C-reactive protein (CRP) is a hallmark acute-phase reactant and is widely used as a blood marker for inflammation. Substantial roles of serum CRP levels in the pathogenesis of diseases have been suggested, and investigation of the mechanisms that regulate serum CRP levels would have a substantial clinical impact. Here, through genome-wide association and replication studies performed using 12 854 Japanese subjects, we identified a significant association between serum CRP levels and a single nucleotide polymorphism in the promoter region of interleukin-6 (IL6) (rs2097677, P = 4.1 × 10−11), a typical pleiotropic pro-inflammatory cytokine. Our study also replicated the associations in the CRP (rs3093059, P = 3.5 × 10−21) and HNF1A loci (rs7310409, P = 2.7 × 10−8). Pleiotropic association analysis with hematological and biochemical traits using 30 466 Japanese subjects demonstrated that the CRP-increasing allele of rs2097677 in the IL6 locus was significantly associated with an increased white blood cell count, platelet count and serum globulin and a decreased mean corpuscular hemoglobin and mean corpuscular hemoglobin concentration (P < 5.0 × 10−4), although no pleiotropic association was observed in the CRP or HNF1A locus (α = 0.01). Our study demonstrated the pivotal role of the IL6 locus in the regulation of serum CRP levels and inflammatory pathways.

INTRODUCTION

C-reactive protein (CRP) is an acute-phase reactant and a hallmark of systemic inflammation (1,2). Plasma levels of CRP acutely rise during inflammatory processes, and CRP is widely used as a blood marker of inflammation in medical treatment (2). It is well known that CRP is a heritable trait and that genetic factors explain the variation in serum CRP levels along with environmental factors such as gender, smoking and obesity (3). Several epidemiologic studies demonstrated that the serum concentration of CRP is associated with the risk of a variety of diseases (3,4). Although the causality of CRP on these diseases has not been established and is still being discussed (57), investigation of the mechanisms of CRP regulation would have a substantial clinical impact.

Recent developments in genome-wide association studies (GWASs) have identified several genetic loci that affect serum CRP levels including those in CRP itself (6,810). However, only small fractions of the variations in CRP levels were explained (6). Moreover, previous GWASs for CRP levels were performed in Caucasian populations, and the studies in Asian populations have not been assessed further. Ethnic differences in the distributions of CRP levels have been noted (1113). It is known that Asian populations, including a Japanese population, have lower CRP levels than Caucasian populations (11,13). These observations suggest the ethnically heterogeneous genetic backgrounds of CRP levels, and that GWAS for CRP levels in Asian population might lead to the identification of novel associated loci.

As CRP plays a substantial role in acute-phase inflammation responses and interacts with a variety of cytokines and immune cells (1,2), it would be of great interest whether the genetic loci associated with CRP levels are associated with hematological or biochemical traits. However, pleiotropic association analysis of CRP-associated loci had not been performed in previous GWASs (6,810).

We report herein a large-scale GWAS for CRP levels in 10 112 Japanese subjects enrolled in the BioBank Japan Project (14). We also performed pleiotropic association analysis of the identified CRP-associated loci with hematological and biochemical traits by using 30 466 Japanese subjects.

RESULTS

Genome-wide association study

A GWAS for CRP levels was performed by genotyping >590 000 single nucleotide polymorphism (SNP) markers. We applied stringent quality control criteria, including principal component analysis (PCA) (15) evaluating potential population stratification, and obtained the genotype data of 10 112 Japanese subjects (Supplementary Material, Table S1) for 477 784 autosomal SNPs. To extend the genomic coverage, we then performed the whole-genome imputation of the SNPs, using HapMap Phase II genotype data of Japanese (JPT) and Han Chinese (CHB) individuals as references (16). After the imputation, 2 178 018 autosomal SNPs that satisfied the criteria of minor allele frequency (MAF) ≥0.01 and imputation score [Rsq value by MACH software (17)] ≥0.7 were obtained. To account for the non-normal distribution of serum CRP levels and to make the results of the study robust, rank-based inverse normal transformation (18) was applied for the CRP levels. The associations of the SNPs and the transformed values of the CRP levels were evaluated using a linear regression model.

A quantile–quantile plot of P-values indicated no remarkable discrepancy from the null hypothesis [inflation factor (19), λGC = 1.039, Fig. 1]. We identified significant associations in three chromosomal loci (1q23, 7p15 and 12q24) that satisfied the genome-wide significance threshold of P < 5.0 × 10−8 (Fig. 2A). Among them, two loci had been reported in previous GWASs for CRP levels (6,810) (P = 5.7 × 10−18 for rs3093059 in the CRP locus at 1q23 and P = 3.7 × 10−9 for rs7310409 in the HNF1A locus at 12q24) (Table 1). In the 7p15 chromosomal region, our study newly identified a significant association of the SNP located in the promoter region of interleukin-6 (IL6) with CRP levels (P = 1.6 × 10−9 for rs2097677; Table 1 and Fig. 2B). According to PCA plots of the subjects, existence of sub-populations was suggested on eigenvector 1 and 2 (Supplementary Material, Fig. S1). However, no significant association with CRP levels was observed for any of the eigenvectors (α = 0.05). The associations of the identified loci with CRP levels were significant when conditioned with the eigenvectors (P = 4.1 × 10−17, 1.8 × 10−9 and 3.8 × 10−9 for the CRP, IL6 and HNF1A loci, respectively). These results suggested that the effect of the potential population stratification would not be substantial in our GWAS for CRP levels.

Table 1.

Results of the genome-wide association study for serum CRP levels

SNP Chr Positiona Gene Allele 1/2b Allele 1 frequencyc GWAS (n = 10 112)
 
Replication study (n = 2742)
 
Meta-analysis (n = 12 854)e
 
      Effect size (SE)d P Effect size (SE)d P Effect size (SE)d P 
rs3093059 157 951 760 CRP G/A 0.12 0.160 (0.019) 5.7 × 10−18 0.174 (0.037) 3.3 × 10−6 0.161 (0.017) 3.5 × 10−21 
rs2097677 22 699 364 IL6 A/G 0.19 0.104 (0.017) 1.6 × 10−9 0.098 (0.034) 0.0040 0.101 (0.015) 4.1 × 10−11 
rs7310409 12 119 909 244 HNF1A G/A 0.47 0.080 (0.014) 3.7 × 10−9 0.028 (0.026) 0.30 0.070 (0.013) 2.7 × 10−8 
SNP Chr Positiona Gene Allele 1/2b Allele 1 frequencyc GWAS (n = 10 112)
 
Replication study (n = 2742)
 
Meta-analysis (n = 12 854)e
 
      Effect size (SE)d P Effect size (SE)d P Effect size (SE)d P 
rs3093059 157 951 760 CRP G/A 0.12 0.160 (0.019) 5.7 × 10−18 0.174 (0.037) 3.3 × 10−6 0.161 (0.017) 3.5 × 10−21 
rs2097677 22 699 364 IL6 A/G 0.19 0.104 (0.017) 1.6 × 10−9 0.098 (0.034) 0.0040 0.101 (0.015) 4.1 × 10−11 
rs7310409 12 119 909 244 HNF1A G/A 0.47 0.080 (0.014) 3.7 × 10−9 0.028 (0.026) 0.30 0.070 (0.013) 2.7 × 10−8 

CRP, C-reactive protein; GWAS, genome-wide association study.

aPositions of the SNPs were derived from dbSNP build 130.

bThe allele that increased serum CRP levels was denoted as allele 1.

cBased on all of the subjects enrolled in the study.

dEffects of allele 1 on the standardized trait.

eMeta-analysis was performed using an inverse-variance method.

Figure 1.

Quantile–quantile plot of P-values in the GWAS for serum CRP levels. The horizontal axis indicates the expected −log10 (P-values). The vertical axis indicates the observed −log10 (P-values). The gray line represents y = x. The λGC represents the inflation factor of the test statistics.

Figure 1.

Quantile–quantile plot of P-values in the GWAS for serum CRP levels. The horizontal axis indicates the expected −log10 (P-values). The vertical axis indicates the observed −log10 (P-values). The gray line represents y = x. The λGC represents the inflation factor of the test statistics.

Figure 2.

Results of the genome-wide association study (GWAS) for serum CRP levels. (A) A Manhattan plot showing the −log10 (P-values) of SNPs in the GWAS for serum CRP levels in 10 112 Japanese subjects. The gray horizontal line represents the genome-wide significance threshold of P = 5.0 × 10−8. (B) Regional plots of SNPs in the IL6 locus. Diamond-shaped dots represent −log10 (P-values) of SNPs; green indicates the most significantly associated SNP, and the density of the red color represents the r2 value of the most significantly associated SNP. The blue line shows recombination rates given by the HapMap database. The gray horizontal line represents the genome-wide significance threshold of P = 5.0 × 10−8. The lower part indicates RefSeq genes in the locus. The plot of (B) was drawn using SNAP version 2.1 (39).

Figure 2.

Results of the genome-wide association study (GWAS) for serum CRP levels. (A) A Manhattan plot showing the −log10 (P-values) of SNPs in the GWAS for serum CRP levels in 10 112 Japanese subjects. The gray horizontal line represents the genome-wide significance threshold of P = 5.0 × 10−8. (B) Regional plots of SNPs in the IL6 locus. Diamond-shaped dots represent −log10 (P-values) of SNPs; green indicates the most significantly associated SNP, and the density of the red color represents the r2 value of the most significantly associated SNP. The blue line shows recombination rates given by the HapMap database. The gray horizontal line represents the genome-wide significance threshold of P = 5.0 × 10−8. The lower part indicates RefSeq genes in the locus. The plot of (B) was drawn using SNAP version 2.1 (39).

The associations in the previously reported CRP-associated loci are summarized in Table 2. We observed the associations of the SNPs in the IL6R (6,8), CRP (6,810), HNF1A (6,810) and AOPE-CI-CII cluster (6,8,9) loci with the same directional effects (P < 0.01). A suggestive same directional effect was also observed in the GCKR locus (8) (P = 0.035).

Table 2.

Replication analysis of the previously-reported CRP-associated loci

SNP Chr Positiona Gene Band Allele 1/2b Allele 1 frequencyc Effect size (SE)d P Reference 
rs6700896 65 862 370 LEPR 1p31 T/C 0.90 0.016 (0.023) 0.50 (6,8,10
rs8192284 152 693 594 IL6R 1q21 A/C 0.60 0.038 (0.014) 0.0074 (6,8
rs11265260 157 966 663 CRP 1q23 G/A 0.14 0.161 (0.020) 1.4 × 10−16 (6,810
rs780094 27 594 741 GCKR 2p23 T/C 0.56 0.029 (0.014) 0.035 (8
rs10778213 12 102 019 281 Unknown 12q23 C/T 0.74 0.010 (0.016) 0.54 (8
rs1183910 12 119 905 190 HNF1A 12q24 G/A 0.53 0.080 (0.014) 8.7 × 10−9 (6,810
rs4420638 19 50 114 786 APOE-CI-CII cluster 19q13 A/G 0.90 0.135 (0.026) 2.9 × 10−7 (6,8,9
SNP Chr Positiona Gene Band Allele 1/2b Allele 1 frequencyc Effect size (SE)d P Reference 
rs6700896 65 862 370 LEPR 1p31 T/C 0.90 0.016 (0.023) 0.50 (6,8,10
rs8192284 152 693 594 IL6R 1q21 A/C 0.60 0.038 (0.014) 0.0074 (6,8
rs11265260 157 966 663 CRP 1q23 G/A 0.14 0.161 (0.020) 1.4 × 10−16 (6,810
rs780094 27 594 741 GCKR 2p23 T/C 0.56 0.029 (0.014) 0.035 (8
rs10778213 12 102 019 281 Unknown 12q23 C/T 0.74 0.010 (0.016) 0.54 (8
rs1183910 12 119 905 190 HNF1A 12q24 G/A 0.53 0.080 (0.014) 8.7 × 10−9 (6,810
rs4420638 19 50 114 786 APOE-CI-CII cluster 19q13 A/G 0.90 0.135 (0.026) 2.9 × 10−7 (6,8,9

CRP, C-reactive protein.

aPositions of the SNPs were derived from dbSNP build 130.

bThe allele that increased serum CRP levels in the referenced study was denoted as allele 1.

cBased on all of the subjects enrolled in the GWAS for CRP levels.

dEffects of allele 1 on the standardized trait.

Replication study

To validate the associations identified in our GWAS for CRP levels, we performed a replication study with independent 2742 Japanese subjects (Supplementary Material, Table S1). The associations in the CRP locus and the IL6 locus were replicated (P = 3.3 × 10−6 for rs3093059 and P = 0.0040 for rs2097677), and the subsequent meta-analysis confirmed the associations in these two loci (P = 3.5 × 10−21 for rs3093059 and P = 4.1 × 10−11 for rs2097677). The association in the HNF1A locus was not replicated (P = 0.30 for rs7310409) but indicated the same directional effect of the allele, and the result of the meta-analysis satisfied the genome-wide significance threshold (P = 2.7 × 10−8). The combinations of the SNPs in these newly identified or previously reported CRP-associated loci explained 1.4% of the variations of common-log transformed serum CRP levels in our study populations.

Pleiotropic association study for CRP-associated loci

Finally, we evaluated the pleiotropic association of the CRP-associated loci identified in our GWAS by using 30 466 Japanese subjects (Table 3 and Supplementary Material, Table S2). We evaluated the associations with eight hematological traits [white blood cell count (WBC), red blood cell count (RBC), hemoglobin (Hb), hematocrit (Ht), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC) and platelet count (PLT)], three inflammation-related traits [erythrocyte sedimentation rate (ESR), ferritin and serum iron (Fe)] and three serum protein-related traits [serum total protein (TP), serum albumin (ALB) and serum globulin (GLB)]. The CRP-increasing allele (A allele) of rs2097677 in the IL6 locus was significantly associated with increased levels of WBC, PLT and GLB and decreased levels of MCH and MCHC (P < 5.0 × 10−4). The allele also had suggestive associations with increased levels of Fe and decreased levels of Hb and Ht (P < 0.01). Contrary to rs2097677 in the IL6 locus, rs3093059 in the CRP locus and rs7310409 in the HNF1A locus did not have significant associations with the evaluated traits (α = 0.01).

Table 3.

Pleiotropic association analysis of the genetic loci associated with serum CRP levels

Trait No. subjects rs3093059 (CRP)
 
rs2097677 (IL6)
 
rs7310409 (HNF1A)
 
  Effect size (SE)a P Effect size (SE)a P Effect size (SE)a P 
Hematological traits 
 WBC 29 886 −0.0005 (0.011) 0.96 0.039 (0.010) 1.5 × 10−4 0.009 (0.008) 0.26 
 RBC 30 278 0.006 (0.011) 0.57 −0.010 (0.010) 0.31 −0.013 (0.008) 0.090 
 Hb 31 490 0.009 (0.010) 0.40 −0.029 (0.010) 0.0026 −0.016 (0.007) 0.029 
 Ht 31 563 0.008 (0.010) 0.44 −0.026 (0.010) 0.0069 −0.018 (0.007) 0.015 
 MCV 29 845 −0.008 (0.011) 0.46 −0.024 (0.010) 0.015 −0.003 (0.008) 0.71 
 MCH 30 257 −0.0001 (0.011) 0.99 −0.043 (0.010) 1.2 × 10−5 −0.006 (0.008) 0.40 
 MCHC 30 185 0.011 (0.011) 0.31 −0.041 (0.010) 5.5 × 10−5 −0.007 (0.008) 0.39 
 PLT 29 547 −0.012 (0.011) 0.29 0.036 (0.010) 3.7 × 10−4 0.017 (0.007) 0.015 
Inflammation-related traits 
 ESR 2669 −0.005 (0.036) 0.90 0.081 (0.034) 0.017 0.013 (0.026) 0.64 
 Ferritin 214 0.268 (0.141) 0.059 0.237 (0.150) 0.12 0.015 (0.123) 0.90 
 Fe 479 0.003 (0.101) 0.98 0.273 (0.103) 0.0087 −0.005 (0.075) 0.94 
Serum protein-related traits 
 TP 9067 −0.008 (0.021) 0.71 0.049 (0.019) 0.0095 0.005 (0.015) 0.76 
 ALB 9067 0.016 (0.020) 0.43 −0.038 (0.019) 0.039 −0.009 (0.014) 0.52 
 GLB 9065 −0.021 (0.021) 0.30 0.087 (0.019) 3.6 × 10−6 0.013 (0.015) 0.39 
Trait No. subjects rs3093059 (CRP)
 
rs2097677 (IL6)
 
rs7310409 (HNF1A)
 
  Effect size (SE)a P Effect size (SE)a P Effect size (SE)a P 
Hematological traits 
 WBC 29 886 −0.0005 (0.011) 0.96 0.039 (0.010) 1.5 × 10−4 0.009 (0.008) 0.26 
 RBC 30 278 0.006 (0.011) 0.57 −0.010 (0.010) 0.31 −0.013 (0.008) 0.090 
 Hb 31 490 0.009 (0.010) 0.40 −0.029 (0.010) 0.0026 −0.016 (0.007) 0.029 
 Ht 31 563 0.008 (0.010) 0.44 −0.026 (0.010) 0.0069 −0.018 (0.007) 0.015 
 MCV 29 845 −0.008 (0.011) 0.46 −0.024 (0.010) 0.015 −0.003 (0.008) 0.71 
 MCH 30 257 −0.0001 (0.011) 0.99 −0.043 (0.010) 1.2 × 10−5 −0.006 (0.008) 0.40 
 MCHC 30 185 0.011 (0.011) 0.31 −0.041 (0.010) 5.5 × 10−5 −0.007 (0.008) 0.39 
 PLT 29 547 −0.012 (0.011) 0.29 0.036 (0.010) 3.7 × 10−4 0.017 (0.007) 0.015 
Inflammation-related traits 
 ESR 2669 −0.005 (0.036) 0.90 0.081 (0.034) 0.017 0.013 (0.026) 0.64 
 Ferritin 214 0.268 (0.141) 0.059 0.237 (0.150) 0.12 0.015 (0.123) 0.90 
 Fe 479 0.003 (0.101) 0.98 0.273 (0.103) 0.0087 −0.005 (0.075) 0.94 
Serum protein-related traits 
 TP 9067 −0.008 (0.021) 0.71 0.049 (0.019) 0.0095 0.005 (0.015) 0.76 
 ALB 9067 0.016 (0.020) 0.43 −0.038 (0.019) 0.039 −0.009 (0.014) 0.52 
 GLB 9065 −0.021 (0.021) 0.30 0.087 (0.019) 3.6 × 10−6 0.013 (0.015) 0.39 

CRP, C-reactive protein; GWAS, genome-wide association study; WBC, white blood cell; RBC, red blood cell; Hb, hemoglobin; Ht, hematocrit; MCV, mean corpuscular volume; MCH, mean corpuscular hemoglobin; MCHC, mean corpuscular hemoglobin concentration; PLT, platelet; ESR, erythrocyte sedimentation rate; Fe, serum iron; TP, serum total protein; ALB, serum albumin; GLB, serum globulin.

aEffects of CRP-increasing alleles of each of the SNPs on the standardized traits.

Since the existence of correlated traits could modulate the pattern of the pleiotropic associations of the locus, we evaluated the correlations between the traits enrolled in the study (Supplementary Material, Table S3). We observed moderate degrees of correlations (R2 ≥ 0.1) between CRP and ESR, between WBC and PLT, among erythrocyte-related traits (RBC, Hb, Ht, MCV, MCH and MCHC), and among serum protein-related traits. ESR and ferritin also demonstrated wide ranges of moderate correlations with other traits. When the potential effects of the correlations between the traits were considered, 59, 0, 47, 0, 42 and 36% of the observed significant associations of the IL6 locus with CRP, WBC, PLT, MCH, MCHC and GLB were considered to be independent of the associations with the other traits, respectively.

DISCUSSION

Through the GWAS and replication study consisting of 12 854 Japanese subjects, our study identified novel significant associations of the IL6 locus with serum CRP levels. Pleiotropic associations of the variant in the IL6 locus with hematological and biochemical traits (WBC, MCH, MCHC, PLT and GLB) were also demonstrated.

IL6 is a typical pleiotropic pro-inflammatory cytokine that regulates a variety of immune responses (20). At the time of inflammation insult, IL6 promotes the production of acute-phase reactants such as CRP or WBC. IL6 promotes B-cell maturation and immuno-globulin production, which increases the fraction of GLB. IL6 also induces the differentiation of megakaryocytes, which produce platelets (21). Excess hepcidin induced by IL6 causes functional iron deficiency and results in anemia (22), as symbolized by the decreased levels of MCH and MCHC. Therefore, our finding that the variant in the promoter region of IL6 had pleiotropic associations, including with CRP levels, is plausible from biological aspects. As for genetic loci associated with erythrocyte-related traits, Ganesh et al. (23) reported several classification patterns of the pleiotropic associations considering the correlations among the traits. Interestingly, the associations observed in our study population (associated with MCH and MCHC but not associated with RBC) seemed to be not entirely compatible with the reported patterns, which motivate further accumulation of the studies.

IL6 has been recognized as an essential mediator in the pathology of inflammatory diseases such as rheumatoid arthritis and Crohn's disease (24). Inhibition of IL6 is a promising strategy in the treatment of the inflammatory diseases, and the clinical benefits of humanized anti-IL6 receptor antibody have been revealed (25). Although several functional variants located ∼0.5 kb upstream of IL6 had been investigated, their effects on CRP levels or other phenotypes had not been established with substantial evidence (2629). Most of these variants were known to be monomorphic in Asian populations, and in fact, we did not observe significant associations with CRP levels in these variants in our GWAS (α = 0.05; Table 4). These results suggested that the variant in the IL6 locus identified in our GWAS, rs2097677, would be a promising target variant when considering the functional etiology of IL6 in disease pathogenesis or the pharmacogenomics of anti-IL6 therapy.

Table 4.

Association analysis of the previously-investigated SNPs in the IL6 locus with serum CRP levels

SNPa Chr Positionb Allele 1/2 Allele 1 frequencyc Effect size (SE)d P 
rs2069827 22 731 981 T/G 0.00 – – 
rs1800797 (−596 A > G) 22 732 746 A/G 0.00 – – 
rs1800796 (−572 G > C) 22 732 771 C/G 0.77 −0.015 (0.016) 0.37 
rs1800795 (−174 G > C) 22 733 170 C/G 0.00 – – 
rs2069830 22 733 662 T/C 0.00 – – 
SNPa Chr Positionb Allele 1/2 Allele 1 frequencyc Effect size (SE)d P 
rs2069827 22 731 981 T/G 0.00 – – 
rs1800797 (−596 A > G) 22 732 746 A/G 0.00 – – 
rs1800796 (−572 G > C) 22 732 771 C/G 0.77 −0.015 (0.016) 0.37 
rs1800795 (−174 G > C) 22 733 170 C/G 0.00 – – 
rs2069830 22 733 662 T/C 0.00 – – 

CRP, C-reactive protein.

aSNPs that had been investigated in the IL6 locus (2629) were indicated.

bPositions of the SNPs were derived from dbSNP build 130.

cBased on all the subjects enrolled in the GWAS for seum CRP levels.

dEffects of allele 1 on the standardized trait.

Interestingly, pleiotropic associations were not observed in the HNF1A and CRP loci, although their effect sizes on CRP levels were as strong as or even stronger than that in the IL6 locus. Binding of HNF-1α to promoter regions of CRP is involved in the regulation of CRP synthesis in the liver (30). Compared with the pivotal role of IL6, HNF-1α and CRP are specifically involved in downstream events of the synthesis pathway of serum CRP, which would not permit pleiotropic associations with the HNF1A and CRP loci. Previous studies on the CRP-associated loci evaluated the causality of CRP itself based on the Mendelian randomization concept (5,6), assuming the homogeneous contributions of the CRP-associated loci via elevated serum CRP levels. However, our study highlighted the necessity of accounting for heterogeneous contributions on the phenotypes among CRP-associated loci. As for the HNF1A locus, the association in the replication study was not significant (α = 0.05), and its confidence interval of the effect size was not overlap the effect size in the GWAS. Although no significant heterogeneity of the effect was demonstrated when stratified with gender, smoking status or disease groups (α = 0.05 for P-values of Cochran's Q statistics), further evaluations of the HNF1A locus considering gene–gene or gene–environment interactions would be important.

Ethnically different genetic background of CRP is a topic to be investigated since the IL6 locus has not been detected in previous GWASs for serum CRP levels in Caucasian populations (6,810). As well as serum CRP levels, several epidemiological studies have reported the ethnically different distributions of the hematological and biochemical traits enrolled in this study (3133). It is known that Japanese populations have, on average, lower levels of Hb and Ht (33). Thus, assessment of whether the pleiotropic associations of the IL6 locus are observed in other populations would provide useful insight into the etiology and the ethnic heterogeneity of the traits.

In summary, our study demonstrated that the variant in the IL6 locus was significantly associated with serum CRP levels. Pleiotropic associations of the IL6 locus with hematological and biochemical traits were also identified. Our study should contribute to our understanding of the etiology of CRP and inflammatory pathway regulations.

MATERIALS AND METHODS

Subjects

The subjects enrolled in the GWAS and the replication study for CRP levels consisted of 12 854 Japanese patients in 15 disease groups (Supplementary Material, Table S1). The subjects enrolled in the pleiotropic association study for hematological and biochemical traits consisted of 30 466 Japanese patients in 28 disease groups (Supplementary Material, Table S2). All subjects were collected under the support of BioBank Japan Project (http://biobankjp.org) (14). The subjects who were determined to be of non-Japanese origin by self-report or by PCA in the process of the GWAS or our previous studies (3436) were not included. All participants provided written informed consent as approved by the ethical committees of the BioBank Japan Project (14) and the University of Tokyo. Clinical information of the subjects including age (mean ± SD; 64.3 ± 10.8), gender (32.2% for female), smoking history (63.0% for ever-smoker) and BMI (23.1 ± 3.6) were collected by self-report on the questionnaire. The laboratory data including the serum CRP levels measured using high-sensitivity assays [mean ± SD of common log-transformed CRP levels (mg/dl); −0.612 ± 0.595] and other hematological and biochemical traits (Supplementary Material, Table S2) were collected from medical records.

Genotyping and quality control

In the GWAS, 592 232 SNPs were genotyped using the Illumina HumanHap610-Quad Genotyping BeadChip (Illumina, CA, USA). In the process of genotyping, subjects with call rates <0.98 were excluded. Then, SNPs with call rates <0.99 or with ambiguous clustering of the intensity plots, or non-autosomal SNPs, were excluded. We excluded closely related subjects by using identity-by-descent (IBD). We calculated forumla, a genome-wide estimate of IBD, for each pair of subjects by using the ‘-genome' option implemented in PLINK version 1.06 (37). For each pair with forumla > 0.4, we excluded the member of the pair with lower call rates. We also excluded subjects whose ancestries were estimated to be distinct from the other subjects, using PCA performed by EIGENSTRAT version 2.0 (15). We performed PCA for the genotype data of our study along with the genotype data of 60 unrelated European (CEU), 60 unrelated African (YRU) and 89 unrelated East-Asian (44 Japanese and 45 Han Chinese; JPT + CHB) individuals obtained from the Phase II HapMap database (release 24) (16). Based on the PCA plot of the subjects that separated the subjects into three clusters as previously anticipated in the Japanese population (38), we visually identified and excluded the outliers in terms of ancestry from JPT + CHB clusters. We excluded the SNPs with MAF < 0.01 or the SNPs with exact P-value of the Hardy–Weinberg equilibrium test <1.0 × 10−7. Finally, 477 784 SNPs for 10 112 subjects were obtained. PCA was performed again using only the subjects enrolled in the GWAS to obtain eigenvectors used in the following analysis.

Imputation of genotypes was performed using MACH 1.0 (17). The genotype data of JPT and CHB individuals obtained from Phase II HapMap database (release 24) (16) were used as references. In the first step of the imputation, recombination and error rate maps were estimated using 500 subjects randomly selected from the GWAS data. In the second step, imputation of the genotypes of all subjects was performed using the rate maps estimated in the first step. The imputed SNPs with MAF < 0.01 or Rsq values <0.7 were excluded.

In the replication and pleiotropic association studies, we obtained the genotype data of the most significantly associated SNPs in the GWAS from genome-wide screening data of the BioBank Japan Project (14) (Supplementary Material, Tables S1 and S2). The genotyping was performed using the Illumina HumanHap610-Quad Genotyping BeadChip or Illumina HumanHap550v3 Genotyping BeadChip (Illumina, CA, USA), and the same quality control criteria and imputation procedure were applied.

Statistical analysis

Rank-based inverse normal transformation (18) was applied to the serum CRP levels of the subjects. The transformed values were adjusted for gender, age, smoking history, body mass index (BMI) and the affection statuses of the diseases by using linear regression with R statistical software (http://cran.r-project.org). Associations of the SNPs with CRP levels were assessed by linear regression, assuming the additive effects of the allele dosages (bound between 0.0 and 2.0) on the residuals by using mach2qtl software (17). Conditional associations of the SNPs with CRP levels were assessed by additionally incorporating the top 10 eigenvectors into the covariates. Associations of the top 10 eigenvectors with adjusted CRP levels were assessed by multivariate linear regression. For the SNPs that satisfied the genome-wide significance threshold of P < 5.0 × 10−8, their associations with CRP levels in the replication subjects and other hematological and biochemical traits were evaluated in the same manner. Transformation methods for the hematological and biochemical traits are summarized in Supplementary Material, Table S2. Correlation between each pair of the transformed values of the traits was evaluated using correlation coefficient, r, and coefficient of determination, R2. Meta-analysis of the results of the GWAS and the replication study were performed using an inverse-variance method from the summary statistics β and the standard error. We assessed six SNPs that indicated significant associations in our GWAS or in the replication study for previously reported loci (IL6R, CRP, GCKR, IL6, HNF1A and APOE-CI-CII cluster loci) in the analysis for the combined effects of the CRP-associated loci. Explained variances of common-log transformed CRP levels by the SNP(s) were estimated based on the differences of the multivariate linear regression model including the SNP(s), age, gender, smoking status and BMI as covariates, and the model including the covariates other than the SNP(s). For the evaluation of the proportions of the pleiotropic associations of the IL6 locus independent of the other traits, the traits that demonstrated significant associations at the landmark SNP in the IL6 locus (P < 5.0 × 10−4 at rs2097677) and the subjects who was available with all of these traits were selected. The proportion was estimated as the fraction of the explained variance of each of the transformed traits by rs2097677 in the multivariate linear regression model including the other selected traits, age, gender, smoking status and BMI as covariates to that in the model including the covariates other than the selected traits. When the explained variance by rs2097677 in the former model was negative, the proportion was denoted as 0%.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This study was supported by Ministry of Education, Culture, Sports, Science and Technology, Japan.

ACKNOWLEDGEMENTS

We would like to thank all staff of the Laboratory for Statistical Analysis at RIKEN for technical assistance and Dr Koichiro Higasa for his useful advice on statistical analyses.

Conflict of Interest statement. None declared.

REFERENCES

1
Thompson
D.
Pepys
M.B.
Wood
S.P.
The physiological structure of human C-reactive protein and its complex with phosphocholine
Structure
 
1999
7
169
177
2
Pepys
M.B.
Hirschfield
G.M.
C-reactive protein: a critical update
J. Clin. Invest.
 
2003
111
1805
1812
3
Hage
F.G.
Szalai
A.J.
The role of C-reactive protein polymorphisms in inflammation and cardiovascular risk
Curr. Atheroscler. Rep.
 
2009
11
124
130
4
Erlinger
T.P.
Platz
E.A.
Rifai
N.
Helzlsouer
K.J.
C-reactive protein and the risk of incident colorectal cancer
J. Am. Med. Assoc.
 
2004
291
585
590
5
Lawlor
D.A.
Harbord
R.M.
Timpson
N.J.
Lowe
G.D.
Rumley
A.
Gaunt
T.R.
Baker
I.
Yarnell
J.W.
Kivimaki
M.
Kumari
M.
et al.  
The association of C-reactive protein and CRP genotype with coronary heart disease: findings from five studies with 4,610 cases amongst 18,637 participants
PLoS One
 
2008
3
e3011
6
Elliott
P.
Chambers
J.C.
Zhang
W.
Clarke
R.
Hopewell
J.C.
Peden
J.F.
Erdmann
J.
Braund
P.
Engert
J.C.
Bennett
D.
et al.  
Genetic loci associated with C-reactive protein levels and risk of coronary heart disease
J. Am. Med. Assoc.
 
2009
302
37
48
7
Zhang
S.M.
Buring
J.E.
Lee
I.M.
Cook
N.R.
Ridker
P.M.
C-reactive protein levels are not associated with increased risk for colorectal cancer in women
Ann. Intern. Med.
 
2005
142
425
432
8
Ridker
P.M.
Pare
G.
Parker
A.
Zee
R.Y.
Danik
J.S.
Buring
J.E.
Kwiatkowski
D.
Cook
N.R.
Miletich
J.P.
Chasman
D.I.
Loci related to metabolic-syndrome pathways including LEPR,HNF1A, IL6R, and GCKR associate with plasma C-reactive protein: the Women's Genome Health Study
Am. J. Hum. Genet.
 
2008
82
1185
1192
9
Reiner
A.P.
Barber
M.J.
Guan
Y.
Ridker
P.M.
Lange
L.A.
Chasman
D.I.
Walston
J.D.
Cooper
G.M.
Jenny
N.S.
Rieder
M.J.
et al.  
Polymorphisms of the HNF1A gene encoding hepatocyte nuclear factor-1 alpha are associated with C-reactive protein
Am. J. Hum. Genet.
 
2008
82
1193
1201
10
Sabatti
C.
Service
S.K.
Hartikainen
A.L.
Pouta
A.
Ripatti
S.
Brodsky
J.
Jones
C.G.
Zaitlen
N.A.
Varilo
T.
Kaakinen
M.
et al.  
Genome-wide association analysis of metabolic traits in a birth cohort from a founder population
Nat. Genet.
 
2009
41
35
46
11
Lakoski
S.G.
Cushman
M.
Palmas
W.
Blumenthal
R.
D'Agostino
R.B.
Jr.
Herrington
D.M.
The relationship between blood pressure and C-reactive protein in the Multi-Ethnic Study of Atherosclerosis (MESA)
J. Am. Coll. Cardiol.
 
2005
46
1869
1874
12
Khera
A.
McGuire
D.K.
Murphy
S.A.
Stanek
H.G.
Das
S.R.
Vongpatanasin
W.
Wians
F.H.
Jr.
Grundy
S.M.
de Lemos
J.A.
Race and gender differences in C-reactive protein levels
J. Am. Coll. Cardiol.
 
2005
46
464
469
13
Saito
I.
Sato
S.
Nakamura
M.
Kokubo
Y.
Mannami
T.
Adachi
H.
Konishi
M.
Okada
K.
Iso
H.
Kario
K.
et al.  
A low level of C-reactive protein in Japanese adults and its association with cardiovascular risk factors: the Japan NCVC-Collaborative Inflammation Cohort (JNIC) study
Atherosclerosis
 
2007
194
238
244
14
Nakamura
Y.
The BioBank Japan Project
Clin. Adv. Hematol. Oncol.
 
2007
5
696
697
15
Price
A.L.
Patterson
N.J.
Plenge
R.M.
Weinblatt
M.E.
Shadick
N.A.
Reich
D.
Principal components analysis corrects for stratification in genome-wide association studies
Nat. Genet.
 
2006
38
904
909
16
The International HapMap, Consortium
The international HapMap project
Nature
 
2003
426
789
796
17
Li
Y.
Willer
C.
Sanna
S.
Abecasis
G.
Genotype imputation
Annu. Rev. Genomics Hum. Genet.
 
2009
10
387
406
18
Beasley
T.M.
Erickson
S.
Allison
D.B.
Rank-based inverse normal transformations are increasingly used, but are they merited?
Behav. Genet.
 
2009
39
580
595
19
Devlin
B.
Roeder
K.
Genomic control for association studies
Biometrics
 
1999
55
997
1004
20
Naka
T.
Nishimoto
N.
Kishimoto
T.
The paradigm of IL-6: from basic science to medicine
Arthritis Res.
 
2002
4
Suppl. 3
S233
S242
21
Ishibashi
T.
Kimura
H.
Uchida
T.
Kariyone
S.
Friese
P.
Burstein
S.A.
Human interleukin 6 is a direct promoter of maturation of megakaryocytes in vitro
Proc. Natl Acad. Sci. USA
 
1989
86
5953
5957
22
Ganz
T.
Nemeth
E.
Iron sequestration and anemia of inflammation
Semin. Hematol.
 
2009
46
387
393
23
Ganesh
S.K.
Zakai
N.A.
van Rooij
F.J.
Soranzo
N.
Smith
A.V.
Nalls
M.A.
Chen
M.H.
Kottgen
A.
Glazer
N.L.
Dehghan
A.
et al.  
Multiple loci influence erythrocyte phenotypes in the CHARGE Consortium
Nat. Genet.
 
2009
41
1191
1198
24
Nishimoto
N.
Kishimoto
T.
Inhibition of IL-6 for the treatment of inflammatory diseases
Curr. Opin. Pharmacol.
 
2004
4
386
391
25
Nishimoto
N.
Miyasaka
N.
Yamamoto
K.
Kawai
S.
Takeuchi
T.
Azuma
J.
Kishimoto
T.
Study of active controlled tocilizumab monotherapy for rheumatoid arthritis patients with an inadequate response to methotrexate (SATORI): significant reduction in disease activity and serum vascular endothelial growth factor by IL-6 receptor inhibition therapy
Mod. Rheumatol.
 
2009
19
12
19
26
Walston
J.D.
Fallin
M.D.
Cushman
M.
Lange
L.
Psaty
B.
Jenny
N.
Browner
W.
Tracy
R.
Durda
P.
Reiner
A.
IL-6 gene variation is associated with IL-6 and C-reactive protein levels but not cardiovascular outcomes in the Cardiovascular Health Study
Hum. Genet.
 
2007
122
485
494
27
Qi
L.
Zhang
C.
van Dam
R.M.
Hu
F.B.
Interleukin-6 genetic variability and adiposity: associations in two prospective cohorts and systematic review in 26,944 individuals
J. Clin. Endocrinol. Metab.
 
2007
92
3618
3625
28
Sainz
J.
Perez
E.
Gomez-Lopera
S.
Lopez-Fernandez
E.
Moratalla
L.
Oyonarte
S.
Jurado
M.
Genetic variants of IL6 gene promoter influence on C-reactive protein levels but are not associated with susceptibility to invasive pulmonary aspergillosis in haematological patients
Cytokine
 
2008
41
268
278
29
Ognjanovic
S.
Yamamoto
J.
Saltzman
B.
Franke
A.
Ognjanovic
M.
Yokochi
L.
Vogt
T.
Decker
R.
Le Marchand
L.
Serum CRP and IL-6, genetic variants and risk of colorectal adenoma in a multiethnic population
Cancer Causes Control
 
2010
21
1131
1138
30
Li
S.P.
Goldman
N.D.
Regulation of human C-reactive protein gene expression by two synergistic IL-6 responsive elements
Biochemistry
 
1996
35
9060
9068
31
Bain
B.
Seed
M.
Godsland
I.
Normal values for peripheral blood white cell counts in women of four different ethnic origins
J. Clin. Pathol.
 
1984
37
188
193
32
Pan
Y.
Jackson
R.T.
Ethnic difference in the relationship between acute inflammation and serum ferritin in US adult males
Epidemiol. Infect.
 
2008
136
421
431
33
Tsubakihara
Y.
Nishi
S.
Akiba
T.
Hirakata
H.
Iseki
K.
Kubota
M.
Kuriyama
S.
Komatsu
Y.
Suzuki
M.
Nakai
S.
et al.  
2008 Japanese Society for Dialysis Therapy: guidelines for renal anemia in chronic kidney disease
Ther. Apher. Dial.
 
2010
14
240
275
34
Kamatani
Y.
Matsuda
K.
Okada
Y.
Kubo
M.
Hosono
N.
Daigo
Y.
Nakamura
Y.
Kamatani
N.
Genome-wide association study of hematological and biochemical traits in a Japanese population
Nat. Genet.
 
2010
42
210
215
35
Okada
Y.
Kamatani
Y.
Takahashi
A.
Matsuda
K.
Hosono
N.
Ohmiya
H.
Daigo
Y.
Yamamoto
K.
Kubo
M.
Nakamura
Y.
et al.  
Common variations in PSMD3-CSF3 and PLCB4 are associated with neutrophil count
Hum. Mol. Genet.
 
2010
19
2079
2085
36
Okada
Y.
Kamatani
Y.
Takahashi
A.
Matsuda
K.
Hosono
N.
Ohmiya
H.
Daigo
Y.
Yamamoto
K.
Kubo
M.
Nakamura
Y.
et al.  
A genome-wide association study in 19 633 Japanese subjects identified LHX3-QSOX2 and IGF1 as adult height loci
Hum. Mol. Genet.
 
2010
19
2303
2312
37
Purcell
S.
Neale
B.
Todd-Brown
K.
Thomas
L.
Ferreira
M.A.
Bender
D.
Maller
J.
Sklar
P.
de Bakker
P.I.
Daly
M.J.
et al.  
PLINK: a tool set for whole-genome association and population-based linkage analyses
Am. J. Hum. Genet.
 
2007
81
559
575
38
Yamaguchi-Kabata
Y.
Nakazono
K.
Takahashi
A.
Saito
S.
Hosono
N.
Kubo
M.
Nakamura
Y.
Kamatani
N.
Japanese population structure, based on SNP genotypes from 7003 individuals compared to other ethnic groups: effects on population-based association studies
Am. J. Hum. Genet.
 
2008
83
445
456
39
Johnson
A.D.
Handsaker
R.E.
Pulit
S.L.
Nizzari
M.M.
O'Donnell
C.J.
de Bakker
P.I.
SNAP: a web-based tool for identification and annotation of proxy SNPs using HapMap
Bioinformatics
 
2008
24
2938
2939