Interleukin-6 Receptor Gene, Plasma C-Reactive Protein, and Diabetes Risk in Women

The Nurses’ Health Study began in 1976 with the recruitment of 121,700 female registered nurses (aged 30–55 years). Between 1989 and 1990, 32,826 women provided blood. The medical history and lifestyle information were updated every 2 years using a questionnaire (10). Samples for the present study were selected from women who provided a blood sample and were free from diabetes, cardiovascular disease, stroke, or cancer at the time of blood collection. Incident cases of type 2 diabetes were defined as self-reported diabetes confirmed by a validated supplementary questionnaire and diagnosed at least 1 year after blood collection. We used National Diabetes Data Group criteria (11) to define diabetes, because our subjects were diagnosed before the release of the American Diabetes Association criteria in 1997. The validity of this method has been confirmed (12). We used the American Diabetes Association diagnostic criteria for diagnosis of diabetes cases after the 1998 cycle (13). The incident cases were matched to control subjects who did not report physician-diagnosed diabetes on age, month and year of blood draw, and fasting status (14,15). In total, 1,325 European Caucasian women (633 diabetic patients and 692 control subjects) with plasma IL-6 and CRP measures were included.

Blood sample collection (between 1989 and 1990) and processing were previously described (14,16). The assays were performed in 2003 using the stored blood samples (in the vapor phase of liquid nitrogen freezers; the highest temperature is −130°C). Frozen plasma aliquots from case and control subjects were selected for simultaneous analysis. Study samples were analyzed in randomly ordered case-control pairs to further reduce systematic bias and interassay variation. CRP levels were measured via a high-sensitivity latex-enhanced immunonephelometric assay on a BN II analyzer (Dade Behring, Newark, DE). Plasma concentrations of IL-6 were measured using a quantitative sandwich enzyme immunoassay technique (Quantikine HS Immunoassay kit). The coefficient of variation was 3.8% for CRP and 5.9% for IL-6 (7,16). BMI was calculated as weight in kilograms divided by the square of height in meters. Physical activity was expressed as metabolic equivalent task (MET) hours based on self-reported types and durations of activities over the previous year.

DNA was extracted from the buffy coat fraction of centrifuged blood using the QIAmp Blood kit (Qiagen, Chatsworth, CA). Ten linkage disequilibrium tagging SNPs (rs4845618, rs12083537, rs4075015, rs6684439, rs4845622, rs8192284, rs4329505, rs4240872, rs2229238, and rs4845617) for IL6R were selected as previously described (7). The tagging SNPs cover ∼94% allele variance. The polymorphisms were genotyped using Taqman SNP allelic discrimination by means of an ABI 7900HT (Applied Biosystems, Foster City, CA). Replicate quality-control samples (10%) were included and genotyped with >99% concordance. All genotypes fit Hardy-Weinberg equilibrium (HWE).

A χ² test was used to assess whether the genotypes were in HWE. General linear models were used to compare geometric mean values of quantitative traits across groups. Plasma CRP and IL-6 were logarithmically transformed to improve the normality. We adjusted for covariates, including age (continuous), BMI (<23, 23–24.9, 25–29.9, 30–34.9, or 35 kg/m²), physical activity (<1.5, 1.5–5.9, 6.0–11.9, 12–20.9, or 21.0 MET h/week), smoking (never, past, or current), alcohol intake (nondrinker or drinker [0.1–4.9, 5–10, or >10 g/day]), family history of diabetes, menopausal status (pre- or postmenopausal [never, past, or current hormone use]), and IL-6 levels (in quartiles). Because of the missing data in biomarkers and genotyping, some matching pairs in the case-control design were broken. Therefore, we used unconditional logistic regression as the primary analysis in estimating odds ratios (ORs) for diabetes risk to avoid the loss of unpaired samples. Conditional logistic regression yielded similar results (data not shown). The interactions between polymorphisms and biomarkers were examined using likelihood ratio test, with a comparison of the log likelihood of the two models with and without the interaction terms.

To account for multiple statistical testing, we calculated false discovery rate (FDR) for the analyses on the polymorphisms by the method of Benjamini and Hochberg (17) using SAS procedure PROC MULTTEST. FDR estimates the proportion of results declared positive that are actually false (18). The SAS statistical package was used for the analyses (SAS, version 8.2 for UNIX). Haplotype analysis was conducted based on the Stochastic-EM algorithm using THESIAS program (19). All P values are two sided.

Table 1 shows age and age-adjusted baseline characteristics according to rs8192284 genotypes for women with both CRP and IL-6 measurements available. IL-6 levels were higher in women with AC and CC genotypes compared with those with AA genotype. The genotypes were not associated with any other characteristics.

In nondiabetic women, the three SNPs rs6684439, rs4845622, and rs8192284, which were in strong linkage disequilibrium, were significantly associated with low plasma CRP levels after adjusting for IL-6 and other covariates (Table 2). SNPs rs12083537 and rs4329505 were associated with high plasma CRP levels. We calculated FDR by the method of Benjamini and Hochberg (17) to adjust for the multiple testing. The FDRs for all of these associations were <0.05. In most cases, similar associations between IL6R polymorphisms and CRP levels were observed in the diabetic patients.

We inferred the haplotypes from the polymorphisms within the linkage disequilibrium block (including SNPs rs6684439, rs4845622, rs8192284, rs4329505, rs4240872, rs2229238, and rs4845617). Because rs6684439, rs4845622, and rs8192284 are in nearly perfect linkage disequilibrium (r² > 0.9), only rs8192284 was kept in haplotype inference, together with the other four SNPs in this linkage disequilibrium block (rs4845618, rs4329505, rs4240872, and rs2229238). Five common haplotypes accounted for ∼96% allele variance of the linkage disequilibrium block. Haplotypes 21122 (P = 0.039; 1, the common allele; 2, the minor allele) and 11211 (P = 0.0006) were associated with 0.13 (0.01–0.26) and 0.23 (0.10–0.37) pg/ml higher CRP levels, respectively, compared with the most common haplotype 12111, adjusting for IL-6 and other covariates (Fig. 1). The global test for haplotype associations was statistically significant (P = 0.0002).

Because SNP rs8192284 was a missense variant, related to CRP in the GWAS, and showed the strongest association with IL-6 levels in our previous study (7), we further tested the interactions between this SNP and CRP levels in relation to diabetes risk and observed significant multiplicative interaction (P for interaction = 0.026; Fig. 2). We then examined the stratified associations between CRP levels and diabetes risk by rs8192284 genotypes. To improve power, we grouped the subjects into carriers and noncarriers of the minor allele C. After adjusting for IL-6 levels and other covariates, the ORs across increasing quartiles of CRP were 2.21 (95% CI 1.18–4.12), 3.77 (1.87–7.57), and 5.02 (2.4–10.5) in the noncarriers; whereas in the carriers, the ORs were 2.19 (1.42–3.36), 2.03 (1.27–3.23), and 2.92 (1.77–4.82), respectively.

We found significant associations between IL6R variants and plasma CRP levels. Our results are consistent with the findings from recent GWASs (9,20). We demonstrated that the associations between IL6R variants and CRP levels are independent of IL-6 concentration. In an exploratory analysis, we found IL6R variant rs8192284 significantly interacts with CRP levels in relation to diabetes risk. The associations between CRP levels and increased diabetes risk are more evident in women carrying the wild-type genotype than in those with the minor allele C.

IL-6 is a pleiotropic cytokine that performs as the chief stimulator of the production of CRP from the liver (21). In epidemiological studies, circulating levels of IL-6 and CRP are significantly correlated (22). In our earlier analyses, we found that IL6R variants are significantly related to IL-6 levels (7). The same association was also observed in the Health ABC study (8). Our data from the present study suggest that the association between IL6R variants and CRP levels is unlikely due to the changes in IL-6 levels, because the IL6R gene–CRP association was independent of IL-6 levels.

Because CRP and IL-6 are not the direct products of the IL6R gene, the associations between IL6R variants and these biomarkers are likely mediated by other metabolic changes. SNP rs8192284 in the IL6R gene has been associated with soluble IL6R levels (8). These data suggest that IL6R variant may primarily affect IL6R levels and that the changes in CRP and IL-6 are likely secondary.

The precise mechanisms underlying the opposite associations of IL6R variant (rs8192284) with IL-6 and CRP are not clear. The differing associations are particularly puzzling considering that IL-6 may stimulate the product of CRP in the liver. However, the data from our study are highly consistent with the GWAS and previous studies (8,9,20). We suspect that the genotype-related changes in CRP and IL-6 levels may be parallel changes, rather than sequential events, both induced by the alterations in IL6R products.

Subclinical systemic inflammation is now considered an important mechanism leading to insulin resistance and type 2 diabetes (23). Epidemiological studies have documented that circulating inflammatory markers, including CRP, significantly predict diabetes risk (4–6). Previous studies indicate that polymorphisms affecting CRP levels may also influence the risk of type 2 diabetes (24). Although IL6R SNPs were not significantly associated with the incidence of diabetes in our study sample (7), the exploratory analysis indicated that the genetic variant might modify the association between CRP levels and diabetes risk. The associations between CRP and increased risk of type 2 diabetes are more evident in women carrying genotype AA of rs8192284 compared with those carrying the minor allele C. This observation reflects a synergic effect of IL6R genotype and CRP levels on the development of diabetes. The observed interaction needs to be confirmed in future studies.

Several limitations need to be considered. SNP rs4129267 reported by the GWAS (9) was not typed in the present study. However, this SNP is in near perfect linkage disequilibrium with rs8192284 (D′ = 1 and r² = 0.96; HapMap, CEU). Population stratification arising from ethnic admixture may cause spurious associations. However, the present study was less likely to be influenced by population stratification because the study populations were highly homogeneous, including only European Caucasians. In addition, our analyses were restricted to women and therefore may not be generalized to men.

In summary, we demonstrated that the IL6R variants are significantly associated with plasma CRP levels, independent of IL-6 levels. In addition, IL6R variant interacts with CRP in relation to diabetes risk. Further research is warranted to elucidate the potential mechanisms underlying the associations between IL6R variants and the opposite changes in CRP and IL-6 levels.

L.Q. has received an American Heart Association Scientist Development Award and Boston Obesity Nutrition Research Center Grant DK-46200. This study was supported by National Institutes of Health grants DK-58845 and CA-87969.

No potential conflicts of interest relevant to this article were reported.