Genetic Epidemiology Study Of Obesity And Theory Investigating And Applying Of Bivariate Association Mapping

Recent data suggest that current obesity diagnostic criterion for Caucasians may not be appropriate for Asian populations. Genome wide association studies for obesity have been successfully conducted in Caucassians but none was done in Chinese. In the area of statistical genetics, univariate association analyses have been widely investigated, however, multivariate association analyses for population design have not received sufficient attention.Here, in this study, firstly, we utilized receiver operating characteristic analyses to evaluate body mass index (BMI), waist circumference (WC) and waist-to-hip ratio (WHR) in predicting obestiy using percentage body fat (PBF) as gold standard in 1109 males and 879 females. BMI, WC and WHR showed strong positive correlation with PBF (r=0.47-0.75) in both males and females within both age groups. True-positive rates ranged from 82.4% to 94.1% and 68.8% to 86.3% in males and females, respectively. True-negative rates ranged from 64.1% to 84.7% and from 56.9% to 79.0%, respectively. The areas under the curves (AUCs) for WC and BMI were high (0.76-0.92) in both sexes and divided age groups (20-30-y and 31-45-y), and those for WHR were a little lower (0.74-0.88). Our study suggested that BMI and WC are two important predictors for obesity in Chinese, and WHR is an alternative. We also suggested an obviously lower cut of BMI to define obestity in young Chinese compared with that in Caucasians.Than we conducted a genome wide association study for BMI using Affymetrix 500k chip in 597 northern Chinese and a replication study in 2955 southeren Chinese. After quality control,281,533 single nucleotide polymorphisms (SNPs) were included in the association analysis. Eight SNPs were significantly associated with BMI variation after false discovery rate (FDR) correction (FDR q=0.033-0.048). Among these eight SNPs, rs4633 which is located at the exon of catechol-O-methyltransferase (COMT) gene is the most significant SNP associated with BMI (FDR q=0.033). Two additional contiguous SNPs (rs4432245 & rs711906) in the eukaryotic translation initiation factor 2 alpha kinase 4 (EIF2AK4) gene were significantly associated with BMI (FDR q=0.048). In the follow-up replication study (southern Chinese sample) containing 2955 Chinese Han subjects, we confirmed the associations between BMI and rs4432245, rs711906 in the EIF2AKE gene (p=0.03 & 0.01, respectively).Besides, we performed extensive simulation studies and introduced generalized estimating equation 2 (GEE2) to association mapping for two traits. Through power comparison with univariate analyses, we found that bivariate models can be more powerful than univariate model. The introduced method leads to false-positive rates around 0.05 at the nominal significant level of 0.05, and so the introduced method is reasonably robust. Our real data analyses for obesity and osteoporosis confirmed these findings.The present study is the first attempt to evaluate the diagnostic indices for obesity in southern Chinese, the first genome wide association study for Chinese and identified a new gene for BMI. It is also the first effort to introduce a method applied to association mapping for two traits for population-based design.