is a weaker transcriptional activator than SREBP-1a due to its shorter transactivation domain, but it is the major isoform in liver, muscle, and adipose tissue (1,5). SREBP-1c is mainly regulated by the nutritional environment. Our group and others (2) have shown that SREBP-1c mediates the transcriptional effects of insulin on genes encoding enzymes involved in glycolysis, lipogenesis, and gluconeogenesis. The SREBP-1c rat homolog is also involved in adipogenesis (6). Together, these findings demonstrate that SREBP-1 transcription factors play a key role in energy homeostasis.
Genome scan studies have linked the 17p11 region, comprising the human SREBF-1 gene locus (3), to plasma leptin concentrations in American obese families (7) and to BMI in a combined analysis of four ethnic groups (8). A meta-analysis performed in four European genome-wide screens demonstrated linkage between the 17p11.2-q22 region and type 2 diabetes (9).
We hypothesized that genetic variations of the SREBF-1 gene could be associated with obesity and/or obesity-related metabolic traits such as insulin resistance, type 2 diabetes, and dyslipidemia. The whole SREBF-1 gene was screened in 40 unrelated obese patients for single nucleotide polymorphism (SNP) detection. Case-control and association studies were performed in large French obese, diabetic, and nonobese nondiabetic cohorts.
The present screening identified 19 polymorphisms (SNP1 through SNP19) including previously reported variants (10,11) and 15 new SNPs (Fig. 1). Five SNPs were only found in two Afro-Caribbean screened patients and were not studied in Caucasian cohorts. For the association study, we selected six SNPs based on their allelic frequency or their potential to modify SREBP-1 expression and function as follows: SNP1 (promoter 1a), SNP3 and SNP5 [5′ untranslated region (UTR) exon 1a], SNP6 (promoter 1c), SNP7 (Pro197Leu, exon 3), and SNP17 (Gly952Gly, exon 18c).
Genotypes and allelic frequencies in the French obese and SU-VI-MAX (Supplementation en Vitamines et Minéraux Antioxidant) nonobese groups are presented in Table 1. All polymorphisms are in Hardy-Weinberg equilibrium in both populations. For each SNP, we named “allele 1” the allele described in the published SREBF-1 sequence (NT010718.13), while we used “allele 2” for the newly detected allele. SNP1, SNP6, and SNP7 are characterized by low allelic frequencies (allele 2 <5%) that are not statistically different between the obese and control groups. SNP3, SNP5, and SNP17 demonstrate higher frequencies (allele 2 >50%) and show a decreased allele 2 frequency in the obese group (Table 1). SNP3 and SNP17 show a significant association with morbid obesity (P = 0.038 and P = 0.006). SNP17 is more tightly associated with obesity in a dominant allele 1 model (P = 0.002). Allele 1 carriers (genotypes 1/1 and 1/2) were more prevalent in the obese group compared with the control group (65.8 vs. 55.7%, P = 0.002; odds ratio 1.53, 95% CI 1.17–2.00). We confirmed this association by logistic regression, taking into account variables such as age and sex repartition between the obese and nonobese groups (P = 0.002).
Seven “common” haplotypes (frequency >1%) were differently distributed between the obese and nonobese groups (P = 0.027, see online appendix 1, available from http://diabetes.diabetesjournals.org). SNP3, SNP5, and SNP17 are found in strong linkage disequilibrium (all pairwise D′ values >0.8, P < 0.001). As expected, the frequency of haplotype C/G/G/T/C/G (HAP2) grouping alleles 1 of the six studied SNPs is increased in the obese group, suggesting that HAP2 might be a risk haplotype for morbid obesity (P = 0.003 in dominant HAP2 model). However, the SNP detection method used, single-strand conformation polymorphism (SSCP), does not give complete allele information; consequently, all common gene haplotypes may not have been investigated.
Considering the high linkage disequilibrium between SNP3, SNP5, and SNP17, we focused on SNP17, the most obesity-associated SNP. SNP17 frequency is again significantly different between the obese group and another independent nonobese group (control group 2, n = 247) (1/1, 1/2, and 2/2 genotypes were 15.4, 50.4, and 34.2% for obese vs. 9, 51, and 40% for nonobese, respectively; P = 0.02). Overall, we show here that SNP17 is strongly associated with morbid obesity in French Caucasians. We further evaluated SNP17 frequency in obese children. As in the adult obese group, the allele 1 frequency is increased in the child group, but this does not reach statistical significance (Table 2). Interestingly, allele 1 is associated with an early age of adiposity rebound (AAR) (mean AAR for 1/1, 1/2, and 2/2 carriers were 2.75, 2.87, and 3.22 years, respectively; P = 0.04). Six independent studies have confirmed that early AAR is associated with increased BMI in adulthood (12). SNP17 could therefore contribute to massive adult obesity without any direct effect in children. As transcription factors involved in adipocyte differentiation and lipogenesis (1), SREBP-1 could improve the individual susceptibility for fat storage capacity of adipose tissue and thus obesity development. Indeed, adipose overexpression of mature SREBP-1a in mice resulted in hypertrophy of adipocytes (13). Surprisingly, adipose overexpression of mature SREBP-1c in mice resulted in generalized lipodystrophy (14). In humans, SREBP-1c mRNA expression is decreased in adipose tissue of obese subjects compared with that of lean control subjects (15–18). Thus, the direct consequences of the gain or loss of SREBP-1 function is difficult to conclude.
We further evaluated the influence of SREBF-1 SNPs in the development of type 2 diabetes and dyslipidemia in morbidly obese patients. While SREBP-1c plays a central role in carbohydrate metabolism, mediating insulin transcriptional effects on glycolytic and lipogenic enzymes (2), the SNPs studied were not associated with type 2 diabetes prevalence or modifications of glucose/insulin parameters. It is possible that morbid obesity, known to be associated with insulin resistance (19), masks moderate SNP effects. The 17p11.2-q22 region, where the SREBF-1 gene is located, was recently linked to type 2 diabetes (9). Therefore, SNP17 was further studied in two independent diabetic cohorts: probands of French families with type 2 diabetes (group 1, n = 369) and diabetic patients examined in a diabetology department (group 2, n = 342). Here, we again observe a decreased allele 2 frequency in both diabetic groups compared with control subjects (Table 2). Statistical significance was reached in the diabetic group 2 (P = 0.03). Logistic regression, taking into account age, sex, and BMI, confirms that SNP17 is strongly associated with type 2 diabetes independently of the obesity of diabetic patients (P = 0.003). In group 1, allele 1 carriers develop type 2 diabetes earlier than noncarriers (mean age of onset 43.5 vs. 45.25 years, P = 0.05). In group 2, for which we obtained details regarding renal function, an association with nephropathy was observed for allele 1 carriers compared with noncarriers (36 vs. 26.6% of nephropathy, P = 0.028). These results highlight a contribution of SNP17 to type 2 diabetes.
Male subjects of the obese group show an association between SNP3, SNP5, and SNP17 and allele 2 and hypertriglyceridemia (Fig. 2). Allele 1 is associated with obesity and could protect from dyslipidemia by increasing adipogenesis. However, this was not observed for obese women. Total cholesterol, HDL cholesterol, apolipoprotein (apo)A1 and apoB are not significantly modified (data not shown). Previous genetic studies (10,11,20) analyzing SNP5 and SNP17 individually observed modifications in lipid parameters in predominantly male populations. SNP5 was associated with an atherogenic lipid profile in men at high cardiovascular risk (10). In a study evaluating fluvastatin treatment, patients demonstrated differentially modified lipid parameters according to their SNP5 genotype (20). Another study of highly active antiretroviral therapy (HAART) in HIV-1–infected patients showed that SNP17 allele 2 carriers had a high risk of developing hyperlipoproteinemia (11). Allele 2 carriers seem to develop dyslipidemia in particular pathophysiological contexts, suggesting that gene-gene and/or gene-environment interactions could be necessary for the phenotype expression. For example, the disruption of the LDL receptor in mice overexpressing mature SREBP-1a in liver unmasked hyperlipidemia resulting from increased production of VLDL (21). It could therefore be interesting to look for possible interaction effects with known genetic variants that affect lipoprotein metabolism. The observed male-specific dyslipidemia suggests that sex hormones could interfere with the SREBP-1 pathway. Heemers et al. (22) have shown that androgen-induced upregulation of lipogenic gene expression was accompanied by increased nuclear SREBP-1. However, whether these observations may partly explain the well-known effects of androgens on serum lipid levels is unknown. SNP1 is also associated with modifications of lipid parameters in the obese group. Rare allele 2 carriers showed decreased plasma triglycerides (P = 0.024), apoA1 (P = 0.03), and apoB (P = 0.068) concentrations compared with those of noncarriers. Analysis of haplotypes confirms that this result is not due to confounding effects of SNP3, SNP5, or SNP17 (data not shown).
SNPs linked to obesity or obesity-related traits do not directly change protein structure; however, they could influence various aspects of mRNA metabolism. Alterations of regulatory RNA-binding protein sites and mRNA secondary structures have been implicated in pathophysiological situations (23). Moreover, we cannot exclude that these SNPs are in linkage disequilibrium with unidentified functional mutations, either in the SREBF-1 gene or a gene located in that region. While additional studies are needed to analyze the potential functional role of these SNPs, our results suggest that the SREBF-1 gene may play a role in genetic predisposition to human obesity and obesity-related traits.
RESEARCH DESIGN AND METHODS
Caucasian obese subjects (BMI ≥40 kg/m2) were recruited from the Nutrition Department of the Hôtel-Dieu Hospital (France) as previously described (24). The diagnosis of type 2 diabetes was based on antidiabetic treatment and/or fasting glycemia >7 mmol/l. Diagnosis of dyslipidemia was based on hypolipidemic treatments and/or a fasting triglyceride level ≥1.71 mmol/l and/or total cholesterol level ≥6.45 mmol/l. Nephropathy was defined by elevated microalbuminuria (≥30 mg/24 h or ≥2 mg/dl for two of three samples on 3 consecutive days) or renal failure (creatinine clearance <30 ml/min). Studies were first performed on a cohort of 400 French obese unrelated Caucasians. The analysis of SNPs showing a trend or association with obesity was extended to 857 patients. The 411 French unrelated obese children were recruited at the Institut Pasteur (Lille, France). Nonobese subjects were randomly chosen from the Supplementation en Vitamines et Minéraux Antioxidant (SU-VI-MAX) cohort. We selected never-obese subjects based on weight retrospective history and 8 years’ prospective follow-up. Additional unrelated nonobese and nondiabetic adults (control group 2) were selected from the “Fleurbaix-Laventie” general population. Type 2 diabetic subjects were ascertained from probands of type 2 diabetic families recruited by the CNRS-Institut Pasteur Unit in Lille (group 1) and diabetic patients examined at the Diabetology Department of the Corbeil-Essonne Hospital (group 2). Phenotypes of subjects are available in online appendix 2. Written informed consent was obtained. The appropriate ethics committee (CCPPRB Hôtel-Dieu) approved the protocols of investigation. Metabolic parameters were measured from blood samples collected after an overnight fast.
SNP identification and genotyping.
We divided the SREBF-1 gene into 28 fragments covering proximal promoters, 22 exons, and intron-exon junctions. The conditions for PCR amplification and primer sets are available online (online appendix 3). PCR products were analyzed by PCR-SSCP as previously described (25). Each exon presenting a shift was sequenced in both directions. Different assays based on PCR/restriction fragment–length polymorphism were developed to study SNP1, SNP3, SNP5, SNP6, and SNP17 with BseYI, NaeI, ApaI, MseI, and XmnI (New England Biolabs), respectively. The genotyping of SNP7 was performed using degenerated primers 5′-CCTGCCACTGGCTTCCTCG-3′ and 5′-GGCTGTAAGCTGTGTGTCTGG-3′ and AciI. SNP17 was also genotyped by allele-specific hybridization probes (7900 HT; Applied Biosystems).
Statistical analysis.
Hardy-Weinberg equilibrium was tested by the χ2 test. Genotype frequencies and other categorical clinical variables were compared using the Pearson χ2 test or by logistic regression analysis adjusted for relevant covariates. Continuous variables were analyzed using either one-way ANOVA or Wilcoxon’s test depending on the shapes of the distribution curves. Biological data were adjusted for age, sex, BMI, and other covariates when necessary. Quantitative variables are expressed as means ± SE. The level of statistical significance was set at P < 0.05. The reconstruction of haplotypes was performed by the PHASE program. Linkage disequilibrium, evaluated by D′ coefficients, was calculated by Arlequin software (available from http://anthro.unige.ch/arlequin). D′ is the standardized linkage disequilibrium coefficient that represents the proportion of the maximum possible allelic association given the allele frequencies and the direction of association.
SNPs identified within the SREBF-1 gene. A: SNPs are located within the SREBF-1 gene, which is composed of 22 exons (black boxes). Two specific promoters and alternative splicing of 5′ and 3′ ends are used to produce SREBP-1a (exons 1a, 18a, and 19a) and SREBP-1c (exons 1c, 18c, and 19c) transcripts. Other exons are common to both isoforms. B: SNP identification (ID): SNPs identified were numbered from SNP1 to SNP19 from 5′ to 3′ of the SREBF-1 gene. In the “Description” column, due to the specific structure of the SREBF-1 gene and transcripts, SNPs are described and positioned as follows: SNPs located in the promoter and exon 1 of SREBP-1a or 1-c are positioned according to the ATG of their respective isoform (adenine nucleotide = +1), SNPs located in other exons are positioned according to the beginning of the exon (first exon nucleotide = +1). Allele found in the referenced SREBF-1 sequence (NT010718.13) is listed first (allele 1) and followed by identified allele (allele 2). For deletion, only alleles 2 are listed. In the “Type” column, because SREBP-1a and -1c proteins differ by their NH2-terminal extremity, the SREBP-1a longest protein was used to position coding mutations (first amino acid of SREBP-1a protein = +1). NC, noncoding. In the “mRNA location” column, the GenBank sequence NM_004176 corresponding to the SREBP-1a mRNA was used to position SNPs. *SNPs selected for genotyping, #SNPs only found in two Afro-Caribbean patients. NCBI db is available from http://www.ncbi.nlm.nih.gov/SNP/.
SNPs identified within the SREBF-1 gene. A: SNPs are located within the SREBF-1 gene, which is composed of 22 exons (black boxes). Two specific promoters and alternative splicing of 5′ and 3′ ends are used to produce SREBP-1a (exons 1a, 18a, and 19a) and SREBP-1c (exons 1c, 18c, and 19c) transcripts. Other exons are common to both isoforms. B: SNP identification (ID): SNPs identified were numbered from SNP1 to SNP19 from 5′ to 3′ of the SREBF-1 gene. In the “Description” column, due to the specific structure of the SREBF-1 gene and transcripts, SNPs are described and positioned as follows: SNPs located in the promoter and exon 1 of SREBP-1a or 1-c are positioned according to the ATG of their respective isoform (adenine nucleotide = +1), SNPs located in other exons are positioned according to the beginning of the exon (first exon nucleotide = +1). Allele found in the referenced SREBF-1 sequence (NT010718.13) is listed first (allele 1) and followed by identified allele (allele 2). For deletion, only alleles 2 are listed. In the “Type” column, because SREBP-1a and -1c proteins differ by their NH2-terminal extremity, the SREBP-1a longest protein was used to position coding mutations (first amino acid of SREBP-1a protein = +1). NC, noncoding. In the “mRNA location” column, the GenBank sequence NM_004176 corresponding to the SREBP-1a mRNA was used to position SNPs. *SNPs selected for genotyping, #SNPs only found in two Afro-Caribbean patients. NCBI db is available from http://www.ncbi.nlm.nih.gov/SNP/.
Plasma triglyceride concentrations in the obese men group. For each SNP or haplotype, number of patients for each genotype is indicated under the bars. All patients treated for dyslipidemia by drugs were excluded from the analysis. A linear regression statistical test was performed on log-transformed values adjusted for age, BMI, and type 2 diabetes status. P values are from the test performed on log-transformed values, but untransformed values appear in the figure. Two analytical models were considered: ¶P values from an additive model (allele dose effect) and #P values from an allele 2 or HAP1 dominant model. HAP2 is composed of allele 1 of six SNPs (C/G/G/T/C/G), and HAP1 grouped allele 2 of SNP3, SNP5, and SNP17 (C/A/delG/T/C/C). TG, triglycerides.
Plasma triglyceride concentrations in the obese men group. For each SNP or haplotype, number of patients for each genotype is indicated under the bars. All patients treated for dyslipidemia by drugs were excluded from the analysis. A linear regression statistical test was performed on log-transformed values adjusted for age, BMI, and type 2 diabetes status. P values are from the test performed on log-transformed values, but untransformed values appear in the figure. Two analytical models were considered: ¶P values from an additive model (allele dose effect) and #P values from an allele 2 or HAP1 dominant model. HAP2 is composed of allele 1 of six SNPs (C/G/G/T/C/G), and HAP1 grouped allele 2 of SNP3, SNP5, and SNP17 (C/A/delG/T/C/C). TG, triglycerides.
Genotypes and allelic frequencies of the selected SNPs
| Obese group | SU-VI-MAX nonobese group | P values | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Genotype frequency (%) | Allele 2 frequency (%) | n | Genotype frequency (%) | Allele 2 frequency (%) | Additive | Allele 1 dominant | Allele 1 recessive | |||||||||||||||
| 1/1 | 1/2 | 2/2 | 1/1 | 1/2 | 2/2 | ||||||||||||||||||
| SNP1 −688 C/T | 420 | 93.7 | 6.0 | 0.2 | 3.3 | 417 | 93.9 | 6.2 | 0 | 3.1 | 0.62 | ||||||||||||
| SNP3 −150 G/A | 807 | 20.3 | 50.3 | 29.4 | 54.6 | 342 | 14.5 | 50.3 | 35.2 | 60.4 | 0.038 | 0.060 | 0.025 | ||||||||||
| SNP5 −36delG | 807 | 17.6 | 48.8 | 33.6 | 58.0 | 342 | 13.8 | 47.2 | 39.0 | 62.6 | 0.14 | 0.088 | 0.13 | ||||||||||
| SNP6 −435 T/C | 857 | 95.5 | 4.5 | 0 | 2.3 | 467 | 97.1 | 2.9 | 0 | 1.4 | 0.18 | ||||||||||||
| SNP7 68 C/T | 433 | 98.1 | 1.9 | 0 | 0.93 | 426 | 98.2 | 1.8 | 0 | 0.89 | 0.93 | ||||||||||||
| SNP17 54 G/C | 807 | 15.4 | 50.4 | 34.2 | 59.4 | 342 | 12.0 | 43.7 | 44.3 | 66.2 | 0.006 | 0.002 | 0.14 | ||||||||||
| Obese group | SU-VI-MAX nonobese group | P values | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Genotype frequency (%) | Allele 2 frequency (%) | n | Genotype frequency (%) | Allele 2 frequency (%) | Additive | Allele 1 dominant | Allele 1 recessive | |||||||||||||||
| 1/1 | 1/2 | 2/2 | 1/1 | 1/2 | 2/2 | ||||||||||||||||||
| SNP1 −688 C/T | 420 | 93.7 | 6.0 | 0.2 | 3.3 | 417 | 93.9 | 6.2 | 0 | 3.1 | 0.62 | ||||||||||||
| SNP3 −150 G/A | 807 | 20.3 | 50.3 | 29.4 | 54.6 | 342 | 14.5 | 50.3 | 35.2 | 60.4 | 0.038 | 0.060 | 0.025 | ||||||||||
| SNP5 −36delG | 807 | 17.6 | 48.8 | 33.6 | 58.0 | 342 | 13.8 | 47.2 | 39.0 | 62.6 | 0.14 | 0.088 | 0.13 | ||||||||||
| SNP6 −435 T/C | 857 | 95.5 | 4.5 | 0 | 2.3 | 467 | 97.1 | 2.9 | 0 | 1.4 | 0.18 | ||||||||||||
| SNP7 68 C/T | 433 | 98.1 | 1.9 | 0 | 0.93 | 426 | 98.2 | 1.8 | 0 | 0.89 | 0.93 | ||||||||||||
| SNP17 54 G/C | 807 | 15.4 | 50.4 | 34.2 | 59.4 | 342 | 12.0 | 43.7 | 44.3 | 66.2 | 0.006 | 0.002 | 0.14 | ||||||||||
The SNPs were genotyped in a first cohort of 400 French unrelated obese Caucasians. Then, for validation purposes, the genotyping of SNPs showing association was extended to a maximum 857 subjects. Three analytical models were considered to test association between SNPs and morbid obesity: an additive model (allele dose effect) and a dominant or recessive effect for allele 1. Pearson χ2 tests were performed on genotype data. Significant P values are in boldface type.
SNP17 genotypes and allelic frequencies in additional cohorts
| Obese or diabetic groups | SU-VI-MAX + control group 2 | P values | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Genotypes frequency (%) | Allele 2 frequency (%) | n | Genotypes frequency (%) | Allele 2 frequency (%) | Additive | Allele 1 dominant | Allele 1 recessive | |||||||||||||||
| 1/1 | 1/2 | 2/2 | 1/1 | 1/2 | 2/2 | ||||||||||||||||||
| Adult obese group | 807 | 15.4 | 50.4 | 34.2 | 59.4 | 589 | 10.5 | 46.9 | 42.6 | 66.0 | 0.001 | 0.001 | 0.008 | ||||||||||
| Child obese group | 411 | 13.9 | 45.7 | 40.4 | 63.2 | 0.190 | 0.110 | 0.110 | |||||||||||||||
| Diabetic group 1 | 369 | 15.7 | 43.6 | 40.7 | 62.0 | 0.110 | 0.110 | 0.018 | |||||||||||||||
| Diabetic group 2 | 342 | 15.0 | 44.0 | 41.0 | 61.0 | 0.036 | 0.037 | 0.019 | |||||||||||||||
| Obese or diabetic groups | SU-VI-MAX + control group 2 | P values | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Genotypes frequency (%) | Allele 2 frequency (%) | n | Genotypes frequency (%) | Allele 2 frequency (%) | Additive | Allele 1 dominant | Allele 1 recessive | |||||||||||||||
| 1/1 | 1/2 | 2/2 | 1/1 | 1/2 | 2/2 | ||||||||||||||||||
| Adult obese group | 807 | 15.4 | 50.4 | 34.2 | 59.4 | 589 | 10.5 | 46.9 | 42.6 | 66.0 | 0.001 | 0.001 | 0.008 | ||||||||||
| Child obese group | 411 | 13.9 | 45.7 | 40.4 | 63.2 | 0.190 | 0.110 | 0.110 | |||||||||||||||
| Diabetic group 1 | 369 | 15.7 | 43.6 | 40.7 | 62.0 | 0.110 | 0.110 | 0.018 | |||||||||||||||
| Diabetic group 2 | 342 | 15.0 | 44.0 | 41.0 | 61.0 | 0.036 | 0.037 | 0.019 | |||||||||||||||
The two independent nonobese groups were pooled to form a single control group, as they were not statistically different (P = 0.2). Each obese or diabetic group was compared with the pooled control group. P values correspond to Pearson χ2 tests performed on genotype data. Significant P values are in boldface type.
Additional information for this article is available in an online appendix found at http://diabetes.diabetesjournals.org.
Article Information
This work was supported by the Direction de la Recherche Clinique/Assistance Publique-Hôpitaux de Paris, the Programmes Hospitaliers de Recherche Clinique (AOM 96088 and AOR 02076), the Servier Research Institute (IRIS), the Claude Bernard Association, and INSERM. D.E. received a grant from the Ministère de l’Education nationale et de la Recherche.
We greatly thank Geneviève Bonhomme for technical help. We are indebted to Serge Hercberg and Pilar Galan for giving access to the Danone/SU-VI-MAX DNA Bank, Dr. Aline Charles for access to the Fleurbaix-Laventie database, Dr. Guillaume Charpentier for access to the Corbeil patient cohort, J. Weill from the Pediatric Endocrinology Department of the Hôpital Jeanne de Flandre in Lille, and Dr. M. Tauber from the Endocrinology Department of the Hôpital des Enfants in Toulouse. The authors thank Dr. Bronwyn Hegarty for critical reading of the manuscript.
