Correspondence to: Professor T. W. Evans, Royal Brompton Hospital, Sydney Street, London SW3 6NP, UK. E-mail: email@example.com
Albumin is the main determinant of plasma oncotic pressure and it plays a pivotal role in modulating the distribution of fluids between compartments. Moreover, it has many other biological properties which may be important not only for its physiological actions but also for its therapeutic effects. Among the non-oncotic properties are its capacity of molecule transportation and free radical scavenging, its ability to modulate capillar permeability, neutrophil adhesion and activation and its haemostatic effects. The following article reviews these biological effects as well as its structure, synthesis, catabolism and distribution.
Human albumin constitutes some 50% of the protein present in the plasma of normal healthy individuals. It is a 66 kDa protein, which is small relative to other plasma proteins. Albumin is highly soluble and of elliptical shape with a low intrinsic viscosity. Albumin is a very stable protein, although more than 50 slight variants of the 585-amino acid sequence that comprises human albumin have been described. These are termed allo-albumins, and can co-exist in a manner akin to haemoglobins in sickle cell trait.1 Albumin solutions (4.5% w/v) have been used clinically as volume replacement, and as plasma expanders in a more concentrated form (20% w/v) for many years. The high concentration of albumin in plasma, and its strong net negative charge, means that it is responsible for around 70% of plasma oncotic pressure and therefore plays a pivotal role in modulating the distribution of fluid between compartments. However, albumin has many other biologically important properties which may be relevant to its actions under physiological circumstances and in disease.
Albumin is synthesized in polysomes bound to the endoplasmic reticulum of hepatocytes at a rate of between 9 and 12 g per day in healthy adults. Albumin is not stored hepatically and there is therefore no reserve for release on demand.2 However, under physiological circumstances only 20–30% of hepatocytes produce albumin and synthesis can therefore be increased on demand by a factor of 200–300%. Changes in the rate of production are governed primarily by alterations in colloid osmotic pressure, and the osmolality of the extravascular hepatic space. However, hormonal changes, for example raised concentrations of insulin, thyroxine and cortisol, can influence albumin synthesis.3 Surprisingly, growth hormone has no such effect. Albumin production can be rate limited by amino acid deficiencies, particularly of leucine, arginine and isoleucine and valine, but these are rarely seen clinically, except in states of extreme malnutrition. Whether or not synthesis can be enhanced by amino acid supplementation in the absence of deficiency is unclear.
Catabolism probably occurs in, or immediately adjacent to, the vascular endothelium of tissues, again at a rate of 9–12 g per day. Albumin is pinocytosed into cells at a rate which is related to atrial natriuretic peptide (ANP) concentrations, but is not excessively catabolized in starvation and deficiency states, possibly because it represents a poor source of essential amino acids, being very low in tyrosine residues.
Albumin is predominantly an extravascular protein and its serum concentration is around 40 g/L, suggesting a total intravascular mass of about 120 g. The interstitial concentration is lower (14 g/L) and varies in different anatomical regions. However, the total extravascular mass is around 160 g. Some of this albumin can easily be mobilized from loose interstitial tissues, whilst some is tightly bound (particularly in the skin). There appears to be a circulation of albumin from the intravascular to extravascular space, returning via the lymphatic vessels. Such movement has been measured and represents a circulation half-life of around 16–18 h.
Some 4–5% of intravascular albumin leaves the intravascular compartment per hour in healthy individuals, a figure termed the transcapillary escape rate, TER. The transcapillary escape rate is determined by the capillary and interstitial free albumin concentrations, microvascular permeability to albumin, the movement of solvents and solutes, and transcapillary electrical charge. Clearance of albumin and other proteins from the interstitium is dependent upon lymphatic flow which is itself determined by interstitial fluid pressure, intrinsic pumping by the lymphatic vessels, and the effects of muscular contraction or external compression (e.g. arterial pulsation).
Non-oncotic properties of albumin
Albumin has a strong negative charge, but binds weakly and reversibly to both cations and anions. It therefore functions as a circulating depot and transports molecules for a large number of metabolites including fatty acids, ions, thyroxine, bilirubin and amino acids (Table 1). Albumin also binds covalently and irreversibly with d-glucose and d-galactose. The glycosylation of albumin, which is to a certain extent age-dependent, has effects upon its charge and therefore may influence capillary permeability characteristics. Glycosylated albumin is thought to have a major role in the pathogenesis of atherosclerotic disease in diabetics.
Table 1. Albumin is a transport vehicle for a variety of substances and binds a number of drugs
Albumin is a transport vehicle for:
○ Bile pigments
○ Nitric oxide
○ Fatty acids
Albumin interacts with:
Free radical scavenging
Albumin is the major extracellular source of the reduced sulphydryl groups which are present on a single exposed cysteine residue at position 31 in the molecule. These sulphydryl groups, termed thiols, are avid scavengers of reactive oxygen and nitrogen species, especially the superoxide hydroxyl and peroxynitrite radicals. Albumin can also limit the production of these reactive species by binding free copper Cu2+, an ion known to be particularly important in accelerating the production of free radicals. In sepsis, the administration of human albumin (200 mL, 20% w/v) has been demonstrated to lead to significantly increased levels of plasma albumin, which remained significantly increased 4 h following administration.4 Total plasma thiol levels at the same time showed similar trends. Unlike the albumin measurements, however, where a significant fall between 5 min and 4 h following administration occurred, thiol remained significantly elevated for up to 18 h following albumin administration. These results have several possible implications. First, they strongly suggest that an increase in plasma protein thiols associated with albumin administration is sustained long-term compared with plasma albumin levels, which is indicative of an albumin-mediated thiol exchange in the plasma of these patients. Second, given that albumin accounts for most of the total plasma thiol content in normal healthy individuals (the remainder being associated particularly with gamma globulins, plasmin and fibrinogen), albumin may in this way influence redox balance (Figure 1), which has a number of important implications for other indices of critical illness, including capillary permeability, rheological changes, cell signalling processes, antihaemostatic effects and drug metabolism and transport.
The optimal fluid for resuscitation in patients with critical illness has yet to be designed, but in addition to providing optimal intravascular filling, it would ideally favourably modulate the inflammatory processes that characterize sepsis, the systemic inflammatory response syndrome (SIRS) and their vascular sequelae. Patients with these syndromes develop increased vascular permeability, leading to tissue oedema formation, dysfunctional vasomotor function with disordered delivery of cellular nutrients, and rheological changes characterized by increased neutrophil rolling, adherence and activation (Figure 2).
Under physiological circumstances there is a net movement of albumin from the intravascular to the interstitial space and back, via the lymphatic vessels (transcapillary escape rate, see above).5 Albumin may itself directly influence vascular integrity, by binding in the interstitial matrix and sub-endothelium and by altering permeability of these layers to large molecules and solutes.6,7 Indirect effects may be mediated firstly by the binding of arachidonic acid (AA), which itself increases capillary permeability.8 Second, polynitroxylated albumin (PNA) protects tissues against ischaemic reperfusion injury, possibly by enhancing tissue redox activity. Certainly, polynitroxylated albumin has been shown to be a potent inhibitor of xanthine–xanthine oxidase mediated adhesion of human neutrophils to cultured human endothelial cells.9 The beneficial effects of polynitroxylated albumin may therefore be attributed to the attenuation of leucocyte–endothelial cell interaction. The influence of other colloids on changes in microvascular integrity are less well characterized. However, increased vascular permeability induced by endotoxemia can be attenuated by hypertonic saline, with or without dextran.10 Hypertonic saline–dextran also improves intestinal perfusion and survival in porcine septic shock.11 In experimental models of ischaemic reperfusion, hydroethylstarch solutions reduce capillary permeability and tissue oedema formation compared to crystalloids, and decrease both pulmonary and splanchnic injury and xanthine oxidase release after hepatoenteric ischaemic reperfusion.12 Moreover, the increased permeability of the mesenteric capillary bed in rodent endotoxaemia is attenuated equally by resuscitation using either albumin or crystalloid,13 suggesting that changes in endothelial integrity may be favourably influenced by volume repletion, independent of changes in oncotic pressure.
Rheological changes, neutrophil adhesion and activation
Hydroethylstarch solutions decrease endothelial cell activation in vitro compared to albumin.14 Such effects may be due to a free radical scavenging capacity or to a beneficial effect on cytokine release.15 Moreover, a moderate increase in the expression of complement receptors on the surface of polymorphonuclear leucocytes has been described following the incubation of whole blood with colloids. Neutrophil oxidative burst activity is also markedly increased following incubation with artificial colloids and crystalloids, although little such activation was seen using albumin.16 Indeed, human serum albumin has been shown to suppress the respiratory burst of neutrophils in response to exposure to cytokines relevant to the pathogenesis of critical illness (tumour necrosis factor—TNF) and complement components (e.g. C5A). Moreover, human serum albumin selectively and reversibly inhibits tumour necrosis factor-induced neutrophil spreading and the associated fall in cAMP.17 By contrast, albumin, gelatin or hydroethylstarch in moderate amounts show no short-term effects on adhesion or granulocyte activation in patients undergoing anaesthesia for orthopaedic surgery.18
Cell signalling processes
Albumin in the reduced state contains a single exposed thiol group,19 which is the principal extracellular antioxidant and chiefly responsible for maintaining the redox state of plasma. Redox regulation at a transcriptional level of the ubiquitous cell signalling moiety nuclear factor kappa B (NF-κB) has been described. Moreover, free thiols have been shown to be important factors in determining the DNA binding activity of active transcription factors including NF-κB,20 thereby potentially influencing processes determining cellular fate or apoptosis (Figure 3). Albumin administered to critically ill patients increases plasma thiol levels, even after it is cleared from the circulation, presumably by virtue of exchange mechanisms with as-yet unidentified plasma constituents.21 This may initiate cascades of thiol oxidative–reductive reactions that ultimately influence cellular signalling processes.
Albumin has an antithrombotic, anticoagulant effect, possibly because of its capacity to bind nitric oxide (NO) to form S-nitrosothiols,22 thereby inhibiting the rapid inactivation of NO and allowing prolongation of its anti-aggregatory effects on platelets. Thus, priming of the cardiopulmonary bypass circuits with albumin solutions may reduce platelet deposition to 4–5% of that observed for an equivalent pre-treatment with normal saline. Other studies have shown that this practice has no clinically detectable advantage in terms of haemostasis, chest tube drainage or requirement for blood transfusion.22,23 In the USA, bleeding complications have been reported following the administration of hydroethylstarch 480/0.7, although when given in doses below 1.5 L these are no greater than expected with other colloid solutions.24 However, hydroethylstarch with a high initial molecular weight or with a high in vivo molecular weight seems to have more unfavourable effects on coagulation than medium or lower molecular weight hydroethylstarch that is easier to degrade,25 although the duration of effect is less, through its faster elimination. Finally, using thromboelastography, in vitro studies have suggested that whereas gelatin solutions were less intrinsically anticoagulant than hydroethylstarch, 10% dextran 40 had the strongest effect.26
Pharmacological interactions, drug binding
There are four discrete binding sites on the albumin molecule, which each have varying specificity for different substances.27 Ligands can compete at a single site, or may compete by altering the affinity of remote sites by conformational changes to the tertiary structure of the molecule. Thus, drugs binding at the same site compete for occupancy and are likely to displace one another (e.g. warfarin, phenytoin), whilst others drugs known to be highly albumin-bound in plasma but binding at separate sites may not displace each other (e.g. warfarin, diazepam). Drugs with which albumin interacts in a highly clinically significant fashion owing to their highly protein-bound state and low margins of safety include warfarin, phenytoin, non-steroidal anti-inflammatory drugs and digoxin. Midazolam, thiopental and a number of antibiotics also interact with albumin in this fashion. The volume of distribution of drugs bound to albumin may increase in hypoalbuminaemic states, thereby reducing their efficacy. The administration of mixtures of loop diuretics (e.g. frusemide (furosemide)) with albumin has therefore been advocated, although this has been shown to be ineffective in cirrhotic patients with ascites.28