A total of 2029 individuals, including 1024 patients with T2D and 1005 control subjects, were included in the present study. All participants were unrelated and of Northern Han Chinese ancestry residing in the Beijing metropolitan area. The patients were recruited from the Endocrinology and Metabolism Outpatient Clinics of Peking University People's Hospital in Beijing, China, and were diagnosed with T2D in accordance with the 1999 WHO criteria (fasting plasma glucose ≥7.0 mmol/l and/or 2-h plasma glucose ≥11.1 mmol/l) 16 . Patients diagnosed with T2D before 30 years of age, those with an elevated body mass index (BMI; > 35 kg/m²), or those with other clinical and genetic features for specific-type diabetes (e.g., maturity onset diabetes of the young [MODY]) were excluded from the study 17 18 . The control subjects were selected from communities near Peking University People's Hospital. The inclusion criteria for the control subjects in this study were as follows: (1) normal glucose regulation confirmed by a 75 g oral glucose tolerance test (OGTT) according to the 1999 WHO criteria (fasting plasma glucose <6.1 mmol/l and 2-h plasma glucose <7.8 mmol/l), (2) no family history of T2D, (3) >40 years of age, and (4) a BMI ≤ 35 kg/m². The clinical features of the participants are summarized in Table 1. Informed consent was obtained from every participant, and the study protocol was approved by the Ethics Committee of Peking University People's Hospital.

All of the patients with T2D and control subjects were examined in the morning after an overnight fast for at least 8 hours, with measurements of height, weight, and blood pressure. Blood samples were collected for biochemical measurements of fasting plasma glucose for all the participants, and 2-hour plasma glucose during a 75 g OGTT for the controls. The plasma glucose concentrations were measured by the glucose oxidase-peroxidase method. The fasting serum insulin concentrations were measured by an electrochemiluminescence immunoassay on a Roche Elecsys 2010 Chemistry Analyzer. The homeostasis model assessment of the insulin resistance index (HOMA-IR) and beta-cell function (HOMA-ί) were assessed, as described previously 19 .

Nine SNPs representing nine genes identified in recent GWAS or large candidate gene studies 1 2 3 4 5 6 7 8 were selected, including SLC30A8-rs13266634, HHEX- rs5015480, CDKAL1-rs10946398, CDKN2A/B-rs10811661, IGF2BP2-rs4402960, FTO-rs8050136, TCF2-rs7501939, KCNQ1-rs2237892, and WFS1- rs6446482. Genotyping of these SNPs was performed with an ABI SNaPshot^® Multiplex System (Applied Biosystems, Foster City, CA, USA) in the Chinese National Human Genome Center in Shanghai. Genotyping accuracy, as determined from the genotype concordance between duplicate samples (130 samples), was 100%. The genotyping success rate was >95% for the patient and control groups.

Data are given as the means ± SD for quantitative variables with a normal distribution, and as medians (interquartile range) for non-normally distributed data. Fasting serum insulin, HOMA-IR, and HOMA-ί were natural logarithm-transformed to normal distributions before statistical analysis.

Characteristics were compared and tested for significant differences between patients and controls using Student's t test for continuous variables, and using χ² test for categorical variable. Chi-squared tests were used to determine whether or not individual polymorphisms were in Hardy-Weinberg equilibrium (p > 0.05). The differences in allele and genotype frequencies between the patients with T2D and controls were analyzed using Pearson's χ² test. Logistic regression analysis was performed to calculate risk allele-specific odds ratios (ORs), 95% confidential intervals (CIs), and corresponding p -values after adjustment for gender, age, and BMI as covariates; under an additive model, 10,000 permutations were used to obtain a true estimate for statistical significance.

Linear regression was applied to test quantitative variables for differences between the genotype groups. SNPs that showed nominal evidence for an association between the age of diagnosis (AOD) for T2D, fasting plasma glucose, fasting serum insulin, HOMA-ί, and HOMA-IR were adjusted by age, gender and BMI. Additionally, BMI was adjusted for age and gender. All analyses were adjusted using linear regression on a priori genotypic models (dominant, additive. and recessive). Due to a lack of validity of the large sample χ2 test statistic, only the dominant model was considered for SNPs with ≤10 individuals homozygous for the minor allele (for WFS1 and FTO loci in the present study).

All statistical tests were performed by PLINK, version 1.06 http://pngu.mgh.harvard.edu/~purcell/plink or SPSS, version 11.5 for Windows (SPSS, Inc., Chicago, IL, USA). A p-value < 0.05 was considered statistically significant (two-tailed).

Power calculations were performed using Quanto software (available at http://hydra.usc.edu/gxe/), and the powers shown in Additional file 1, Table S1 were calculated using ORs from published studies, sample sizes and minor allele frequencies (MAF) in the present study, the prevalence of T2D in Chinese of mainland China 20 , and the type 1 error rate (0.05).

Pairwise linkage disequilibrium (LD) coefficients (r²) between SNPs were calculated by Haploview software, version 4.0 (Daly Lab at the Broad Institute, Cambridge, MA, USA).

To identify studies eligible for meta-analysis, MEDLINE citations (January 2000-January 2010) were surveyed using the National Library of Medicine's PubMed on-line search engine with each susceptibility gene name (CDKAL1, IGF2BP2, SLC30A8, CDKN2A/B, HHEX, KCNQ1, FTO, TCF2, and WFS1) and Chinese as key words. The retrieved references were read to identify studies that examined the allelic associations between the SNPs within the above nine genetic regions and T2D in Chinese Han living in China. For meta-analysis of WFS1, FTO, and TCF2, the C allele of WFS1-rs6446482, the A allele of FTO-rs8050136, and the T allele of TCF2- rs7501939 were used as surrogates for the minor A allele of WFS1- rs10010131, the A allele of FTO-rs9939609, and the G allele of TCF2-rs4430796 because they are in strong LD (r² = 1) in a HapMap Chinese Han population. In the meta-analysis, Cochran's x²-based Q-statistic test was performed to assess heterogeneity in all of the studies included, and a random-effect model was adopted when heterogeneity existed, otherwise a fixed-effects model was appropriate. Combined ORs were calculated using the Mantel-Haenszel (fixed-effects) and DerSimonian and Laird (random-effects) tests. The significant P value of overall ORs was determined using the Z-test. All calculations for the meta-analysis were performed on Revman42 (provided by Cochrane-Collaboration, available at http://ims.cochrane.org/revman/download/revman-4).

We have successfully genotyped all nine SNPs. There were no significant departures from the Hardy-Weinberg equilibrium for all SNPs in the control group (p > 0.05) as assessed by a χ² test. In Table 2, we present for each SNP the genotype and allele distributions, risk allele-specific ORs, and p-values under an additive genetic model after correction for gender, age, and BMI. The allele frequencies of all SNPs in our studied population (Additional file 1, Table S1) are comparable to those of the Chinese Han population and different from Europeans in HapMap (http://www.hapmap.org/, accessed Feb 2009). The associations of 7 SNPs with T2D were observed, which included, as follows: CDKAL1-rs10946398 (OR, 1.27; 95% CI, 1.11-1.45; empirical p = 0.0008); IGF2BP2-rs4402960 (OR, 1.26; 95% CI, 1.08-1.47; empirical p = 0.003); SLC30A8-rs13266634(OR, 1.19; 95% CI,1.04-1.37; empirical p = 0.009); CDKN2A/B-rs10811661(OR, 1.22; 95% CI, 1.06-1.41; empirical p = 0.005); HHEX-rs5015480 (OR,1.20; 95% CI, 1.01-1.42; empirical p = 0.03); and KCNQ1-2237892 (OR, 1.37; 95% CI, 1.19-1.69; empirical p = 1.0 × 10^-4). Interestingly, FTO-rs8050136 was also associated with T2D (OR, 1.25; 95% CI, 1.02-1.52; empirical p = 0.029), which remained significant after adjustment for age, gender, and BMI (OR,1.24; 95% CI, 1.01-1.52; empirical p = 0.046). However, no association with diabetes was found for TCF2-rs7501939 and WFS1 -rs6446482.

All published studies and references are shown in Additional file 1, Table S2, which indicates the inconsistency for associations between HHEX, IGF2BP2, TCF2, and FTO with T2D, the low MAF of WFS1-rs6446482 and FTO-rs8050136, and the small sample sizes in all studies in Chinese compared with those in Caucasian populations. Except for the Li study 13 due to the lack of allelic frequencies, the data from other studies and the current study were included in this meta-analysis to improve the power in evaluating the collective evidence on the relationships of these nine genetic loci to T2D, especially HHEX, IGF2BP2, TCF2 FTO, and WFS1 genetic loci. Table 3 shows the outcome of the meta-analysis. Overall, the pooled ORs for T2D were significant for HHEX-rs5015480 (pooled OR,1.24; p < 0.00001), IGF2BP2-rs4402960 (pooled OR, 1.14; p = 0.0002), TCF2-rs7501939 (pooled OR, 1.16; p < 0.00001), FTO-rs8050136 (pooled OR, 1.17; p < 0.00001), WFS1-rs6446482 (pooled OR, 1.26; p = 0.009), CDKAL1-rs10946398 (pooled OR, 1.24; p < 0.00001), SLC30A8-rs13266634 (pooled OR, 1.19; p < 0.00001), CDKN2A/B-rs10811661 (pooled OR, 1.28; p < 0.00001), and KCNQ1-2237892 (pooled OR, 1.33; p < 0.00001).

Genotype-age of diagnosis (AOD) for diabetes For the patients with T2D, multiple linear regression was used to evaluate the effect of nine representative SNPs on AOD with adjustment for gender and BMI under a priori models. The analysis revealed that only WFS1-rs6446482 was negatively related to AOD under a dominant model (p = 0.02), and the patients with one or two risk alleles of WFS1- rs6446482 were diagnosed approximately 3 years earlier than those without risk alleles (49 ± 11 years versus 52 ± 11 years, p = 0.03). Prompted by this observation, we performed a subgroup analysis of 406 early-onset T2D patients (age of diagnosis ≦45 years) and 973 control subjects, showing that WFS1- rs6446482 was significantly associated with early-onset T2D (risk allele-specific OR, 1.96; 95% CI, 1.05-4.76; nominal p = 0.03; empiric p = 0.03) after adjustment for gender, age, and BMI.

Genotype-BMI For analysis of the genotype-BMI relationship, only the control subjects were examined because the treatment for T2D might distort the relationship. Only FTO-rs8050136 of nine genetic loci was found to be significantly associated with BMI (p = 0.04) after adjusting for gender and age as covariates. The individuals with one or two risk alleles had a higher BMI (25. ± 3.3 kg/m², n = 205) than those without a risk allele (24.8 ± 3.4 kg/m², n = 790; p = 0.04).

Genotype-beta cell function and insulin resistance Because the insulin level might be influenced by a hyperglycemic agent, 493 non-diabetic control subjects with fasting insulin concentrations available were studied. Only at the FTO and CDKAL1 loci was a relationship between SNP and phenotype observed. As shown in Table 4, the individuals with a risk allele of FTO-rs8050136 had a higher HOMA-ί, HOMA-IR, BMI, and fasting serum insulin level than those without the risk allele(under a dominant model), and the effects of FTO on HOMA-ί, HOMA-IR, and the fasting serum insulin level were independent of gender, age, and BMI. In addition, CDKAL1-rs10946398 significantly affected beta cell function independent of gender, age, and BMI, and the individuals with 2 risk alleles had a lower HOMA-ί than those without a risk allele (under the recessive model).