Association of high-sensitivity C-reactive protein with fatal and non-fatal cardiovascular events in working-age people: data from the ESSE-RF study

Aim. To study the relationship of different levels of high-sensitivity C-reactive protein (hs-CRP) with cardiovascular events and assess its contribution to the development of outcomes in Russian regions. Material and methods. The work used the data from the multicenter study ESSE-RF — a representative sample of male and female population aged 25-64 years. All participants signed informed consent. The study included 10421 people (women, 6399 (61,4%)). The cohort was followed up from 2012 to 2019 (median follow-up period, 5,5 years). A hard endpoint (cardiovascular mortality and nonfatal myocardial infarction (MI)) was determined in 187 people, while a soft endpoint (nonfatal MI, stroke, revascularization, heart failure progression and cardiovascular mortality) — in 319 people. Results. The results showed that hs-CRP is significantly associated with the main risk factors (with the exception of low-density lipoproteins). At the same time, it was found that optimal hs-CRP level for predicting the risk of cardiovascular events (CVE) in Russian population is significantly lower than 3 mg/L, but higher than 1 mg/L (1,54/1,89 mg/dL for men and women, respectively). Adding hs-CRP to sex and age significantly improved risk prediction (AUC, 79,7; 95% CI, 77,8-81,7). At the same time, adding a wide list of confounders to hs-CRP, sex and age does not improve the model’s predictive value (AUC, 79,7; 78,2-82,1). Conclusion. This study for the first time showed a significant independent contribution of hs-CRP to CVEs development in the Russian population, and the addition of hs-CRP to sex and age significantly increased the predictive value of model. cm for men and women, respectively), glucose ≥5,6 mmol/L. — adjusted for study regions. Abbreviations: HTN — hypertension, hs-CRP — highly sensitive C-reactive protein, CI — confidence interval, TG — triglycerides, HDL-C — high density lipoprotein cholesterol, LDL-C — low density lipoprotein cholesterol, AUC — area under the ROC curve.

Based on numerous studies, now there is a concept of additional or novel risk factors (RFs), which, together with the classical factors, increase the predictive power. In recent years, the researchers ac tively searched for novel biomarkers.
One of these biomarkers is C-reactive protein (CRP), which is a marker of inflammation. This unusual, "bifacial" biomarker brings us more and more surprises. Dozens of years have passed since the CRP discovery by Tilet W and Francis T (1930) [1], but a new CRP role was identified after the discovery of a highsensitivity technique for measuring CRP (1994CRP ( -1997, which allowed researchers to look at the problems of nonspecific inflammation from a different angle. Subsequently, it became known that high-sensitivity CRP (hs-CRP) level increases sharply during the inflammatory response, causing coronary artery damage due to the direct activation of endothelial cells. Factors causing artery spasm contribute to the formation of microthrombi and impaired microcirculation, which has a strong effect on plaque development. In an experimental study, it was shown that this process requires a transition from pentameric CRP to its monomeric conformation (mCRP) [2].
Given the discovery of vesicular transport system, we now know that exosomes can carry different cell components, including proteins and lipids [3]. The study by Melnikov I. S. et al. (2019) suggested that in patients with coronary artery disease (CAD), blood increase in circulating exosomes carrying mCRP and leukocyte markers on their surface, as well as the mCRP in atherosclerotic areas, may indicate the participation of mCRP in the pathogenesis of coronary atherosclerosis [4].
In 2001, the American Heart Association published guidelines for determining hs-CRP to stratify the risk of cardiovascular diseases (CVD), in particular CAD, in the moderate risk group [5].
Finally, the Justification for the Use of statins in Primary prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER study) showed that statins, regardless of total cholesterol (TC) levels, reduced hs-CRP concentration, which decreased the risk of cardiovascular events (СVE) [6].
However, despite the evidence from numerous studies, the data are contradictory, especially when it comes to the relationship between hs-CRP and the death risk among population.
The aim was to study the relationship of different levels of high-sensitivity C-reactive protein (hs-CRP) with cardiovascular events and assess its contribution to the development of outcomes in Russian regions.

Material and methods
The object of the ESSE-RF multicenter study was representative samples of male and female population aged 25-64 years from 6 Russian regions (Vladivostok, Vologda, Ivanovo, St. Petersburg, Tomsk, Tyumen). A multistage stratified sample was selected using the Kish grid [7]. The study was approved by the local ethics committee. All subjects signed informed consent. In general, the response rate was 80,0%. A detailed protocol for the ESSE-RF study was published earlier [8].
The presence of a prior disease was considered with a positive answer to the question: "Have a doctor ever told you that you have/had the following diseases?: hypertension (HTN), CAD (angina), myocardial infarction (MI), stroke (cerebral vessel thrombosis or hemorrhage), rheumatoid arthritis (RA), cancer, chronic bronchitis, Parkinson's disease (PD)". Smoking status was assessed: never smoked, former smoking, current smoking.
Instrumental diagnostic tests. Blood pressure (BP) was measured twice on the right arm in a sitting posture with a brief in between break (2 min) using an automatic BP monitor OMRON M3 Expert (Japan).
All anthropometric measurements were performed routinely, as in most studies. In the study, the body mass index (BMI) was calculated using the Quetelet's equation (BMI = Height (m) / Weight (kg 2 )). Abdominal obesity (AO) was considered as a waist circumference ≥102/88 cm for men and women.
Laboratory diagnostic tests. In all centers, blood was drawn from the cubital vein after 12 hours of fasting. Blood serum was obtained by low-speed centrifugation at 900g for 20 min at a temperature of +4  C. Biological material samples were frozen and stored at a temperature not exceeding -20  C until they were sent to the federal center. The transportation of biomaterials was carried out by specialized services. The lipid profile parameters, including the levels of triglycerides (TG), glucose and hs-CRP, were determined using an Abbott Architect c8000 analyzer using diagnostic kits from Abbott Diagnostic (USA). Standardization and quality control of the analysis was carried out in accordance with the requirements of the Federal system for external quality control of clinical laboratory procedures.
Hs-CRP was studied in grades of low, moderate and high cardiovascular risk. The hs-CRP values of 3-10 mg/L were taken as the increased level [9].
Prospective follow-up of the cohort from 2012 to 2019 (median follow-up, 5,5 years) revealed hard endpoint (CVD mortality and nonfatal MI) in 187 (1,79%) people and soft endpoint (nonfatal MI and stroke, revascularization, heart failure (HF) progression and CVD mortality) in 319 (3,06%) people. The HF progression was considered as the hospitalization due to the disease severity (increase in stage and/or decrease in functional class).
The central results of descriptive analysis were presented earlier [10].
Statistical data analysis. Statistical analysis was carried out using the R 3.6.1 environment. To assess the associations of hs-CRP and RFs, a linear regression model was used, where the logarithm of hs-CRP was used as a dependent variable. The Kaplan-Meier survival curves were used to estimate the probability of survival at a certain point in time. Comparison of two survival curves was carried out using the log-rank test. When comparing three or more survival curves, the Holm correction for multiple comparisons was applied. The Cox proportional hazards model Table 1 Multivariate Note: * -how many times higher is the level of hs-CRP in the risk group or when the continuous parameter is increased by the specified value.

regression analysis of the association of hs-CRP with risk factors, adjusted for regions
Abbreviations: AHT -antihypertensive therapy, hs-CRP -high-sensitivity C-reactive protein, CI -confidence interval, CADcoronary artery disease, BMI -body mass index, MI -myocardial infarction, LP(a) -lipoprotein (a), SBP -systolic blood pressure, TG -triglycerides, HDL-C -high density lipoprotein cholesterol, LDL-C -low density lipoprotein cholesterol, log -logarithm.  significantly positively associated with an increase in age, systolic ИЗ, BMI, TG, glucose (p<0,0001), antihypertensive therapy (p<0,006) and is inversely associated with HDL-C (p<0,0001). No association with LDL-C was found. In the multivariate model, significant associations with prior diseases persisted only for RA (p=0,017) ( Table 1). It should be noted separately that the association of CRP level with age is different for men and women. Dependence of the average CRP logarithm on sex and age: loghs-CRP = -1,09+0,267 (men) + 0,023 (age in years, if a man) + (age in years, if a woman). That is, at 25 years of age, hs-CRP is 10% higher in men than in women. Every ten years, the level of hs-CRP rises by 12% in men and 15% in women. At 39 years old, hs-CRP levels became equal, and by age 65, hs-CRP levels in women are 50% higher than in men.
Survival curves for soft endpoints with Holm correction demonstrate a significant relationship with (Cox regression) was used to predict the risk of cardiovascular death or adverse event (CAD, including MI, stroke, revascularization, HF) and to assess the effect of predetermined independent variables on this risk. Additionally, age was included as an independent variable in each of the Cox regressions presented. If the results were presented without separating by sex, then the variable "sex" was also added to the independent variables. The relative risk (RR) and 95% confidence interval [95% CI] were calculated. The Receiver Operator Characteristic (ROC) analysis was carried out. For all types of analysis described, the differences were considered significant at p<0,05.

Results
Multivariate regression analysis, adjusted for regions, revealed that hs-CRP level in men is 20% lower than in women, and in current smokers, regardless of sex − 10% higher. The hs-CRP level is    hs-CRP has a significantly higher risk of CVE than at lower levels, and slightly higher than at an hs-CRP ≥3 mg/L. Notably, adding hs-CRP to sex and age significantly improves risk prediction. In other words, the negative effect of increased hs-CRP begins at its lower values It is important to emphasize that adding a broad list of confounders to hs-CRP, sex and age does not improve the predictive power of the model. Over the past decade, more than 20 prospective cohort studies have shown that hs-CRP levels are independently associated with future risk of MI, stroke, metabolic syndrome, and type 2 diabetes. Most studies with adequate sample sizes have demonstrated that hs-CRP adds predictive information to the risk stratification. Some studies have reported conflicting data on the association between hs-CRP and mortality risk in the population. Thus, the meta-analysis of 14 studies (up to October 2016, n=83995), studying the effect of elevated baseline hs-CRP levels on cancer, CVD, or all-cause mortality in the general population, showed that when comparing the highest versus lowest hs-CRP level, RR was 1,25 [95% CI, 1,13-1,38] for cancer mortality, 2,03 [95% CI, 1,65-the hs-CRP level. The analysis was carried out in three groups of CRP: low risk -up to 1 mg/L, moderate -1-2,99 mg/L, and high risk -≥3 mg/L. Pairwise analysis showed that all curves are pairwise different, and the higher the hs-CRP level, the lower the survival rate ( Figure 1).
In order to clarify the hs-CRP risk levels in population, a ROC analysis was performed, which showed that the optimal cut-off point for predicting a soft endpoint is 1,49 mg/L for men and 1,83 mg/L for women, which is close to the fourth quintile (Q4: men -1,54 mg/L; women -1,89 mg/L). Therefore, in further analysis, we used the Q4 as a cut-off point.
Cox regression showed significant associations of hs-CRP with hard endpoints regardless of the model (M), with the exception of M4 with hs-CRP >1 mg/L, where no significant association was found (RR, 1,28 [0,91-1,8], p>0,05). The highest risk was already noted in M1 at Q4, which is very slightly inferior to currently accepted point of the increased hs-CRP level (>3 mg/L). Comparison of M1 and M2 did not show the significant RR change, which speaks in favor of a pronounced association of isolated hs-CRP with hard endpoint ( Table 2).
A similar data was obtained when analyzing various levels of hs-CRP with soft endpoint, where significant associations were found in all models ( Table 3).
The analysis of proportional hazard models showed a rather pronounced and independent role of CRP in the development of both hard and soft endpoints.
ROC analysis for soft endpoints revealed that isolated hs-CRP has an area under the ROC curve equal to 64,2%, which significantly increases to 79,7 when sex and age are added to the model. Further addition of a wide list of confounders did not significantly increase the area under the ROC curve. In other words, it does not significantly increase the predictive power of the model (Figure 2).

Discussion
The results obtained in our study including more than 10 thousand people showed that an increased level of hs-CRP is a reliable and independent predictor of CVE in the Russian population, even with significant relationships with the main RFs and RA, which is one of the most prominent representatives of chronic inflammatory diseases, and its relationship with hs-CRP is not in doubt among most researchers [10]. At the same time, it was found that the hs-CRP level, which is optimal for predicting the CVE risk, in Russian population is significantly lower than 3 mg/L, but higher than 1 mg/L (Q4 -1,54/1,89 mg/L in men and women, respectively). This shows that above these levels,  This meta-analysis demonstrated that the relationship between hs-CRP and mortality was nonlinear for all-cause and CVD mortality and linear for cancer and non-cardiovascular mortality [13].
Recently published results of the Brazilian study showed a significant association of hs-CRP with allcause mortality in a multivariate model [14].
The studies with an Asian population demonstrated that hs-CRP level in Asians is reduced in comparison with white Europeans [15,16]. Thus, the median hs-CRP level in the Korean population is 0,6 (95% CI, 0,3-1,3) in men and 0,4 (0,1-1,1) in women (p<0,001). Only 8,6% of men and 6,2% of women reached the standard cut-off point for hs-CRP >3 mg/L, which is the upper tertile in the Caucasian population. Moderate-risk hs-CRP models improved the reclassification of CVD mortality risk by 24,9% (p=0,04). Standard cutoff points for CRP in the Asian population may lead to risk underestimation [17]. In addition, both Asian and European studies indicate the need for additional analysis to study the sex contribution at different age periods to unfavorable risk.

Conclusion
This study for the first time showed a significant independent contribution of hs-CRP to CVEs development in the Russian population, and the addition of hs-CRP to sex and age significantly increased the predictive value of model.

Relationships and Activities.
The ESSE-RF prospective study was carried out within the state assignment № AAAA-A17-117070760036-6.