Introduction
Despite aggressive supportive treatment and prompt antibiotic administration, sepsis remains a life-threatening complication of infection [
1‐
4]. The endothelium plays a vital role in the pathogenesis of organ dysfunction in sepsis. Endothelial injury and dysfunction lead to the breakdown of the microvascular barrier resulting in increased extravascular fluid, tissue edema, organ dysfunction, and death [
5‐
8].
The endothelial barrier is composed of tight and adherens junctions. Vascular endothelial cadherin (VE-cadherin), an endothelial transmembrane glycoprotein, is the major cell-cell adhesion molecule that forms adherens junctions [
9,
10]. During inflammatory processes, the extracellular domain of VE-cadherin can be cleaved by neutrophil elastase and specific disintegrins and metalloproteinases (ADAMs) and shed into the circulation as soluble VE-cadherin (sVE-cadherin) [
11‐
13]. Several human studies showed elevated serum levels of sVE-cadherin in disorders associated with endothelial dysfunction including systemic vasculitis, malignancy, rheumatoid arthritis, and coronary atherosclerosis [
14‐
18]. Plasma levels of sVE-cadherin are also elevated in sepsis patients, and the degree of shedding of VE-cadherin is correlated with the severity of sepsis and poor outcomes [
19]. However, the relationship between specific organ failures and sVE-cadherin level in sepsis has not been studied.
Acute kidney injury (AKI) is a sudden decrease in renal function, characterized by an increase in serum creatinine level or decrease in urine output that results in significant morbidity and mortality, especially if renal replacement therapy is required (AKI-RRT) [
20‐
22]. There are many causes of AKI during critical illness, including sepsis, trauma, shock, surgery, and nephrotoxic agents [
23]. Although the role of tubular epithelial injury in AKI has been well studied, the contribution of endothelial injury is less well understood [
24,
25]. A previous report demonstrated plasma sVE-cadherin levels were increased in patients on chronic hemodialysis [
26]. However, whether or not shedding of VE-cadherin is associated with AKI in sepsis patients is not known. To test the hypothesis that VE-cadherin shedding is associated with sepsis-related AKI-RRT, we measured plasma sVE-cadherin levels in sepsis patients with or without AKI-RRT. We also evaluated the relationship between the degree of VE-cadherin shedding and other organ dysfunction.
Discussion
Injury and dysfunction of the microvascular endothelium, the pathophysiologic hallmarks of sepsis, lead to shedding of VE-cadherin from the endothelium into the circulation. The degree of shedding of VE-cadherin is correlated with the severity of disease in sepsis by APACHE II score [
19]. In the present study, we found that sVE-cadherin levels are elevated in sepsis patients with more organ failures and are associated with AKI-RRT. These results suggest that disruption of endothelial adherens junctions is a prominent feature of sepsis-induced severe AKI.
Sepsis commonly leads to organ dysfunction or failure, with AKI being one of the most common sepsis-related organ dysfunctions. Decreased renal blood flow in sepsis results in endothelial activation and increased renal microvascular permeability [
24,
35]. Several studies have investigated the mechanisms underlying increased microvascular permeability in sepsis. Flemming et al. found that treating human dermal microvascular endothelial cells with tumor necrosis factor-α (TNF-α) or bacterial lipopolysaccharide (LPS) caused VE-cadherin shedding from the endothelium and increased sVE-cadherin levels in the cell culture supernatant [
17]. In a mouse model of intraperitoneal LPS injection, VE-cadherin mRNA levels in the kidney increased at the 24th hour after LPS injection [
36], but the expression of VE-cadherin on renal endothelium was not changed significantly. Herwig et al. reported that the expression of VE-cadherin on pulmonary endothelium was decreased in septic ARDS patients compared with patients receiving lobectomy for lung malignancy [
37]. Furthermore, Carden et al. reported that the plasma levels of sVE-cadherin were elevated in ARDS patients compared with healthy subjects [
13]. Taken together, these studies suggest that shedding of VE-cadherin may exceed the production of VE-cadherin in sepsis, leading to disruption of adherens junctions and increased endothelial permeability. In a small study of 38 consecutive patients with and without sepsis, Zhang et al. reported that sVE-cadherin levels were elevated in sepsis patients and were associated with mortality [
19]. We extend these findings in the present study of 228 subjects with sepsis to study the association between organ dysfunction and sVE-cadherin with a specific focus on acute kidney injury. In the present study, to enrich for patients likely to develop AKI-RRT, we only enrolled severe sepsis patients with APACHE II score greater than 25 and respiratory failure requiring mechanical ventilation. We found that the plasma levels of sVE-cadherin, as an indicator of endothelial injury, were increased in sepsis patients with severe AKI. Furthermore, AKI-RRT was correlated with the level of sVE-cadherin even after adjusting for other AKI risk factors. A rise in sVE-cadherin could be a biomarker for early detection of severe AKI, but larger confirmatory studies are needed. To our knowledge, this is the first study of sVE-cadherin as a biomarker for sepsis patients with severe AKI.
Injury and dysfunction of the liver endothelium may also result in shedding of VE-cadherin into the circulation. In a mouse hemorrhagic shock model, mRNA levels of VE-cadherin in liver cryoresections were elevated within 90 min and persisted after resuscitation [
38]. Using a biochip-based, organoid model of human liver sinusoid with human umbilical vascular endothelial cells, monocytes and hepatocytes, Groger et al. found that the expression of VE-cadherin in the liver endothelium and MRP-2, a biliary transport protein, in hepatocytes was decreased after LPS stimulation [
39]. In the current study, we found that the plasma levels of sVE-cadherin were elevated in sepsis patients with liver dysfunction. Taken together, these studies suggest that shedding of VE-cadherin may also exceed the production of VE-cadherin in injured liver endothelium.
Aslan et al. measured VE-cadherin expression in a mouse model with intraperitoneal injection of LPS and reported discrepancies in the degree of VE-cadherin expression between the lung and the kidney [
36]. The mRNA levels of VE-cadherin in kidney were decreased at the 4th and 8th hour but increased at the 24th hour after LPS injection. The mRNA levels of VE-cadherin in the lung were decreased at the 24th hour. For VE-cadherin protein expression, there was no significant difference in kidney within 24 h. But there was a decrease in the protein expression of VE-cadherin in the lung at the 24th hour. Beyond these data, little is known about the degree of VE-cadherin shedding from the endothelium in different organs or species and this is an important topic for future studies.
Sepsis is a heterogeneous clinical syndrome that can be caused by a variety of underlying infections. Calfee et al. reported that biomarkers of endothelial injury, including angiopoietin-2, were higher in patients with ARDS due to non-pulmonary sepsis than ARDS due to pulmonary sepsis [
40]. Murphy et al. reported greater elevation of plasma syndecan-1, a degradation product of the endothelial glycocalyx, in ARDS patients with non-pulmonary sepsis compared to those with pulmonary sepsis [
27]. The current study is concordant with the prior reports, demonstrating that plasma sVE-cadherin levels were higher in non-pulmonary sepsis patients compared with pulmonary sepsis patients. Taken together, these studies provide a robust body of evidence that clinical non-pulmonary sepsis is characterized by more severe endothelial injury than pulmonary sepsis. Furthermore, Zhang et al. reported that the levels of sVE-cadherin were correlated with the severity of sepsis [
19]. Our study builds on that finding, showing that sepsis patients with higher plasma levels of sVE-cadherin had greater numbers of dysfunctional organs. Profound and widespread endothelial injury may be a potential contributor to the number and severity of organ failures. The role of endothelial injury in the distinct molecular phenotypes of pulmonary and non-pulmonary sepsis and the association between the degree of endothelial injury and the severity of organ dysfunction in sepsis could have important implications for the development of sepsis therapies that target endothelial injury.
Our study has some limitations. First, this was an observational, single-center study. Measurement of plasma sVE-cadherin in a larger multi-center study should be performed to validate these results. Second, we only evaluated the sVE-cadherin level at enrollment on the morning of ICU day 2; whether the change of plasma sVE-cadherin over time is in accordance with the disease progression is not known. Third, in order to enrich for more severely ill patients, this study enrolled only mechanically ventilated sepsis patients with an APACHE II score greater than 25. For this reason, these findings cannot be generalized to other less severely ill sepsis patients, patients without respiratory failure, or critically ill patients without sepsis. The high severity of illness in this cohort may be one reason that we did not detect an association between higher sVE-cadherin and mortality. Fourth, we could not directly measure the integrity of adherens junctions in our study, so it is not possible to determine whether the elevated levels of sVE-cadherin in this study are reflective of decreased adherens junction integrity in vivo. In our study, the plasma levels of sVE-cadherin were not associated with fluid accumulation during the study period. Whether other biomarkers of increased vascular permeability are associated with shedding of sVE-cadherin is unknown. Finally, little is known about the mechanisms of clearance of plasma sVE-cadherin. Doulgere et al. reported plasma sVE-cadherin levels were still elevated when serum creatinine levels decreased or were close to normal range in patients with hemolytic uremic syndrome due to Shiga toxin 2 producing
Escherichia coli infection (STEC-HUS) [
41]. Ebihara et al. reported that there was no difference in plasma sVE-cadherin levels before and after renal replacement therapy in sepsis patients [
42]. Whether elevated levels of sVE-cadherin are in part due to impaired clearance in the setting of other organ dysfunction needs to be further investigated.