Intravenous fluid resuscitation is associated with septic endothelial glycocalyx degradation

Since its introduction during the cholera epidemics of the nineteenth century, intravenous fluid resuscitation has served as a mainstay of supportive sepsis care [1, 2]. Today, there is increasing concern that intravenous fluids may unexpectedly augment septic endothelial dysfunction, potentially negating the beneficial hemodynamic effects of fluid resuscitation [3]. Such iatrogenic injury could explain the findings of several recent randomized trials which demonstrated that early bolus intravenous fluids worsened sepsis survival [4, 5], as well as observational studies that identified associations between fluid administration [6, 7], fluid balance [8‐12], and adverse outcomes.

The mechanisms by which intravenous fluid resuscitation could cause harm are uncertain. Preclinical studies suggest that intravenous crystalloids promote degradation of the endothelial glycocalyx [13], a ubiquitous endothelial cell-surface layer composed of transmembrane or membrane-anchored proteoglycans (such as syndecan-1) covalently decorated with sulfated glycosaminoglycans (predominantly heparan sulfate, Fig. 1a). The glycocalyx is essential to microvascular homeostasis, as it contributes to the endothelial barrier, mediates shear-induced vasorelaxation, and opposes leukocyte-endothelial adhesion [14]. During sepsis, tumor necrosis factor-α [15] and angiopoietin-2 [16] induce endothelial expression and activation of heparanase, an endoglucuronidase that degrades glycocalyx heparan sulfate, inducing endothelial dysfunction and consequent organ injury. Heparanase and inflammatory stimuli may additionally induce metalloproteinase-mediated shedding of syndecan-1 [17], potentially augmenting glycocalyx collapse. Therefore, the presence of circulating glycocalyx constituents such as heparan sulfate or syndecan-1 fragments indicates a loss of glycocalyx integrity and associated endothelial injury [18].

Preclinical studies have suggested that atrial natriuretic peptide (ANP), a hormone released in response to volume loading-induced atrial stretch, is sufficient to degrade the endothelial glycocalyx in non-septic animals and humans [19‐21]. Similarly, a preclinical study of ovine endotoxemia observed that intravenous fluid resuscitation induced a simultaneous rise in circulating glycocalyx fragments and plasma ANP, coincident with worsened septic vasoplegia [13]. These concordant observations suggest that ANP upregulation could be a mechanism for volume overload-related glycocalyx degradation independent of tumor necrosis factor-α and angiopoietin-2-related degradation. The association between ANP and glycocalyx degradation in septic humans, however, has not been explored.

To explore the potential importance of the glycocalyx in human sepsis pathophysiology as well as the association between intravenous fluid resuscitation and glycocalyx degradation, we measured circulating glycocalyx constituents in (a) a subgroup of septic shock patients enrolled in the Protocolized Care for Early Septic Shock (ProCESS) trial and (b) sepsis patients presenting to emergency departments (EDs) at the Beth Israel Deaconess Medical Center (BIDMC, Boston, MA, USA) or St. Vincent’s Hospital (Worchester, MA, USA). We hypothesized that (1) the degree of glycocalyx degradation, as measured by circulating heparan sulfate, is associated with sepsis severity and mortality and (2) the volume of intravenous fluids administered early in resuscitation is independently associated with the degree of glycocalyx degradation.

Our report supports preclinical observations [13] that both sepsis severity and the volume of intravenous fluids administered during sepsis resuscitation are associated with glycocalyx degradation. While our study design is unable to prove causality, the associations are consistent across patient cohorts. An injurious effect of fluid resuscitation on the endothelial glycocalyx would be expected to worsen sepsis outcomes, given the importance of glycocalyx integrity to vascular homeostasis. In healthy vessels, the intact endothelial glycocalyx functions to oppose transvascular fluid flux and leukocyte adhesion. Accordingly, pathologic loss of glycocalyx integrity during sepsis might directly contribute to the tissue edema characteristic of septic organ injury [14]. Furthermore, as the glycocalyx plays a critical role in regulating nitric oxide synthesis, septic glycocalyx degradation may contribute to the microvascular heterogeneity and vasoplegia characteristic of septic shock [13]. Exacerbation of sepsis-induced glycocalyx degradation by intravenous fluids would therefore be expected to worsen inflammatory organ injury and microcirculatory dysfunction, significantly impacting patient outcomes.

The mechanisms by which intravenous fluids could induce glycocalyx degradation are uncertain. Pre-clinical animal and human studies have shown that ANP degrades the endothelial glycocalyx [19‐21], suggesting that fluid resuscitation may cause iatrogenic glycocalyx degradation in septic patients via the induction of volume overload. While our study does support an association between the volume of fluid resuscitation and glycocalyx degradation, we did not observe an association between plasma ANP (or BNP) and heparan sulfate. These findings, while observational and limited to plasma samples available from ProCESS cohort, do not support the hypothesis that natriuretic peptide-mediated degradation in response to volume overload is the primary mechanism responsible for intravenous fluid-associated glycocalyx degradation. Rather, intravenous fluids may be capable of directly inducing endothelial injury and endothelial shedding independently of fluid balance. One possible mechanism is fluctuations in endothelial shear stress caused by fluids. Sudden vascular stretch from fluid boluses paired with the presence of inflammatory mediators may stimulate endothelial expression of glycocalyx-shedding matrix metalloproteinases [34]. Furthermore, oscillatory shear stress may promote cathepsin L activation, an enzyme implicated in post-translational activation of endothelial heparanase [35]. Alternatively, isotonic fluid administration could directly activate circulating leukocytes [36, 37], potentially inducing neutrophil elastase-mediated glycocalyx degradation [38]. These speculative mechanisms will require additional translational investigation.

Intriguingly, the ProCESS and BIDMC/St. Vincent’s cohorts relied upon intravenous saline as the primary volume resuscitation agent, accounting for greater than 90% of the fluid administered in both cohorts. Emerging clinical and preclinical studies suggest that saline may be injurious when compared to balanced crystalloids [39], albumin [40], and fresh frozen plasma [41] in the resuscitation of critical illness. Future human studies will be necessary to determine if these resuscitation agents are differentially associated with glycocalyx degradation.

Our study has several strengths, including the use of two independent patient cohorts. Additionally, our findings are based upon state-of-the-art mass spectrometry analyses (LC-MS/MS MRM) of plasma heparan sulfate [28], an approach that is highly sensitive to both septic and non-septic endothelial glycocalyx degradation [31]. Circulating heparan sulfate is also highly specific to endothelial glycocalyx degradation: while heparan sulfate exists external to the vascular lumen (e.g. in the basement membrane and interstitium), negatively-charged fragments produced by degradation of extravascular heparan sulfates would be repelled by the negative charge of an intact glycocalyx, preventing plasma penetration. As such, the presence of circulating heparan sulfate fragments necessitates a breach of glycocalyx integrity. The validity of mass spectrometry measures of heparan sulfate as an index of glycocalyx degradation in our cohort is supported by the observed correlation with shed syndecan-1 (Fig. 1d), a commonly-used ELISA-based assay of glycocalyx degradation [18, 42], as well as shed thrombomodulin (Fig. 1e), an endothelial-surface chondroitin sulfate proteoglycan.

Our study also has several limitations. First, we only included a convenience sample of patients in the study cohorts and it is possible that this introduced a selection bias. Second, despite our study being amongst the largest to employ mass spectrometry to investigate glycocalyx degradation in humans, the overall sample size and number of deaths is still relatively low; thus, larger studies are needed to validate our findings. Third, it is possible that other measures of glycocalyx degradation are needed above and beyond circulating levels of heparan sulfate to have a comprehensive readout of glycocalyx degradation. Finally, it is important to emphasize that despite our use of multivariable modeling to account for measured confounders, our cohorts were underpowered to address additional pertinent variables that could affect glycocalyx integrity, such as underlying comorbidities [43], use of stress-dose glucocorticoids [44], and appropriate antibiotic choices (Fig. 2e). Our observational study is therefore unable to exclude the contribution of these and other unrecognized confounders of the observed association between fluid resuscitation and glycocalyx degradation. As such, we are unable to conclude causality.

Our observational findings therefore require confirmation by prospective, randomized studies. The Prevention and Treatment of Acute Lung Injury (PETAL) Network has recently initiated a large study of volume resuscitation practices in sepsis. This Crystalloid Liberal or Vasopressors Early Resuscitation in Sepsis (CLOVERS) study will compare the use of a liberal fluids protocol (larger volume of fluids prior to the initiation of vasopressors) with a restrictive fluids protocol (smaller amounts of intravenous fluids and early use of vasopressors) in patients with sepsis-induced hypotension [45]. Studies such as CLOVERS will provide opportunities to leverage randomized assignment of resuscitation strategies, allowing for better insight into the causal relationship between intravenous fluids and glycocalyx degradation.

Our report demonstrates an association between fluid resuscitation and glycocalyx degradation in sepsis, supporting observations made using preclinical models of endotoxemia [13]. Future randomized controlled studies will provide an opportunity to confirm a causal association.