Plasma levels of high-sensitivity C-reactive protein (hsCRP) are an important predictor of cardiovascular disease, and achievement of lower targets of hsCRP with rosuvastatin treatment was associated with improved cardiovascular outcomes. The aim of this study was to examine whether hsCRP levels were related to genetic variants and traditional cardiovascular risk factors in Chinese patients treated with rosuvastatin 10 mg/day. The relations were analyzed between on-treatment plasma hsCRP concentrations and cardiovascular risk factors and 14 single-nucleotide polymorphisms in CRP and other candidate genes. In 281 patients with a median plasma hsCRP level of 0.81 mg/L (interquartile range 0.46 to 1.86), higher hsCRP levels were significantly associated with female gender, greater waist circumference (WC), having diabetes, higher triglycerides, and lower high-density lipoprotein (HDL) cholesterol. Three single-nucleotide polymorphisms (rs1205, 3872G>A and rs2808630, 5237A>G in CRP and rs1169288, I27L in HNF1A) were independently associated with hsCRP levels before and after adjustment for other variables. WC and the CRP rs1205 polymorphism showed the strongest relations with hsCRP, and in multiple regression analysis, gender, WC, diabetes, triglycerides, HDL cholesterol, and the 3 genetic variants explained 35.5% of the variance in hsCRP levels. The 2 CRP polymorphisms, female gender, higher WC, and lower HDL cholesterol were associated with risk for having CRP concentrations ≥1 mg/L. In conclusion, central obesity, low HDL cholesterol, and CRP polymorphisms are major determinants of higher hsCRP levels in Chinese patients receiving treatment with rosuvastatin.
Previous studies have shown that even during statin therapy, plasma high-sensitivity C-reactive protein (hsCRP) levels were still associated with multiple cardiovascular risk factors. However, the effect of CRP gene polymorphisms on hsCRP levels during treatment with statins has not been reported. Therefore, we examined the associations of hsCRP with genetic variants and cardiovascular risk factors in a group of Chinese patients taking rosuvastatin to explore whether the previously reported relations persisted during statin treatment.
Methods
The methods of this study have been described in detail previously. In brief, Han Chinese patients aged ≥18 years with established coronary heart disease or coronary heart disease risk equivalent including some with familial hypercholesterolemia (FH) who were recruited for a pharmacogenetic study of the lipid response to rosuvastatin 10 mg/day were included for the analysis. Patients with uncontrolled diabetes mellitus, hypertension or thyroid disease, and significant renal or hepatic dysfunction; those who had experienced cardiovascular disease (CVD) events within the 3 months before recruitment; and those with poor adherence with rosuvastatin therapy (reporting taking <80% or >120% of the prescribed number of tablets) were excluded. Six subjects with hsCRP levels >10 mg/L were also excluded from the analysis because this high level of hsCRP is likely to be due to acute illness. Patients were interviewed by research staff members after ≥4 weeks of treatment with rosuvastatin 10 mg/day (>97% of patients had ≥6 weeks of treatment with rosuvastatin) to assess drug adherence by asking patients about their medication-taking behavior in a nonjudgmental manner and tablet counting. Subjects with poor compliance with therapy were included if they agreed to improve compliance and have a further assessment after another 4 weeks of treatment. Anthropometric measurements, including body weight, waist circumference (WC), hip circumference, and estimation of percentage body fat using an impedance device (Tanita Body Composition Analyzer BF-350; Tanita Corporation of America, Arlington Heights, Illinois) were performed by a research nurse, and fasting blood samples were taken for deoxyribonucleic acid extraction, lipid profile, laboratory safety tests, and hsCRP measurements at the study visit. The study protocol was approved by the local clinical research ethics committee, and all participants gave written informed consent.
The lipid and laboratory safety parameters were measured by routine methods. Rosuvastatin treatment was well tolerated in all patients, and there were no symptomatic side effects or abnormal laboratory safety test results. The plasma hsCRP concentration was determined using an immunonephelometric method (Siemens Dade Behring CardioPhase hsCRP assay; Dade Behring, Newark, Delaware) on the Siemens BN ProSpec System (Siemens Medical Solutions USA, Inc., Mountain View, California) with a detection limit of 0.146 mg/L (the measurement range was 0.146 to 9.35) and interassay coefficients of variation of 2.5%, 3.8%, and 2.1% at hsCRP concentrations of 0.5, 1.3, and 2.1 mg/L, respectively.
Two common single-nucleotide polymorphisms (SNPs) in CRP (3872G>A [rs1205] and 5237A>G [rs2808630]) were selected on the basis of previous findings. Some common SNPs in genes related to inflammation and/or CRP were also selected, including the Ile27Leu (rs1169288) polymorphism in HNF1A, Cys112Arg (rs429358, 334T>C) and Cys158Arg (rs7412, 472C>T) in APOE, rs4420638 in the APOE-CI-CII cluster, and Lys109Arg (rs1137100) and Gln223Arg (rs1137101) in LEPR. In addition, 6 common SNPs in genes potentially related to the pharmacokinetics of rosuvastatin (CYP2C9*#x002A;3 [1075A>C, rs1057910], CYP2C19*#x002A;2 [681G>A, rs4244285] and CYP2C19*#x002A;3 [636G>A, rs4986893], ABCG2 421C>A [rs2231142], SLCO1B1 388A>G [rs2306283], and 521T>C [rs4149056]) were also chosen to examine whether these variants might affect hsCRP levels by altering the pharmacokinetics of rosuvastatin. The ABCG2 421C>A polymorphism is a major determinant of rosuvastatin pharmacokinetics and the higher plasma levels in East Asians. Subjects with the 421A variant allele have reductions in low-density lipoprotein (LDL) cholesterol and on-treatment LDL cholesterol levels similar to those seen with double the dose of rosuvastatin in those with the 421C/C genotype. Genotyping was performed at the Genome Research Centre, University of Hong Kong, using the mass spectroscopy–based, high-throughput MassARRAY iPLEXTM platform (Sequenom, San Diego, California). All SNPs genotyped were in Hardy-Weinberg equilibrium (chi-square test p >0.05).
For subjects with hsCRP levels less than the limit of detection, a value of 0.1 mg/L was assigned. HsCRP levels were log transformed for analysis to fulfill the model assumption of residual normality. Univariate and multivariate regression analysis were performed to identify the clinical covariates associated with log hsCRP. Clinical covariates examined in the study included age, gender, body mass index, percentage of total body fat, WC, waist-to-hip ratio, smoking status, previous CVD, diabetes mellitus, hypertension, high-density lipoprotein cholesterol, LDL cholesterol, triglycerides, and concomitant medications. Clinical variables with p values <0.05 were retained in the stepwise selection models. Genetic covariates were added to the model to examine the association between genetic polymorphisms and log hsCRP. Effects of genetic polymorphisms on log hsCRP were assessed by analysis of covariance followed by a post hoc Bonferroni test after correction for the effect of clinical variables and other genetic variables with p values <0.05 in the multivariate model. Finally, multivariate logistic regression analysis was performed to examine covariates associated with having an elevated hsCRP level of ≥1 mg/L. To adjust for multiple testing, we calculated an experiment-wide significance level using the conservative method of Bonferroni (0.05/14 = 0.0036), and only those polymorphisms with probability values less than this level were considered statistically significant. Data were analyzed using SPSS version 17.0 (SPSS, Inc., Chicago, Illinois).
Results
The clinical characteristics of 281 subjects with good adherence to rosuvastatin and having hsCRP levels <10 mg/L, including 13 subjects with hsCRP levels less than the limit of detection, are listed in Table 1 .
| Characteristic | Total | Male | Female | p Value |
|---|---|---|---|---|
| (n = 281) | (n = 137) | (n = 144) | ||
| Age (years) | 57 (49–64) | 55 (48–64) | 58 (50–65) | 0.077 |
| Body mass index (kg/m 2 ) | 24.9 (22.7–27.1) | 25.2 (23.3–27.0) | 24.5 (22.0–27.2) | 0.130 |
| Waist circumference (cm) | 86.0 (79.0–93.8) | 89.0 (83.2–99.0) | 83.5 (76.3–90.5) | <0.001 |
| Waist-to-hip ratio | 0.89 (0.84–0.94) | 0.91 (0.87–0.97) | 0.87 (0.81–0.92) | <0.001 |
| Body fat (%) | 28.7 (23.6–35.3) | 24.5 (21.7–28.4) | 34.2 (29.2–39.4) | <0.001 |
| Systolic blood pressure (mm Hg) | 123 (113–135) | 123 (114–135) | 123 (110–134) | 0.300 |
| Diastolic blood pressure (mm Hg) | 75 (66–82) | 78 (72–86) | 71 (63–80) | <0.001 |
| Heart rate (beats/min) | 69 (63–77) | 70 (63–76) | 68 (62–77) | 0.566 |
| Diabetes mellitus | 71 (25.2%) | 38 (27.7%) | 33 (22.9%) | 0.410 |
| Hypertension | 151 (53.7%) | 77 (56.2%) | 74 (52.1%) | 0.473 |
| Previous cardiovascular disease | 48 (17.1%) | 30 (21.9%) | 18 (12.5%) | 0.040 |
| Current smokers | 36 (12.8%) | 32 (23.4%) | 4 (2.8%) | <0.001 |
| High-sensitivity C-reactive protein (mg/L) | 0.81 (0.46–1.86) | 0.81 (0.48–1.76) | 0.83 (0.40–1.87) | 0.909 |
| Low-density lipoprotein cholesterol | ||||
| mmol/L | 2.4 (1.9–3.3) | 2.4 (1.8–3.3) | 2.4 (1.9–3.2) | 0.815 |
| mg/dl | 93 (73–127) | 93 (69–127) | 93 (73–124) | |
| High-density lipoprotein cholesterol | ||||
| mmol/L | 1.50 (1.24–1.76) | 1.33 (1.20–1.55) | 1.60 (1.38–1.98) | <0.001 |
| mg/dl | 58 (48–68) | 51 (46–60) | 62 (53–76) | |
| Triglycerides | ||||
| mmol/L | 1.34 (0.92–1.88) | 1.40 (0.98–1.99) | 1.20 (0.90–1.79) | 0.067 |
| mg/dl | 119 (81–167) | 124 (87–176) | 106 (80–159) | |
| Total cholesterol | ||||
| mmol/L | 4.7 (4.0–5.5) | 4.6 (3.8–5.6) | 4.85 (4.2–5.4) | 0.092 |
| mg/dl | 181 (154–212) | 178 (147–216) | 187 (162–208) | |
| Fasting plasma glucose | ||||
| mmol/L | 5.4 (4.9–6.3) | 5.4 (4.9–6.5) | 5.4 (5.0–6.3) | 0.803 |
| mg/dl | 97 (88–113) | 97 (88–117) | 97 (90–113) | |
| Concomitant medications | ||||
| Antihypertensive and antiarrhythmic drugs | 151 (53.7%) | 80 (58.4%) | 71 (49.3%) | 0.151 |
| Antidiabetic drugs | 41 (14.6%) | 21 (15.3%) | 20 (13.9%) | 0.739 |
| Aspirin | 41 (14.6%) | 28 (20.4%) | 13 (9.0%) | 0.011 |
| Non-steroidal anti-inflammatory drugs | 21 (7.5%) | 6 (4.4%) | 15 (10.4%) | 0.069 |
In univariate analysis, all covariates except smoking, previous CVD, LDL cholesterol level, and concomitant use of aspirin or nonsteroidal anti-inflammatory drugs were significantly associated with serum hsCRP levels (p <0.05). WC (p <0.001), body mass index (p <0.001), waist-to-hip ratio (p <0.001), and percentage body fat (p <0.001) were significantly associated with log hsCRP in univariate analysis, but the associations of body mass index, waist-to-hip ratio, and percentage body fat with hsCRP were not significant after adjustment for WC and other variables, suggesting collinearity between these obesity indexes, so only WC was included in the multivariate analysis. Multivariate analysis showed that female gender, greater WC, having diabetes, higher triglycerides, and lower high-density lipoprotein cholesterol level were nongenetic variables significantly associated with higher hsCRP levels.
Regression analysis was then performed for each SNP with adjustment for nongenetic variables, which showed that among all the SNPs examined, the rs1169288 in HNF1A (p <0.003) and rs1205 (p <0.001) and rs2808630 (p <0.001) in CRP were independently associated with log hsCRP levels after correction for multiple testing. Subjects with 1 or 2 copies of variant alleles of any of these SNPs had reduced hsCRP levels compared to those with wild-type alleles ( Figure 1 ). HsCRP levels differed across 6 diplotypes of rs1205 and rs2808630 in CRP (p <0.001) with subjects with wild-type alleles (GG/AA) having higher hsCRP levels, those with 2 mutated alleles having the lowest hsCRP levels, and carriers of 1 mutated allele having intermediate levels ( Figure 1 ). None of the polymorphisms in genes potentially related to the pharmacokinetics of rosuvastatin was found to be associated with hsCRP levels ( Table 2 ).
| Polymorphism | Plasma hsCRP Level (mg/L) | ||
|---|---|---|---|
| 11 ⁎ | 12 ⁎ | 22 ⁎ | |
| CYP2C9 ⁎ 2 | 0.814 (0.440–1.860) | 0.832 (0.478–2.508) | |
| (n = 265) | (n = 16) | ||
| CYP2C19 ⁎ 2 | 0.853 (0.439–1.860) | 0.804 (0.430–1.950) | 0.730 (0.343–1.690) |
| (n = 111) | (n = 132) | (n = 19) | |
| CYP2C19 ⁎ 3 | 0.808 (0.457–1.780) | 0.956 (0.475–2.230) | |
| (n = 255) | (n = 23) | ||
| ABCG2 421C>A | 0.732 (0.377–1.848) | 0.847 (0.474–1.720) | 1.040 (0.574–2.255) |
| (n = 132) | (n = 112) | (n = 37) | |
| SLCO1B1 388A>G | 0.648 (0.230–1.471) | 0.695 (0.423–1.660) | 0.884 (0.473–1.920) |
| (n = 16) | (n = 93) | (n = 151) | |
| SLCO1B1 521T>C | 0.788 (0.463–1.900) | 0.800 (0.410–1.485) | 0.100 (0.967–2.220) |
| (n = 212) | (n = 61) | (n = 6) | |
WC and the rs1205 polymorphism were the most significant factors associated with hsCRP in both genders in multivariate models. HsCRP concentrations increased with increasing WC within each rs1205 genotype group. Similarly, within each WC tertile, increasing numbers of the rs1205 variant allele were associated with lower hsCRP levels (overall p <0.001; Figure 2 ). In addition, there was no difference in WC or all other obesity parameters among rs1205 genotype groups, suggesting that there was no relation between obesity and the rs1205 polymorphism. These 3 polymorphisms and the cardiovascular risk factors explained totally 35.5% of the variance in hsCRP levels. WC explained the largest proportion of the variance in hsCRP concentrations (19.1%), and the 3 genetic variables together explained 8.5% of the variance. Including body mass index, waist-to-hip ratio, or percentage body fat in the multivariate model neither explained a higher proportion of variance in hsCRP nor changed the associations between genetic factors and hsCRP.
Multivariate logistic regression analysis showed that compared to those with homozygous wild-type alleles of CRP rs1205, subjects with variant alleles had a reduced risk for having hsCRP levels ≥1 mg/L (n = 117; adjusted odds ratio 0.249, 95% confidence interval 0.103 to 0.605 for 1 variant allele, p <0.005; odds ratio 0.114, 95% confidence interval 0.039 to 0.336 for 2 variant alleles, p <0.001). The CRP rs2808630 polymorphism was also inversely associated with having hsCRP concentrations ≥1 mg/L. Female gender, higher WC, and lower high-density lipoprotein cholesterol level were also predictors for having hsCRP levels ≥1 mg/L. Every 10-cm increase in WC was associated with a 1.8-fold increased risk for having hsCRP level ≥1 mg/L. Smoking, hypertension, and diabetes also tended to increase the risk for having an hsCRP level ≥1 mg/L ( Table 3 ). Similar predictors were observed when using 2 mg/L as a cut-off point.
