Introduction

Shock, defined as the clinical expression of circulatory failure, leads to a mismatch in the oxygen demand and delivery to tissue that is initially reversible [1]. However, progressive cellular and tissue hypoxia rapidly becomes irreversible [2], leading to a cascade of multi-organ failure and, ultimately, death if untreated [3]. Endothelial dysfunction is the putative underlying common pathway leading to this cascade. Johansson et al. recently described shock-induced endotheliopathy (SHINE), a profound sympathoadrenal hyperactivation in shock states that leads to endothelial stimulation, glycocalyx damage, and eventual compromise of end-organ perfusion [4].

The breakdown and shedding of the glycocalyx layer in shock, termed endotheliopathy, trigger a cascade of inflammatory and coagulation responses that can lead to uncoupling of the macro- and microcirculation [5]. Invasive arterial access for pressure monitoring and acute resuscitative interventions for shock, including venous lines for rapid volume expansion and fluid bolus administration, non-pulsatile blood flow (such as mechanical circulatory support with extracorporeal membrane oxygenation (ECMO), and other extracorporeal support therapies, such as renal replacement therapy (RRT), have traditionally targeted restoring well-validated macro-circulatory endpoints such as improving mean arterial blood pressure and urine output. Microcirculatory endpoints, however, have been difficult to quantify objectively, particularly in critical illness [6,7,8], yet recent data has shown that these strongly correlate with patient outcomes [9, 10].

Despite several direct and indirect techniques for assessing endothelial integrity and microcirculatory flow, they have yet to be standardised and adopted for evaluating microcirculation during acute resuscitation [11]. Therefore, inference on endothelial dysfunction is usually obtained from a singular or a combination of different techniques of microcirculatory assessment.

While SHINE provides a biological plausible concept for endotheliopathy observed in shock, it is incomplete and needs to be further expanded to include haemodynamic resuscitation. Resuscitative interventions may either cause endothelial dysfunction by themselves, augment the dysfunction initially instigated by shock, and may inhibit the natural defensive mechanisms that repair endothelial integrity. We coined the term resuscitation-associated endotheliopathy (RAsE) to encompass this phenomenon beyond SHINE.

Some literature supports this theory and has explored the mechanistic effects of aggressive resuscitation protocols, including rapidly administered fluid boluses for volume expansion [12]. In addition, increased circulating volume and pressure within blood vessels may generate shearing forces, leading to glycocalyx shedding and subsequent endothelial dysfunction [13, 14]. Therefore, it is plausible that aggressive volume-expansion resuscitation exacerbates endotheliopathy, predisposing to progressive worsening of end organs and ultimately adverse outcomes in the context of a shock state.

This comprehensive review aimed to systematically examine the literature for studies describing endothelial dysfunction following resuscitation therapies administered in shock states. We sought to synthesise recommendations for reporting standards in this fast-expanding research are by quantifying published literature.

Methods

A predetermined systematic search, registered in the prospective international register of systematic reviews—PROSPERO (ID: CRD42022349074), was performed. Two online indexed medical databases, PubMed/MEDLINE and Excerpta Medica Database (Embase), were searched per the PRISMA guidelines [15] (Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram showing study selection, inclusion, and exclusion for the systematic review and meta-analysis

Full size image

All studies that reported on endothelial biomarkers data from critically ill humans who underwent resuscitation interventions were searched. The search terms used in [MeSH Terms] or [All Fields] were the keywords [‘resuscitation’] AND [‘endothelial dysfunction’ OR ‘endotheliopathy’ OR ‘endothelial damage’ OR ‘endothelial activation’]. The initial search was conducted from 1 January 2011 to 31 December 2021 and updated until 31 July 2023. A detailed description of the search strategy is included in Supplementary Table S1. All abstracts retrieved from the searches were filtered for duplicates, compiled in EndNote® (Thomson Reuters), and screened for relevance.

Eligible studies for inclusion were original clinical studies (including randomised controlled clinical trials, observational studies, case series, and case reports) published in English (or with an English translation). In these studies, the population was patients in shock states, and the intervention was administration of resuscitation fluid for treatment of the shock. Shock typically presents with a reduction in blood pressure, and most clinical guidelines recommend administration of volume expansion resuscitation to restore the microcirculatory parameters such as blood pressure and urine output. Therefore, some studies did not specify a comparator population, and it was assumed that all patients presenting in shock states were treated in accordance with the clinical resuscitation guidelines. The outcome of interest was a description of endotheliopathy following resuscitation for shock states by the following: (a) direct imaging for assessment of the microcirculatory function, measurement and quantification of (b) glycocalyx-breakdown biomarkers, and (c) mediators released from endothelial cells circulating in plasma. Relevant studies had their full manuscripts retrieved and reviewed by two independent reviewers in duplicate (NGO, DPS, assisted by RR, BS) (Supplementary Table S2). Assessment for risk of bias was performed based on the Cochrane risk of bias for randomised controlled trials [16] and the Newcastle–Ottawa scale (NOS) for observational studies [17] (Supplementary Table S3). Disagreements were resolved by consensus and additional senior review (SR, LESH). Reference lists and citations of the retrieved articles were also screened for relevance. The review articles and studies excluded did not describe endothelial dysfunction following resuscitation for circulatory shock or were conducted in animals.

Data analysis

Meta-analysis of eligible studies presenting means and standard deviations with comparable microcirculatory and endothelial assessments was performed. Transformational analysis based on the method by Wan et al. [18] was performed for comparison of studies assessing similar markers of endothelial and microvascular dysfunction but reporting medians and interquartile ranges. A random-effects meta-analysis model was used, and all analysis was performed using STATA (ver. 17).

Results

One-hundred and ninety-five articles were identified from the database searches, reference lists of key publications, and contact with authors. After an initial screening to remove duplicates and articles of no relevance, 102 studies were screened for eligibility, of which 68 full-text articles were accessed and reviewed. Thirty-two studies met the inclusion criteria and were included in the final qualitative synthesis as shown in Fig. 1. The details of these studies, including the population, intervention, control, and endothelial assessment, are presented in Table 1 below (additional details in Supplementary Table 2).

Table 1 Table showing population, intervention, control, and endothelial assessment of the included studies
Full size table

Of these thirty-two studies examining resuscitation and associated endotheliopathy included, there were 11 (34%) each on patients with trauma and haemorrhagic shock [19,20,21,22,23,24,25,26,27,28,29], and septic shock [30,31,32,33,34,35,36,37,38,39,40], and 6 (19%) on cardiogenic shock patients [41,42,43,44,45,46]. Two studies were case reports on systemic capillary leak syndrome (SCLS) [47, 48]. One study reported post-resuscitation endothelial dysfunction in acute respiratory failure [49] and another in dengue shock syndrome [50].

Twenty-four of the included studies (75%) were conducted in adults ≥ 18 years old (75%, i.e. 24/32) [20, 22, 23, 25,26,27,28, 30, 32,33,34,35,36,37,38, 40,41,42,43,44,45,46,47,48]. One study (3%) reported enrolment of a mix of adults and children [31].

Endotheliopathy assessment

A combination of mediators released from endothelial cells and biomarkers of glycocalyx breakdown was reported in 47% [15/32] of studies [19,20,21,22,23,24,25,26, 34, 37, 39,40,41, 44, 45]. The number of studies reporting only endothelial cell mediators in plasma was 25% (8/32) [28, 29, 33, 35, 42, 43, 49, 50], while those that exclusively reported biomarkers of glycocalyx shedding were 12.5% (4/32) [30, 36, 38, 48] (Supplementary Table 2).

The microcirculatory flow was assessed in 22% (7/32) of the studies, 57% (4/7) of which used orthogonal polarisation spectroscopy (OPS) [23, 31, 46], with one study only reporting the perfused boundary region (PBR) rather than microvascular flow [39]. One study used laser Doppler flowmetry [32] and flow-mediated dilatation (FMD) [27], while another study did not clearly describe the technique used to evaluate microvascular flow [47].

Meta-analysis

Ten unique studies were included in the quantitative meta-analysis. However, it was only possible to analyse comparative endothelial assessments performed. For the glycocalyx biomarker syndecan-1, eight studies were included [20, 24, 31, 34, 36, 38,39,40], with a heterogeneity index I2 = 75.85% and pooled effect size mean = 0.27 (95% CI − 0.07 to 0.60; p = 0.12) (Fig. 2a). Four studies were included for the endothelial cell mediator thrombomodulin [20, 24, 34, 42], with a heterogeneity index I2 = 78.93% and pooled effect size mean = 0.41 (95% CI − 0.10 to 0.92; p = 0.12) (Fig. 2b). Comparable data was available for three studies for the endothelial cell mediator E-selectin [20, 35, 40] (I2 = 50.29%; mean =  − 0.15; 95% CI − 0.64 to 0.33; p = 0.53) (Fig. 2c) and only two studies for the microvascular flow index (MFI) [31, 32] (I2 = 0%; mean =  − 0.80; 95% CI − 1.37 to − 0.24; p < 0.01) (Fig. 2d). Graphical summaries of the meta-analyses and publication bias are presented as Funnel plots (Supplementary Fig. 1a–d) and Galbraith plots (Supplementary Fig. 2a–d), respectively.

Fig. 2
figure 2

a Forest plot showing meta-analysis of eight studies that described syndecan-1 release post-resuscitation. Despite the high heterogeneity index, I2 = 75.87%, resuscitation caused a statistically significant release of syndecan-1 (pooled effect size; mean = 0.27; 95% CI - 0.07 to 0.60; p = 0.12). b Forest plot showing meta-analysis of four studies that described thrombomodulin release post-resuscitation. There was a high heterogeneity index between the studies with no significant effect of resuscitation on thrombomodulin release (I2 = 78.93%; mean = 0.41; 95% CI − 0.10 to 0.92; p = 0.12). c Forest plot of e-selectin release post-resuscitation. Only three studies described e-selectin release with high homogeneity but no significant effect of resuscitation on e-selectin release (I2 = 50.29%; mean =  − 0.15; 95% CI − 0.64 to 0.33; p = 0.53). d Forest plot of microvascular flow index (MFI) post-resuscitation. Only two-studies described MFI with high homogeneity but showed resuscitation significantly reduced the MFI (I.2 = 0%; mean =  − 0.80; 95% CI − 1.37 to − 0.24; p < 0.01)

Full size image

Discussion

Since the description of the SHINE phenomenon [4], this study systematically reviews post-resuscitation endotheliopathy in different types of circulatory shock. In summary, two-thirds of the studies included were published in 2017 or later, with equal numbers reporting on septic and haemorrhagic-trauma-related shock forming the bulk of the studies. While there are several biomarkers and techniques for assessing and quantifying endothelial and microcirculatory dysfunction, there are no standardised criteria for use in critically ill patients in shock. Only a few studies yielded comparable measurements for inclusion in the meta-analysis. Eight studies quantitatively compared syndecan-1, and four compared thrombomodulin that were included in the meta-analysis. However, these had relatively elevated heterogeneity indices indicative of underlying variability in the original studies. E-selectin had three comparable studies, while microvascular flow index (MFI) had only two comparable studies with high homogeneity. In the meta-analysis, only MFI reached the statistical threshold of significance. Four other studies meeting the inclusion criteria describing endothelial dysfunction post-resuscitation were not included in the meta-analysis as they did not fit within the framework of SHINE described by Johanssen et al. Two of these studies were case reports on systemic capillary leak syndrome: one on acute respiratory failure and another on dengue shock syndrome.

Knowledge gap in endothelial biomarker research

While there are several biomarkers and techniques for assessing and quantifying endothelial and microcirculatory dysfunction, there are no standardised clinical criteria. Therefore, clinical assessment and quantification of endothelial dysfunction in critical illness and during resuscitation need a consistent approach. Despite the previous description of microvascular dysfunction in critical illness [13], only a few studies yielded comparable measurements for inclusion in the meta-analysis. Injury to the endothelium and shedding of the glycocalyx trigger the inflammatory-coagulopathy cascade leading to progressive microvascular dysfunction [13, 51,52,53]. Most of the studies included in this review used a combination of biomarkers, including glycocalyx breakdown products and mediators released from endothelial cells [19,20,21,22,23,24,25,26, 34, 37, 39,40,41, 44, 45]. Syndecan was the most described glycocalyx breakdown product in the reviewed studies. It is a transmembrane proteoglycan that undergoes cytokine-mediated release during inflammation [54], with levels circulating in plasma increasing during shock states [4]. Of the syndecans classified, syndecan-1 is the most common in shock-induced inflammation and has been extensively described in a recent literature review [55]. Our results highlight the variability in reporting practices. Future reporting guidelines need to be more prescriptive to enable progress in this field of research.

Relationship between biomarkers and resuscitation practices

The typical clinical presentation of shock states is reduced blood pressure, indicative of impaired perfusion to match tissue requirements. Tissue hypoxia activates neutrophils in microvessels, and the subsequent neutrophil accumulation induces endothelial damage [56, 57].

Different shock aetiologies could impact the endothelium differently (Supplementary Fig. 3a–c). For instance, in septic shock, diverse pathogens may cause varied profiles of endotheliopathy [51]. Despite these differences, shock types share similar phenotypic features as the shock progresses, including sympatho-adrenal activation, catecholamine-induced glycocalyx damage, and pro-coagulant profile [4]. A recent review highlighted contradictions between basic, preclinical, and clinical studies on the significance of glycocalyx damage as a marker of vascular permeability [58].

Figure 3 shows the hypothesised exacerbation of endotheliopathy during resuscitation for septic, haemorrhagic, and cardiogenic shock. Preclinical evidence has demonstrated that aggressive volume expansion in acute critical illness resuscitation leads to the progression and exacerbation of microcirculatory and endothelial dysfunction of endotoxaemic shock [12]. Based on this conceptual framework, it is plausible that initial damage to the endothelial-glycocalyx layer from the underlying shock could be further exacerbated by subsequent resuscitative interventions, thus predisposing to additional end-organ injury. Currently, there is limited clinical evidence for variation in endothelial injury following aggressive resuscitation for different shock states.

Fig. 3
figure 3

Illustration showing progression from normal healthy endothelium through shock-induced endotheliopathy (SHINE) following shock to resuscitation-associated endotheliopathy (RAsE) post-resuscitation. RBC, red blood cell; IV, intravenous

Full size image

A summary of the pathophysiological mechanisms associated with the three different types of shock discussed are presented in Supplementary Fig. 3a–c. In septic shock, there is a relative reduction in the effective circulating volume due to vasodilatation and leakage into interstitial tissue. In contrast, in haemorrhagic shock, the decrease in the intravascular volume is due to blood loss. Volume replacement is the current standard of resuscitation for both these shock types [59,60,61]. In cardiogenic shock, global tissue hypoxia is secondary to poor perfusion, inducing a systemic inflammatory response syndrome (SIRS) comparable to sepsis [53, 62]. A similar SIRS response is seen with the initiation of extracorporeal membrane oxygenation (ECMO) [63]. The endothelial damage during shock and its subsequent exacerbation following resuscitation and reperfusion could have potential implications on clinical outcomes. It has been shown that disruption of the endothelial glycocalyx in cardiogenic shock is associated with worse patient outcomes [53, 64, 65]. Tsai et al. (2019) reported significantly higher levels of circulating vascular endothelial growth factor (VEGF), an endothelial survival factor associated with angiogenesis, at 72 h in patients who survived compared to those who died.

While higher resuscitation fluid volumes have been reported to correlate with higher levels of biomarkers such as syndecan-1 circulating in plasma [38], the clinical utility of syndecan-1 measurements during resuscitation remains limited. Thus, more research is required to correlate endothelial dysfunction biomarkers with the progression of end-organ dysfunction in shock states and during resuscitation [66]. Additionally, attainment of uniformity in analysing biomarkers of endotheliopathy requires some degree of standardisation of the time when they are measured since their relative abundance circulating in blood will vary over time. A proposed framework showing the major domains for the assessment and quantification of endotheliopathy in clinical studies is presented in Supplementary Table 4. With a better description of the endothelial injury biomarkers in the various resuscitation scenarios, investigators might be able to delineate the ‘epiphenomenon’ from real correlation and even risk factors.

This study has some limitations. One major limitation is the paucity of literature on resuscitation-associated endothelial dysfunction. Therefore, in order to understand which markers have been used clinically, studies were included on the basis of their reporting of endothelial markers post-resuscitation. Assessment of bias was performed in light of the fact that different study designs have been considered. As expected, the risk of bias was much lower in the randomised controlled studies than in the observational studies as reported in Supplementary Table 3. Funnel plots have been included to highlight the effects of smaller and non-randomised studies. Further description of the heterogeneity seen is provided by Galbraith plots that show the potential outliers mainly being non-randomised studies. However, the studies included in the meta-analysis are few, and more randomised studies investigating resuscitation-associated endothelial dysfunction are therefore required to address the knowledge gaps in this field.

Another limitation of this study was the lack of consistency in reporting of results in the studies that were reviewed. The lack of comparable biomarkers led to a reduction in the final number of studies that could be included in the quantitative meta-analysis resulting in high heterogeneity indices. Additionally, the application of transformations for the estimation of means and standard deviations from medians and interquartile ranges in the original publications were based on methods described by Wan et al. and are subject to mathematical assumptions [18]. Therefore, clearer reporting guidelines are necessary to achieve comparable and scientifically reproducible outcomes.

Conclusion

In this review, we conceptualise the term resuscitation-associated endotheliopathy (RAsE) in relation to worsening endothelial dysfunction resulting from acute resuscitative therapies administered in shock states described as shock-induced endotheliopathy (SHINE). Unfortunately, there is neither consensus nor consistency in the definition of microvascular biomarkers in critically ill patients. Thus, additional research and standardisation of the ideal assessment and panel of biomarkers are urgently needed.