Distribution and determinants of serum high-sensitive C-reactive protein in a population of young adults. The Cardiovascular Risk in Young Finns Study

Authors


Olli T. Raitakari, Department of Clinical Physiology, PO Box 52, 20521 Turku, Finland.
(fax: +358 2 313 1666; e-mail: olli.raitakari@utu.fi).

Abstract.

Objectives.  Elevated C-reactive protein (CRP) is a suggested risk marker for cardiovascular disease. We aimed at investigating the distribution and determinants of CRP levels in young adults.

Design.  Population-based study.

Subjects.  A total of 2120 participants aged 24–39 years.

Main outcome measures.  Distribution of CRP, and the relationship between CRP and risk factors.

Results.  CRP concentration (mean ± SD) was 1.43 ± 3.26 mg L−1 in men, 1.36 ± 2.36 mg L−1 in women who did not use oral contraceptives (OC) and 3.69 ± 6.01 mg L−1 in women who used OCs. In total, 8.8% of men, 10.3% of non-OC user women and 35.3% of OC user women had CRP concentration >3 mg L−1 (recommended cut-off point of high risk for cardiovascular disease). In univariate analysis, CRP was associated with obesity indices and physical activity amongst both sexes. In men, the multivariate correlates of CRP included waist circumference (P < 0.0001), smoking (<0.0001) and HDL cholesterol (P = 0.024) (inverse association). These three variables explained 21.9% (model R2) of the total variation in CRP, waist circumference having the greatest influence (partial R2 = 19.6%). In women, the multivariate correlates of CRP included OC use (P < 0.0001), body mass index (BMI) (P < 0.0001), triglycerides (<0.0001) and physical activity (P = 0.025) (inverse association). These four variables explained 38.2% (model R2) of the total variation in CRP, with OC use (partial R2 = 18.4%) and BMI (partial R2 = 18.0%) having the greatest influence.

Conclusions.  The determinants of CRP level include obesity and smoking in men, and obesity, OC use and physical activity in women. About one in three of healthy women who use OCs have CRP concentration exceeding 3 mg L−1.

Introduction

In recent years the pathophysiology of atherosclerosis has been shown to include local inflammation as a promoter to plaque formation [1], as well as a trigger to plaque instability [2]. Within various markers of inflammation, high-sensitive C-reactive protein (CRP) has shown to be a predictor of the risk of cardiovascular events [3]. CRP has assay characteristics of detecting very low concentrations, and increases in CRP within the normal range seem to predict future vascular events in apparently healthy asymptomatic individuals [4]. In a meta-analysis of prospective studies, individuals in the top tertile of baseline CRP concentration had a risk ratio of about 1.5 for the development of cardiovascular disease compared with those in the bottom tertile [5]. CRP has shown to retain an independent association with coronary events after adjusting for age, total cholesterol, HDL cholesterol, smoking, body mass index (BMI), history of hypertension or diabetes, and family history of coronary disease [3]. This may suggest a direct involvement in the pathophysiology of atherosclerosis. A statement for healthcare professionals was recently published from the Centers for Disease Control and Prevention and the American Heart Association addressing recommendations for the use of CRP in assessing cardiovascular risk [6]. The cut-off points for low, average or high risk are based on distributions of CRP in over 15 population studies performed in adults [6]. The currently recommended plasma CRP cut-off points are <1.0 mg L−1 for low risk, 1.0–3.0 mg L−1 for average risk and >3.0 mg L−1 for high risk.

In the present study, we investigated the distribution of CRP and the population determinants of CRP levels, including lifestyle variables and conventional risk markers, in a cohort of 2021 subjects aged 24–39 years participating in the Cardiovascular Risk in Young Finns Study. Age- and sex-specific distribution profiles for CRP were determined and the associations between CRP and body size indices, serum lipids, blood pressure and lifestyle were analysed.

Methods

Subjects

The Cardiovascular Risk in Young Finns Study is an ongoing, five-centre follow-up study of atherosclerosis precursors in Finnish children and adolescents. The first cross-sectional survey was conducted in 1980, when 3596 participants, aged 3, 6, 9, 12, 15 and 18 years were randomly chosen from the national population register [7]. In 2001, we re-examined 2283 of these individuals, now aged 24–39 years. Subjects gave written informed consent and the study was approved by the local ethics committees.

Clinical characteristics and risk factors

Height and weight were measured, and BMI was calculated [8]. Waist and hip circumference was measured with an accuracy of 0.1 cm. Blood pressure was measured with a random zero sphygmomanometer, and an average of three measurements was used in the analysis. Smoking habits, alcohol use, oral contraceptive (OC) use, physical activity, history of recent infection and chronic rheumatic disease use were enquired with a questionnaire. Physical activity index was constructed by combining the information on the frequency, intensity and duration of physical activity, including leisure time physical activity and commuting to the workplace. Participants were asked about the frequency of participation in physical activity and its intensity outside school and working hours. Participants were offered multiple-choice answers. A metabolic equivalent index (hours/week) for physical activity was calculated from the product of intensity, frequency, duration of leisure time physical activity, as well as commuting to the workplace. When estimating physical activity during the journey to work, we considered the length of the journey and whether it was travelled by foot or by bicycle. The coefficients for the variables were estimated from existing tables [9].

For the determination of serum lipoprotein levels, venous blood samples were drawn after an overnight fast. All lipid determinations were carried out using standard methods [8]. LDL cholesterol concentration was calculated by the Friedewald formula. Serum CRP was analysed by an automated analyser (Olympus AU400; Tokyo, Japan) and a highly sensitive turbidimetric immunoassay kit (‘CRP-UL’-assay; Wako Chemicals, Neuss, Germany). Detection limit of the assay was 0.06 mg L−1. The inter-assay coefficient of variation was 3.33% at the mean level of 1.52 mg L−1 (n = 116) and 2.65% at the mean level of 2.51 mg L−1 (n = 168). Glucose concentrations were analysed enzymatically (Olympus Diagnostica GmbH, Hamburg, Germany).

We excluded subjects with diabetes (n = 13, CRP 4.29 ± 5.46 mg L−1), chronic rheumatic disease (n = 34, CRP 5.14 ± 7.31 mg L−1), history of recent infection (n = 113, CRP 3.34 ± 6.62 mg L−1), pregnant women (n = 61, CRP 4.63 ± 4.87 mg L−1) and lactating women (n = 53, CRP 2.48 ± 5.27 mg L−1) from all analyses. We did not define any arbitrary cut-off points to exclude subjects with high CRP values (for example, values exceeding 10 mg L−1).

Statistical analyses

Values for CRP and triglycerides were log-transformed prior to analyses due to skewed distributions. The univariate associations between CRP and other variables were examined using regression analysis. Stepwise multiple regression analyses were performed thereafter to test the variables that contributed significantly to the variation of CRP concentrations. Smoking was modelled as a dichotomous variable (smoking/nonsmoking). Alcohol use was categorized as the number of standard drinks. Comparisons between the several groups were performed using analysis of variance with Bonferroni multiple comparison procedure to allow pair-wise testing (Fig. 1). Sex was analysed separately. All analyses were performed using Statistical Analysis System (version 8.1), and statistical significance was inferred at a two-tailed P-value <0.05.

Figure 1.

Women who use oral contraceptives (OC users) had significantly higher C-reactive protein (CRP) concentration than non-OC users, who had CRP concentration comparable to men. Median CRP values and interquartile range are shown. P-values are derived from statistical tests comparing log-transformed CRP values across the groups.

Results

The characteristics of the study subjects are shown in Table 1. The mean CRP was higher in women than in men (P < 0.0001). This sex difference was largely due to OC use. Women who used OCs had an average of 2.4 mg L−1 higher CRP concentration than nonusers, who had CRP levels comparable to men (Fig. 1). The nonuser group included 111 women with levonorgestrel-releasing intrauterine device. In these women, CRP concentration was comparable to nonusers (1.48 ± 2.05 mg L−1 vs. 1.33 ± 1.51 mg L−1, P = 0.88).

Table 1.  Characteristics of the study subjects [mean (SD)]
 Men (n = 975)Women (n = 1046)
  1. CRP, C-reactive protein. aData available: men, n = 839; women, n = 806.

Age (years)31.7 (5.1)31.7 (5.1)
CRP (mg L−1)1.43 (3.26)2.01 (3.90)
CRP (mg L−1) median (interquartile range)0.59 (0.29–1.38)0.81 (0.34–2.17)
BMI (kg m−2)25.7 (4.1)24.4 (4.7)
Waist circumference (cm)89.6 (10.7)78.8 (11.4)
Waist-to-hip ratio0.90 (0.06)0.79 (0.06)
Systolic blood pressure (mmHg)121 (12)113 (13)
Diastolic blood pressure (mmHg)73 (11)69 (10)
Total cholesterol (mmol L−1)5.3 (1.0)5.0 (0.9)
LDL cholesterol (mmol L−1)3.4 (0.9)3.1 (0.8)
HDL cholesterol (mmol L−1)1.17 (0.28)1.39 (0.30)
Triglycerides (mmol L−1)1.5 (0.1)1.4 (0.3)
Physical activity index (%) (range)a16 (0–113)17 (0–76)
Daily smoking (%)3020.4
Oral contraceptive use (%) 28.2

The univariate correlates of CRP are shown in Table 2. In both sexes, CRP correlated directly with obesity indices, blood pressure, total cholesterol, LDL cholesterol and triglycerides, and inversely with physical activity index. In addition, in men, CRP correlated directly with smoking and inversely with HDL cholesterol. Age was directly correlated with CRP levels in men, and inversely in women. Alcohol use was not associated with CRP in either sex. We also tested for potential nonlinear relationship between alcohol use and CRP, as reported previously [10, 11]. CRP concentration was 1.66 ± 2.89 mg L−1 in nondrinkers (n = 514), 1.78 ± 4.07 mg L−1 in subjects (n = 1265) drinking one to two drinks per day, 1.52 ± 2.15 mg L−1 in subjects (n = 178) drinking three to four drinks per day, and 1.86 ± 2.86 mg L−1 in subjects (n = 43) drinking greater than four drinks per day. Quadratic term for alcohol use (P = 0.91) did not support a nonlinear relationship between CRP and alcohol consumption.

Table 2.  Univariate associations between log10-transformed CRP and other study variables
VariableMen (n = 975)Women (n = 1046)
Regression coefficientSE PRegression coefficientSE P
  1. CRP, C-reactive protein; BMI, body mass index; SE, standard error of regression coefficient.

Age (years)0.00900.00310.0036−0.01160.00330.0004
BMI (kg m−2)0.05050.0035<0.00010.04980.0033<0.0001
Waist (mm)0.00200.0001<0.00010.00190.0001<0.0001
Waist-to-hip ratio2.86740.2326<0.00012.01620.2599<0.0001
Systolic blood pressure (mmHg)0.00650.0013<0.00010.00720.0013<0.0001
Diastolic blood pressure (mmHg)0.00710.0014<0.00010.00980.0016<0.0001
Total cholesterol (mmol L−1)0.07310.0149<0.00010.10070.0189<0.0001
LDL cholesterol (mmol L−1)0.07020.0175<0.00010.04470.02220.0444
HDL cholesterol (mmol L−1)−0.31220.0550<0.00010.03620.05610.5190
Triglycerides (mmol L−1)0.10460.0155<0.00010.24600.0239<0.0001
Smoking (daily)0.11200.03420.00110.00690.04120.8987
Oral contraceptive use 0.45630.0345<0.0001
Exercise−0.00200.00100.0393−0.00270.00130.0402
Alcohol (no. drinks per week)0.00220.00150.1534−0.00130.00290.6413

The correlates of CRP that were significant in the univariate analysis were included in stepwise multivariate analysis. In men, the multivariate correlates of CRP included waist circumference (P < 0.0001), smoking (<0.0001) and HDL cholesterol (P = 0.024) (inverse association). Together these three variables explained 21.9% (model R2) of the total variation in CRP, with waist circumference having the greatest influence (partial R2 = 19.6%). In women, the multivariate correlates of CRP included OC use (P < 0.0001), BMI (P < 0.0001), triglycerides (P < 0.0001) and physical activity (P = 0.025) (inverse association). These four variables explained 38.2% (model R2) of the total variation in CRP, with OC use (partial R2 = 18.4%) and BMI (partial R2 = 18.0%) having the greatest influence.

Amongst men, 65.4% had CRP concentration <1 mg L−1, 25.7% had concentration between 1 and 3 mg L−1, and 8.8% had concentration >3 mg L−1. This distribution was similar amongst women who did not use OCs; 66.8% had CRP concentration <1 mg L−1, 22.9% had concentration between 1 and 3 mg L−1, and 10.3% had concentration >3 mg L−1. Amongst women who used OCs, 29.5% had CRP concentration <1 mg L−1, 35.3% had concentration between 1 and 3 mg L−1, and 35.3% had concentration >3 mg L−1. The distribution of CRP in the study population presented as percentiles is shown in Table 3.

Table 3.  Mean CRP concentrations (mg L−1) in young adults at selected percentile cut-off points
nAge (years)Mean ± SDPercentiles
01510255075909599100
  1. CRP, C-reactive protein; OC, oral contraceptives. Subjects with diabetes, recent infection, rheumatic disease, as well as pregnant women and lactating women excluded.

Women (no OC use)
80241.40 ± 2.880.070.070.100.150.220.521.062.875.8318.5518.55
108271.43 ± 2.190.060.100.110.140.360.631.753.344.9712.0014.25
114301.61 ± 2.960.070.080.130.180.280.582.013.304.9715.9021.45
148331.33 ± 2.560.050.080.130.160.260.531.472.754.0111.5024.00
149361.38 ± 2.070.100.110.140.180.290.591.423.384.8111.8012.00
152391.10 ± 1.610.050.080.110.150.240.571.252.834.038.6412.40
Women (OC users)
75243.38 ± 8.110.120.120.140.460.841.913.338.5013.5564.9064.90
70272.90 ± 4.260.190.190.340.460.971.703.714.658.4032.3532.35
50303.20 ± 4.140.190.190.240.370.691.474.267.589.7520.9520.95
45334.84 ± 6.240.230.230.400.571.152.684.6514.7515.0030.8030.80
28363.87 ± 4.110.330.330.410.480.973.204.6512.6013.2016.5516.55
27394.01 ± 7.250.140.140.150.230.490.912.8720.3024.0526.0026.00
Men
155241.33 ± 3.130.070.070.100.140.260.531.192.324.0520.5028.05
140271.45 ± 3.180.030.070.130.160.270.521.212.885.7219.5023.50
177301.24 ± 2.490.050.070.130.150.270.551.252.363.0915.5024.70
164331.49 ± 3.580.070.070.150.180.300.721.342.453.7529.0030.75
178361.59 ± 4.510.070.110.130.200.310.571.383.504.6810.6556.40
161391.46 ± 2.040.060.070.130.200.370.781.693.384.2412.6013.70

Discussion

With the recent development of a high-sensitivity CRP assay, increased CRP levels in otherwise healthy normal adults have been linked to increased risk for cardiovascular disease independently of traditional risk factors [12]. CRP is an acute-phase reactant produced mainly by the liver and for some extent by the adipose tissue, and released in response to cytokine production induced by inflammatory stimuli [13]. In the present study, we examined the distribution profiles and population determinants of CRP in Finnish young adults. We found that 9–10% of young adults had CRP concentration above 3 mg L−1, which is the currently recommended cut-off point for high risk. The prevalence was higher amongst women who used OCs, as 35% of these women had CRP >3 mg L−1.

The distribution of CRP values in healthy subjects in Europe has been previously studied by Imhof et al. [14]. They found a trend towards higher CRP concentrations with increasing age. In a sample of 13 527 individuals aged 25–74 years from the former West Germany, France and Scotland, the median CRP values for men and women (not taking OCs or hormone replacement therapy) up to 44 years of age were 0.6–1.1 mg L−1, and 1.2–1.7 mg L−1 amongst those 45 years and older. In our population of young adults aged 24–39 years, the median values of CRP varied between 0.5 and 0.8 mg L−1 in different categories stratified by age and sex.

As in several previous studies [13], the CRP levels were independently associated with obesity indices in both men and women. The possible mechanism for this association may include increased interleukin-6 production by adipocytes, stimulating hepatic CRP production [15]. In men, other independent correlates of CRP included smoking and HDL cholesterol. The underlying mechanism for smoking to increase CRP concentrations is unknown, but indirect consequences of the tissue-damaging effects of smoking and smokers’ increased susceptibility to respiratory infections have been suggested [13]. The inverse association between HDL cholesterol and CRP, also observed in previous studies [13, 16, 17], may be related to the upregulation of pro-inflammatory mechanisms induced by low levels of HDL cholesterol [17, 18].

In women, the most important determinant of CRP concentration was OC use. Women who used OCs had about 2.4 mg L−1 higher CRP levels compared with nonusers. Both OC use [19, 20] and hormone replacement therapy in postmenopausal women [21–24] have been shown to increase the level of CRP. The mechanisms by which hormones affect CRP concentrations, as well as the clinical significance of these changes are unknown. However, oral oestrogens have been suggested to be relevant in the pathophysiology of both venous and arterial thromboembolism [25, 26]. This may reflect the possible pro-inflammatory effects of oral oestrogens. Some previous studies have reported higher CRP levels in women compared with men [13]. In the present study, however, young women who did not use OCs had comparable CRP concentrations compared with men. In agreement, Imhof et al. [14] found that the distribution of CRP was similar between men and women who did not use OCs or hormone replacement therapy.

We found that the CRP levels were inversely associated with the degree of physical activity in univariate analysis in both sexes. The inverse relationship between CRP and physical activity has also been reported previously [27–30]. The mechanism(s) of this association are unknown, but part of the effect of physical activity is likely to be mediated by changes in fat mass and decreased adipocyte production of interleukin-6 [31, 32]. In women, however, this association remained significant also in multivariate analysis, thus suggesting an obesity-independent effect of physical activity on CRP concentrations.

The mechanisms as to how CRP may be involved in the pathophysiology of cardiovascular diseases are being actively studied. In addition to being a marker of inflammation, CRP has proinflammatory effects on endothelial cells [33]. CRP induces a dose-dependent decrease in the expression of endothelial nitric oxide synthase, which may lead to endothelial dysfunction promoting the development of atherosclerosis [34]. Along with stimulation of monocyte chemotaxis [35], CRP may be able to opsonize native LDL and mediate the uptake of CRP/LDL by macrophages [36]. CRP may also influence atherosclerotic vessels by activation of the complement system, thereby promoting inflammation and thrombosis [37]. Thus elevated CRP levels may reflect increased atherosclerotic burden and/or higher tendency for plaque rupture/thrombosis. Studies examining the associations between CRP and direct measures of atherosclerosis have given contradictory results. Higher CRP has been related to increased carotid artery atherosclerosis in some studies [38, 39], but not all [40]. In the Young Finns population, we have not been able to show a significant association between CRP and carotid artery intima-media thickness [41].

In summary, we found that in young adults obesity is an important lifestyle determinant of CRP, explaining about 20% of the variation in both men and women. In women, another important determinant is contraceptive use, which also explains about 20% in variation in CRP levels. Weaker lifestyle correlates include physical activity, which is inversely associated with CRP in both sexes, and smoking, which turned out to be a significant correlate in men.

Conflict of interest statement

No conflict of interest was declared.

Acknowledgements

This study was financially supported by the Academy of Finland (grant nos 53392 and 34316), the Social Insurance Institution of Finland, the Turku University Foundation, the Juho Vainio Foundation, the Yrjö Jahnsson Foundation, Research fund from the Turku University Hospital and the Finnish Foundation of Cardiovascular Research.

Ancillary