Recently, common noncoding variants in the TCF7L2 gene were strongly associated with increased risk of type 2 diabetes in samples from Iceland, Denmark, and the U.S. We genotyped 13 single nucleotide polymorphisms (SNPs) across TCF7L2 in 8,310 individuals in family-based and case-control designs from Scandinavia, Poland, and the U.S. We convincingly confirmed the previous association of TCF7L2 SNPs with the risk of type 2 diabetes (rs7903146T odds ratio 1.40 [95% CI 1.30–1.50], P = 6.74 × 10−20). In nondiabetic individuals, the risk genotypes were associated with a substantial reduction in the insulinogenic index derived from an oral glucose tolerance test (risk allele homozygotes have half the insulin response to glucose of noncarriers, P = 0.003) but not with increased insulin resistance. These results suggest that TCF7L2 variants may act through insulin secretion to increase the risk of type 2 diabetes.

Type 2 diabetes is highly heritable, but known variants explain only a small fraction of the overall genetic risk in the population. Recently, Grant et al. (1) reported a strong association of variants in TCF7L2 with increased risk of type 2 diabetes in an Icelandic sample, and this association was confirmed in Caucasian samples from Denmark and the U.S. (combined odds ratio [OR] 1.56, P = 4.7 × 10−18). Testing of this association in other well-phenotyped samples is needed to 1) validate the association, 2) estimate the true effect size, and 3) identify effects on intermediate traits that may suggest how TCF7L2 variants act (e.g., through changes in insulin secretion, insulin resistance, BMI, waist-to-hip ratio).

TCF7L2 has been implicated as a member of the Wnt signaling pathway and was previously well studied only in colon cancer. However, based on its role in intestinal cells (2), Grant et al. (1) proposed that variants of TCF7L2 may alter levels of glucagon-like peptide 1, which influences insulin secretion from the β-cells of the pancreas. Thus, one hypothesis is that TCF7L2 might influence the risk of type 2 diabetes by influencing insulin secretion. Alternatively, a gene increasing the risk of diabetes could act through insulin action or through currently unknown mechanisms.

To evaluate these questions, we selected tag SNPs to capture common variation in a 64.6-kb region of strong linkage disequilibrium surrounding the most significant association signal and spanning intron 3, exon 4, and intron 4 of TCF7L2. We genotyped 13 tag SNPs that capture 32 of 44 variants with r2 > 0.8 (mean r2 > 0.985) in the phase II HapMap CEU population (3); all 5 SNPs that were most highly correlated with the DG10S478 allele X in the original report (1) were directly genotyped.

The tag SNPs were genotyped in well-characterized family-based and case-control samples from Scandinavia, Poland, and the U.S.; phenotypic characteristics of all samples are described in Table 1. We included five previously described patient samples that have formed the basis of multiple previous publications from our research group: 333 Swedish and Finnish trios; 2 Scandinavian case-control samples with 918 and 1,010 subjects, respectively; a Polish case-control sample with 2,018 subjects; and a U.S. Caucasian case-control sample with 2,452 individuals (4–11). A smaller fraction of the samples studied have not previously been described: 1) a case-control sample (444 subjects) and 130 discordant sibpairs from Botnia (a Swedish-speaking isolate of Finland) and 2) a case-control sample (266 subjects) and 106 discordant sibpairs from Sweden and Finland. All Scandinavian case-control samples were matched for age, sex, BMI, and geographic region (described in more detail in research design and methods).

Association of SNPs in TCF7L2 with type 2 diabetes was strongly confirmed in these samples (Table 2 and online appendix Table 1 [available at http://diabetes.diabetesjournals.org]). We found that rs7903146 was most significantly associated with the risk of type 2 diabetes (OR 1.40 [95% CI 1.30–1.50], P = 6.74 × 10−20) in agreement with the original report (best SNP) (1). Heterozygous and homozygous carriers of the risk allele had genotype relative risks of 1.40 (1.27–1.55) (P = 3.22 × 10−11) and 1.86 (1.55–2.23) (P = 1.38 × 10−11) relative to noncarriers, which is consistent with an additive model. Overall, given that the original report was highly significant (P = 10−18) (1), our results provided an independent P value of 10−20, and Groves et al. (unpublished observations) observed a P value of 10−14; this association is by far the most convincing and broadly relevant risk factor for type 2 diabetes yet found in the human population.

The strongest single variant (rs7903146) was individually significant (P < 0.05) in the six largest samples (including all previously published trio and case-control samples) and trended in the expected direction in the three smaller remaining samples (Table 3). We have previously published a lack of association to many candidate genes in these samples (4–8,10,11); replication of TCF7L2 association in each subsample provides a positive control for those previous studies, clearly demonstrating that the samples can be used to distinguish true diabetes genes of this magnitude and robustness from statistical fluctuations. To test if the best result was the only signal of association observed at this locus, we performed logistic regression analysis conditional on rs7903146. No additional signal of association was observed (data not shown), suggesting that the entire signal observed stems from rs7903146 or from a closely correlated variant.

We tested for epistasis between TCF7L2 rs7903146 and two other common variants known to be causally associated with the risk of type 2 diabetes: peroxisome proliferator–activated receptor γ P12A and Kir6.2 E23K (4,5,12). No significant genetic interactions were seen (online appendix Table 2).

We next examined the correlation of TCF7L2 genotypes with covariates of sex, age of onset, and BMI in case and control subjects to test if TCF7L2 variants contribute to risk of type 2 diabetes through an effect on these covariates. No heterogeneity by sex was observed in the case-control samples (male [n = 3,288] OR 1.46 [95% CI 1.31–1.62], P = 8.75 × 10−12 and female [n = 3,424] 1.32 [1.19–1.47], P = 3.79 × 10−7; Phomogeneity = 0.20). Furthermore, in case subjects, TCF7L2 genotypes did not associate with the age of onset of diabetes (n = 1,856, mean [±SD] TT = 55 ± 11, TC = 55 ± 12, and CC = 56 ± 12; P = 0.64). In addition, no significant association to BMI was observed (Table 4). Thus, while some variants may prove to be heterogeneous, associated only in substrata of sex, BMI, or other genotypes, common variation in TCF7L2 is associated with type 2 diabetes across the measured covariates.

The order and relative contributions of defects in insulin resistance and insulin secretion to the pathogenesis of type 2 diabetes remain controversial (13,14). Some postulate that type 2 diabetes is caused primarily by defects in insulin resistance, followed by a failure of pancreatic β-cells to compensate for increased insulin demand (15). Conversely, type 2 diabetes genes identified thus far (the maturity-onset diabetes of the young genes that cause monogenic forms of type 2 diabetes and Kir6.2) act through reduced insulin secretion without insulin resistance in carriers before the onset of disease (16,17). Each additional gene that is truly associated with type 2 diabetes helps inform the relative contributions of these two mechanisms.

We found no effect of TCF7L2 rs7903146 on oral glucose tolerance test (OGTT)-derived measures of insulin resistance in 995 nondiabetic individuals (Table 4). We did see a dramatic effect, however, on insulin secretion as measured by the OGTT (Table 4). The OGTT provides rough measures of insulin secretion and resistance and has been widely used in clinical investigations of type 2 diabetes. Individuals homozygous for the rs7903146 risk allele have a significant 50% reduction in insulinogenic index (P = 0.003) and insulin disposition index (P = 0.004). We also observed a significant reduction in the area under the curve (AUC) for insulin during the OGTT (P = 0.0007) and AUCinsulin/AUCglucose (P = 0.002), suggesting that the polymorphism not only influences the early insulin response to glucose but also could have an effect on the capacity of the β-cells to secrete insulin. These associations are stronger in risk allele homozygotes than in heterozygous carriers, and a trend is seen in heterozygous carriers compared with noncarriers.

Our results replicate the strong association of TCF7L2 variants with the risk of type 2 diabetes. The risk model for the most significantly associated SNP, rs7903146, was slightly weaker than the original report, perhaps because the magnitude of the risk effect may have been overestimated by Grant et al. (1), which was expected due to the winners curse. Consistent with this, Groves et al. (unpublished observations) estimate an effect size similar to that in our study (OR 1.36 and 1.40, in both replications, vs. 1.54 for the same SNP in the original report). Nevertheless, TCF7L2 variation contributes more powerfully to increased risk of type 2 diabetes than any other gene identified thus far. Consistent replication across European populations confirms that the causal TCF7L2 variant influences disease risk reproducibly, without the need to yet invoke population-specific effects.

We observed an insulin secretion defect in nondiabetic individuals homozygous for risk alleles of TCF7L2, suggesting that as in maturity-onset diabetes of the young and the common E23K polymorphism in Kir6.2, the primary defect attributable to a common variation in the TCF7L2 region is reduced insulin secretion rather than insulin resistance (12,16,17). However, OGTT measures used here provides only rough estimates of insulin secretion, and follow-up work will be necessary to understand the nature of the defect in insulin secretion and any possible effects on insulin action.

TCF7L2 has not previously been implicated in type 2 diabetes and would not have been an obvious diabetes gene candidate. (A PubMed search for the keywords “TCF7L2” or “TCF4” reveals 218 articles but none that share the term “diabetes” before the Grant et al. [1] publication.) The discovery of this gene reinforces that type 2 diabetes is an endocrine disease with its origin in multiple organ systems, now possibly including the intestine. Post hoc, because TCF7L2 activates glucagon-like peptide 1 in a cell-specific manner, a putative mechanism to influence blood glucose homeostasis could be proposed, and reduced insulin secretion observed here is consistent with this mechanism (1). However, the causal variant or functional defect in TCF7L2 has not yet been found. (It could be rs7903146 itself, a proxy within a broader region of this gene, or a proxy even in an adjacent gene.) More extensive genotyping and sequencing is clearly warranted, as are functional studies of the most associated alleles to document that they function through TCF7L2 rather than some adjacent gene.

The identification of this gene has interesting implications for the several diabetes whole-genome association studies planned in the coming year. There has been much speculation that common variants responsible for common diseases such as type 2 diabetes, at most, exert extremely modest effects, owing to an underlying complex genetic architecture (18–20). Some have even questioned whether disease-influencing common variants exist or can be found with linkage disequilibrium approaches (21,22). Grant et al. (1) discovered an association to TCF7L2 during a follow-up of a putative linkage peak, but as they state, the finding is likely coincidental since “the median allele-sharing LOD score generated with our previous familial samples is less than 0.1.” That is, this variant could not possibly have generated the linkage peak but could be (and was) found by systematic studies of common variation for the association with type 2 diabetes. Unless TCF7L2 is the only such gene in the genome, which is unlikely given that a focused search of ∼10.5 Mb led coincidentally to its discovery, more diabetes genes of similar effect are likely to be found by ongoing whole-genome association studies.

The diabetic samples include a previously described sample of 333 Scandinavian parent-offspring trios (163 offspring with impaired fasting glucose, 36 with impaired glucose tolerance [IGT], and 134 with type 2 diabetes) (4) and two previously described Scandinavian case-control samples consisting of 954 and 1,028 subjects, individually matched for age, sex, BMI, and geographic region of origin (4,5). World Health Organization 1998 definitions of type 2 diabetes, IGT, impaired fasting glucose, and normal glucose tolerance (NGT) were used for these samples, and severe IGT was defined as 2-h OGTT plasma glucose >8.5 mmol/l but <10 mmol/l. This study also includes two previously described case-control samples from Poland (2,018 subjects) and the U.S. (2,452 subjects) individually matched for sex, age, and geographic origin, both collected by Genomics Collaborative (6,11). For these Polish and U.S. samples, case subjects met the American Diabetes Association 2003 criteria for type 2 diabetes, and control subjects had fasting plasma glucose <7 mmol/l. Finally, the study included four newly selected Scandinavian samples: 1) a case-control sample (444 subjects) from Botnia (a Swedish-speaking isolate of Finland) individually matched by age, sex, and BMI; 2) 130 sibpairs from Botnia, discordant for type 2 diabetes; 3) a case-control sample from Sweden and Finland (266 subjects) individually matched by age, sex, BMI, and geographic origin; and 4) 106 sibpairs discordant for type 2 diabetes from Sweden and Finland. For the new case-control and family-based samples, diabetic case subjects were defined by the American Diabetes Association 2003 criteria, had an age of onset ≥35 years, and were GAD antibody negative. Control subjects had NGT (fasting plasma glucose <6.1 mmol/l and 2-h OGTT plasma glucose <7.8 mmol/l). Age matching required that control subjects be normal glucose tolerant at age <5 years from onset age of matched case. For discordant sibpairs, the youngest sibling who fulfilled the case inclusion criteria was matched with the eldest normal glucose tolerant sibling. Case and control subjects were recruited from the previously described Botnia Study, which includes families from the Botnia region on the western coast of Finland and families from other parts of Finland and Sweden (23). All patient samples were approved by the human subject institutional review board at respective institutions, and informed consent was obtained from all subjects. Insulin measures during the OGTT were available for a subset of individuals as previously described (7,11); genotype-phenotype correlation was examined for rs7903146 with fasting insulin, insulinogenic index as a measure of early insulin response to glucose ([Ins30 − fasting insulin]/[Gluc30 − fasting glucose]), homeostatis model assessment of insulin resistance ([fasting glucose × fasting insulin]/22.5), insulin disposition index (insulinogenic index × homeostatis model assessment of insulin resistance), and AUCinsulin, AUCglucose, and AUCinsulin/AUCglucose (AUCs determined by the trapezoidal method).

Genotyping.

Patient DNA was isolated from whole blood, whole genome amplified using REPLI-G (Qiagen), and purified using the Nucleofast (Clontech). Genotyping was performed by primer extension of multiplex products with detection by MALDI-TOF mass spectroscopy using a Sequenom platform. Genotyping success rate was 99%, and concordance rate, based on 889 duplicate comparisons for each of the 13 SNPs, was 99.58%.

Statistical analysis.

Tag SNPs were selected using Tagger (24); 12 SNPs were untested. Simple χ2 analysis was used to test the association of SNPs with type 2 diabetes in the matched case-control subjects, a transmission disequilibrium test was performed in the trios (25), and the discordant allele test was carried out in the sibpairs (26). Results from each sample were combined by Mantel-Haenszel meta-analysis of ORs; homogeneity was tested using the Breslow Day test, and no heterogeneity was found. Logistic regression for each SNP with diabetes-affected status was performed conditionally on rs7903146 using Whap (http://pngu.mgh.harvard.edu/∼purcell/whap). For epistasis analysis, pairwise combinations of SNPs rs7903146 with peroxisome proliferator–activated receptor γ P12A and Kir6.2 E23K were tested for association with type 2 diabetes using PLINK (http://pngu.mgh.harvard.edu/∼purcell/plink). Log-transformed quantitative traits were compared by ANOVA across three genotypic classes of rs7903146; two-tailed t tests were performed for other models/risk estimates. Unadjusted Pnominal values are reported; results were similar after adjustment for sex, age, and BMI.

TABLE 1

Clinical characteristics of study samples

SampleSex (M/F)Age (years)BMI (kg/m2)Fasting plasma glucose (mmol/l)Plasma glucose at 2-h OGTT (mmol/l)Reference
Scandinavian trios       
    Diabetes/IGT/IFG 176/157 39 ± 9 27 ± 5 7.2 ± 2.6 8.5 ± 2.9 4  
    NGT 187/192 31 ± 10 24 ± 5 5.2 ± 0.5 5.6 ± 1.1  
Scandinavian C/C       
    Diabetes/sIGT 247/212 61 ± 10 28 ± 5 9.8 ± 3.4 15.0 ± 5.3 4  
    NGT 247/212 60 ± 10 26 ± 4 6.2 ± 1.8 6.8 ± 2.8  
Swedish C/C       
    Diabetes/sIGT 267/247 66 ± 12 28 ± 4 8.5 ± 2.5 ND* 5  
    NGT 267/247 66 ± 12 28 ± 4 4.8 ± 0.7 ND  
Genomics Collaborative Poland C/C       
    Diabetes 422/587 62 ± 10 30 ± 5 8.9 ± 4.0 ND† 6  
    NGT 422/587 59 ± 7 26 ± 4 4.8 ± 1.2 ND  
Genomics Collaborative U.S. C/C       
    Diabetes 644/582 63 ± 11 33 ± 7 9.8 ± 3.0 ND‡ 6  
    NGT 644/582 61 ± 10 27 ± 5 5.1 ± 0.9 ND  
Botnia C/C       
    Diabetes 101/121 65 ± 10 28 ± 4 9.1 ± 2.8 14.7 ± 5.1 New 
    NGT 101/121 57 ± 10 27 ± 4 5.2 ± 0.5 5.2 ± 1.4  
Swedish/Finnish C/C       
    Diabetes 63/70 66 ± 10 28 ± 4 9.6 ± 3.1 16.3 ± 4.4 New 
    NGT 63/70 55 ± 9 26 ± 4 5.3 ± 0.6 5.6 ± 1.0  
Swedish/Finnish sibs       
    Diabetes 57/49 60 ± 11 30 ± 5 10.8 ± 3.7 15.2 ± 4.8 New 
    NGT 40/66 61 ± 12 27 ± 4 5.3 ± 0.4 6.1 ± 1.0  
Botnia sibs       
    Diabetes 55/75 64 ± 12 28 ± 5 9.0 ± 2.7 14.8 ± 5.3 New 
    NGT 56/74 62 ± 12 26 ± 4 5.5 ± 0.5 6.1 ± 1.2  
SampleSex (M/F)Age (years)BMI (kg/m2)Fasting plasma glucose (mmol/l)Plasma glucose at 2-h OGTT (mmol/l)Reference
Scandinavian trios       
    Diabetes/IGT/IFG 176/157 39 ± 9 27 ± 5 7.2 ± 2.6 8.5 ± 2.9 4  
    NGT 187/192 31 ± 10 24 ± 5 5.2 ± 0.5 5.6 ± 1.1  
Scandinavian C/C       
    Diabetes/sIGT 247/212 61 ± 10 28 ± 5 9.8 ± 3.4 15.0 ± 5.3 4  
    NGT 247/212 60 ± 10 26 ± 4 6.2 ± 1.8 6.8 ± 2.8  
Swedish C/C       
    Diabetes/sIGT 267/247 66 ± 12 28 ± 4 8.5 ± 2.5 ND* 5  
    NGT 267/247 66 ± 12 28 ± 4 4.8 ± 0.7 ND  
Genomics Collaborative Poland C/C       
    Diabetes 422/587 62 ± 10 30 ± 5 8.9 ± 4.0 ND† 6  
    NGT 422/587 59 ± 7 26 ± 4 4.8 ± 1.2 ND  
Genomics Collaborative U.S. C/C       
    Diabetes 644/582 63 ± 11 33 ± 7 9.8 ± 3.0 ND‡ 6  
    NGT 644/582 61 ± 10 27 ± 5 5.1 ± 0.9 ND  
Botnia C/C       
    Diabetes 101/121 65 ± 10 28 ± 4 9.1 ± 2.8 14.7 ± 5.1 New 
    NGT 101/121 57 ± 10 27 ± 4 5.2 ± 0.5 5.2 ± 1.4  
Swedish/Finnish C/C       
    Diabetes 63/70 66 ± 10 28 ± 4 9.6 ± 3.1 16.3 ± 4.4 New 
    NGT 63/70 55 ± 9 26 ± 4 5.3 ± 0.6 5.6 ± 1.0  
Swedish/Finnish sibs       
    Diabetes 57/49 60 ± 11 30 ± 5 10.8 ± 3.7 15.2 ± 4.8 New 
    NGT 40/66 61 ± 12 27 ± 4 5.3 ± 0.4 6.1 ± 1.0  
Botnia sibs       
    Diabetes 55/75 64 ± 12 28 ± 5 9.0 ± 2.7 14.8 ± 5.3 New 
    NGT 56/74 62 ± 12 26 ± 4 5.5 ± 0.5 6.1 ± 1.2  

Data are means ± SD. Plasma glucose was measured at fasting and 2 h after an OGTT.

*

HbA1c (A1C) = 6.5 ± 1.5%;

†

A1C = 7.9 ± 1.3%;

‡

A1C = 8.0 ± 3.1%. C/C, case/control subjects; ND, not determined; sIGT, severe IGT.

View Large
TABLE 2

Association of TCF7L2 SNP rs7903146 with risk of type 2 diabetes and model-free estimates of genotype relative risks

AlleleOR (95% CI)P
Total sample* T 1.40 (1.30–1.50) 6.74 × 10−20 
Case/control subjects only† T 1.39 (1.29–1.50) 1.55 × 10−17 
 		Genotype
 		Genotype relative risks (95% CI)
 		P
 
Heterozygotes† TC vs. CC 1.40 (1.27–1.55) 3.22 × 10−11 
Homozygotes† TT vs. CC 1.86 (1.55–2.23) 1.38 × 10−11 
Homozygotes vs. Heterozygotes† TT vs. TC 1.30 (1.08–1.57) 0.00269 
AlleleOR (95% CI)P
Total sample* T 1.40 (1.30–1.50) 6.74 × 10−20 
Case/control subjects only† T 1.39 (1.29–1.50) 1.55 × 10−17 
 		Genotype
 		Genotype relative risks (95% CI)
 		P
 
Heterozygotes† TC vs. CC 1.40 (1.27–1.55) 3.22 × 10−11 
Homozygotes† TT vs. CC 1.86 (1.55–2.23) 1.38 × 10−11 
Homozygotes vs. Heterozygotes† TT vs. TC 1.30 (1.08–1.57) 0.00269 

Genotype relative risks are relative to noncarriers or to heterozygote carriers as indidcated. One-tailed P value:

*

results from all 9 samples or

†

6 case-control samples combined by Mantel-Haenszel meta-analysis.

View Large
TABLE 3

Association of rs7903146 with type 2 diabetes in each subsample

SamplenOR (95% CI)One-tailed P valueMAF
Case-control group     
    Scandinavian 946 1.27 (1.03–1.58) 0.0277 0.23 
    Swedish 966 1.45 (1.18–1.77) 2.84 × 10−4 0.28 
    Genomics Collaborative Polish 1,942 1.38 (1.20–1.59) 1.04 × 10−5 0.27 
    Genomics Collaborative U.S. 2,246 1.45 (1.27–1.64) 1.54 × 10−8 0.32 
    Botnia 430 1.47 (1.06–2.03) 0.0195 0.22 
    Swedish/Finnish 260 1.02 (0.69–1.51) 0.921 0.25 
    All* 6,790 1.39 (1.29–1.50) 1.55 × 10−17 0.25 
Family-based group     
    Botnia sibs 260 1.83 (0.75–1.63) 0.0863  
    Swedish/Finnish sibs 212 1.56 (0.84–2.91) 0.160  
    Scandinavian trios† 756 1.42 (1.09–1.86) 0.0101  
    All 1,228 1.48 (1.17–1.87) 4.69 × 10−4  
Combined (all samples)* 8,018 1.40 (1.30–1.50) 6.74 × 10−20  
SamplenOR (95% CI)One-tailed P valueMAF
Case-control group     
    Scandinavian 946 1.27 (1.03–1.58) 0.0277 0.23 
    Swedish 966 1.45 (1.18–1.77) 2.84 × 10−4 0.28 
    Genomics Collaborative Polish 1,942 1.38 (1.20–1.59) 1.04 × 10−5 0.27 
    Genomics Collaborative U.S. 2,246 1.45 (1.27–1.64) 1.54 × 10−8 0.32 
    Botnia 430 1.47 (1.06–2.03) 0.0195 0.22 
    Swedish/Finnish 260 1.02 (0.69–1.51) 0.921 0.25 
    All* 6,790 1.39 (1.29–1.50) 1.55 × 10−17 0.25 
Family-based group     
    Botnia sibs 260 1.83 (0.75–1.63) 0.0863  
    Swedish/Finnish sibs 212 1.56 (0.84–2.91) 0.160  
    Scandinavian trios† 756 1.42 (1.09–1.86) 0.0101  
    All 1,228 1.48 (1.17–1.87) 4.69 × 10−4  
Combined (all samples)* 8,018 1.40 (1.30–1.50) 6.74 × 10−20  
*

Data combined by Mantel-Haenszel meta-analysis; no heterogeneity was observed (P = 0.60 by Breslow Day test).

†

Diabetes trios (n = 91) OR 1.79 (95% CI 1.09–2.93) (P = 0.02); impaired fasting glucose trios (n = 163) 1.43 (0.98–2.09) (P = 0.06); IGT trios (n = 36) 1.18 (0.53–2.64) (P = 0.68). MAF, minor allele frequency.

View Large
TABLE 4

Mean trait values by genotype

PhenotypenMean trait values (±SD) by genotype
Two-tailed P value
TTCTCCAdditiveRecessiveTT vs. CCTT vs. TCTC vs. CC
BMI (kg/m2) 8,258 28.4 ± 5.4 28.4 ± 5.3 28.2 ± 5.2 0.31     
n 995 42 322 631      
    Ins index (mU/mmol)  10.9 ± 12.7 16.5 ± 50.5 18.1 ± 33.1 0.0030 0.0010 0.0009 0.0031 0.3767 
    Disp index (mU2/l2)  22.5 ± 28.9 35.8 ± 112.9 42.6 ± 79.9 0.0044 0.0011 0.0012 0.0019 0.6278 
    Fasting insulin (mU/l)  7.27 ± 4.24 9.19 ± 6.66 8.94 ± 6.02 0.3423 0.1405 0.1537 0.1340 0.7685 
    HOMA-IR (mmol · mU/l2)  1.88 ± 1.29 2.30 ± 2.06 2.19 ± 1.63 0.2561 0.1286 0.1590 0.1007 0.5191 
n 721 27 236 454      
    AUCins*  3,911 ± 3,658 4,971 ± 3,176 5,229 ± 3,248 0.0007 0.0004 0.0002 0.0048 0.1250 
    AUCglu*  339.8 ± 262.8 271.4 ± 214.5 270.0 ± 195.3 0.1775 0.0841 0.0778 0.1266 0.9297 
    AUCins/AUCglu*  23.46 ± 39.75 24.33 ± 71.06 32.14 ± 243.05 0.0023 0.0025 0.0015 0.0135 0.1640 
PhenotypenMean trait values (±SD) by genotype
Two-tailed P value
TTCTCCAdditiveRecessiveTT vs. CCTT vs. TCTC vs. CC
BMI (kg/m2) 8,258 28.4 ± 5.4 28.4 ± 5.3 28.2 ± 5.2 0.31     
n 995 42 322 631      
    Ins index (mU/mmol)  10.9 ± 12.7 16.5 ± 50.5 18.1 ± 33.1 0.0030 0.0010 0.0009 0.0031 0.3767 
    Disp index (mU2/l2)  22.5 ± 28.9 35.8 ± 112.9 42.6 ± 79.9 0.0044 0.0011 0.0012 0.0019 0.6278 
    Fasting insulin (mU/l)  7.27 ± 4.24 9.19 ± 6.66 8.94 ± 6.02 0.3423 0.1405 0.1537 0.1340 0.7685 
    HOMA-IR (mmol · mU/l2)  1.88 ± 1.29 2.30 ± 2.06 2.19 ± 1.63 0.2561 0.1286 0.1590 0.1007 0.5191 
n 721 27 236 454      
    AUCins*  3,911 ± 3,658 4,971 ± 3,176 5,229 ± 3,248 0.0007 0.0004 0.0002 0.0048 0.1250 
    AUCglu*  339.8 ± 262.8 271.4 ± 214.5 270.0 ± 195.3 0.1775 0.0841 0.0778 0.1266 0.9297 
    AUCins/AUCglu*  23.46 ± 39.75 24.33 ± 71.06 32.14 ± 243.05 0.0023 0.0025 0.0015 0.0135 0.1640 

Traits were log transformed.

*

Four negative values of AUCins were removed, and 26 negative values of AUCins/AUCglu were removed. Insulin and glucose units used for all calculations were mU/l and mmol/l, respectively. AUC, AUC during OGTT; Disp index, insulin disposition index; Ins index, insulinogenic index.

View Large

L.C.G. has served on advisory boards for and received consulting fees from sanofi-aventis, Bristol-Myers Squibb, GlaxoSmithKline, Kowa, and Roche.

Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

D.A. is a Charles E. Culpeper Scholar of the Rockefeller Brothers Fund and a Burroughs Wellcome Fund Clinical Scholar in Translational Research. R.S. is a recipient of a National Institutes of Health (NIH) National Research Service Award fellowship. This work was funded by The Richard and Susan Smith Family Foundation/American Diabetes Association Pinnacle Program Project Award (to D.A., J.N.H., and M.J.D.); preparation of the new DNA samples was supported by the Diabetes Genetics Initiative (Broad Institute-Novartis Institute of BioMedical Research-Lund University collaboration). J.C.F. is supported by NIH Research Career Award 1 K23 DK65978-03. L.G., T.T., and the Botnia Study are supported by the Sigrid Juselius Foundation, the Academy of Finland, the Finnish Diabetes Research Foundation, the Swedish Medical Research Council, the Juvenile Diabetes Foundation Wallenberg Foundation, and the Novo Nordisk Foundation.

We thank Mark McCarthy and colleagues for kindly sharing data on association analyses of TCF7L2, Ryan Tewhey for excellent technical assistance, the Botnia and Skara research teams for clinical contributions, and colleagues at Massachusetts General Hospital and the Broad Institute for helpful discussions and comments on the manuscript.

1.
Grant SF, Thorleifsson G, Reynisdottir I, Benediktsson R, Manolescu A, Sainz J, Helgason A, Stefansson H, Emilsson V, Helgadottir A, Styrkarsdottir U, Magnusson KP, Walters GB, Palsdottir E, Jonsdottir T, Gudmundsdottir T, Gylfason A, Saemundsdottir J, Wilensky RL, Reilly MP, Rader DJ, Bagger Y, Christiansen C, Gudnason V, Sigurdsson G, Thorsteinsdottir U, Gulcher JR, Kong A, Stefansson K: Variant of transcription factor 7-like 2 (TCF7L2) gene confers risk of type 2 diabetes.
Nat Genet
 
38
:
320
–323,
2006
2.
Yi F, Brubaker PL, Jin T: TCF-4 mediates cell type-specific regulation of proglucagon gene expression by beta-catenin and glycogen synthase kinase-3beta.
J Biol Chem
 
280
:
1457
–1464,
2005
3.
Altshuler D, Brooks LD, Chakravarti A, Collins FS, Daly MJ, Donnelly P: A haplotype map of the human genome.
Nature
 
437
:
1299
–1320,
2005
4.
Altshuler D, Hirschhorn JN, Klannemark M, Lindgren CM, Vohl MC, Nemesh J, Lane CR, Schaffner SF, Bolk S, Brewer C, Tuomi T, Gaudet D, Hudson TJ, Daly M, Groop L, Lander ES: The common PPARgamma Pro12Ala polymorphism is associated with decreased risk of type 2 diabetes.
Nat Genet
 
26
:
76
–80,
2000
5.
Florez JC, Burtt N, de Bakker PI, Almgren P, Tuomi T, Holmkvist J, Gaudet D, Hudson TJ, Schaffner SF, Daly MJ, Hirschhorn JN, Groop L, Altshuler D: Haplotype structure and genotype-phenotype correlations of the sulfonylurea receptor and the islet ATP-sensitive potassium channel gene region.
Diabetes
 
53
:
1360
–1368,
2004
6.
Florez JC, Sjogren M, Burtt N, Orho-Melander M, Schayer S, Sun M, Almgren P, Lindblad U, Tuomi T, Gaudet D, Hudson TJ, Daly MJ, Ardlie KG, Hirschhorn JN, Altshuler D, Groop L: Association testing in 9,000 people fails to confirm the association of the insulin receptor substrate-1 G972R polymorphism with type 2 diabetes.
Diabetes
 
53
:
3313
–3318,
2004
7.
Florez JC, Agapakis CM, Burtt NP, Sun M, Almgren P, Rastam L, Tuomi T, Gaudet D, Hudson TJ, Daly MJ, Ardlie KG, Hirschhorn JN, Groop L, Altshuler D: Association testing of the protein tyrosine phosphatase 1B gene (PTPN1) with type 2 diabetes in 7,883 people.
Diabetes
 
54
:
1884
–1891,
2005
8.
Florez JC, Wiltshire S, Agapakis CM, Burtt NP, de Bakker PI, Almgren P, Bengtsson Bostrom K, Tuomi T, Gaudet D, Daly MJ, Hirschhorn JN, McCarthy MI, Altshuler D, Groop L: High-density haplotype structure and association testing of the insulin-degrading enzyme (IDE) gene with type 2 diabetes in 4,206 people.
Diabetes
 
55
:
128
–135,
2006
9.
Sun MW, Lee JY, de Bakker PI, Burtt NP, Almgren P, Rastam L, Tuomi T, Gaudet D, Daly MJ, Hirschhorn JN, Altshuler D, Groop L, Florez JC: Haplotype structures and large-scale association testing of the 5′ AMP-activated protein kinase genes PRKAA2, PRKAB1, and PRKAB1 with type 2 diabetes.
Diabetes
 
55
:
849
–855,
2006
10.
Winckler W, Burtt NP, Holmkvist J, Cervin C, de Bakker PI, Sun M, Almgren P, Tuomi T, Gaudet D, Hudson TJ, Ardlie KG, Daly MJ, Hirschhorn JN, Altshuler D, Groop L: Association of common variation in the HNF1α gene region with risk of type 2 diabetes.
Diabetes
 
54
:
2336
–2342,
2005
11.
Winckler W, Graham RR, de Bakker PI, Sun M, Almgren P, Tuomi T, Gaudet D, Hudson TJ, Ardlie KG, Daly MJ, Hirschhorn JN, Groop L, Altshuler D: Association testing of variants in the hepatocyte nuclear factor 4α gene with risk of type 2 diabetes in 7,883 people.
Diabetes
 
54
:
886
–892,
2005
12.
Gloyn AL, Weedon MN, Owen KR, Turner MJ, Knight BA, Hitman G, Walker M, Levy JC, Sampson M, Halford S, McCarthy MI, Hattersley AT, Frayling TM: Large-scale association studies of variants in genes encoding the pancreatic β-cell KATP channel subunits Kir6.2 (KCNJ11) and SUR1 (ABCC8) confirm that the KCNJ11 E23K variant is associated with type 2 diabetes.
Diabetes
 
52
:
568
–572,
2003
13.
Groop L: Pathogenesis of type 2 diabetes: the relative contribution of insulin resistance and impaired insulin secretion.
Int J Clin Pract Suppl
 
113
:
3
–13,
2000
14.
Chiasson JL, Rabasa-Lhoret R: Prevention of type 2 diabetes: insulin resistance and β-cell function.
Diabetes
53 (Suppl 3):S34–S38,
2004
15.
Lowell BB, Shulman GI: Mitochondrial dysfunction and type 2 diabetes.
Science
 
307
:
384
–387,
2005
16.
t Hart LM, van Haeften TW, Dekker JM, Bot M, Heine RJ, Maassen JA: Variations in insulin secretion in carriers of the E23K variant in the KIR6.2 subunit of the ATP-sensitive K+ channel in the β-cell.
Diabetes
 
51
:
3135
–3138,
2002
17.
Fajans SS, Bell GI, Polonsky KS: Molecular mechanisms and clinical pathophysiology of maturity-onset diabetes of the young.
N Engl J Med
 
345
:
971
–980,
2001
18.
Pritchard JK: Are rare variants responsible for susceptibility to complex diseases?
Am J Hum Genet
 
69
:
124
–137,
2001
19.
Reich DE, Lander ES: On the allelic spectrum of human disease.
Trends Genet
 
17
:
502
–510,
2001
20.
Pritchard JK, Cox NJ: The allelic architecture of human disease genes: common disease-common variant or not?
Hum Mol Genet
 
11
:
2417
–2423,
2002
21.
Terwilliger JD, Hiekkalinna T: An utter refutation of the “Fundamental Theorem of the HapMap.”
Eur J Hum Genet
 
14
:
426
–437,
2006
22.
Weiss KM, Terwilliger JD: How many diseases does it take to map a gene with SNPs?
Nat Genet
 
26
:
151
–157,
2000
23.
Groop L, Forsblom C, Lehtovirta M, Tuomi T, Karanko S, Nissen M, Ehrnstrom BO, Forsen B, Isomaa B, Snickars B, Taskinen MR: Metabolic consequences of a family history of NIDDM (the Botnia Study): evidence for sex-specific parental effects.
Diabetes
 
45
:
1585
–1593,
1996
24.
de Bakker PI, Yelensky R, Pe’er I, Gabriel SB, Daly MJ, Altshuler D: Efficiency and power in genetic association studies.
Nat Genet
 
37
:
1217
–1223,
2005
25.
Spielman RS, McGinnis RE, Ewens WJ: Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM).
Am J Hum Genet
 
52
:
506
–516,
1993
26.
Boehnke M, Langefeld CD: Genetic association mapping based on discordant sib pairs: the discordant-alleles test.
Am J Hum Genet
 
62
:
950
–961,
1998
DIABETES
2006

Supplementary data

Analysis of epistasis between TCF7L2 SNP rs7903146 and PPARã-P12A / KIR6.2-E23K- doc file
Association of TCF7L2 SNPs with risk of type 2 diabetes- doc file