If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Clinical Sciences Malmö, Lund University, SE-202 13, Malmö, SwedenFaculty of Health and Society, Malmö University, SE-205 06, Malmö, SwedenBiofilms – Research Center for Biointerfaces, Malmö University, SE-205 06, Malmö, Sweden
Higher CCL21 & CCL19 levels at baseline in patients who suffered a coronary event.
•
High levels of CCL21 are associated with increased risk of incident coronary events.
•
High levels of CCL19 are associated with increased risk of mortality & heart failure.
Abstract
Background and aims
The homeostatic chemokines CCL21 and CCL19 have been explored as biomarkers in cardiovascular disease prediction in patients with established cardiovascular disease, but associations between these chemokines and first-time coronary event incidence have not been investigated before. Here, we explored associations between CCL21 or CCL19 and first-time incident coronary events in the general population-based Malmö Diet and Cancer cohort with two decades of follow-up.
Methods
CCL21 and CCL19 levels in plasma were analysed with ELISA and proximity extension assay and associations with disease incidence were explored with conditional logistic regression in a nested case-control cohort (CCL21; n = 676) and with Cox regression in a population-based cohort (CCL19; n = 4636).
Results
High CCL21 levels in plasma were associated with incident first-time coronary events independently of traditional risk factors (odds ratio of 2.64 with 95% confidence interval 1.62–4.31, p < 0.001, comparing the highest versus the lowest tertile of CCL21), whereas CCL19 was not. CCL19 was, however, associated with incident heart failure, as well as increased all-cause, cardiovascular and cancer mortality independently of age and sex.
Conclusions
Even though CCL21 and CCL19 both signal through CCR7, these chemokines may not be interchangeable as disease predictors and CCL21 could be used for prediction of future coronary events in individuals without any previous coronary heart disease history.
In many countries the mortality caused by coronary heart disease, including myocardial infarction (MI), has decreased during the last decades. This can be attributed both to improved medical care and changes in lifestyle factors, such as a reduced smoking rate. However, coronary heart disease still remains the most common cause of death world-wide [
]. Consequently, continued research into prevention and early detection of the disease remains paramount, especially considering the continuous increase in obesity and diabetes [
]. This includes identifying biomarkers that can be used to predict the risk of future disease with high accuracy.
Currently, prediction is still mainly based on lifestyle factors, including the traditional Framingham risk factors: age, male sex, cholesterol, high density lipoprotein (HDL), diabetes, smoking, systolic blood pressure (BP) and use of BP-lowering treatment. Some of these factors are modifiable (e.g. cholesterol), while others are not (e.g. age) [
]. With time, more focus has been put on inflammatory biomarkers as an early prediction method. One of these is C-reactive protein (CRP), which has been found to predict future cardiovascular disease (CVD) in large meta-studies [
The Emerging Risk Factors Collaboration C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis.
]. Still, no inflammatory marker is currently in wide clinical use for disease prediction.
The homeostatic chemokines C–C motif chemokine ligand (CCL) 19 and CCL21 are continuously expressed and are essential for the homing of immune cells both to and within lymphoid organs [
]. Both CCL19 and CCL21 signal via the C–C motif chemokine receptor 7 (CCR7), a G protein-coupled receptor expressed on several subsets of T cells, mature dendritic cells, naïve B cells and macrophages [
]. CCL19, CCL21 and CCR7 have all been detected in atherosclerotic plaques and CCR7 has been found to co-localise with both T cells, macrophages and smooth muscle cells (SMCs) in human plaques [
Enhanced expression of the homeostatic chemokines CCL19 and CCL21 in clinical and experimental atherosclerosis: possible pathogenic role in plaque destabilization.
]. The results from atherosclerosis studies using CCR7-deficient mice are somewhat conflicting, with Wan et al. showing an atheroprotective role for CCR7 [
]. Both studies found more T cells in plaques from CCR7-deficient mice compared to controls, but while Luchtefeld et al. showed a decrease in macrophages, Wan et al. found no difference in macrophages in their study.
CCL19 and CCL21 have been investigated in several human cohorts as potential biomarkers for CVD.
In studies of patients with acute or chronic heart failure plasma CCL21 has been found to predict both all-cause and CVD mortality [
]. In acute coronary syndrome (ACS) patients, CCL21 has been found to predict future major adverse cardiovascular events (MACE) and MI, while CCL19 has been associated with heart failure [
Even though these studies support a role for CCL19 and CCL21 in CVD prediction, associations between these homeostatic chemokines and disease incidence have not been investigated in a prospective general population cohort. Previous studies have focused on patients with established cardiovascular disease and assessing if CCL19 and CCL21 have associations with recurring events or mortality. In this study, we investigated the predictive power of CCL19 and CCL21 on future incidence of coronary events (CE), in a prospective population-based cohort.
2. Patients and methods
2.1 Study design
The Malmö Diet and Cancer study (MDC) is a prospective study based on the general population of Malmö, Sweden (Fig. 1). Participants 45–73 years of age were recruited to the study 1991–1996 (n = 28 449). Between 1991 and 1994 every other newly recruited participant in the MDC was invited to participate in a substudy focused on cardiovascular outcomes (MDC-CV, n = 6103). Participants underwent a physical examination and answered a questionnaire regarding diet, lifestyle habits and medical history.
Plasma levels of CCL21 were analysed in the nested-case control cohort (left arm) and plasma levels of CCL19 were analysed in the population-based MDC-CV cohort (right arm).
Fasting blood samples were collected in median 8 months after study inclusion (n = 5533). Aliquots for chemokine analysis were kept at −80 °C for future analysis. The study was approved by the ethics committee at Lund University (LU 51–90 and LU 532–2006) and performed in accordance with the Declaration of Helsinki. All participants gave written informed consent.
A coronary event (CE) was defined as fatal or non-fatal MI or ischemic heart disease, international classification of diseases (ICD)-9 codes 410, 412 and 414, corresponding to ICD-10 codes I21, I22, I23 and I25. Stroke was defined by ICD9 codes 430, 431, 434 and 436, corresponding to ICD10 codes I60, I61, I63 and I64. Heart failure was defined by ICD9 code 428, corresponding to ICD10 code I50. Information about incident disease and cause of death were found through the Swedish hospital discharge register and the Cause of Death Registry of Sweden.
2.2 Population-based MDC-CV cohort
In the MDC-CV (with available CCL19 results, n = 4721) participants were followed until December 31, 2014. During this period there were 436 (9.4%) incident CE after the exclusion of participants with a prevalent CE (n = 85), making the total number of participants 4636 (Fig. 1). In the MDC-CV the median time from study inclusion to incident CE was 14.2 years (range 1.3–21.2). There were also 411 (8.9%) incident stroke cases, excluding 30 prevalent stroke cases (n = 30), and 187 (4.1%) incident heart failure cases, excluding prevalent heart failure (n = 3) and incident heart failure within 24 h of MI (n = 35). The median time from study inclusion to incident stroke or heart failure was 14 years (range 0.25–22.9) and 15.9 years (range 0.5–22.5), respectively.
Causes of death and corresponding ICD-9/ICD-10 codes are presented in Supplemental Table 4. During the follow-up period 1215 (26%) participants died. Information about the causes of death were retrieved from the Cause of Death Registry of Sweden.
2.3 Nested case-control cohort
In the nested case-control part of the MDC-CV there were 434 incident CE until the end of the follow-up period on June 30, 2009. Thirty-eight participants with a history of CE before blood sampling were excluded. The remaining 379 cases with available plasma were cumulatively matched to a control from the baseline study with no history of CE based on age (±1 year), sex and inclusion in the baseline study (±6 months). After the exclusion of 41 pairs, due to low sample volumes or measurement errors in the ELISA, 338 matched case-controls pairs were included in the study (total n = 676, Fig. 1). In the nested case-control study the median time from study inclusion to CE was 10.8 years (range 0.6–17.6).
2.4 Laboratory analyses
Triglycerides, total cholesterol, HDL, white blood cells (WBC), haemoglobin A1c (HbA1c), creatinine and blood glucose were analysed in fasting blood samples at the Department of Clinical Chemistry, Skåne University Hospital, Malmö. Low-density lipoprotein (LDL) was calculated with Friedewald's formula. The estimated glomerular filtration rate (eGFR) was calculated with the MRDR formula.
CRP was analysed using the Tina-quant CRP latex high sensitivity assay (Roche Diagnostics, Basel, Switzerland) on an AVIDA 1650 Chemistry System (Bayer Healthcare, NY, USA). The average coefficient of variation in the assay was 4.6% [
Elevated Lp-PLA2 levels add prognostic information to the metabolic syndrome on incidence of cardiovascular events among middle-aged nondiabetic subjects.
CCL21 was analysed in EDTA-plasma using an ELISA-kit (R&D systems, Minneapolis, USA), according to the manufacturer's instructions. The samples were diluted 1:2.5 prior to analysis and absorbance was measured at 450 nm with correction at 570 nm on a Tecan Sunrise plate reader (Tecan Trading AG, Switzerland). Samples with values below the detection limit n = 122, 18% (46.8–124.2 pg/mL, depending on the plate) were recoded as the value of the plate-specific detection limit. Setting the detection limit to 124.2 pg/mL regardless of the plate-specific detection limit only caused minor alterations in the results and did not change any of the interpretations or conclusions (data not shown). The mean intra- and inter-assay variability were 9.0% and 18.8%, respectively.
CCL19 was analysed in EDTA-plasma using proximity extension assay (PEA). The analysis was performed at the Clinical Biomarkers Facility, Science for Life Laboratory, in Uppsala, Sweden, using the Proseek Multiplex Oncology I v.296×96 reagents kit (Olink Proteomics, Uppsala, Sweden). The lower and upper limit of quantification in the analysis were 15 and 62500 pg/mL, respectively. The intra- and inter-assay variability were 5% and 27%, respectively. The results from the PEA are expressed as arbitrary units (AU).
2.5 Statistics
Statistical analyses were performed in SPSS Statistics 27 (IBM, Armonk, USA) and in R version 4.1.3 [
]. Figures were made in GraphPad Prism 9.1 (GraphPad Software, San Diego, USA) and drawio diagrams.net 18.0.6 (JGraph Ltd, UK). Continuous baseline data is presented as median and inter-quartile range (IQR) and categorical data is presented with count (n) and percentage of the total. Differences in variables between cases and controls were tested using the Mann-Whitney test or the chi-square test. Differences in variables between tertiles of CCL19 or CCL21 were tested using the Kruskal-Wallis test or the chi-square test. Spearman correlations were performed to analyse correlations between continuous variables. Cox regression was used to analyse associations between plasma CCL19 (either stratified into tertiles or as standardised continuous variables) and incident CE, stroke, heart failure or mortality in the MDC-CV population-based cohort. Conditional logistic regression conditioned for matched case-control pairs was used to analyse associations between plasma CCL21 levels (either stratified into tertiles or as standardised continuous variables) and incident CE. Regression models were unadjusted or adjusted for Framingham risk factors (total cholesterol, HDL, diabetes, use of BP-lowering medication, systolic BP and smoking [
]. Traditional risk factors or relevant variables significantly associated across CCL21 or CCL19 tertiles were considered as potential confounders and included in adjusted models. In all models tertile 1 with the lowest levels of CCL19 or CCL21 was used as the reference. Natural logarithm transformation was applied to non-normally distributed variables (CRP, HDL, fasting glucose, CCL19). Pair-wise differences between conditional logistic regression models were analysed using likelihood ratio tests. Receiver operating characteristic (ROC) curves were generated and possible differences in the area under the curve (AUC) from different models were analysed using DeLong's test.
3. Results
3.1 Increased CCL19 levels in incident coronary event cases
CCL19 levels were measured by PEA technology in the population-based MDC-CV cohort (n = 4636; Fig. 1). Incident CE cases had higher plasma levels of CCL19 than controls, a median of 714 [IQR: 508–1048] AU in cases versus a median of 641 [IQR: 471–945] AU in controls, p = 0.002 (Fig. 2A). Baseline characteristics for cases and controls are presented in Supplemental Table 1. Cases were, at baseline, older, had higher values for; body mass index (BMI), lipids, glucose, HbA1c, BP, CRP, WBC and had more often been diagnosed with diabetes and currently smoked.
Fig. 2Plasma CCL19 and CCL21 concentrations in controls and CE cases.
(A) CCL19 was measured in the population-based cohort MDC-CV with 436 incident CE cases and 4196 controls. CCL19 levels of each participant are presented as log2 transformed with median and IQR. Statistical significance was calculated with the Mann-Whitney test. (B) CCL21 levels were measured in the nested case-control cohort with 338 age and sex-matched pairs. CCL21 levels of each participant are presented with median and IQR. Statistical significance was calculated with the Mann-Whitney test. Samples with values below the detection limit (n = 122) were recoded as the plate-specific detection limit (46.8–124.2 pg/mL).
Baseline characteristics for participants in the MDC-CV cohort stratified into tertiles based on plasma levels of CCL19 are presented in Table 1. With increasing CCL19 levels, more participants had an incident CE and prevalent diabetes. Risk factors for CVD, such as age and systolic BP, also increased with increasing CCL19 levels, whereas HDL decreased. Notably, smoking and total cholesterol were not associated with CCL19 levels. Taken together, CCL19 levels were associated with incident CE as well as with many traditional cardiovascular risk factors. Correlations analyses between CCL19 and continuous variables from Table 1 are presented in Supplemental Table 2.
Table 1Baseline characteristics for participants in the population-based and nested case-control cohorts by CCL19 or CCL21 tertiles.
CCL19 levels in population-based cohort
CCL21 levels in nested case-control cohort
Tertile 1 (n = 1545)
Tertile 2 (n = 1546)
Tertile 3 (n = 1545)
p-value
Total N
Tertile 1 (n = 224)
Tertile 2 (n = 226)
Tertile 3 (n = 226)
p-value
Total N
Incident CE
118 (8%)
149 (10%)
169 (11%)
0.007
4636
84 (38%)
108 (48%)
146 (65%)
<0.001
676
Mortality
345 (23%)
385 (25%)
485 (32%)
<0.001
4603
46 (21%)
52 (23%)
89 (39%)
<0.001
676
Age
57 [52–62]
57 [52–63]
59 [53–63]
<0.001
4636
59 [53–63]
62 [58–65]
63 [60–65]
<0.001
676
Sex (women)
979 (63%)
899 (58%)
944 (61%)
0.012
4636
83 (37%)
85 (38%)
100 (44%)
0.221
676
BMI (kg/m2)
24.3 [22.4–26.9]
25.3 [23.1–27.7]
25.9 [23.4–28.7]
<0.001
4636
25.1 [22.9–28.2]
26.0 [24.0–28.9]
26.1 [24.2–29.2]
0.009
675
Smoking
317 (21%)
327 (21%)
360 (23%)
0.141
4629
47 (21%)
53 (24%)
62 (29%)
0.187
659
Diabetes
88 (6%)
119 (8%)
174 (11%)
<0.001
4636
17 (8%)
38 (17%)
49 (22%)
<0.001
674
Fasting glucose (mmol/L)
4.8 [4.5–5.2]
4.9 [4.6–5.3]
4.9 [4.6–5.3]
<0.001
4636
4.9 [4.6–5.3]
5.0 [4.7–5.5]
5.2 [4.8–5.8]
<0.001
675
HbA1c (%)
4.7 [4.4–5.0]
4.8 [4.5–5.1]
4.9 [4.6–5.2]
<0.001
4636
4.8 [4.5–5.1]
4.9 [4.6–5.2]
5.0 [4.6–5.4]
<0.001
669
Triglycerides (mmol/L)
1.1 [0.8–1.4]
1.2 [0.9–1.6]
1.2 [0.9–1.7]
<0.001
4635
1.1 [0.8–1.5]
1.4 [1.0–1.9]
1.5 [1.1–2.1]
<0.001
673
Total cholesterol (mmol/L)
6.1 [5.4–6.8]
6.1 [5.4–6.8]
6.2 [5.4–6.9]
0.055
4634
6.0 [5.2–6.6]
6.2 [5.5–6.9]
6.5 [5.7–7.2]
<0.001
673
HDL (mmol/L)
1.4 [1.2–1.7]
1.3 [1.1–1.6]
1.3 [1.1–1.5]
<0.001
4635
1.3 [1.1–1.6]
1.2 [1.0–1.5]
1.2 [0.9–1.4]
<0.001
661
LDL (mmol/L)
4.0 [3.4–4.7]
4.1 [3.5–4.8]
4.2 [3.5–4.8]
0.002
4630
4.0 [3.4–4.6]
4.2 [3.6–5.0]
4.5 [3.7–5.2]
<0.001
650
eGFR (mL/min/1.73 m2)
72 [64–81]
74 [66–83]
74 [66–84]
<0.001
4584
77 [69–86]
75 [66–85]
70 [60–79]
<0.001
660
Systolic BP (mmHg)
138 [125–150]
140 [128–150]
140 [130–158]
<0.001
4636
142 [130–155]
148 [131–160]
150 [138–160]
0.001
676
Diastolic BP (mmHg)
85 [80–90]
86 [80–92]
88 [80–95]
<0.001
4636
88 [80–95]
90 [82–96]
90 [82–96]
0.179
676
CRP (mg/L)
1.0 [0.5–2.0]
1.3 [0.7–2.6]
1.7 [0.8–3.7]
<0.001
4534
1.3 [0.7–2.6]
1.5 [0.8–3.3]
2.1 [1.0–4.0]
0.001
626
WBC (109 cells/L)
5.6 [4.8–6.7]
5.8 [5.0–6.9]
6.1 [5.1–7.2]
<0.001
4633
5.5 [4.8–6.7]
6.0 [5.2–7.0]
6.4 [5.5–7.3]
<0.001
674
Neutrophils (109 cells/L)
3.4 [2.8–4.2]
3.5 [2.9–4.3]
3.7 [3.0–4.5]
<0.001
4633
3.5 [2.8–4.3]
3.6 [2.9–4.4]
3.8 [3.1–4.6]
0.001
674
Lymphocytes (109 cells/L)
1.7 [1.4–2.1]
1.8 [1.5–2.2]
1.8 [1.5–2.2]
<0.001
4633
1.7 [1.4–2.0]
1.8 [1.6–2.2]
2.0 [1.6–2.3]
<0.001
674
BP lowering medication
182 (12%)
243 (16%)
270 (17%)
<0.001
4636
45 (20%)
38 (17%)
50 (22%)
0.358
676
Statin treatment
11 (1%)
19 (1%)
24 (2%)
0.089
4636
5 (2%)
7 (3%)
5 (2%)
0.791
676
Continuous variables are presented as median [interquartile range] and categorical variables are presented as N (percent of total). Differences in continuous variables between tertiles were calculated with Kruskal Wallis tests and Chi-2 tests for categorical variables.
3.2 CCL19 levels do not predict future first-time CE independently of common risk factors
Cox regression revealed that CCL19 levels were significantly associated with incident CE with a HR for the highest vs lowest tertile of 1.57 (95% CI 1.24–1.98, p < 0.001, Table 2). When CCL19 values were standardised the HR per SD increase was 1.17 (95% CI 1.07–1.28, p < 0.001, Supplemental Table 3). However, when the model was adjusted for the Framingham risk factors the HR for the highest vs. lowest tertile was 1.09 (95% CI 0.85–1.39, p=0.497) and per SD increase 1.04 (95% CI 0.94–1.14, p=0.474), indicating that the association between CCL19 levels and incident CE was not independent of traditional risk factors (Table 2, Supplemental Table 3).
Table 2Cox regression with hazard ratios and 95% CI for the association between CCL19 in tertiles and incident CE, HF, stroke, CVD- and all-cause mortality an unadjusted and adjusted model.
We went on to investigate if CCL19 levels are associated with two other cardiovascular outcomes, incident stroke and heart failure, with cox regression analysis. We did not find a statistically significant association between CCL19 and incident stroke independent of the Framingham risk factors, HR 1.07 for the highest vs lowest tertile (95% CI 0.85–1.36, p=0.555; Table 2). In contrast, CCL19 was independently associated with incident heart failure, HR 1.60 for the highest vs lowest tertile (95% CI 1.09–2.34, p=0.016), after the regression model was adjusted for the Framingham risk factors (Table 2).
3.3 CCL19 levels predict all-cause mortality
In total, 1215 participants died during the follow-up period. Most participants died of cancer (n = 533; 44% of all deaths) or CVD (n = 350; 29% of all deaths; Supplemental Table 4). CCL19 plasma levels were higher in participants who died of any cause, CVD, or cancer during follow-up (Supplemental Table 5).
Cox regression revealed that CCL19 levels were significantly associated with all-cause mortality, independently of age and sex, with a HR for the highest vs lowest tertile of 1.35 (95% CI 1.17–1.55, p < 0.001; Table 2). CCL19 was also associated with CVD mortality when adjusting for age and sex, but the association did not remain statistically significant after adjustment for the Framingham risk factors with a HR for the highest vs lowest tertile of 1.15 (95% CI 0.88–1.51, p=0.283; Table 2, Supplemental Table 6). In addition, we found that CCL19 was associated with cancer mortality (n = 533) independently of age and sex with HRs for the highest vs lowest tertile of 1.26 (95% CI 1.02–1.55, p=0.030) and also in an unadjusted model digestive systems disease (n = 27) with an HR of 3.47 (95% CI 1.13–10.64, p=0.030; Supplemental Table 6).
Taken together, CCL19 levels were associated with CVD mortality in age and sex adjusted models, although this association could not withstand further adjustment for risk factors. CCL19 was, however, also associated with all-cause and cancer mortality in age and sex adjusted models. Consequently, CCL19 was not exclusively associated with death by cardiovascular causes but may reflect general longevity.
3.4 Increased CCL21 levels in incident coronary event cases
Next, to evaluate associations between CCL21, the other CCR7 binding homeostatic chemokine, and incident CE we measured CCL21 levels with ELISA. We opted for a nested case-control study design (Fig. 1), which retains much of the statistical power inferred for the whole cohort while using a smaller number of samples [
]. Incident CE cases in the nested case-control cohort had higher plasma levels of CCL21 than the controls (median 416 [IQR 291–542] pg/mL vs. median 326 [IQR 105–441] pg/mL, p < 0.001; Fig. 2B). Baseline characteristics for cases and controls are presented in Supplemental Table 7. Cases had, at baseline, higher BMI, lipids, fasting glucose, HbA1c, BP, CRP, WBC and had more often been diagnosed with diabetes and currently smoked. Cases and controls were matched for age and sex, and also did not differ in total cholesterol levels, eGFR and BP lowering or statin treatment.
Baseline characteristics for participants in the nested case-control cohort, stratified into tertiles based on plasma levels of CCL21, are presented in Table 1. With increasing CCL21 levels, more participants had an incident CE during follow-up and baseline diabetes diagnosis. BP lowering medication or statin use did not differ across CCL21 tertiles. Levels of risk factors for CVD, except the number of smokers, increased with increasing CCL21 levels, whereas HDL decreased. Taken together, the associations between CCL21 and incident CE and risk factors mostly mirror the association between CCL19 and incident CE and risk factors found in the whole MDC-CV cohort. Amiable to this idea, we found a weak correlation between CCL19 and CCL21 (Spearman's rho 0.26, p < 0.001, n = 528) in the nested case-control study (Supplemental Table 8). Correlations analyses between CCL21 and continuous variables from Table 1 are presented in Supplemental Table 8.
3.5 CCL21 levels predict future first-time CE independently of common risk factors
Associations between plasma levels of CCL21 and incident CE, the traditional Framingham risk factors and potential confounders were assessed by conditional logistic regression. After adjusting for the Framingham risk factors, high CCL21 levels (tertile 3) remained independently associated with incident CE with an OR of 2.64 (95% CI 1.62–4.31, p < 0.001), compared with the lowest CCL21 tertile (Fig. 3, Supplemental Table 9). CCL21 levels were also significantly associated with incident CE independently of the Framingham risk factors when all continuous variables were standardised; OR per SD 1.74 (95% CI 1.40–2.16, p < 0.001; Fig. 3, Supplemental Table 9). To further elucidate connections between CCL21 and the possible confounders identified in Table 1 the regression model was adjusted for the Framingham risk factors, as well as CRP, HbA1c, fasting glucose and eGFR. Likewise, in this extended model, high CCL21 levels remained independently associated with incident CE; the OR for the highest vs. lowest CCL21 tertile was 2.12 (95% CI 1.18–3.81 p=0.012; Fig. 3, Supplemental Table 9) and in the continuous variable model the OR per CCL21 SD was 1.54 (95% CI 1.19–1.98, p=0.001; Fig. 3, Supplemental Table 9).
Fig. 3Conditional logistic regression with odds ratios (OR) and 95% CI for the associations between CCL21 in tertiles (T1-T3) or standardised (Z) and incident CE in an unadjusted model (n = 338 pairs, black), a model adjusted for the Framingham risk factors (total cholesterol, HDL, diabetes, use of BP lowering medication, systolic BP and smoking) (n = 305 pairs, white) and an extended model adjusted for the Framingham risk factors as above and CRP, HbA1c, fasting glucose and eGFR (n = 257 pairs, grey).
Both CCL21 and CRP had a significant association to incident CE, indicating that CCL21 and CRP and not interchangeable inflammatory markers (Supplemental Table 9). Likelihood ratio tests comparing conditional logistic regression models adding CCL21 to models adjusted for the Framingham risk factors with and without CRP further supports this notion (Supplemental Table 10). ROC curve analysis followed by calculation of the Youden index provided a cut-off point of 405 pg/mL for CCL21 resulting in a sensitivity of 51.1% and a specificity of 69.6%. Addition of CCL21 to a model that combined the Framingham risk factors increased the sensitivity and specificity to 63.0% and 68.7%, respectively, which corresponds to 0.3% and 1.6% increases over the model without CCL21. While CCL21 had a larger AUC than CRP, there were no statistical differences between AUCs from different models (Supplemental Table 11). Additional adjusting of the extended model with total WBC counts did not appreciably alter the OR in any of the models (data not shown).
All participants with prevalent CE were excluded from the analysis of the nested case-control cohort (Fig. 1), but there were 28 participants (16 cases and 12 controls) with a history of other types of CVD, such as stroke and atrial fibrillation (Supplemental Table 12). The exclusion of these participants did not change the association between CCL21 and incident CE and the risk factors, with only slightly increased ORs for the highest vs. lowest CCL21 tertile in all three regression models (Supplemental Table 12).
4. Discussion
In this study we found that high plasma levels of CCL21 at baseline were associated with an increased risk of first-time incident CE in a population-based cohort followed for more than 15 years. CCL21 levels were associated with incident CE independently of both Framingham risk factors and additional potential confounding variables (Framingham risk factors plus CRP, HbA1c, fasting glucose and eGFR). Plasma levels of CCL19 were similarly associated with first-time incident CE, stroke, heart failure and CVD mortality independently of age and sex. These associations to the cardiovascular outcomes were, however, not independent of traditional risk factors, with the exception of incident heart failure which remained significant even after risk factor adjustment. Notably, CCL19 was associated not only with CVD mortality, but also with all-cause and cancer mortality in regression models adjusted for age and sex.
CCL21 and CCL19 have not previously been investigated as biomarkers for incident CE in a population-based prospective cohort of healthy individuals, but high serum CCL21 levels have been associated with increased risk of incident MACE and MI in ACS patients [
]. Even though CCL19 and CCL21 share binding to the same receptor, CCR7, we did not find an association between CCL19 and incident CE or stroke after adjustment, indicating that CCL19 may not contribute to or reflect the risk of suffering an acute cardiovascular event. Interestingly, in ACS patients, CCL19 was not independently associated with incident MACE or MI, but notably and similar to the results of our study, CCL19 was associated with incident heart failure [
]. Taken together, our study in initially CE-free individuals found that CCL21 was independently associated with incident CE, while CCL19 was not, which is in line with findings in patients with established CVD. CCL19 was, however, associated with incident heart failure, which is also in keeping with previous findings in patients with established CVD.
Currently the Framingham risk factors; age, male sex, total cholesterol, HDL, diabetes, systolic BP, use of BP lowering medication and current smoking, are used to identify persons at risk of coronary heart disease. In our study, using likelihood ratio tests, we found that adding CCL21 to the Framingham risk factors significantly improved the predictive ability of the regression model for incident CE, although the difference in ROC curve AUC was not significant. Similar results, i.e., observing that a variable has an independent association to a disease outcome in multivariate regression models, but no statistically significant improvement in AUC is not uncommon, even for markers that have clear clinical relevance, such as HDL [
]. A reason for this is that the test used to compare AUCs is quite conservative and is additionally considered inappropriate to use for nested models, such as the models used in our paper [
Our finding that CCL19 and CCL21 may have different roles to play in the clinical manifestation of coronary heart disease is supported by in vitro studies of the structure and function of these chemokines. CCL19 and CCL21 are structurally similar with the exception that CCL21 has a C-terminal tail peptide which allows it to bind to glycosaminoglycans (GAGs) expressed in the extracellular matrix or on cells [
]. The tail peptide can be cleaved by plasmin, allowing for the regulated release of soluble CCL21, making cleaved CCL21 structurally more similar to CCL19 [
Plasmin and regulators of plasmin activity control the migratory capacity and adhesion of human T cells and dendritic cells by regulating cleavage of the chemokine CCL21.
]. As long as CCL21 is intact and non-bound the tail peptide folds in such a way that it inhibits efficient interaction with CCR7, but attachment to GAGs or proteolytic cleavage of the tail peptide triggers a conformation change, that increases the CCR7-activation potency of CCL21 [
The C-terminal peptide of CCL21 drastically augments CCL21 activity through the dendritic cell lymph node homing receptor CCR7 by interaction with the receptor N-terminus.
In addition to the structural differences, there is evidence that CCL19 and CCL21 are functionally different. CCL19 is a more potent activator of chemotaxis in human dendritic cells, while CCL21 requires a 10-fold higher concentration to elicit a similar response [
]. Stimulation of human lymphocytes and dendritic cells with CCL19 also prevents further activation by inducing internalisation of CCR7 to a higher degree than stimulation with CCL21 [
]. Furthermore, CCL19 and CCL21 differ in their activation of intracellular signalling pathways as, compared with CCL19, CCL21 stimulation causes a stronger activation of ERK1/2 in human dendritic cells [
]. Stronger activation of ERK1/2 after CCL21 stimulation has also been seen in THP-1 macrophages and aortic SMCs, whereas CCL19 stimulation resulted in stronger AKT activation in aortic SMCs [
]. In the same study, effects of the difference in activation included that aortic SMCs stimulated with CCL19 exhibited higher rates of proliferation compared with cells stimulated with CCL21. Also, in THP1 macrophages, stimulation with CCL21, but not CCL19, caused an increased binding of acetylated LDL to cells, as well as accumulation of intracellular lipids [
Subsequently, it is possible that these well-established differences regarding the properties of CCL21 and CCL19 could be related to the differences in predictive ability observed in our study. For example, the GAG-binding ability of CCL21 that CCL19 lacks may be advantageous in the recruitment of immune cells into atherosclerotic plaques. It is also possible that the association of CCL21 to incident CE reflects a general increase in inflammation, which induces CCL21 release [
]. On the other hand, Yndestad et al. has proposed that CCL21, but not CCL19, may have a directly pathogenic role in HF, in addition to predicting future mortality in HF patients [
A strength of our study is the large number of participants, and the large number of cardiovascular events accumulated over the long follow-up time. The nested case-control study was designed to study CE as the outcome, allowing us to perform the plasma analysis of CCL21 in only a subset of participants, while retaining much of the statistical power inferred from the entire cohort [
]. Although this matched study design is tailored for this outcome, it is not possible to study other outcomes as the sampling of participants may be biased. Thus, a limitation of the study is that we could not investigate if there is an association between CCL21 and mortality or other types of CVD similar to the associations between CCL19 and these outcomes. Another potential limitation is that the relative and not the absolute expression levels were obtained for CCL19, which does not allow direct comparisons between the absolute levels of CCL19 and CCL21 and makes comparison of the CCL19 levels between studies difficult.
In conclusion, in the current study we found that high plasma levels of CCL21 were independently associated with first-time incident CE in a population-based cohort, similar to what has previously been described in patients with established CVD. In contrast, plasma levels of CCL19 were not independently associated with first-time incident CE in this population-based cohort, but high plasma levels of CCL19 were independently associated with incident heart failure. These finding indicate that CCL21 and CCL19 may have different roles to play in the clinical manifestation of coronary heart disease and that CCL21 could be a potential biomarker for future coronary events regardless of previous disease history.
Financial support
This study was supported by the Swedish Research Council (Strategic Research Area Exodiab Dnr 2009-1039, Linnaeus grant Dnr 349-2006-237, Project Research Grant Dnr 2018-02939), The Swedish Heart Lung Foundation (Project Research Grants 20170617, 20200760 and 20190337), and The Albert Påhlsson Foundation.
CRediT authorship contribution statement
Pernilla Katra: Formal analysis, Visualization, Writing – original draft. Viktoria Hennings: Investigation, Writing – review & editing. Jan Nilsson: Resources, Writing – review & editing. Gunnar Engström: Writing – review & editing, Data curation. Daniel Engelbertsen: Writing – review & editing, Supervision. Eva Bengtsson: Writing – review & editing, Supervision. Harry Björkbacka: Conceptualization, Methodology, Writing – review & editing, Supervision.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
We thank Marlena Maziarz for her valuable statistical advice, Linda Andersson for managing the sample collection and Greg Markby for reviewing the language of the paper.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
Enhanced expression of the homeostatic chemokines CCL19 and CCL21 in clinical and experimental atherosclerosis: possible pathogenic role in plaque destabilization.
Elevated Lp-PLA2 levels add prognostic information to the metabolic syndrome on incidence of cardiovascular events among middle-aged nondiabetic subjects.
Plasmin and regulators of plasmin activity control the migratory capacity and adhesion of human T cells and dendritic cells by regulating cleavage of the chemokine CCL21.
The C-terminal peptide of CCL21 drastically augments CCL21 activity through the dendritic cell lymph node homing receptor CCR7 by interaction with the receptor N-terminus.
00005-9&id=ga1.jpg){kind=link}
00005-9&id=gr1.jpg){kind=link}
00005-9&id=gr2.jpg){kind=link}
00005-9&id=gr3.jpg){kind=link}