Main

The albumin:globulin ratio (A/G ratio) is a biochemical parameter that is used to interpret changes in serum proteins that accompany disease.1 The primary clinical manifestation of the A/G ratio is a reduction due to a decrease in serum albumin, and the subsequent increase in serum globulins.2 This pattern is the classical change that is expected to accompany liver disease when serum albumin decreases below normal levels.3 A low A/G ratio is associated with all-cause mortality after non-ST (electrocardiogram S and T wave interval) elevation myocardial infarction,4 nephrotic syndrome5 and autoimmune disease.6

Recent genome-wide association studies (GWASs) have identified many loci that are linked to common diseases and quantitative traits.7, 8 Previous GWASs have reported candidate genes for total serum proteins, serum albumin and globulins, but A/G ratio was not considered the target phenotype.9, 10, 11 Thus, we performed a GWAS using two population-based cohorts, comprising 8842 Korean individuals.

Study subjects were selected from an ongoing population-based cohort, as part of the Korean Genome and Epidemiology Study. Participants were recruited from residents in two cities (Ansung and Ansan) in Gyeonggi-do province, Korea. The Ansan cohort consisted of 4637 individuals and was used as the discovery data set for GWAS. The Ansung cohort comprised 4205 persons and was the validation data set for the replication study. Written informed consent was obtained from all participants, and this research project was approved by the institutional review board of the Korea National Institute of Health.

Total serum protein and serum albumin were measured on an ADIVA1650 (Siemens, Tarrytown, NY, USA) and analyzed at a commercial analytical service center (SCL Co., Seoul, Korea). Globulin levels were calculated as the difference between total protein and albumin.

Genomic DNA was genotyped using the Affymetrix Genome-wide Human single-nucleotide polymorphism (SNP) array 5.0 at final concentrations of 100 g ml−1 and 500 ng (Affymetrix, Inc., Santa Clara, CA, USA). The quality control steps have been described.12 Briefly, SNPs with a missing genotype call rate >0.1 or a minor allele frequency <0.01 and in Hardy–Weinberg equilibrium (P<1 × 10−6) were excluded; ultimately, we used 333 651 SNPs that were genotyped in the Korean Association Resource study.12 For this study, we also removed SNPs on the X and Y chromosomes and those with minor allele frequency <5%; thus, 290 659 markers were used for the GWAS. For the fine mapping of the identified loci, we included the imputed SNPs in the signal plots. The detailed imputation procedure has previously been described in Hong et al.,13 which were previously used in the Korean Association Resource study. Briefly, the genotypes of the individuals were imputed using IMPUTE14 based on the International HapMap Phase II JPT (Japanese in Tokyo)+HCB (Han-Chinese in Beijing) panel.

The demographics and serum protein traits between Ansan and Ansung subjects were compared by χ2-test and t-test. The GWAS on the A/G ratio of the Ansan cohort was performed by linear regression, controlling for covariates, such as age, sex, body mass index, smoking status and alcohol drinking. SNPs that had a P-value<1 × 10−4 were selected. The selected SNPs were validated in the Ansung cohort. Combined β- and P-values were obtained for the SNPs with P-values<0.05 in Ansung cohort. Most association analyses were performed using PLINK (version 1.07; http://pngu.mgh.harvard.edu/~purcell/plink/) and SAS (version 9.1; SAS institute inc., Cary, NC, USA).

Table 1 describes the demographic features, body mass index, smoking status, alcohol drinking status and serum protein traits of the cohorts. The concentrations of all protein traits were higher in the Ansan cohort compared with the Ansung cohort. The mean age of the Ansan cohort was significantly lower than that of Ansung subjects. Thus, the difference in serum protein traits might be attributed to the older mean age in Ansung subjects, as reported by a Japanese epidemiological study, in which decreasing tendency of serum protein concentrations.15

Table 1 Basic characteristics of biochemical traits in the Ansan and Ansung cohorts
Full size table

Quantile–quantile plots and minus log10 (P) values were plotted against chromosomal position on Manhattan plots, as shown in Supplementary Figure 1 and Figure 1, respectively. The measured genomic inflation factor was 1.014. Thirty-eight SNPs in 18 loci had a P-value<10−4, two of which had genome-wide significance (P-value<5 × 10−8). The results of the 38 SNPs can be found in Supplementary Table 1.

Figure 1
figure 1

Mahattan plot for the GWAS of A/G ratio. SNPs were plotted based on their physical chromosomal positions (horizontal axis) together with their −log10 (P-values) in the GWAS (vertical axis). The dotted horizontal line is the suggestive threshold of P=1 × 10−4, and the solid horizontal line is the genome-wide significance threshold of P=5 × 10−8.

Full size image

Subsequently, 38 SNPs were extracted from the genome-wide SNPs in the Ansung cohort and examined in the replication study. In all, 5 of the 38 SNPs were significantly replicated in Ansung subjects, with P<1.3 × 10−3 (Bonferroni correction P-value criteria=0.05/38), and 6 SNPs were weakly replicated with P<0.05. The 11 SNPs lay in six loci (GALNT2, IRF4, HLA-DPB1, SLC31A1, FADS1 and TNFRSF13B), and the top SNPs in each locus are listed in Table 2. The signal plots for six replicate loci are shown in Supplementary Figures 2A–F.

Table 2 Repeat associated SNPs in the Ansan and Ansung cohorts and the further association results with total proteins, albumin and globulins
Full size table

Of the six loci, TNFRSF13B (rs4561508) and FADS1 (rs174548) showed genome-wide significance (overall P<5 × 10−8) in the discovery cohort, and remained significant in the replication data set. To compare the genetic effects on the other protein traits, we tested the six SNPs to the total serum protein, serum albumin and globulins in Table 2. The TNFRSF13B (rs4561508) and FADS1 (rs174548) revealed significant association with globulins, and the combined P-values were 3.36 × 10−21 and 1.37 × 10−8, respectively. These results were similar to the association with A/G ratio.

In this study, we identified two significant loci and four suggestive loci that affect the A/G ratio. The most significant SNP resided in the TNFRSF13B gene, which was reported in two Japanese GWAS studies as the significant loci with regard to total serum protein concentration9 or globulins.11 The SNPs rs4273077 and rs4985726 lie ∼500 bp and ∼5 kbp upstream, respectively, of rs4561508—identified in the current study. The linkage disequilibrium (LD) was moderate between rs4985726 andrs4561508 (r2=0.608), and perfect LD between rs4985726 and 4561508 (r2=0.1.000).

The TNFRSF13B gene encodes a lymphocyte-specific member of the tumor necrosis factor receptor superfamily. It interacts with calcium modulator and cyclophilin ligand, activates the transcription factors NFAT, AP1 and NF-kappa-B, and regulates humoral immunity by interacting with a tumor necrosis factor ligand. TNFRSF13B mutations have been reported to be associated with common variable immunodeficiency16 and hypogammaglobulinemia.17 Among the suggestive loci, IRF4 and HLA-DBP1 are important inflammation factors. Thus, our study suggests that A/G ratio predicts genetic predisposition to susceptibility for liver inflammation.

The FADS1 (FAD synthetase 1) gene was first identified in this study as a candidate gene for the A/G ratio. FAD synthetase is a critical member of the electron transport chain and is believed to mediate the cellular oxidative phosphoylation pathway.18, 19 This gene was identified as the antigenic target of immunoglobulin A.20 The rs174548 SNP in FADS1 has been reported to have robust associations with high-density lipoprotein cholesterol and triglyceride levels in European studies.21, 22 Cholesterol in the blood is synthesized in the liver and is a factor of cerebral infarction, myocardial infarction and liver function. Thus, our association results for rs174548 and A/G ratio are relevant to the previous two GWAS studies.

Serum protein measurements can be expected to provide greater detail on the potentially affected disease pathway. Thus, we hope that our findings increase our understanding of clinical end points, such as coronary artery disease and kidney disease.