Introduction

According to World Health Organization cardiovascular diseases (CVD) globally represent the main cause of death1. Therefore, prevention and early recognition is highly relevant. Previous research findings have shown an association of cardiovascular risk with socioeconomic factors. Men and women of lower socioeconomic position (SEP) do not only suffer more frequently from coronary artery disease2, but also from higher levels of coronary artery calcification as a marker of subclinical atherosclerosis3. Diverse approaches exist in order to explain this relationship; material, behavioural and psychosocial factors have to be considered4. For gaining knowledge on how socioeconomic factors influence biological pathways of cardiovascular risk, it is important to quantify the impact of SEP on biomarkers indicating subclinical stages of cardiovascular outcomes. In this context, determining the role of known and modifiable cardiovascular risk factors may help to take early preventive and therapeutic actions.

There is evidence from epidemiological studies for an association between increased cystatin C levels, as diagnostic marker of subclinical renal dysfunction, and cardiovascular mortality5, which has been described as being stronger compared to the association with creatinine6. However, after adjusting for traditional cardiovascular risk factors and hs-CRP, an independent association between cystatin C and cardiovascular disease has been observed in most but not all studies7,8,9. Additionally, Mendelian randomization analyses did not support a causal role of cystatin C in the ethology of CVD9,10. The relationship between cystatin C and CVD could likely be explained (1) by cystatin C being a marker for renal dysfunction, which itself contributes to increased risk of cardiovascular disease or (2) by its correlation with long-term exposure to CVD risk factors such as hypertension or diabetes10. In a cohort study by Kestenbaum et al. cystatin C has been associated with a greater incidence of hypertension11. In Shankar and Teppala this association has been found for women, but not for men in a cross-sectional study design12. A systematic review and meta-analysis by Ma et al. demonstrated that type 2 diabetes mellitus patients had higher cystatin C compared to healthy controls13. In a meta-analysis by van der Laan et al. cystatin C was associated with several cardiovascular risk factors and traits like LDL, HDL, BMI, diastolic blood pressure and smoking. In their genetic analysis however, the cystatin C associated SNP (rs911119) showed no significant association with these traits9. A mendelian randomization study by Rasheed et al. did not observe a causal effect of kidney function measured by estimated glomerular filtration rate (GFR) on lipids14.

On the other hand, increased levels of cystatin C seem to be associated with systemic inflammation which may indicate an increase in mortality risk independently of GFR15. Some authors have described the prognostic benefit of cystatin C as marker of coronary artery disease severity and long-term mortality (both cardiovascular and non-cardiovascular) in populations with normal renal function16,17, even though therapeutics targeted at lowering circulating cystatin C are unlikely to be effective in preventing CVD9.

It is assumed, that the effect of SEP on cardiovascular morbidity and mortality is not direct, but mediated through a complex interplay of social inequalities in risk factors18. Evidence from longitudinal studies had shown that the CVD risk factors smoking19,20, and systolic blood pressure21,22, are associated with cystatin C, both in the presence and absence of kidney disease. An association of BMI and cystatin C could not be seen in longitudinal23 but in cross-sectional study design24. A mendelian randomization study has given some evidence for a weak causal effect of triglycerides, HDL but not LDL on estimated GFR using cystatin C14. As people with low SEP tend to have a worse CVD risk profile including higher prevalence of smoking and high blood pressure, the role of these risk factors as mediators in a possible relationship between SEP and cystatin C as a marker of renal disease and predictor for CVD is of interest.

While there are some studies describing the relationship between indicators of SEP and chronic kidney disease25, information on the association between SEP and cystatin C considering the impact of established cardiovascular risk factors is sparse. The aim of the present study was to investigate the impact of educational attainment and household income on levels of cystatin C in a population-based cohort study separately for women and men, while including cardiovascular risk factors as potential mediators in the analysis.

Methods

Study population

Baseline data from the Heinz Nixdorf Recall Study were used, including 4814 men and women aged 45–75 years recruited in three large cities in the western part of Germany (Bochum, Essen and Mülheim/Ruhr) during 2000–200326,27. The participants were selected by random sampling from registration offices with a baseline response rate of 56%28. In the course of the study medical histories, clinical examinations, cardiological functional diagnostics, imaging as well as information about cardiovascular risk factors, psychosocial and socioeconomic factors were gathered. Furthermore, urine and blood samples were collected and subsequently frozen in a biobank to enable future examinations. All participants provided written informed consent and the study was approved by the ethical committee of the University of Duisburg-Essen. The Heinz Nixdorf Recall Study was certified and recertified by DIN ISO 9001:2000/2008 and is performed in accordance with the Declaration of Helsinki.

Cystatin C

Cystatin C at study baseline was measured in 2012 from frozen blood samples at the laboratory of the University Hospital in Düsseldorf (Germany). Cystatin C was quantified by particle-enhanced immunonephelometric assay (Siemens, Marburg, Germany). The manufacturer indicated the analytical sensitivity of the assay by 0.05 mg/l29.

Socioeconomic position

Education and income were used as SEP indicators assessed at study baseline by standardized face-to-face interviews. Education was defined according to the International Standard Classification of Education (ISCED) as total number of educational years consisting of highest school and professional degree30. In order to categorize the educational variable four groups were created with the lowest category including 10 or less years of educational training and the highest 18 or more years.

Income was calculated as monthly household income in Euros by dividing the total net income of a household by a weighting factor for each household member31. For a categorized presentation income was divided into four groups based on the calculated quartiles of the analyzed population.

Cardiovascular risk factors

Established cardiovascular risk factors (i.e. body mass index (BMI), blood pressure, diabetes, smoking, total cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides and high-sensitive C-reactive protein (hs-CRP)) assessed at study baseline were included in the analyses. BMI was determined based on standardized direct measurement of weight and height (kg/m2). Blood pressure was measured using an oscillometric device (Omron, HEM-705CP, OMRON Corporation, Hoofdrop, the Netherlands) and the mean of the second and third of three measurements was calculated. Blood pressure was classified in accordance with the guidelines of JNC7 (The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure)32. Hypertension was defined as high blood pressure stage 1 and 2 according to JNC7 or the intake of antihypertensive medication. Diabetes was considered as self-reported diagnosis, the intake of antidiabetic drugs or an elevation of fasting or non-fasting glucose. Blood glucose was determined by enzymatic hexokinase method. Smoking was assessed in face-to-face interviews and divided in three categories including ‘current smoker’, ‘ex-smoker’ and ‘never smoked’. Total cholesterol, HDL, LDL and triglycerides were assessed from serum samples by standardized enzymatic methods. Hs-CRP was measured from serum samples with the BN-II nephelometer (Dade Behring/ Siemens Healthcare Diagnostics, Eschborn, Germany).

Statistical analyses

Out of the 4814 participants of the Heinz Nixdorf Recall Study serum cystatin C levels were missing for 339. Furthermore, we excluded three participants with cystatin C levels > 4 mg/l in order to avoid outliers. Some observations on education (n = 12) and income (n = 282) were missing, but no correlation between missing cystatin C and SEP was found. There were also missing data for some of the regarded risk factors (see Table 1).

Table 1 Characteristics of the analysis population (M: mean; SD: standard deviation; IQR: interquartile range; [nmiss = number of missings]).
Full size table

Describing the data, for quantitative variables mean ± standard deviation or median and interquartile range (IQR) were stated. Qualitative variables were indicated by absolute (N) and relative frequencies (%). Multiple linear regression was applied to quantify the association between SEP, cardiovascular risk factors and cystatin C. Income and education were included separately as continuous as well as categorical independent variables in the regression models. The highest SEP category served as a reference in the categorical analysis. Different regression models were calculated including a crude model (unadjusted) and a basic model (adjusted for the confounders age and sex) to estimate the total effect of SEP on cystatin C. Using education and income as continuous variables, the basic model was separately supplemented by each cardiovascular risk factor hypothesized as potential mediators for the effect of SEP on cystatin C. Finally, a full model adjusted for BMI, hypertension, diabetes, hs-CRP, total cholesterol, HDL, triglycerides and smoking was calculated to estimate the direct effect of SEP on cystatin C.

Sensitivity analyses were performed by excluding participants with coronary artery disease, stroke or GFR < 60 ml/min/1.73 m2 at baseline from the study population in order to examine the robustness of the results in a population without manifest cardiovascular diseases or impaired renal function.

All analyses were stratified by sex. All calculations were computed using the IBM software SPSS Statistics Version 25 (SPSS Inc., Chicago, Illinois, USA).

Results

In men, cystatin C showed higher levels compared to women (Table 1). Men also held a higher educational level as well as higher income. Most cardiovascular risk factors were more pronounced in men compared to women at the same average age.

Considering the lowest category of education, in the overall group a 0.039 mg/l higher level of cystatin C was observed compared to the highest category after adjusting for age and sex (Fig. 1). Across all subgroups the effect estimates were strongest for the lowest compared to the highest educational category and decreased almost gradually with increasing categories. There was a successive decline in effects from the crude model over the basic model to the full model. After adjustment for cardiovascular risk factors the effect of education on cystatin C was substantially reduced in both men and women. The difference in effect estimates between the crude and the basic model in the female subgroup was strongest, especially in the lowest educational category.

Figure 1
figure 1

Beta estimates and 95% confidence intervals (95% CI) for the association of education categories (≥ 18 years of education as reference) with cystatin C in linear regression models (basic model: adjusted for age (+ sex); full model: age (+ sex), BMI, hypertension, diabetes, hs-CRP, total cholesterol, HDL, triglycerides, smoking).

Full size image

Concerning income as SEP indicator, similar observations were made (Fig. 2). For the overall group a 0.024 mg/l higher level of cystatin C was observed for the lowest income quartile compared to the highest after adjusting for age and sex. The effect estimates in the lowest quartile presented strongest. Using the full model, the effect of income on cystatin C was substantially reduced.

Figure 2
figure 2

Beta estimates and 95% confidence intervals (95% CI) for the association of income quartiles (4th quartile as reference) with cystatin C in linear regression models (basic model: adjusted for age (+ sex); full model: age (+ sex), BMI, hypertension, diabetes, hs-CRP, total cholesterol, HDL, triglycerides, smoking).

Full size image

For the crude linear regression model using the continuous education variable as SEP indicator, effect estimates were considerably higher in women than in men (Table 2). Here, cystatin C was on average 0.064 mg/l lower per five educational years. After adjustment for age (and sex in the overall group) the effect estimates diminished with the strongest decrease in women. Adjustment for single risk factors displayed different results. After adjustment for BMI, hypertension, HDL, triglycerides and smoking a reduction of effect strength was observed with BMI showing the highest impact. In the full model the strength of association between education and cystatin C was further reduced.

Table 2 Beta estimates and 95% confidence intervals (95% CI) for the association of education (per 5 years of education) with cystatin C in linear regression models (basic model: adjusted for age (+ sex); full model: age (+ sex), BMI, hypertension, diabetes, hs-CRP, total cholesterol, HDL, triglycerides, smoking).
Full size table

Analogous results for income (continuous) are described in Table 3. Again, the crude model showed stronger effect estimates in women than in men, which were reduced strongest in the highest income quartile by adjustment for age. In the overall group, cystatin C was on average 0.014 mg/l lower per 1000 euros income in the basic model. After adjusting for single risk factors in the overall group, for BMI, hypertension, diabetes, hs-CRP, HDL and smoking a reduction of effect strength was observed. Adjustment for all considered risk factors caused further reduction of effects.

Table 3 Beta estimates and 95% confidence intervals (95% CI) for the association of income (per 1000 Euro/month) with cystatin C in linear regression models (basic model: adjusted for age (+ sex); full model: age (+ sex), BMI, hypertension, diabetes, hs-CRP, total cholesterol, HDL, triglycerides, smoking).
Full size table

Tables S1–S4 in the supplement show the results of sensitivity analysis which refers to those participants without coronary artery disease or stroke and a GFR ≥ 60 ml/min/1.73 m2 at baseline. The trends observed in the main analysis population can also be found in the sensitivity analysis, although effect estimates were less strong.

Discussion

The results of the present study demonstrated an association between SEP indicators education and income with cystatin C. Participants with lower SEP presented higher cystatin C levels after adjustment for confounding by age and sex. After adjusting for traditional cardiovascular risk factors, this association was strongly attenuated suggesting that known social inequalities in cardiovascular risks mediate the association between SEP and cystatin C. Accordingly, people with lower SEP more frequently hold adverse risk factor profiles which also seemed to be associated with higher levels of cystatin C. Substantial sex-differences in the association between SEP and cystatin C were not observed. Sensitivity analysis demonstrated that the results were not strongly influenced by participants with manifest cardiovascular diseases or impaired renal function, demonstrating that the observed association between SEP indicators education and income with cystatin C is not mainly triggered by social inequalities in Cystatin C-related clinical endpoints, but is also indicated at the subclinical stage.

Previous studies have produced convincing evidence for an inverse relationship between SEP and manifest kidney disease33, but few studies have examined the association between SEP and cystatin C as a subclinical biomarker for early kidney disease. Peralta et al. (2006) have reported an association between income and cystatin C in African Americans consistent to the results of the present study34. Tamrat et al. have used cystatin C based estimated GFR for their study in African Americans and reported an association with income corresponding to the results of the present study35. With regard to European study populations, Thio et al. have published results from the Netherlands showing a longitudinal association between education and impaired renal function (estimated GFR including serum creatinine and cystatin C)36. Consistent to the results presented here, the association of education and renal dysfunction has been largely explained by mediation through traditional cardiovascular risk factors as diabetes, overweight, smoking and hypertension36. In a population-based cross-sectional study in Ireland Canney et al. (2018) have found a strong association between low childhood SEP and chronic kidney disease defined by a GFR < 60 ml/min/1.73 m2 in women including cystatin C for GFR estimation. In contrast to the study presented here, an association between SEP and CKD in men was not observed37. Adding the results presented here to previous evidence, early recognition of high levels of cystatin C combined with reduction of modifiable cardiovascular risk factors seems to offer great potential to lower the impact of socioeconomic inequalities on Cystatin C as a marker for renal dysfunction on the subclinical level.

While having the strength of a population-based study design with available information on different SEP indicators and a wide range of cardiovascular risk factors, this study also has some limitations. First, the cross-sectional design has to be considered. This prohibits strong conclusions about the causal direction of the association between income and cystatin C. However, concerning education reversed causation is very unlikely, because it represents a stable indicator for SEP across the life course, usually determined in young adulthood. Using subclinical measures of health further reduces the potential extent of reverse causation. However, reverse causation in the relationship of cystatin C and blood pressure is possible, as there is some evidence from previous studies that cystatin C may be caused by hypertension and vice versa. Therefore, the cross-sectional design of the present study does not allow to draw conclusion about the potential role of hypertension mediating the association between SEP and cystatin C. Results of the sensitivity analysis showed that prevalent cases of cardiovascular disease or renal impairment did not strongly affect the observed associations. Another limitation was that at the time of laboratory analysis standardized assays for cystatin C blood level determination were not yet available. This may limit the comparability of absolute cystatin C values with other studies.

Despite these limitations the results of the present study demonstrate the impact of socioeconomic factors on cystatin C and show that the association is also apparent when cystatin C is used as a subclinical marker in participants without clinical endpoints such as renal impairment, coronary artery disease and stroke. Consideration of traditional cardiovascular risk factors suggests that prevention strategies to modify adverse health behaviors could also be helpful to reduce levels of cystatin C, especially in groups of lower SEP.