Platelet versus plasma CXCL14, coronary artery disease, and clinical outcomes

Background Platelets express CXCL14, while platelet-derived CXCL14 induces monocyte chemotaxis and exerts an angiostatic effect on endothelial cells. Objectives This study investigated both platelet surface–associated and circulating levels of CXCL14 in patients with heart disease and associations of this chemokine with myocardial function and outcomes in patients with coronary artery disease (CAD). Methods This prospective study enrolled 450 patients with symptomatic heart disease. Platelet surface–associated and plasma CXCL14 levels were analyzed. All patients were followed up for 360 days for a primary composite outcome consisting of all-cause mortality, myocardial infarction, and/or ischemic stroke. Secondary outcomes consisted of the single events of all-cause mortality or myocardial infarction. Results Baseline platelet-associated but not circulating CXCL14 levels were significantly lower in patients with chronic coronary syndrome (mean fluorescence intensity logarithmized, 1.35 ± 0.35) when compared to those with acute coronary syndrome (1.47 ± 0.38) and without CAD (1.51 ± 0.40). Platelet CXCL14 levels were significantly lower (1.37 ± 0.37 vs 1.48 ± 0.39) and circulating CXCL14 levels were significantly higher (lg, 2.88 ± 0.20 pg/mL vs 2.82 ± 0.26 pg/mL) in patients with normal baseline left ventricular ejection fraction (LVEF) when compared to those with impaired LVEF. Low baseline circulating CXCL14 (hazard ratio, 2.33; 1.00-5.46) but not platelet CXCL14 was associated with worse outcome in patients with CAD. Conclusion Platelet-associated and circulating CXCL14 levels show differential regulation in patients with and without CAD. Although platelet-associated CXCL14 increased and circulating CXCL14 decreased with impairment of LVEF, only lower circulating CXCL14 upon admission was associated with worse prognosis in patients with CAD.


| I N T R O D U C T I O N
Besides their role in thrombosis and hemostasis, platelets are critically involved in inflammatory and immunomodulatory processes and thus progression of atherosclerosis [1]. Platelets store a variety of chemokines, which are released upon activation. Acting as both autocrine and paracrine mediators, they potentially promote inflammation in their immediate microenvironment and, thus, atherosclerosis or participate in its resolution [2]. Some platelet-derived chemokines like CXCL14 [3][4][5] and PF4 [6] are angiostatic, while SDF-1α [6] is both angiogenic and regenerative [7]. Therefore, activated platelet-derived mediators may decide on the delicate balance between vascular inflammation and regeneration [3,4,8]. The chemokine CXCL14 (BRAK, BMAC, Mip-2γ) has been shown to be expressed in leukocytes, endothelial cells [3], as well as in both human and murine platelets at protein levels and secreted upon activation [4]. CXCL14 expression in cells can be governed by genetic modulation in terms of methylation state and single-nucleotide polymorphisms (SNPs), which may have a bearing on its pathophysiologic actions, as recently shown by us in case of junctional adhesion molecule-A (F11R) in patients with coronary artery disease (CAD) [9]. Although CXCL14 SNPs have not been explored in the context of CAD, there are reports from studies on patients with influenza A (H1N1). Influenza infection is associated with increased cardiovascular risk and mortality particularly during the influenza season, also shown to be a predisposing factor for atherosclerosis. Influenza can also be a trigger for acute coronary syndrome (ACS) [10,11]. On the contrary, administration of influenza vaccine can reduce cardiovascular events in patients with CAD [12].
The influence of methylation in the CXCL14 promoter region and SNPs rs2237061, rs2237062, and rs2547 have been assessed with respect to disease severity in patients with influenza A (H1N1). This study revealed that CXCL14 expression is decreased and CXCL14 methylation is enhanced, even more so with respect to disease severity in patients with HIN1 infection, and correlates with decreased number of T lymphocytes. SNP analysis further revealed that although the frequency distribution of genotypes and alleles of rs2237061 and rs2237062 show no significant differences, there is a significant difference observed for rs2547. Compared with the GGwild-type carriers of rs2547, GA-mutant and AA-mutant carriers show increased risk of H1N1, therefore assigned as risk genotypes for H1N1 [13]. No doubt, similar studies in the context of CAD might reveal important insights.
CXCL14 mRNA is significantly upregulated in the left anterior descending arteries of obese pigs [14]. Furthermore, CXCL14 shows a strong chemotactic effect on monocytes, suggesting a potential role of this chemokine in monocyte migration and differentiation [4,15]. Besides, monocytes represent key players in atherosclerosis since they promote vascular inflammation [16]. This chemotactic influence has also been specifically shown for activated platelet-derived CXCL14 [4]. Platelet-derived CXCL14 imposes inhibition on the proliferation and migratory potential of human umbilical vein endothelial cells in scratch assays in vitro [4,17]. Platelet-derived CXCL14, on one hand, may prevent angiogenesis and, on the other hand, may foster atheroprogression in patients with pre-existing atherosclerotic disease.
CXCL14 may also modulate platelet functionality. Thrombus stability is affected in CXCL14-deficient mice, which can be recovered to normal levels upon ex vivo supplementation with recombinant CXCL14 [5]. CXCL14-mediated impact on platelet function could be exerted through CXCR4, which probably represents a major receptor for CXCL14, although conflicting evidence both in favor of and against this possibility has been reported [4,5]. Nevertheless, platelets do express CXCR4 [18,19]. We have previously shown that CXCR4 platelet surface exposure is higher in patients with CAD when compared to healthy individuals [18]. Additionally, we have demonstrated that lower platelet CXCR4 is associated with worse prognosis in patients with CAD [20]. We have also shown that recombinant CXCL14-mediated chemotaxis is affected in platelets derived from CXCR4-deficient murine and human induced pluripotent stem cell (iPS) culture-derived CXCR4-negative platelets [5].
This suggests that CXCR4 may well be a functional receptor for CXCL14 in platelets. Considering the potential implications of platelet-derived or platelet-associated CXCL14 and its receptor CXCR4 on functional recovery following myocardial infarction (MI), this study aimed to investigate both platelet surface-associated and circulating levels of CXCL14 in patients with heart disease. We sought to investigate differential expression patterns between CXCL14 levels in patients diagnosed with ACS, chronic coronary syndrome (CCS), and without CAD. Finally, we aimed to assess potential associations of CXCL14 with myocardial function and prognosis in patients with CAD.

Essentials
• CXCL14 may be associated with recovery of myocardial function in coronary artery disease (CAD).
• We investigated both platelet and circulating CXCL14 in 450 patients with heart disease.
• CXCL14 differed significantly between patients with CAD and those without CAD and correlated with myocardial function.
• Low circulating CXCL14 levels were associated with worse outcomes in patients with symptomatic heart disease.
This was a prospective study of patients with symptomatic heart disease. In this study, 450 patients were enrolled. The reasons for admission consisted of elective cardiac catheterization and acute chest pain.
All patients received cardiac catheterization immediately before study enrollment. Based on the results of cardiac catheterization, the patients were grouped into those having CAD vs those not having CAD.

| Blood collection
Ten milliliters of blood was collected at the cardiac catheterization laboratory from the arterial sheath in the femoral or radial artery during cardiac catheterization.

| Flow cytometry
Platelets in whole blood were analyzed for surface expression of CD62P, CXCR4, and CXCL14, gating for the platelet-specific marker CD42b. Blood was collected in citrate phosphate dextrose adenine,

| Impedance platelet aggregometry
We applied the Multiplate analyzer to study platelet aggregation levels. Furthermore, 600 μL of blood samples acquired in hirudinized tubes (Sarstedt) was used to perform adenosine diphosphate and thrombin receptor activating peptide tests. The area under the aggregation curve was used to determine overall platelet aggregation response.

| Survival outcomes and prognostic associations
All patients were followed up for 360 days for a primary composite clinical outcome consisting of all-cause mortality (ACM), MI, and/or ischemic stroke. Secondary outcomes consisted of the single events ACM or MI. Acute MI and ischemic stroke were defined as described previously [24]. Follow-up was performed by telephonic interview and/or review of patients' charts on readmission by investigators blinded to laboratory results. Forty-five (10.0%) patients were lost to follow-up.

| Statistical analyses
All statistical analyses were performed using SPSS, version 27.0 (IBM), and GraphPad Prism software (GraphPad Software, Inc).
Data are presented as median with 25th and 75th percentiles, mean ± SD, or count and percentage. Student's t-tests and Mann-Whitney U-tests were applied as appropriate to analyze differences between the 2 groups. Analysis of variance tests were applied for comparison between >2 groups. Correlations of normally distributed data were assessed by Pearson rank correlation coefficient (r).
Regression analyses were applied to test independent associations.
Cox proportional hazard regression was applied to investigate the associations between survival outcomes and both platelet-

| Platelets as a potential source of CXCL14 in heart disease
The study flow chart is presented in Figure 1. We characterized the surface association of CXCL14 on circulating platelets and ascertained plasma levels of circulating CXCL14 in patients with symptomatic heart disease. Baseline characteristics of the complete clinical cohort stratified according to CAD vs non-CAD are presented in Table 1.
For the CAD cohort (n = 388), the number of events and incidence rates per 100 person-years are presented in Table 2. CXCL14 platelet surface-associated levels were not associated with outcome. On the other hand, there was a marked trend among patients with low circulating CXCL14 levels to have worse outcomes (composite outcome and ACM) when compared with those with higher circulating CXCL14 levels ( Figure 5 and Table 3). As we did not see clear linear trends, we had to analyze the respective CXCL14 as categorical data.
F I G U R E 1 Study flow chart. CAD, coronary artery disease. *Due to staffing limitations and logistical challenges, we could not enroll more eligible patients. **Forty-five patients were lost to follow-up due to impossibility to establish contact to patients, treating physicians, or patients' relatives.
We, thus, had to primarily refer to the overall test with 3 degrees of freedom. In these tests, we did not find significant results. The results of multivariable Cox regression analyses are presented in Table 3.     Previous studies and experiments have suggested that CXCL14 may play a role in the pathophysiology of CAD [4,5]. Activated platelet-derived CXCL14 shows prominent thromboinflammatory F I G U R E 2 Baseline estimation of platelet surface-associated CXCL14 and circulating CXCL14 levels stratified according to chronic coronary syndrome, acute coronary syndrome, and noncoronary artery disease. *P < .05. **P < .01. ACS, acute coronary syndrome; CAD, coronary artery disease; CCS, chronic coronary syndrome; MFI, mean fluorescence intensity; ns, not significant.
F I G U R E 3 A, Correlation between baseline platelet surface-associated CXCL14 and circulating CXCL14 levels. B, Correlation between baseline CD62p platelet surface exposure and baseline CXCL14 platelet surface exposure. C, Correlation between baseline platelet surfaceassociated CXCL14 and baseline platelet surface-associated CXCR4 levels. D, Correlation between baseline creatine kinase and circulating CXCL14 levels. LVEF, left ventricular ejection fraction; MFI, mean fluorescence intensity.
influence in triggering monocyte migration [4,5,15], while it also exerts an angiostatic effect on endothelial cells and counteracts the angiogenic response of vascular endothelial growth factor and CXCL12 [4], which may impair vascular regeneration or reendothelialization after MI. These proinflammatory and angiostatic properties of CXCL14 taken together may foster progression of atherosclerosis, eventually leading to CAD, and also hinder functional recovery of the affected myocardium following ischemia-reperfusion injury by retarding angiogenesis. Therefore, we sought to investigate platelet-associated and circulating CXCL14 levels in patients with heart disease in this translational study. We found that platelet  Therefore, we reinvestigated the aggregation profile of 446 patients after ADP and TRAP stimulation. Considering that ADP does not activate platelets to the extent that TRAP does, besides TRAP-induced aggregation cannot be completely abolished by P 2 Y 12 inhibitors, unlike ADP, we created a ratio between ADP-and TRAP-induced aggregation. A higher ratio denotes that platelets are more prone to ADP activation. In the overall cohort, we observed higher plateletassociated CXCL14 levels in patients exhibiting a higher ADP/TRAP ratio. However, we could not find significant differences in plateletassociated CXCL14 levels when patients with CCS were stratified according to treatment with clopidogrel, ticagrelor, or prasugrel.
Interestingly, we found a much lower ADP/TRAP ratio in patients taking the more potent P 2 Y 12 inhibitors ticagrelor or prasugrel when compared with those taking clopidogrel, suggesting better inhibition of platelet activation. In patients with potent P 2 Y 12 inhibitors, we could show a strong correlation between ADP/TRAP ratio and platelet-associated CXCL14, indirectly suggesting that strong inhibition of platelet activation with potent P 2 Y 12 inhibitors results in lower platelet-associated CXCL14 levels ( Supplementary Figure). However, besides platelets, circulating leukocytes and inflamed endothelium may contribute significantly to circulatory CXCL14 levels. [3,15,28] Platelet-associated CXCL14 and circulating CXCL14 correlated inversely, which may indicate that platelets bind circulating CXCL14 onto their surface. Hyperactive platelets may act as a prime source of circulatory CXCL14, particularly in ACS, as they engage in mounting acute myocardial inflammation and subsequent regenerative or fibrotic processes following MI [29].
Unexpectedly, low circulating CXCL14 levels at baseline were associated with worse event-free survival. However, similar to LVEF, prognosis may be favorable if circulating CXCL14 levels decline over time. Intriguingly, circulating CXCL14 levels at baseline were lower in To conclude, we provide evidence on the differential regulation of platelet-associated and circulating CXCL14 levels and their association with myocardial function in patients with heart disease. Finally, we demonstrate associations of CXCL14 with the prognosis of patients with CAD. These findings provide interesting insights into the barely investigated effects of CXCL14 on the cardiovascular system and encourage further research in this direction.

| Study limitations
Being a translational study, adequate explanations for the clinical findings remain to be verified through experimental investigations in CXCL14 deficient cellular systems (ie, iPS-derived platelets) or murine models (eg, megakaryocyte-platelet lineage-specific CXCL14-deficient mice). Platelets are not the exclusive source of circulating CXCL14 and could be derived from circulating leukocytes; this was not explored in the current cohort and must be attended to in future studies. Another limitation of the current study is that angiogenesis was not investigated.
Although we did observe a correlation between platelet surfaceassociated CXCL14 and platelet surface expression of its receptor CXCR4, we did not verify this association with biochemical evidence from the current clinical cohort as we have previously demonstrated in our experimental studies through confocal microscopy, coimmunoprecipitation, and human iPS culture-derived platelets [5]. Such experimental analyses would have required copious amounts of blood, which could not be attempted due to ethical limits. Furthermore, the sample size of our cohort was moderate, which impaired the power to investigate secondary outcomes. We could not provide sequential biomarker measurements for the complete study collective, which would have been highly desirable to further delineate the chronic effects of platelet-associated and circulating CXCL14 on prognosis in CAD. In addition, the current study has several possible sources of bias, including a selection bias, since only a minority of eligible patients was enrolled in the study. Ten percent of patients were lost to follow-up. To our knowledge, we considered all possible confounders and adjusted for them. However, we cannot exclude residual confounding by unknown variables not included in our analysis. Furthermore, we performed 6 analyses with 2 predictors (platelet surface-associated and circulating CXCL14) and 3 outcomes (combined outcome, ACM, and MI). The Bonferroni correction was not feasible due to the limited power of the study. Thus, we cannot exclude the findings by chance.

ACKNOWLEDGMENTS
The technical assistance of Mrs Lydia Laptev is greatly appreciated in flow cytometry measurements and biobanking of plasma samples. We acknowledge support from the Open Access Publication Fund of the University of Tübingen.

RELATIONSHIP DISCLOSURE
There are no competing interests to disclose.