Introduction

Arteriosclerosis is associated with increased cardiovascular events [1], and also an independent predictor of these events [2, 3]. Arteriosclerosis decreases arterial elasticity and increases arterial stiffness, which shows that the pulse wave velocity (PWV) increases clinically. Carotid-femoral PWV is frequently considered as the standard measurement of arterial stiffness [4]. Inflammation plays an important role in the genesis and development of arteriosclerosis. And C-reactive protein (CRP) is one of the most sensitive markers of inflammation and tissue damage [5, 6]. Furthermore, numerous studies have shown that CRP is an independent risk factor of increased PWV among populations with metabolic syndrome, acute coronary syndrome, hypertension, and severe asthma [7,8,9,10,11,12].

The association between inflammation marker CRP and PWV has been confirmed; however, most of the related studies only using a single CRP concentration measurement to explore their association. Furthermore, serum CRP can be influenced by many factors, such as age, gender, race, lifestyle, and infections [13, 14]. Therefore, using a single CRP would not accurately reflect the association between CRP and PWV. Cumulative C-reactive protein (cumCPR) considers exposure dose and time sufficiently, and it is more precise than a single CRP. To the best of our knowledge, few studies concerning the association between cumCRP and brachial–ankle pulse wave velocity (baPWV). The Kailuan study (Trial identification: ChiCTR-TNC-11001489) was based on the risk factors of cardiovascular diseases in a functional community. The community received comprehensive cardiovascular risk factor assessments including CRP and baPWV, which can be used to analyze the association between cumCRP and baPWV.

Materials and methods

Population

We collected the data from the Kailuan community in Tangshan, a city in northern China. This community represented the Chinese population from a socioeconomic perspective. The Kailuan study is composed of independent data set from the staff of Kailuan community and also is an ongoing prospective cohort study which medical examination started in 2006–2007 in Tangshan. Detailed study design and procedures have been published previously [15]. Briefly, between 2006–2007 and 2007–2010 (visit 1), there were 101,510 participants (81,110 males and 20,400 females) were recruited from 11 hospitals of Kailuan community. All participants were then underwent repeated questionnaire assessment, clinical and laboratory tests every 2 years, i.e., in the years 2008 and 2009 (visit 2), in the years 2010 and 2011 (visit 3). Among them, 22,622 participants underwent baPWV assessment between 2010 and 2015. In our study, we excluded 7190 participants who lack of data for any two CRP values of the three medical examinations. Finally, 15,432 participants included in our analysis.

The protocol for this study was performed according to the guidelines of the Helsinki Declaration and approved by the Ethics Committee of the Kailuan Medical Group, Kailuan Company. All of the participants provided written informed consent.

Epidemiological survey and anthropometric parameters

The epidemiological survey and anthropometric parameters were in accordance with the previous published articles by our research group [16]. CRP was measured by high-sensitivity nephelometry assay (Cias Latex CRP-H, Kanto Chemical, Tokyo, Japan). All blood samples were tested using an auto-analyzer (Hitachi 747; Hitachi, Tokyo, Japan) in the central laboratory of Kailuan General Hospital. The laboratory proficiency testing value was 100%, as assessed by the Ministry of Health in 2006–2009. The two other samples were tested twice a day with 2 h intervals and lasted for 20 days. The precision was evaluated using a Hitachi 7600 automatic biochemical analyzer. Smoking was defined as having at least one cigarette a day in the recent years; drinking was defined as having 100 ml/day (alcohol content > 50%) for more than 1 year; physical training was defined as having aerobic exercise (e.g., walking, jogging, balls, and swimming) for ≥ 3 times/week at ≥ 30 min/time.

Assessment of baPWV and ankle brachial index (ABI)

All the baPWV and ABI data were read from BP-203RPEIII Internet (Oumulong). The temperature of the examination room was stable at 22 °C to 25°. And smoking was not allowed and more than 5 min of rest was essential prior to the test. Age, gender, height, and weight of forehead were recorded. At the beginning of the measurement, the participant stayed calm and laid horizontally, and then keep the blood pressure cuffs which were attached to the upper arm and ankle, at the same time, with the balloon sign of the upper arm cuff aligned with the brachial artery, and guaranteed the balloon sign of the leg cuff placed preaxially. The cardiechema collecting device was placed at the precordial region, with the electrocardiography acquisition device clipped to left and right wrists. The specific measurement method in detail can be found in the published paper of our research group [17]. The test was repeated twice and the second readings were considered the final value. The larger values of the left-side and right-side baPWV and the smaller values of the left-side and right-side ABI were used for further analysis. BaPWV < 1400 cm/s was considered as normal and baPWV ≥ 1400 cm/s was considered as arterial stiffness [18]. And the reproducibility of PWV measurement was evaluated by professionals trained by Kailuan Research Kailuan General Hospital.

Assessment of cumCRP and mean arterial pressure (MAP)

CumCRP was calculated according to the method of cumulative exposure to heart rate [19] and blood pressure [20]. The formula was as follows: cumCRP = [(CRP1 + CRP2)/2 × time1–2] + [(CRP2 + CRP3)/2 × time2–3], where CRP1, CRP2 and CRP3 are the first, second, and third physical examinations for CRP, respectively; time1–2 and time2–3 are the time intervals of two adjacent measurements; and cumCRP is the cumulative C-reactive protein. Mean CRP (meanCRP) was calculated as follows: meanCRP = (CRP1 + CRP2 + CRP3)/3, where CRP1, CRP2, and CRP3 are the first, second and third physical examinations for CRP, respectively. MAP was calculated as follows [21]: MAP = (systolic pressure + 2×diastolic pressure)/3.

Statistical analysis

Participants were divided into four groups according to the quartile of cumCRP exposure levels. The physical examination data from 2006 to 2007, 2008 to 2009, and 2010 to 2011 were extracted by professional workers and used in creating an Oracle 10.2 database (Oracle, USA). BaPWV data were established into a database by Epidata 3.0 (EpiData Association, Denmark), and analyses were performed using SPSS 13.0 (IBM, USA). Data in normal distribution were presented as mean ± standard deviation (x ± s). Measurement data in multiple groups were compared with one-way ANOVA and LSD-t/Dunnett’s T test. When the participants lost any CRP measurement of the three examinations (n = 3499), we used the average value of the other two results. Mean CRP was in a skewed distribution and logarithmically transformed. Some CRP values presented 0 and could not be logarithmically transformed. Thus, we used 0.05 mg/L for calculation. These values showed normal distribution after logarithmic transformation and were analyzed by variance. Enumeration data were presented as N (%). Intergroup and rate comparisons were conducted using Chi-square test. Multivariable logistic regression analysis was applied to analyze the influence of different cumCRP groups and the increase in each standard deviation on baPWV. P < 0.05 was considered statistically significant. To avoid the influence of acute inflammation, infection, lipid-lowering agent, diabetes, hypertension, cardiovascular disease, and peripheral arterial diseases, we applied sensitivity analysis as follows: (i) exclusion of CRP > 10 mg/L, (ii) exclusion of those taking lipid-lowering agents, (iii) exclusion of those with hypertension, diabetes, and cardiovascular diseases when taking the first physical examination in 2006 and new onset during 2006–2010, (iv) exclusion of ABI < 0.9 and (v) CRP values were completed in three examinations.

Results

Participant characteristics

The participants (n = 15,432) with 61.2% male (n = 9443) and 38.8% female (n = 5989) had a mean age of 51.52 ± 11.11 years at the baseline. The four groups of cumCRP were: Q1: cumCRP < 2.72 mg/L year; Q2: 2.72 < cumCRP < 5.08 mg/L year; Q3: 5.08 < cumCRP < 9.60 mg/L year; Q4: cumCRP3 > 9.60 mg/L year. When the cumCRP increased from Q1 to Q4, the age, heart rate, systolic blood pressure (SBP), diastolic blood pressure (DBP), MAP, Visit1_lgCRP, Visit2_lgCRP, Visit3_lgCRP, lgmeanCRP, body mass index (BMI), fasting blood glucose (FBG), TC, low-density lipoprotein cholesterin (LDL-C), physical training, hypertension, diabetes, and use of hypotensive drugs and lipid-lowering drugs were significant increases (P < 0.001, respectively) (Table 1). And we also compared the characteristics of participants between inclusion and exclusion population at the baseline (Supplementary Table 1).

Table 1 Baseline characteristics of the study population by quartile of cumCRP exposure levels
Full size table

The mean baPWV and corresponding incidence of arterial stiffness of each group

The mean baPWV values of each group were 1425.70, 1499.54, 1557.62, and 1626.48 cm/s, respectively. From Q1 to Q4, the corresponding incidence of arterial stiffness was 44.7%, 55.5%, 61.5%, and 70.1% (P < 0.001). Results were the same in the male and female groups and showed statistical significance (P < 0.001) (Table 2).

Table 2 The incidence of arterial stiffness according to the quartile of cumCRP exposure levels
Full size table

Multivariable logistic regression analysis of the arterial stiffness

Using arterial stiffness as the dependent variable, cumCRP or cumCRP (+1SD) was the independent variable (Q1 was the control group) in the multivariable logistic regression analysis. In model 3, after adjusted for age, gender, BMI, LDL-C, FBG, baseline CRP, MAP, smoking, alcohol intake, physical exercise, the OR was 1.42 (95% CI 1.24–1.63) of Q4. Of note, in a fully adjusted regression model, when the cumCRP increased per SD, the OR was 1.10 (95% CI 1.02–1.18) (Table 3).

Table 3 Logistic regression analyses for the presence of arterial stiffness according to the quartile of cumCRP exposure levels
Full size table

In addition, the baseline CRP quartiles also exhibited a predictive value for arterial stiffness (Supplementary Table 2).

Sensitivity analysis

We conducted the sensitivity analysis as follows: (i) exclusion of CRP > 10 mg/L, (ii) exclusion of those taking lipid-lowering agents, (iii) exclusion of hypertension, diabetes, and cardiovascular diseases when taking the first physical examination in 2006 and the new onset during 2006–2010, (iv) exclusion of ABI < 0.9, and (v) CRP values were completed in three examinations. Results were in agreement with those previously described (Table 4).

Table 4 The sensitivity analysis: adjusted odds ratios and 95% confidence intervals for arterial stiffness according to the quartile of cumCRP exposure levels
Full size table

Discussion

Age, SBP, FBG, smoking, HDL-C, and other related factors were well established associated with arterial stiffness [22,23,24,25], and inflammation played a vital role in the development and progression of article stiffness. We discovered that with cumCRP increased, the incidence of article stiffness and mean value of baPWV also increased. The incidence of arterial stiffness of Q1 and Q4 was 44.7% and 70.1%, respectively. The mean value of Q1 was 1425.70 cm/s and that of Q4 was 1626.48 cm/s. The trend existed in different gender groups. These results were similar to an early cross-sectional study about the association between single CRP and baPWV. The cross-sectional study researched the middle-aged and elderly people in Japan, and found that the mean baPWV in males increased from 1358 cm/s (hs-CRP Q1) to 1381 cm/s (hs-CRP Q4) (P < 0.01) and that in females also increased from 1241 to 1266 cm/s [26].

After adjusting the confounding factors, the risk of arterial stiffness in the cumCRP exposure group increased by 1.42-fold more in Q4 than that in Q1. In addition, cumCRP increased per standard deviation, and the risk of arterial stiffness increased by 1.10-fold. Furthermore, the risk of arterial stiffness remained unchanged when cumCRP increased, except for those with CRP > 10 mg/L; taking lipid-lowering drugs; with hypertension, diabetes, and cardiovascular diseases; ABI < 0.9; and lacking any one of the CRP values. A prospective study concluded that in rheumatoid arthritis patients, the risk of arterial stiffness was high in the CRP ≥ 5.31 mg/L group, which is 4.84-fold higher than that in the CRP < 5.31 mg/L group [27].

In addition, we also found that there was association between the baseline CRP and arterial stiffness (Supplementary Table 2), but when baseline CRP was introduced into cumCRP quartiles (Table 3), no significance was observed in predicting arterial stiffness (OR= 1.0; 95% CI 0.99–1.02). Similarly, a study included 107 middle-aged males found that the baseline CRP has no relationship with baseline baPWV, and the results remained the same after 1 year of follow-up [28]. However, our results were contradictory to those of previous studies about the association between baseline CRP and arterial stiffness. McEniery et al. [29] studied 825 middle-aged males and pointed out that baseline CRP was both related to baseline aortic pulse wave velocity (aPWV) and aPWV of a 20-year follow-up. In another study that included 3769 Europeans which followed up 16 years, the baseline CRP was correlated with the aPWV of both genders [30]. Thus, single CRP measurement showed an unidentified relationship with PWV. This result might be because the single-time point measurement of CRP does not consider the long-lasting effects on the individual body. Our study applied the cumCRP measurement, which is made up of deficiencies, and might have precisely indicated the association between CRP exposure and arterial stiffness.

We discovered that higher cumCRP could promote the increase in baPWV and increase the incidence of arterial stiffness, and its possible pathogenesis is as follows: (1) as a biomarker of inflammation, CRP is related to the increased risk factors in cardiovascular diseases; in addition, baPWV is a sensitivity index of arteriosclerosis [31, 32] that could predict cardiovascular events [33,34,35]. Increased cumCRP expedited the baPWV, possibly by enhancing the risk factors of cardiovascular diseases. (2) CRP could suppress the mobilization and differentiation of endothelial progenitor cells (EPCs) and newly formed blood vessels by down-regulating the expression of endothelial nitric oxide synthesis. In addition, CRP could promote the apoptosis of endothelial cells. Endothelial cells receive no repairs for a prolonged period, which results in endothelial dysfunction and increases aortic stiffness [36, 37]. (3) CRP might be related with medial vascular calcification, which increases aortic stiffness and PWV [38,39,40,41]. CRP induces vascular endothelial injury in different ways. The damaged endothelial tissues affected with inflammatory factors exhibit hyaline degeneration and necrosis. Elastic fiber thickening and muscle fiber fracture may increase aortic stiffness. Such injury is a chronic pathological process, because it will take years for vascular sclerosis to occur.

Although we confirmed the association between cumCRP and baPWV among Chinese population, our analysis still has some limitations. First, the study population showed a weak balance in both genders, which might have influenced the representative effect of our cohort. The popularity remains to be confirmed. Second, although we adjusted the impact of various confounding factors in multivariable logistic regression analysis, several factors influencing the results were remain unadjusted, such as the white coat effect and the environmental changes. Third, the absence of CRP values (n = 3499) might have influenced the results accordingly, but we carried out a sensitivity analysis, and the results of which had no effect on our findings. Finally, our study only confirmed the association between cumCRP and arterial stiffness, but did not specifically determine their causal relationship.

In conclusion, our study provided that further evidence of high cumCRP exposure is associated with increased arterial stiffness.