Meta-analysis Added Power to Identify Variants in FTO Associated With Type 2 Diabetes and Obesity in the Asian Population

Authors

  • Yun Liu,

    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    3. Institutes of Biomedical Sciences, Fudan University, Shanghai, PR China
    Search for more papers by this author
  • Zhe Liu,

    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Yiqing Song,

    1. Division of Preventive Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
    Search for more papers by this author
  • Daizhan Zhou,

    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Di Zhang,

    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Teng Zhao,

    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Zhuo Chen,

    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Lan Yu,

    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Yifeng Yang,

    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Guoyin Feng,

    1. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Jun Li,

    1. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    Search for more papers by this author
  • Jie Zhang,

    1. Department of Thoracic Surgery, Shanghai Cancer Hospital, Fudan University, Shanghai, PR China
    Search for more papers by this author
  • Simin Liu,

    1. Department of Epidemiology, University of California at Los Angeles, Los Angeles, California, USA
    2. Department of Medicine, University of California at Los Angeles, Los Angeles, California, USA
    3. Center of Metabolic Disease Prevention, University of California at Los Angeles, Los Angeles, California, USA
    Search for more papers by this author
  • Zuofeng Zhang,

    1. Department of Epidemiology, University of California at Los Angeles, Los Angeles, California, USA
    Search for more papers by this author
  • Lin He,

    Corresponding author
    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    3. Institutes of Biomedical Sciences, Fudan University, Shanghai, PR China
    Search for more papers by this author
  • He Xu

    Corresponding author
    1. Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Graduate School of the Chinese Academy of Sciences, Shanghai, PR China
    2. Bio-X Center, Shanghai Jiao Tong University, Shanghai, PR China
    3. Department of Epidemiology, University of California at Los Angeles, Los Angeles, California, USA
    4. Department of Medicine, University of California at Los Angeles, Los Angeles, California, USA
    5. Center of Metabolic Disease Prevention, University of California at Los Angeles, Los Angeles, California, USA
    Search for more papers by this author

  • The first two authors contributed equally to this work.

(hexuc@yahoo.com.cn)

(helin@bio-x.cn)

Abstract

Several common variants in the intron 1 of FTO (fat mass and associated obesity) gene have been reliably associated with BMI and obesity in European populations. We analyzed two variants (rs9939609 and rs8050136) in 4,189 Chinese Han individuals and conducted a meta-analysis of published studies in Asian population to investigate whether these variants are associated with type 2 diabetes (T2D) and obesity in Asian population. In this study, both the minor allele A of rs9939609 and the minor allele A of rs805136 were associated with increased risk of T2D, independent of measures of BMI; the odds ratios (ORs) per copy of the risk allele were 1.19 for rs9939609 (95% confidence interval (CI), 1.04–1.37; P = 0.01) and 1.22 for rs8050136 (95% CI, 1.07–1.40; P = 0.004) after adjusting for age, sex, and BMI. Our results also showed association with risk of obesity (rs9939609: OR = 1.39 (95% CI 1.04–1.85), P = 0.02; rs8050136: OR = 1.45 (95% CI 1.09–1.93), P = 0.01) but no association with overweight. These results were consistent with the pooled results from our meta-analysis study (for diabetes, rs8050136, P = 1.3 × 10−3; rs9939609, P = 9.8 × 10−4; for obesity, rs8050136, P = 2.2 × 10−7; rs9939609, P = 9.0 × 10−9). Our findings indicate that the two variants (rs9939609 and rs8050136) in the FTO gene contribute to obesity and T2D in the Asian populations.

Introduction

Obesity is a complex disorder involving a host of heritable and behavioral features and it is a major risk factor for type 2 diabetes (T2D). The prevalence of obesity and T2D has increased dramatically in China (1). Recently, genome-wide association studies carried by Frayling et al. identified a common variant, rs9939609, in the FTO (fat mass and associated obesity) gene that was strongly associated with BMI and obesity in European population (2,3). This association was subsequently confirmed by other independent studies based on European cohorts (3,4,5,6,7). However, it remains controversial whether the association is replicable in other ethnic populations, especially when there is evidence to show ethnic differences in the allele frequencies of FTO variants.

Hotta et al. confirmed that single-nucleotide polymorphisms (SNPs) in FTO including rs9939609 were associated with severe obesity in the Japanese (8). Horikoshi et al. found that rs8050136 in FTO showed a nominal significance in terms of the association with T2D in the Japanese population (9), although Omori et al. did not replicate the association of the SNPs in FTO and T2D but found a modest association between the SNPs in FTO and BMI (10). Among Chinese Han populations, Chang et al. found that BMI was higher in individuals who had the A allele of rs9939609 (ref. 11); however, this finding was not confirmed by Li et al. (12).

The aim of this study is to determine whether or not the variations in FTO are associated with the susceptibility to obesity and T2D in a sample comprising 4,189 individuals of Chinese Han origin. We also undertook a meta-analysis to provide a quantitative assessment of the collective evidence on the relationship between T2D (covering a total of 22,181 individuals), and obesity (covering a total of 10,422 individuals) and FTO variations in Asian populations.

Methods and Procedures

Subject and DNA preparation

The clinical characteristics of the participants enrolled in this study are summarized in Supplementary Table S1 online. In our case–control study, 1,912 unrelated T2D patients and 236 IFG/IGT individuals were recruited from Shanghai, China. T2D and IFG/IGT status were defined in accordance with World Health Organization criteria. A total of 2,041control individuals with fasting plasma glucose concentration <6.1 mmol/l were enrolled from the same region. The anthropometry of individuals were measured by two well-trained clinic staff independently; the average value of these two measurements was recorded. Blood pressure measurements were acquired using a conventional mercury sphygmomanometer, blood glucose, lipids, and HbA1c were determined on Roche modular P800 autoanalyzer (Mannheim, Germany).

A standard informed consent in the protocol, which was reviewed and approved by the ethics committee of the Shanghai Institute for Biological Sciences, was given by the participants after the nature of the study had been fully explained. High-molecular-weight genomic DNA was prepared from venous blood using the QuickGene 610L Automatic DNA/RNA Extraction System (Fujifilm, Tokyo, Japan).

Genotyping

The FTO gene variants rs9939609 and rs8050136 have been reported to be most associated with T2D and were thus chosen and genotyped in our Chinese Han sample using TaqMan technology on an ABI7900 system (Applied Biosystems, Foster City, CA). The standard 5-µl PCR reactions were carried out using TaqMan Universal PCR Master Mix reagent kits under the guidelines provided. Genotype data were obtained in about 95% of the DNA samples and replicate quality control samples (32 samples) were included and genotyped with 100% concordance.

Statistical analysis

We first used SHEsis to perform the Hardy–Weinberg equilibrium test and to compare the differences of allele frequencies between cases and controls (13). The association of the FTO SNPs with T2D was assessed by logistic regression after adjusting for sex, age and natural-log transformed BMI (lnBMI). Logistic regression was also used to test for association between risk of obesity and/or overweight and FTO SNPs after adjusting for sex and age.

The differences in the BMI and waist-to-hip ratio according to genotype were analyzed using a linear regression with BMI and waist-to-hip ratio as the dependent variable and genotype as the independent variable with sex as a covariate. The statistical analyses were performed using the SPSS (SPSS, Chicago, IL) program. Population attributable risk was calculated as population attributable risk = (X − 1)/X, assuming the multiplicative model where X = (1 − f)2 + 2f(1 − f)γ + f2γ2; γ is the estimated odds ratio (OR) and f is the frequency of the risk allele (14).

We performed power calculations on Power Calculator for Genetic Studies software developed by Skol and Abecasis based on the method of Skol et al. (15). In the meta-analysis, Cochran's χ2-based Q-statistic test was performed to assess heterogeneity in all the studies included. The random-effects model, which yielded wider confidence intervals (CIs), was adopted when heterogeneity existed; otherwise both the fixed-and random-effects models were deemed appropriate. Combined ORs were calculated using the Mantel–Haenszel (fixed-effects) and DerSimonian and Laird (random-effects) methods and 95% CIs were constructed using Woolf's method. The significant P value of overall ORs was determined using the Z-test. All the calculations for the meta-analysis were carried out on Comprehensive Meta-Analysis software (version 2.0; BIOSTAT, Englewood, NJ).

Results

Neither of the two SNPs showed statistical deviation from Hardy–Weinberg equilibrium among the controls (Table 1). The minor allele frequencies of the two SNPs were 0.12 in our Chinese Han population, which contrasts with data previously reported in European whites (0.38–0.42) but is consistent with those for Chinese/Japanese samples in the Hapmap database (0.12/0.17). The two SNPs are located in intron 1 and are in strong linkage disequilibrium (r2 = 0.96). Both the minor allele A of rs9939609 and the minor allele A of rs8050136 were associated with increased risk of T2D. The ORs per copy of the risk allele were 1.19 for rs9939609 (95% CI, 1.04–1.37; P = 0.01) and 1.22 for rs8050136 (95% CI, 1.07–1.40; P = 0.004) in our sample. The association between the two SNPs and T2D persisted after correction for age, gender, and lnBMI under the additive model (for rs9939609, OR = 1.19 (95% CI 1.03–1.38), P = 0.02; for rs8050136, OR = 1.22 (95% CI 1.05–1.41), P = 0.008) (Table 1).

Table 1.  Association study of type 2 diabetes susceptibility variants in the Chinese Han population
inline image

Based on the Chinese adult physical characteristics, a BMI of 24 with best sensitivity and specificity for identification of the risk factors of related chronic diseases including diabetes was recommended as the cutoff point for overweight; a BMI of 28, was recommended as the cutoff point for obesity (16). We tested the association between the FTO SNPs and overweight or obesity (as binary outcomes) under the additive model in the controls and found that both SNPs showed positive association with risk of obesity (rs9939609: OR = 1.39 (95% CI 1.04–1.85), P = 0.02; rs8050136: OR = 1.45 (95% CI 1.09–1.93), P = 0.01), independent of measures of BMI, but showed no association with overweight (Table 2). Assuming a population frequency of 12% for the risk allele A in rs7754840 as observed in our control subjects, population attributable risk were 4.8% for T2D and 9.9% for obesity in our sample.

Table 2.  Associations of SNPs with obesity or overweight
inline image

We also examined the association of these polymorphisms with BMI and waist-to-hip ratio as continuous outcomes in the control individuals using general linear model analysis after adjustment for age and sex (see Supplementary Table S2 online). There was a trend toward a BMI increment associated with each additional copy of the minor alleles in both SNPs, although such associations were not statistically significant.

We performed power calculations on Power Calculator for Genetic Studies software developed by Skol and Abecasis based on the method of Skol et al. (15). Based on the minor allele frequency of 0.12, an OR of 1.15, a prevalence of 0.1 and a significance level at 0.05, the approximate total sample size of 3,000 (1,500 T2D individuals and 1,500 control individuals) can only get 52% power. Given the relatively low frequencies of the risk alleles of these two common FTO variants in Asian populations, we conducted a meta-analysis of all published studies in these populations and our present data to attain sufficient power to quantitatively assess the collective evidence on the relation of FTO variants with T2D (9,10,11,12,17,18,19,20,21) and obesity (8,11,12,20), the associations with T2D in the meta-analysis were adjusted for age, gender, and BMI except Ng's study, which was not adjusted for BMI in individual population but gave adjusted combined result of rs8050136 (Table 3). There was no heterogeneity of ORs across these studies (Figure 1a). Figure 1b–e shows the outcome of the meta-analysis. Overall, the pooled ORs for obesity and T2D were statistically significant for both the risk alleles of rs9939609 and rs8050136 of the FTO gene (for diabetes, rs8050136, pooled OR = 1.11, P = 1.3 × 10−3; rs9939609, pooled OR = 1.12, P = 9.8 × 10−4; for obesity, rs8050136, pooled OR = 1.30, P = 2.2 × 10−7; rs9939609, pooled OR = 1.31, P = 9.0 × 10−9). The solid associations did not change fundamentally whether we included our present data or not (data not show), indicating that these association were not from this study.

Table 3.  OR and 95% CI of studies used in meta analysis
inline image
Figure 1.

Meta-analysis of the association of FTO SNPs with type 2 diabetes and obesity. For each study, the point estimate of the OR with 95% CI is shown. (a) Cochran's χ2-based Q-statistic test: Q, df, and P value for these four statistics. (b) rs8050136 and type 2 diabetes; (c) rs8050136 and obesity; (d) rs9939609 and type 2 diabetes; (e) rs9939609 and obesity. CI, confidence interval; FTO, fat mass and associated obesity gene; OR, odds ratio; SNP, single-nucleotide polymorphism.

Discussion

A significant epidemic of obesity and T2D currently represents a major public health problem both in China and worldwide. Approximately 58% of T2D globally can be attributed to overweight and obesity (22). In this study, we identified positive associations of SNPs rs9939609 and rs8050136 within the FTO gene with susceptibility to T2D in the Chinese Han population. The risk alleles in our study were in the same direction as in previous studies and the OR (1.22 in this study, under additive model) was similar to the results of FUSION-DGI-WTCCC/UKT2D (OR = 1.17, under additive model) (ref. 6).

As shown in Table 2, both SNPs showed significant association with obesity in the control individuals, but neither of them showed evidence for association with overweight, which suggested that the risk alleles of FTO might be inclined to concentrate in the individuals of the high tail of the BMI distribution, so the power to detect an association in the severe obesity population might be larger than in the overweight population, this finding was corroborated with the discovery of another original work conducted by Dina et al. (23) and other two replication studies (5,23,24) that also detected the suspect FTO gene but with different SNPs strongly contributing to early-onset and severe obesity.

The cutoff point for adiposity in the Chinese population is different from that among Europeans (25). BMI measures of 24 and 28 were selected for overweight and obesity, respectively, as a BMI of 24 gives the best sensitivity and specificity for identification of the risk factors and a BMI of 28 will identify the risk factors with specificity around 90% (ref. 16). In our study, the association between the two SNPs and T2D remained significant after adjusting for BMI. This is in contrast with the previous studies based on European populations, which showed that the association of these SNPs with T2D risk is mediated through BMI/adiposity (2,3). One explanation for this difference is that the BMI in our samples was inadequately adjusted for adiposity and thus BMI did not reflect the adiposity level in Chinese population generally (16). Sanghera et al. and Yajnik et al. also reported an independent BMI effect of FTO and T2D in Asian Indians (19,26). BMI is not considered an accurate measure for obesity in Asian populations where at a given BMI, the muscle mass is low and visceral and subcutaneous fat is high (27,28,29). Therefore, measures of BMI or waist and hip may not provide accurate estimates of total fat in the Han population and in Asians in general (27,28,29). However, this does not necessarily apply to the FTO effect on BMI and adiposity in other populations such as Europeans. We also considered the possibility that BMI is a better measure of adiposity in relatively high-BMI individuals than in low-BMI individuals. This would be consistent with our finding that the FTO gene showed a positive association with risk of obesity, but not with risk of overweight. In view of the low proportion of obesity in our population, the negative result in the BMI quantitative trait analysis was understandable (7).

According to our meta-analysis of diabetes, only two of the seven published studies and data sets for rs8050136 were consistent with our own study in obtaining positive association for this variant. Most of them showed negative results including the study by Li et al. in another Chinese Han population, which obtained a negative result (12) (Figure 1b). The overall OR (1.11, 95% CI: 1.04–1.18, P = 1.3 × 10−3) indicates significant overall effect in the eight studies. The meta-analysis of rs8050136 and obesity also produced strong evidence of association in the combined data although some studies showed either no association or only a weak association. It was similar for the rs9939609 variant in the meta-analysis. Recently, a replication study of FTO polymorphisms on BMI population conducted by Cha et al. in a Korean study also confirmed the association (30). Although they used different tag SNPs, their results further confirmed that the FTO may be one of the worldwide obesity-risk genes.

There are a number of possible underlying reasons for the differences found among the studies. One explanation is that the nonreplication may reflect a lack of sample power in individual studies. The minor allele frequency in the Chinese Han population and other non-Europeans included in previous studies was much smaller than that in the European population, and a larger sample size is required to achieve sufficient power under a smaller minor allele frequency. We can only get 52% power to detect the association between FTO genotype and the individuals phenotype in this study. The sample size in the studies of Asian populations was modest, giving a possible explanation of the inconsistency in results. However, although these individually underpowered studies each contributed a small effect on the identification of the susceptibility of FTO variants on T2D and obesity, they may have combined to generate a significant overall association. Another explanation is that variations may be due to differences in sample components and diagnosis including factors such as age, distribution of males and females, and the criteria for defining obesity or overweight.

In conclusion, these results from a Chinese Han population further support the involvement of FTO in the susceptibility of T2D and obesity and may have potentially important scientific, clinical, and public health implications. Further genetic and functional studies in the FTO variants in other large independent populations may help to clarify the physiological mechanisms by which they affect the pathogenesis of T2D and obesity.

SUPPLEMENTARY MATERIAL

Supplementary material is linked to the online version of the paper at http:www.nature.comoby

Acknowledgments

We appreciate the contribution of all participants in this study, as well as that of the doctors who helped us in the diagnosis. This work was supported by the Knowledge Innovation Program of Shanghai Institute for Biological Sciences, the Chinese Academy of Sciences (2007KIP210), the National Key Technology R&D Program (2006BAI05A05), the Chinese Nutrition Society (05015), the Dannon Institute, the Shanghai-Unilever Research and Development Fund (06SU07007), the Shanghai Municipality Science & Technology Commission (05JC14090), the Shanghai Leading Academic Discipline Project (B205), and the Knowledge Innovation Program of the Chinese Academy of Sciences (KSCX2-YW-R-01).

Disclosure

The authors declared no conflict of interest.

Ancillary