The rationale for fluid therapy

The administration of fluids as a therapeutic approach dates back at least to the 19th century when Thomas Latta infused a fluid into a woman suffering from cholera and associated hypovolemic shock who ‘had apparently reached the last moments of her earthly existence’ . Although his efforts showed only transient beneficial effects and his patient did not survive, fluid therapy became an essential part of supportive therapy for patients with critical illness. Presently, administration of fluids is recommended in various international guidelines for life-threatening haemodynamic conditions such as sepsis and haemorrhagic shock . These conditions are assumed to be mainly caused by oxygen deficiency leading to cell dysfunction . Therefore, the ultimate goal of fluid therapy and other haemodynamic interventions is to restore and maintain oxygen delivery to the individual cell to ensure organ function.

Fluid administration aims to improve the stroke volume of the heart and thereby increase the cardiac output via the Frank–Starling mechanism (i.e. by optimizing pre-load) . Notably, research from the last decades repeatedly demonstrates that despite haemodynamic optimization, organ function and microcirculation may not improve . Ince defines haemodynamic coherence as the situation in which an improvement in macrohaemodynamic variables (e.g. cardiac output) correlates with the beneficial effects of microcirculation and vice versa . If macrohaemodynamic optimization and microcirculation do not respond in the same direction, haemodynamic coherence is lost. Thus, improvement of macrohaemodynamics is a condition to enable haemodynamic coherence. In this context, Pottecher et al. showed that in patients with sepsis and septic shock in need of volume expansion, passive leg raising test and pulse pressure variation predicted an improvement in microcirculation .

The present review summarizes the current data on fluid therapy and its impact on haemodynamic coherence. Blood products are not addressed because they are discussed in another article in this issue.

Basic considerations in fluid therapy and microcirculation

Because physiology suggests that fluid therapy increases microcirculatory blood flow only if global blood flow is increased (i.e. cardiac output), macrocirculatory fluid responsiveness is essential to improve microcirculation with fluids. There are four distinct scenarios in critically ill patients. In the first scenario, macro- and microcirculation are stable, and no change in haemodynamic management is indicated. In the second scenario, which is probably common, macrohaemodynamics are unstable (e.g. a certain drop in blood pressure), but microcirculatory analysis (e.g. sublingual) suggests maintained microvascular blood flow. In this setting, it can be hypothesized that cardiac output is sufficient and blood pressure is at least appropriate for the index tissue in which the microcirculation is monitored. If blood pressure is thought to be too low in other areas (e.g. the carotid area), then a vasopressor may be the treatment of choice. In the third scenario, macro- and microcirculation are both impaired (e.g. in untreated haemorrhagic shock), and haemodynamic interventions (fluid therapy, catecholamines, transfusion, etc.) will determine whether haemodynamic coherence persists after macrohaemodynamic stabilization. In the (probably most interesting) fourth scenario, macrohaemodynamics have been stabilized, but microcirculation remains impaired. As a first approach, macrohaemodynamic fluid responsiveness may be assessed. If the patient is fluid responsive, the associated increase in cardiac output may or may not improve the microcirculation. In the authors experience, the administration of fluids rarely improves microcirculation in this setting. In contrast, other interventions such as red blood cell transfusion, vasodilators and antioxidants may prove effective in improving microcirculation . However, all these interventions are experimental; they will be effective only on an individual patient basis and necessitate a close and concomitant monitoring of the microcirculation.

Basic considerations in fluid therapy and microcirculation

Because physiology suggests that fluid therapy increases microcirculatory blood flow only if global blood flow is increased (i.e. cardiac output), macrocirculatory fluid responsiveness is essential to improve microcirculation with fluids. There are four distinct scenarios in critically ill patients. In the first scenario, macro- and microcirculation are stable, and no change in haemodynamic management is indicated. In the second scenario, which is probably common, macrohaemodynamics are unstable (e.g. a certain drop in blood pressure), but microcirculatory analysis (e.g. sublingual) suggests maintained microvascular blood flow. In this setting, it can be hypothesized that cardiac output is sufficient and blood pressure is at least appropriate for the index tissue in which the microcirculation is monitored. If blood pressure is thought to be too low in other areas (e.g. the carotid area), then a vasopressor may be the treatment of choice. In the third scenario, macro- and microcirculation are both impaired (e.g. in untreated haemorrhagic shock), and haemodynamic interventions (fluid therapy, catecholamines, transfusion, etc.) will determine whether haemodynamic coherence persists after macrohaemodynamic stabilization. In the (probably most interesting) fourth scenario, macrohaemodynamics have been stabilized, but microcirculation remains impaired. As a first approach, macrohaemodynamic fluid responsiveness may be assessed. If the patient is fluid responsive, the associated increase in cardiac output may or may not improve the microcirculation. In the authors experience, the administration of fluids rarely improves microcirculation in this setting. In contrast, other interventions such as red blood cell transfusion, vasodilators and antioxidants may prove effective in improving microcirculation . However, all these interventions are experimental; they will be effective only on an individual patient basis and necessitate a close and concomitant monitoring of the microcirculation.

Fluid therapy and haemodynamic coherence in patients with haemorrhagic shock

Patients with haemorrhagic shock are usually treated by optimizing the intra-vascular volume through fluid or blood administration after the bleeding is controlled . Haemorrhagic shock reduces microvascular flow and capillary density . Usually, microvascular analysis shows a homogeneous reduction of convection and diffusion due to low cardiac output and anaemia. From a physiological point of view, fluid therapy should restore microvascular perfusion by increasing cardiac output, provided that a certain haematocrit is maintained. However, results of experimental and clinical studies in this context vary concerning haemodynamic coherence associated with fluid therapy.

Wu et al. showed in a rat model of haemorrhagic shock that despite restoring mean arterial pressure by fluid therapy with 0.9% sodium chloride solution, there was no haemodynamic coherence after resuscitation. Intestinal blood flow remained compromised . In a hamster window chamber model of haemorrhagic shock, Villela et al. found that functional capillary density improved after resuscitation with Ringer’s lactate solution . In a clinically relevant paediatric large animal model in pigs, González et al. investigated the impact of 0.9% saline solution and different preparations of 5% albumin solutions on sublingual microcirculation. They found no differences between the fluids, but an improved microvascular flow and perfused vessel density was observed after resuscitation . In addition, in an porcine model of haemorrhagic shock, Guerci et al. found that sublingual microvascular flow and tissue oxygen tension improved after fluid resuscitation . In a prospective observational study, in patients suffering from haemorrhagic shock, Tachon et al. showed that haemodynamic coherence between macrocirculation and sublingual microcirculation was initially impaired after controlling bleeding and restoring macrocirculation . Microcirculation remained altered for the next 72 h and then normalized.

From the results of the above-mentioned studies, it appears that although in haemorrhagic shock the vasculature and endothelium are not a primary part of the pathophysiology, microcirculation is not essentially restored with fluid resuscitation. Reasons for these varying findings may be the use of different fluids or amounts of fluids in the treatment of patients with haemorrhagic shock (see also section on haemodynamic coherence and type of fluid). In addition, the frequently accompanying inflammatory response following haemorrhagic shock alters microcirculation. These alterations are based on neutrophil interactions with the endothelium , leading to reactive oxygen species production, vascular leakage and ultimately organ failure . Early resuscitation may lead to haemodynamic coherence but (too) late initiation of resuscitation does not because of inflammatory alterations of microcirculation. Unfortunately, no conclusive evidence is available to answer this question yet. In addition, masked hypovolaemia, e.g. due to endogenous or exogenous vasopressors, may lead to good macrohaemodynamic data but leaves the microcirculation impaired . Venous congestion may also lead to compromised haemodynamic coherence by diminishing the driving pressure of capillaries. This is supported by the findings of Vellinga et al. that high central venous pressure may impair microvascular blood flow . The authors hypothesized that high central venous pressure may serve as an obstructive outflow resistance. More research is needed on the timing of resuscitation in patients with haemorrhagic shock and the link between inflammation and haemodynamic coherence in patients with haemorrhagic shock.

Fluid therapy and haemodynamic coherence in patients with sepsis and septic shock

Microcirculation in patients with sepsis and septic shock is known to be impaired because of inflammatory processes that lead to shunting and vascular leakage. In fact, sepsis has often been labelled as a ‘disease of the microcirculation’ . One of the main problems in sepsis is the extreme heterogeneity in microvascular perfusion of different organs or even within the same organ . Fluid therapy is an important supportive therapeutic approach recommended by the guidelines for sepsis treatment . The common goal of all haemodynamic interventions (not only) in sepsis is to improve microcirculation. Fluid therapy improves microcirculation by the optimization of stroke volume and cardiac output. Clinical experience and evidence show that the heart is often volume responsive in sepsis, i.e. stroke volume increases after fluid challenge. Haemodynamic coherence is required to translate these changes in global blood flow to changes in microvascular blood flow. Notably, a large body of evidence suggests that haemodynamic coherence is absent, dependent, e.g. on the time point in the course of sepsis, or differs in various organs.

van Genderen et al. found that haemodynamic coherence, as measured by microvascular flow and capillary density, was absent in an experimental pig model of endotoxaemic shock in the early phase of shock, whereas in the late hyperdynamic phase, microcirculation seemed to improve along with the macrohaemodynamics . In an experimental ovine model of endotoxaemia, Dubin et al. investigated the impact of fluid therapy on macro- and microcirculation. The results of their study demonstrated that despite improvement in cardiac output and global organ perfusion, such as mesenteric blood flow, microcirculation improved only heterogeneously . Sublingual and serosal intestinal microvascular flow improved, whereas mucosal intestinal perfusion remained impaired, leading to intra-mucosal acidosis.

Heterogeneity of haemodynamic coherence was also demonstrated by Edul et al. in patients with abdominal sepsis. The investigators found that sublingual microcirculation correlated with cardiac output and improved after a fluid challenge . However, intestinal microcirculation did not react in the same manner and stayed deteriorated. Boerma et al. showed that in patients with abdominal sepsis, heterogeneity in haemodynamic coherence is also time dependent . One day after treatment initialization, macrocirculation was improved, but there was no correlation with sublingual or intestinal microcirculation; moreover, both the microcirculation sites did not correlate with each other. Two days later, microcirculatory correlation between sublingual and intestinal region was restored, and sublingual microcirculation correlated with the cardiac index. Time dependency of haemodynamic coherence in sepsis is supported by the results of Ospina-Tascon et al. who studied patients with severe sepsis. In their study, contrary to the findings of Boerma et al., fluid resuscitation improved microcirculation within the first 24 h after the diagnosis of sepsis but not after 48 h, independent of macrohaemodynamic variables and their changes . At first glance, the different findings may be explained by the fact that Boerma et al. investigated the correlation between macro- and microcirculation at different time points regardless of fluid therapy, whereas Ospina-Tascon et al. investigated the effects of a fluid challenge on microcirculation. These data support the pragmatic notion that early fluid therapy is likely to be beneficial, whereas late fluid therapy is rather adverse .

There are also studies demonstrating haemodynamic coherence in patients with sepsis. For example, Dubin et al. demonstrated that protocolized fluid resuscitation improved microcirculation by increasing the cardiac output in patients with severe sepsis .

Loss of haemodynamic coherence in sepsis is associated with adverse outcomes. This has been demonstrated by Edul et al. who showed that patients with septic shock who had more severe alterations in microcirculation had an inferior rate of survival . Sakr et al. also showed that microcirculation in non-survivors in septic shock did not improve as good as it did in survivors . Findings by Trzeciak et al. support these results by demonstrating that in patients with loss of haemodynamic coherence, organ failure occurred more often .

These heterogeneous findings on haemodynamic coherence in patients with sepsis and septic shock may be explained by various mechanisms. Activation of coagulation plays a role in this context, leading to microvascular thrombosis and altering microcirculation . Oedema formation and subsequent organ failure due to impairment of the glycocalyx are amongst other reasons that contribute to microvascular disorder . In addition, the induction and local alteration of nitric oxide species and nitric oxide synthase in inflammation in sepsis contributes to microcirculatory alterations . Heterogeneity in microvascular flow and capillary density may also contribute to the ambivalent results. As microcirculation is not uniformly impaired in every compartment of an organ, resuscitation may also lead to hyperdynamic flow in regions with normal microcirculation, thus diminishing cellular metabolism and oxygen supply. Haemodynamic coherence in patients with sepsis may also be influenced by a number of other factors such as the time-point in the individual course of disease, the organ in which microcirculation is measured and the variables of microcirculation that reflect microcirculation best. It is still unknown which variables have to be measured to arrive at conclusive therapeutic decisions.

Type of fluid and haemodynamic coherence

Various experimental and clinical data suggest that the type of fluid may also influence the incidence of haemodynamic coherence. Different types of fluid with theoretical and practical advantages and disadvantages are available for use in fluid resuscitation. Fluids can be classified as crystalloid and colloid solutions and as balanced and unbalanced solutions. A more comprehensive review of available fluids has been thoroughly described before .

In a rodent model of haemorrhagic shock, Wu et al. found that fluid resuscitation with 0.9% saline was not as effective in restoring intestinal microcirculation compared to hypertonic saline solution, 4% succinylated gelatin and 6% hydroxyethyl starch (HES) 130/0.4. However, more reactive oxygen species were found in the kidney after resuscitation with either colloid. All fluids improved microcirculation . Supporting these findings, the same group demonstrated in another experimental study that resuscitation of haemorrhagic shock with 0.9% saline did not improve intestinal microcirculation . In a rodent model of haemorrhagic shock and resuscitation, Aksu et al. compared the impact of balanced versus 0.9% saline-based fluids. Although macrocirculation improved in both groups, the balanced fluid improved renal oxygen utilization compared to 0.9% saline . Pascual et al. showed in mice suffering from haemorrhagic shock that resuscitation with hypertonic saline solution reduces vascular leakage compared to resuscitation with balanced crystalloid solution (Ringer’s lactate), thereby improving microcirculation . Villela et al. found that resuscitation of haemorrhagic shock in hamsters using Ringer’s lactate solution with increased viscosity increased functional capillary density compared to normal Ringer’s solution .

In an experimental ovine model of abdominal sepsis, Orbegozo et al. found that using 0.9% saline for resuscitation did not improve microcirculation to the same extent as when using two balanced crystalloid solutions . In a rodent model of endotoxaemia, Schäper et al. demonstrated that fluid therapy with colloids, namely 6% HES 130/0.4 and 4% succinylated gelatin, preserved intestinal microcirculation compared to fluid therapy with 0.9% saline . In patients with severe sepsis, Dubin et al. showed that fluid resuscitation with 6% HES 130/0.4 was superior to that with 0.9% saline solution in improving sublingual microcirculation .

The influence of the type of fluid on haemodynamic coherence may be explained by their biochemical properties and interactions with the endothelium. From the results of the above-mentioned studies, 0.9% saline seems to be the least effective in recruiting the microcirculation in coherence with the macrocirculation. Although 0.9% saline is the most commonly used resuscitation fluid worldwide , it has some less favourable effects. The high chloride content in 0.9% saline is known to induce hyperchloridaemic acidosis, thus triggering the formation of active nitrogen species that cause vasodilatation and induce vascular leakage . In the kidney, high chloride concentrations reduce the rate of glomerular filtration by activating the tubuloglomerular feedback mechanism that alters the afferent glomerular and thereby microvascular blood flow of the kidney . These mechanisms may also explain the association of hyperchloridaemic acidosis with adverse outcomes in critically ill patients , possibly due to the loss of haemodynamic coherence. Other effects of fluids that may explain their different behaviour in haemodynamic coherence include the differing intra-vascular volume effect, particularly when comparing crystalloid and colloid solutions. Infusion of colloid solutions leads to decreased oedema formation because of greater intra-vascular volume effect compared to infusion of crystalloid solutions. Interstitial oedema increases the diffusion distance at a microcirculatory level, thus impairing oxygen delivery to the cells . Several other characteristics influence microcirculatory variables, particularly when HES-based solutions are used. HES solutions increase blood viscosity and reduce platelet aggregation and clot firmness . Influence of blood viscosity on microcirculation was shown by Villela et al. as explained above . In addition, decreased neutrophil recruitment and transmigration after infusion of HES-based solutions have been reported and may lead to decreased inflammation , thereby potentially improving the microcirculation.