1. Introduction
In recent decades, chronic heart failure (CHF) has emerged as a complex syndrome that involves a broad array of biological pathways [
1,
2]. In this context, CHF has been associated with endothelial dysfunction and low-grade inflammation [
3]. Moreover, the role of the immune system in the development and progression of CHF has received considerable attention in recent years [
4]. An essential step in this process is the adherence of circulating mononuclear cells to the vascular endothelium through binding of cell adhesion molecules (CAMs) that are expressed on the surface of these mononuclear cells, or on the endothelial cells, or on both [
5]. Binding of the mononuclear cells to the endothelium leads to extravasation of these cells into the involved tissue [
5], promoting structural deterioration, which eventually contributes to reduced cardiac function. Interestingly, enhanced expression of CAMs has been found within the myocardial microvasculature of patients with severe CHF as compared to healthy subjects [
6], providing further support that vascular inflammation might be involved in the propagation and progression of CHF.
Different classes of CAMs have been identified, and among them are selectins, integrins, cadherins and the immunoglobulin superfamily [
7]. In addition, several other molecules are involved in the cell adhesion processes. In more detail, selectins such as platelet (P)-selectin (SELP) are involved in the adhesion of leucocytes to activated endothelium and are known for the typical “rolling” of leucocytes on the surface of the endothelium. Other selectins such as endothelial (E)-selectin (SELE) are involved in the cell extravasation process. Integrins mediate the leucocyte adherence to the vascular endothelium and other cell–cell interactions [
8]. Cadherins are an important family of calcium dependent cell–cell adhesion molecules. In addition to their structural role, they have been implicated in the regulation of signaling events [
7]. For example, cadherin 5 (CDH5) is a major cell–cell adhesion molecule that forms adherens junctions [
9]. Lastly, the immunoglobulin superfamily comprises a diverse group of proteins including intracellular adhesion molecule-1 (ICAM-1), ICAM-2 and ICAM-3, vascular adhesion molecule-1 (VCAM-1), platelet endothelial cell adhesion molecule 1 (PECAM-1) and others, which are expressed on the surface of the endothelial cells and are known for firm adhesion of leucocytes and transendothelial migration [
10].
Shedding of CAMs from the cell surface results in measurable levels in peripheral blood [
11], which can reflect overexpression of their membrane-bound forms. Since CAMs may thus reflect processes involved in CHF, the association of these circulating biomarkers with clinical outcome provokes interest. Temporal patterns of biomarkers of cell adhesion in CHF, and their associations with an adverse disease course, have not yet been examined. Therefore, in this study, we investigated 12 cell adhesion-related biomarkers repeatedly measured with the Olink Multiplex panel, which contains 92 known human cardiovascular biomarkers that have previously been extensively investigated in the literature as well as exploratory candidates that are thought to carry potential as new biomarkers. Specifically, here, we examined biomarkers from this panel related to the above-described mechanisms (SELP, SELE, CDH5, ICAM-2, and PECAM-1) and other potentially interesting biomarkers related to cell adhesion processes (complement component C1q receptor (C1qR), chitinase-3-like protein 1 (CHI3L1), contactin-1 (CNTN1), ephrin type-B receptor 4 (EPHB4), epithelial cell adhesion molecule (Ep-CAM), integrin beta-2 (ITGB2), and junctional adhesion molecule A (JAM-A). The aim of the present study was to evaluate the association between temporal patterns of these biomarkers of cell adhesion and clinical outcomes in stable patients with CHF.
4. Discussion
In the present study, we found that biomarkers of cell adhesion C1qR, CDH5, CHI3L1, EPHB4, ICAM-2 and JAM-A were associated with clinical outcomes in 263 stable patients with CHF. At baseline, levels of biomarkers C1qR, CDH5, CHI3L1, EPHB4 and JAM-A were higher in patients who later experienced the PE compared to patients who remained event-free. Furthermore, the average biomarker evolutions over time of these markers, and additionally of ICAM-2, showed higher levels as the PE approached. Even more important, repeatedly measured levels of these biomarkers of cell adhesion were independently associated with the PE. Even adjusted for clinical factors, biomarkers of cell adhesion served as predictors of clinical adverse events.
Recent studies suggest a pivotal role of CAMs in the processes of HF. Until now, however, research on CAMs in relation to adverse clinical outcomes in patients with CHF is limited. Previous studies have mostly described the value of single measurements of adhesion molecules (e.g., at admission) for prognosis, and studies were relatively small. Our study, which was based on repeated measurements, demonstrates a promising role for several adhesion biomarkers for individual prognostication in CHF patients Temporal patterns shortly before an adverse event occurs have not yet been investigated in detail previously, while this might be a crucial time window for therapeutic interventions.
In our study, CHI3L1 was the biomarker whose association with the PE was numerically the strongest after adjustment for clinical factors. CHI3L1 is a glycoprotein secreted in vitro by cells such as activated macrophages and neutrophils in different tissues with inflammation. Studies on patients with acute myocardial infarction, stable coronary artery disease, atrial fibrillation and CHF have demonstrated elevated levels of CHI3L1 compared with healthy controls [
21]. Moreover, several studies have previously examined CHI3L1 in relation to clinical outcome in CHF; but repeated measurements were never used. Some of these studies showed that CHI3L1 is associated with all-cause mortality [
22] and that it is able to detect patients at high risk for adverse outcomes as well [
23,
24]. Other studies failed to demonstrate such associations. Rathcke at al. examined CHI3L1 levels in patients with CHF and in age-matched controls without cardiovascular disease [
25]. They found higher levels of CHI3L1 at baseline in patients with CHF, but these levels did not predict cardiovascular events or overall mortality. Mathiasen et al. [
21] suggested that, most likely, elevated levels of CHI3L1 in CHF patients are explained by the presence of concomitant diseases. CHF is a complex disorder, often complicated by other comorbidities in which CHI3L1 is known to be elevated, such as arrhythmias, renal dysfunction, diabetes mellitus and hypertension. These concomitant diseases could thus possibly explain the differences in CHI3L1 levels when compared to healthy individuals. Conversely, in our study, we not only adjusted for age, but also for clinical factors, and still we found an association between CHI3L1 and clinical adverse events.
The barrier formed by endothelial cells allows regulated passage of immune cells in the normal state and during inflammatory conditions. This passage is mediated through junctional molecules, such as ICAM-2, CDH5, JAMA, and PECAM-1 [
26,
27]. ICAM-2 participates in the docking of leukocytes to the endothelium, and is likely to be relevant for leukocyte diapedesis [
28]. For example, former research showed that endothelial cell activation leads to neutrophil transmigration, supported by the sequential roles of ICAM-2, JAM-A and PECAM-1 [
26]. We are not aware of previous investigations that link ICAM-2 to prognosis of stable CHF patients. We show that rate of change of ICAM-2 independently predicts adverse clinical outcome. This suggests that prognosis differs between patients with stable ICAM-2 values and patients with increasing ICAM-2 values. CDH5 is an endothelial transmembrane glycoprotein and is the major molecule for cell–cell adhesion that forms adherens junctions [
9]. Shedding of CDH5 into the circulation is associated with severe acute kidney injury and with more severe organ dysfunction in patients with sepsis [
29] and increased levels of soluble CDH5 were associated with poor outcome in severe sepsis [
30]. In cardiovascular research, elevated levels of CDH5 have also been reported to be associated with coronary atherosclerosis [
31]. Based on our results, CDH5 may be of use as a biomarker that reflects on-going inflammation and indicates impending adverse events in CHF patients. JAM-A is involved in the regulation of vascular permeability [
27] and genetic deletion and blockade of JAM-A generally results in increased permeability of endothelial cells [
32]. JAM-A is also thought to be required for movement of leukocytes toward sites of inflammation [
33] and it may be considered as a marker of acute endothelial activation and dysfunction [
34]. This is in line with our findings; we demonstrate that repeatedly measured levels of JAM-A show a numerically strong independent association with the PE. The significant role of PECAM-1 in platelet aggregation and migration of leukocytes through the endothelium [
35] is interesting in the context of CHF. PECAM-1 has been suggested as a sensitive marker providing early diagnostic aid in acute coronary syndromes [
36]. In heart failure research, soluble PECAM-1 was found to be elevated in the majority of patients with severe CHF [
37]. However, we did not find an association of PECAM-1 with prognosis in our CHF cohort.
SELP is of great interest because of its key role in interactions between platelets, leucocytes, and endothelium [
38]. Abnormal surface SELP expression [
39,
40] and soluble SELP levels [
41] have been reported in decompensated heart failure, suggesting persistent platelet activation. Regarding their prognostic value, however, levels of soluble SELP, platelet surface SELP, and total platelet SELP did not determine prognosis [
42] and our results support these findings. Ep-CAM, CNTN1, ITGB2, and SELE also showed negative results in our study.
Less is known about the other biomarkers in relation to CHF. For example, C1qR is a transmembrane receptor once thought to be only a receptor for C1q, but is now thought to play a role in endothelial cell adhesion [
43]. The up-regulation of this receptor by inflammatory mediators and the ability of complement component C1q itself to increase ICAM-1 expression suggest a potential role for the receptor in vascular inflammation and immune injury [
44]. To the best of our knowledge, C1qR has never been linked directly to prognostication in CHF patients. In our study, repeatedly measured levels of this marker were independently associated with the PE. EPHB4 serves as receptor for its transmembrane ligand ephrin-B2. Both are specifically expressed on arterial and venous endothelial cells. Hamada et al. concluded ephrin-B2 forward signaling and EPHB4 reverse signaling differentially affect cell adhesion and migration between arterial and venous endothelial cells [
45]. We found that both level and slope analysis of EPHB4 were significantly associated with the endpoint, even after adjusting for clinical factors.
While the 263 patients included in our investigation were ambulatory and stable, it has been advocated that grouping of HF patients should not be approached only based on symptoms [
46], nor on ejection fraction solely [
47]. Definitions have been described to identify more advanced disease HF (AdHF), i.e., patients with worsening clinical condition, high rates of re-hospitalization and mortality (meaning a condition where standard treatments are inadequate and additional interventions must be applied; these patients are suitable for LVAD), as well as end-stage heart failure (patients for which advanced therapies, such as LVAD, is contraindicated and palliative cares should be pursued) [
48]. In post-hoc analyses, based on our available data, we identified at least 57 patients who might be categorized into these two groups at baseline; given their ambulant condition most likely the AdHF group. Thirty of them eventually experienced an endpoint during follow-up. Compared to the other 206 patients, these 57 patients were older, had a higher heart rate, lower systolic and diastolic blood pressure, had higher NT-proBNP, hsTnT and eGFR levels, and were more likely to have prior CVA/TIA and diabetes mellitus. Malfunction of other organs could affect prognosis [
49], and, therefore, differences in such risk factors should be taken into account (as for example also highlighted in a recent study about the role of oxidative stress and vascular inflammation in diabetic patients which could result in myocardial infarction [
50]). Since we adjusted our current analyses of the association between circulating biomarkers of cell adhesion and clinical outcomes for variables such as diabetes mellitus and atrial fibrillation, we believe we have accounted for this type of confounding as much as we could in this observational study.
Our study has some limitations. First, because of efficiency reasons, we did not use all 1984 available trimonthly samples, but selected 3 samples for patients with a PE (baseline and last 2 prior to the PE), and 2 samples for event-free patients, resulting in 567 samples. Our previous investigations using all samples demonstrated that most of the examined biomarkers show an increase shortly prior to the incident adverse event. Thus, we believe that with our approach we retain the most informative measurements while enhancing efficiency. Second, as described before [
15,
51], our cohort comprised mainly HF patients with a reduced ejection fraction. This can most likely be attributed to the fact that in the Netherlands, most HF patients with a preserved ejection fraction are treated in secondary referral centers or by the general practitioner. Finally, we used biomarker values in Normalized Protein Expression (NPX) Units, i.e., relative units. While these values can be used for comparing patients and changes over time within a patient, for clinical applications absolute concentrations are recommended.
In conclusion, the present study demonstrates that serial measurements of C1qR, CDH5, CHI3L1, EPHB4, ICAM-2 and JAM-A are independently associated with clinical adverse events in patients with CHF, suggesting that markers of cell adhesion could be useful for individual risk profiling. These biomarkers are also interesting for future therapeutic purposes, as CAMs may be used as targets to inhibit vascular inflammation and endothelial dysfunction. Further studies are warranted to confirm these associations, to investigate whether a combination of different markers (for example C1qR, CHI3L1 and JAM-A) may improve prognostication and to better elucidate the pathophysiological role of cell adhesion in CHF.