Introduction

Capillary leak syndrome (CLS) refers to a syndrome of deranged fluid homeostasis, often observed in critically ill [1,2,3]. In clinical practice, CLS is frequently defined by excessive fluid shift from the intravascular to the extravascular space, resulting in intravascular hypovolemia, extravascular edema formation, and hypoperfusion—necessitating further fluid resuscitation [3].

In health, fluid exchange between intravascular and extravascular spaces is vital for maintaining the body’s homeostasis. However, disturbances in this delicate equilibrium, often driven by systemic inflammation (e.g., sepsis) can lead to the clinical picture of CLS [4,5,6].

Despite efforts to define CLS, there is no established clinical definition nor accepted diagnostic criteria [1, 3]. Previously, Marx et al. characterized CLS as a fluid extravasation, resulting in edema and hypoperfusion [3]. The authors studied septic shock patients using various methods such as indocyanine green measurements, chromium-51 labeled erythrocytes, and colloid osmotic pressure measurements, aiming to differentiate CLS from other hypo-oncotic conditions and clinical scenarios associated with fluid retention [3]. The definition of CLS proposed by Marx et al. in 2000 emphasized three main aspects: intravascular hypovolemia despite fluid resuscitation, generalized edema, and hemodynamic instability. This pivotal description, while not universally adopted, offers valuable insight into the key features of CLS, aiding in the differentiation of this syndrome from other conditions that share similar clinical manifestations. It underlines the necessity for an accepted definition of this syndrome to develop targeted and effective therapeutic interventions [3].

This article will review the current understanding adding new aspects of CLS in clinical practice, and give an overview about the pathophysiology, clinical presentation, diagnostic and therapeutic options.

Pathophysiology of CLS and implications

Triggers and key features

A CLS phenotype can be triggered by numerous disease states as well as certain medications and toxins [7]. Depending on the literature source, terms like “generalized hyperpermeability syndrome”, “endothelial permeability” or “capillary leakage” may be used synonymously for CLS. As an important semantic distinction, the “idiopathic systemic capillary leak syndrome”, also referred as Clarkson’s disease [114,115,116].

The destructive process of glycocalyx during sepsis has substantial physiological implications. The loss of this protective barrier directly impacts local tissue, but the degradation products themselves can also circulate and affect distant sites in the body [85]. This leads to a system-wide impact that contributes to fluid shifts and the multiple organ dysfunction often seen in septic patients. The extent and the specific mechanisms through which glycocalyx degradation affects the progression and prognosis of sepsis are still being uncovered. This understanding is critical for the development of therapeutic strategies to preserve the eGC, attenuate the inflammatory response, and ultimately improve the outcomes of sepsis.

Fluid overload and dynamics

The human body contains various fluid compartments, both intravascular and extravascular, which have specific volumes and protein contents. According to indicator dilution studies, a healthy 70 kg adult typically has about 3 L of plasma, containing around 210 g of protein [117]. On the other hand, the same adult will have approximately 12 L of interstitial fluid. This fluid resides in a gel phase and contains 240–360 g of protein. The capillary pressure in this system is higher than the pressure in the interstitial space, which drives the movement of the solvent and its small lipophobic solutes towards the interstitial space [118]. Trans-endothelial fluid shifts are regulated by the vascular barrier in addition to hydrostatic and oncotic forces, as described by the revised Starling equation [119]. In healthy organs, the increased permeability and movement of proteins and plasma fluid are temporary and decrease once the stimulating factor is removed. Edema is traditionally perceived as a consequence of a pressure-driven net outward filtration in the capillary, partially reversed by fluid reabsorption at the venous end by an oncotic pressure gradient [120]. Contrary to traditional perspectives, more recent theories propose that continuous net filtration is the norm in most capillary networks [121]. Apart from an increased pressure gradient, edema can also be caused by hypo-oncotic states, changes in permeability and impaired lympathics. 

The capillary wall includes a glycocalyx layer, which is a complex meshwork of glycosaminoglycans and additional glycoproteins. This layer serves as a filtration barrier, featuring gaps where capillary filtration takes place [121,122,123]. The movement occurs through regulation of the glycocalyx and the occasional breaks in the inter-endothelial junctions. These breaks constitute less than 0.1% of the total endothelial surface area, allowing a highly regulated fluid exchange process [117]. The glycocalyx layer was previously assumed to have an almost perfect reflection coefficient for proteins, particularly albumin. However, albumin diffusion through capillary pores results in about half of the body's albumin content residing extravascularly, with interstitial oncotic pressure reaching 30–60% of plasma oncotic pressure [124]. The complexity of the interstitial space has been underestimated in the past. It actually consists of a triphasic system that includes freely moving fluids, a gel-like phase rich in large polyanionic glycosaminoglycans molecules, and a collagen framework [117, 124]. Albumin is predominantly absent from this luminal surface, leading to a stronger intravascular oncotic pressure than what direct measurements of interstitial albumin concentration would suggest [125]. As a result, the net filtration process is more influenced by the oncotic pressure beneath the endothelial glycocalyx than by the capillary membrane itself [123].

The clinical consequences of these fluid shifts can be manifold, yet not immediately visible to the clinician. The lungs are especially prone to pulmonary edema due to the unfavorable ratio of endothelium per tissue with the clinical potential to impact gas exchange, and predispose the lungs to further infectious complications [126]. Additionally, the gastrointestinal tract may become edematous, leading to paralytic ileus, an increase in intra-abdominal pressure and subsequent tissue hypoxia, and impaired wound healing [127, 128]. It is noteworthy that the endothelium is highly heterogeneous across different vascular beds; for example, CLS commonly affects various organs but is rarely observed in the brain due to the unique properties of the blood–brain barrier, including a higher pericyte-to-endothelial cell ratio that contributes to its greater impermeability [129].

While CLS is widely acknowledged in the critical care settings, there is a surprising lack of clinical studies exploring its impact on organ dysfunction and mortality [1]. This may stem from the current absence of accepted diagnostic criteria for CLS. However, associated conditions like an inflammatory state and positive fluid balance—circumstances inevitably related to CLS—correlate with higher mortality rates in the ICU [130]. For example, elevated levels of serum cytokines are commonly observed in non-survivors of critically illness, and a positive fluid balance is acknowledged as an independent predictor of outcomes in patients with sepsis [131, 132].

Fluid management can be complex in ICU settings, demanding a thorough understanding of body fluid homeostasis [133]. Fluid overload, which comprises whole body water, i.e., extra- and intravascular fluid, can be detrimental and associated with negative outcomes in patients who are critically ill [134,135,136,137,138,139,140,141,142,143,144,3).

Fig. 3
figure 3

Phases of capillary leak with increased vascular permeability on the left leading to distinct clinical manifestation and necessitating aggressive treatment strategies, while the recovery phase on the right consists of stabilizing and optimizing the fluid status with de-resuscitation

Preservation of the endothelial surface layer (ESL)

In this section, the term 'Endothelial Surface Layer (ESL)' will be used to refer to the intricate structure formed by the endothelial glycocalyx (eGC) along with associated plasma proteins. The eGC serves as a luminal mesh that provides endothelial cells with a framework to bind plasma proteins and soluble glycosaminoglycans [181]. While the eGC itself is considered inactive, it becomes physiologically active once it binds with or is immersed in plasma constituents, thereby forming the ESL. It is worth noting that the specific roles and clinical relevance of the eGC as part of the broader ESL are subjects of ongoing research. The ESL is instrumental in maintaining vascular homeostasis, regulating vascular permeability, and acting as a mechanosensor for hemodynamic shear stresses, in addition to displaying antithrombotic and anti-inflammatory characteristics [182]. Plasma proteins, especially albumin, bind within the glycocalyx and aid in stabilizing this layer [183]. Albumin's function is particularly important as it contributes to plasma colloid osmotic pressure (among other, often unmeasured molecules). Moreover, albumin performs a range of roles—from acting as a free radical scavenger and transporting sphingosine-1-phosphate (which has protective effects on the endothelium), to providing immunomodulatory and anti-inflammatory effects [125].

Experimental studies have highlighted the multifunctional nature of albumin, which includes maintaining ESL integrity, partially restoring compromised vascular permeability, exhibiting anti-oxidative properties and anti-inflammatory properties, improving hemodynamics and microcirculation following endotoxemia or hemorrhagic shock, and acting as an effective plasma volume expander [125, 184,185,186,187,188,189,190]. Interestingly, beneficial effects appear to be independent of albumin's oncotic properties. Additional research has shown that the choice of fluid for infusion significantly affects the ESL [125, 191]. For instance, in vivo experiments conducted on anesthetized rats subjected to hemorrhagic shock followed by fluid resuscitation, the use of normal saline failed to restore ESL thickness and plasma levels of syndecan-1 [192]. Conversely, albumin was found to stabilize permeability and leukocyte rolling/adhesion, partially restoring ESL thickness and reducing plasma syndecan-1 to baseline levels [125, 192]. Authors have proposed several mechanisms to elucidate the positive influence of albumin on the endothelium [193]. Primarily, albumin might alleviate sepsis-induced damage to the ESL. As reviewed by Aldecoa et al., albumin, due to its amphoteric properties, has the ability to establish strong bonds with the ESL, while its negative charge aids in maintaining its parietal electrical barrier [125]. In addition, the antioxidant functions of albumin are well-documented [125]. Albumin's free thiol group, carried by a cysteine residue (Cys-34), assists in neutralizing harmful plasma free radicals, which is highly relevant in the septic environment marked by a high oxidative state. Lastly, albumin's capacity to form complexes with heavy metals provides protection against oxidation via the Fenton reaction [193]. Hariri et al. underscore the mounting evidence, both from experimental models and in the context of critically ill patients, that suggests the protective role of albumin on the endothelium during acute injury [193]. Preservation of the ESL using albumin (and fresh frozen plasma) is intriguing, however clinical studies need to confirm these findings. It is anticipated that the ongoing multicenter ARISS trial will further shed light into the effects of albumin on clinical outcomes [194]. It is crucial to note that commercial albumin solutions are often heated to 60 °C for several hours for inactivation of infectious agents [195]. This heat treatment can lead to protein denaturation and alterations in its negative charge [195], raising the question of the comparability of administered albumin with physiologically circulating albumin synthesized by the liver.

Various clinical studies examine the effects of albumin in the clinical context. Zdolsek et al. have shed light on the impact of exogenous albumin administration on fluid dynamics under various clinical conditions [196]. The primary focus of their study was to evaluate the rate at which infused albumin dissipates from the bloodstream, quantified as the half-life (T1/2), under different clinical scenarios. Their research involved intravenously infusing 3 mL/kg of 20% albumin into a varied population that included healthy volunteers, patients after burns, postoperative patients, and patients who underwent surgery with both minor and significant bleeding. The results showed a consistent T1/2 across all groups, except for those who experienced surgery with major bleeding. In the latter case, the infused albumin disappeared faster, indicating a greater loss of albumin in situations of significant hemorrhage. Zdolsek and colleagues further compared the effects of 20% and 5% albumin concentrations on plasma volume expansion [197]. The study was designed in a way that the same mass of albumin was administered under both scenarios. Their findings showed that while both concentrations led to plasma volume expansion, the 5% albumin concentration had a slightly higher rate of volume expansion. However, they found that a third of the 5% albumin solution quickly leaked from the plasma, likely due to the higher colloid osmotic pressure of volunteer plasma than that of the albumin solution. By the 6-h mark, about 42–47% of the administered albumin had leaked from the capillaries, regardless of the concentration used.

Further research by Hahn and colleagues investigated the body fluid shifts when 20% albumin is administered intravenously, with a specific focus on postoperative patients [198]. They found that the infused albumin expanded the plasma volume beyond the volume of the infusion itself by moving non-circulating fluid. However, the same mechanism also increased fluid losses from the system. Despite these dynamics, they observed that the plasma albumin level and plasma volume remained stable for about 2 h post-infusion. Therefore, the effectiveness of albumin as an administered fluid may depend on the specific clinical scenario and the administered concentration.

Microvascular and ESL protection prior to surgeries (i.e., before an anticipated inflammatory insult) presents an interesting area of research, as highlighted by Yanase et al. [199]. In their study, they explored the feasibility, efficacy, and safety of potential protective influence of dexamethasone and albumin on the ESL in patients undergoing abdominal surgery. In this trial, patients were randomly assigned to two groups. One group was given intravenous dexamethasone and 20% albumin at the onset of anesthesia, followed by additional albumin with each subsequent crystalloid administration. The control group, conversely, received only crystalloid fluid without dexamethasone leading to differences in the crystalloid, colloid administration. The outcomes were evaluated based on alterations in plasma syndecan-1 and heparan sulfate levels as markers for eGC damage, and inflammatory markers measured at four perioperative timepoints. Although no significant differences were noted in syndecan-1 levels between the two groups, the group that received the dexamethasone-albumin treatment demonstrated lower heparan sulfate and C-reactive protein levels on the first postoperative day, suggesting a potential protective effect on the glycocalyx. This group also experienced fewer postoperative complications [199]. It remains uncertain if this effect is related to the dexamethasone or albumin administration, or the combination thereof.

It has to be noted that the role of albumin administration in critically ill patients has been studied extensively in the past. The ALBIOS trial conducted by Caironi et al. [200] aimed to evaluate the efficacy of albumin administration in patients with severe sepsis. In this multicenter, open-label trial, 1818 patients with severe sepsis were randomized to receive either a 20% albumin and crystalloid solution or a crystalloid solution alone. The albumin group was targeted to maintain a serum albumin concentration of 30 g per liter or more until discharge from the ICU or 28 days after randomization. During the first 7 days, the albumin group demonstrated a higher mean arterial pressure and a lower net fluid balance compared to the crystalloid group. However, no significant difference was observed in the total daily amount of administered fluid between the two groups. The 28-day and 90-day mortality rates did not show significant differences between the two groups, indicating that albumin replacement in addition to crystalloids did not improve survival rates at these timepoints [200]. These findings do not support the hypothesis that albumin administration has survival benefits in severe sepsis, despite previous studies and experimental evidence for its protective role. However, the ALBIOS trial did confirm some physiological benefits of albumin administration. Patients in the albumin group exhibited superior hemodynamic responses, with a higher mean arterial pressure, lower heart rate, and lower net fluid balance in the first 7 days of treatment [200]. The average cardiovascular SOFA subscore was lower in the albumin group, and the time to suspension of inotropic or vasopressor agents was shorter, suggesting a decreased need for vasopressors [200]. Similar to the ALBIOS trial, the ALBICS trial for albumin use in cardiac surgery did not show a benefit on major adverse events at 90 days [201]. Many unanswered questions remain around the role of albumin administration, e.g., its role in effective de-resuscitation and augmenting loop diuretic effects [202] and the comparability of exogenously administered albumin’s properties compared to that of circulating albumin. Due to these reasons, no final recommendation can be given for the role of albumin administration for CLS treatment.

Lymphatics in ICU patients

Unlike the cardiovascular system, which ensures bidirectional blood flow, the lymphatic system is specifically designed for unidirectional transit from the extracellular space to the venous system [45]. The lymphatic system plays a pivotal role, actively participating in maintaining tissue fluid equilibrium, aiding in the absorption of lipids from the gastrointestinal tract, and playing an important role in the immune response by transporting antigen-presenting cells and lymphocytes to lymphoid organs [203]. Of note, the lymphatic flow can be increased in health and disease. In the context of critical care, the lymphatic system's potential for increased flow offers interesting avenues for research.

In the critical care setting, physical therapy involving manual lymphatic drainage presents an interesting approach as it has been shown to enhance lymphatic outflow and mobilize fluid [204,205,206]. Studies have found that manual lymphatic drainage can significantly improve the transportation of various substances within the lymphatic system [204,205,206,207]. The findings indicated that manual lymphatic drainage can lead to a modest increase in plasma volume, averaging around 1.5 ± 0.8% [207]. This expansion suggests that lymphatic fluid is being mobilized into the bloodstream. Recent research showed an increase in albumin levels following manual lymphatic drainage [207]. These changes were not solely due to fluid shifts, as albumin concentrations were corrected for changes in plasma volume, and hematocrit remained unaffected by the lymphatic drainage. These observations could imply that the mobilized fluid entering the bloodstream after manual lymphatic drainage therapy possesses a higher albumin content than plasma. The long-term implications of these physiological changes are yet to be fully understood. Nonetheless, the potential role of manual lymphatic drainage in influencing fluid balance and lymphatic outflow could have relevant implications for managing conditions in the ICU.

Experimental approach for endothelial stabilization

Phosphodiesterase (PDE) inhibitors exhibit a diverse range of pharmacological effects, encompassing properties such as anti-inflammatory, antioxidant, vasodilatory, cardiotonic, and anticancer activities, alongside enhancing memory. This expansive superfamily of PDEs is categorized into 11 distinct groups, differentiated by their structural characteristics, cellular localization, gene expression patterns, protein attributes, and a variety of pharmacological properties, influenced by both internal and external regulatory factors. Particularly, phosphodiesterase-4 inhibitors (PDE4-Is, e.g., rolipram and roflumilast) have been explored as potential treatment options stabilizing endothelial interaction during systemic inflammation and sepsis [208, 209]. The proposed mechanism is thought to involve the control of the cAMP/Rac1-signaling pathway, which is integral to the stability of intercellular junctions [208, 210,211,212]. The intracellular second messenger cyclic adenosine monophosphate (cAMP) decreases in endothelial cells under inflammatory conditions, associated with the breakdown of endothelial barrier properties in vitro [210]. Experimental studies further suggest that administration of PDE4-Is which increases endothelium-specific cAMP holds the potential to maintain cellular adhesion and endothelial barrier properties during acute inflammation. Schick et al. showed in a rodent model that the application of rolipram or roflumilast effectively attenuated capillary leakage and improved microcirculatory flow by preventing the inflammation-induced loss of endothelial cAMP [208]. Wollborn et al. further confirmed the effects of PDE4-Is in extracorporeal circulation-induced capillary leak [209]. Various other pathways remain under investigation to evaluate means to stabilize vascular endothelium [39].

In addition to PDE4-Is, other PDE-Is also show potential in endothelial stabilization. The PDE1 family, known for its vasodilatory effects and reduced activity in platelet aggregation, may influence endothelial stability by modifying vascular tone and cellular cAMP levels, crucial factors in maintaining endothelial barrier integrity [213,214,215]. Experimental studies suggest that PDE1 inhibitors, by modulating cGMP and cAMP pathways, could potentially reinforce endothelial cell adhesion and barrier properties, similar to the effects observed with PDE4-Is [216,217,218]. PDE2-Is, through their unique mechanism of cGMP-mediated cAMP regulation, may also contribute to endothelial stability. By enhancing intercellular communication and barrier function, they could offer a novel approach to managing endothelial disruption in conditions such as pulmonary hypertension and heart failure [219,220,221]. Furthermore, PDE3-Is, while primarily recognized for their cardiac effects, could indirectly influence endothelial function. Given their role in modulating intracellular cAMP levels, they might impact endothelial cell junction stability and barrier properties, particularly under stress conditions such as sepsis or systemic inflammation [222,223,224]. Among the most promising for endothelial stabilization are PDE5-Is like sildenafil and tadalafil. These agents have shown effectiveness in improving hemodynamics and endothelial function in heart failure and pulmonary arterial hypertension [225,226,227]. Their mechanism, which involves modulating cGMP-dependent signaling, makes them particularly relevant for maintaining endothelial barrier integrity. While primarily associated with visual functions, the role of PDE6 in other cellular processes remains under-investigated in the context of endothelial stabilization [228], PDE7-Is are present in immune cells and cardiac myocytes and might influence endothelial function indirectly through immunomodulatory pathways [229, 230]. Both PDE8 and PDE9 are involved in cAMP and cGMP signaling, respectively. While their direct role in endothelial stabilization is not as prominent, they may offer insights into cardiovascular functions and pathologies [231,232,233]. PDE10 and PDE11 are primarily explored for neurological and psychiatric disorders, and tumor development. Their role in endothelial stabilization is less defined [234, 235].

Recent insights have highlighted the pivotal role of vasodilators, particularly prostaglandins, in regulating endothelial capillary permeability. Prostaglandins, notably prostaglandin E2, play a significant role in this regard. Activation of the prostaglandin E2 receptor signal, which induces vasodilation, could be targeted to enhance endothelial barrier function and counteract capillary leak syndrome [236]. Experimental strategies might involve modulating these pathways to optimize vascular tone and permeability. Endothelium-derived vasodilators, including NO, prostacyclin, and endothelium-derived hyperpolarizing factors, play a central role in maintaining vascular tone. NO, synthesized by endothelial nitric oxide synthase, is instrumental in regulating vascular tone and endothelial function [237,238,239]. For example, strategies that enhance endothelial nitric oxide synthase activity or NO bioavailability could effectively stabilize endothelial function. This might include gene therapy to upregulate endothelial nitric oxide synthase expression, pharmacological agents to increase NO production, or novel compounds to mimic NO’s vasodilatory effects [236]. Additionally, addressing endothelial hyperpolarization through endothelium-derived hyperpolarizing factors could offer a novel experimental avenue. This might involve manipulating calcium-activated potassium channels or exploring the roles of gap junctions and epoxyeicosatrienoic acids in endothelial cell signaling [240, 241]. Prostacyclin, generated by cyclooxygenase in endothelial cells, activates adenylate cyclase, leading to vascular smooth muscle relaxation [242]. Its role in vasorelaxation suggests potential therapeutic applications in managing endothelial dysfunction. Modulating prostacyclin levels or mimicking its action through pharmacological agents could be an experimental approach to stabilize endothelial cells and maintain vascular homeostasis [236].

Recently, therapeutic plasma exchange has been used in clinical trials to modulate the injurious endothelial activation. The rationale behind this combines two aspects in one procedure: the removal of injurious circulating factors (e.g., Ang-2, heparanase-1) and the replacement of protective factors that have been consumed by the disease process (e.g., heparanase-2 or Ang-1) [243]. This concept has been demonstrated both by quantifying these factors before and after and by ex vivo stimulation of endothelial cells with plasma from these patients [95, 244, 245].

Conclusion

This review elucidates the multifaceted nature of CLS, underscoring the importance of recognizing its diverse triggers, including systemic inflammation and endothelial barrier breakdown. While current diagnostic methods, such as bioelectrical impedance analysis and serum markers, provide insights, their limitations highlight the need for more precise and universally accepted diagnostic criteria. Treatment strategies, primarily focusing on fluid management and endothelial stabilization, have shown potential, yet they lack specificity and efficacy for CLS. Innovative approaches, like the exploitation of the angiopoietin–Tie2 signaling axis, preservation of the endothelial surface layer, and experimental therapies like phosphodiesterase inhibitors, offer promising directions. Future research should aim to develop a consensus on CLS definition, establish reliable diagnostic benchmarks, and explore these novel therapeutic strategies to enhance patient outcomes in critical care settings.

Take-home message

CLS presents a diagnostic and therapeutic challenge in critical care due to its complex pathophysiology and the absence of standardized diagnostic criteria. According to the authors of this review, prioritizing research to refine diagnostic tools and explore novel treatments, including endothelial stabilization strategies and experimental pharmacological interventions, is crucial for improving patient management and outcomes in CLS.