Clinical Science

Research article

Volume expansion and plasma protein clearance during intravenous infusion of 5% albumin and autologous plasma

Annika HEDIN, Robert G. HAHN


Autologous plasma may be used to replace plasma volume and plasma proteins during surgery, but its effectiveness is largely unknown. In the present study, the characteristics of predonated frozen and thawed autologous plasma were compared with those of 5% albumin in 15 male volunteers who received 10 ml/kg of body weight of these colloids as intravenous infusions over 30 min. Venous blood was sampled and urine was collected over 8 h to outline the volume expansion and blood–interstitial fluid space transport of three plasma proteins (albumin, fibrinogen and antithrombin) by means of mass balance analysis. The maximum plasma dilution of 5% albumin and autologous plasma averaged 17 and 21% respectively, and their half-lives were 2.5 and 2.9 h respectively (P<0.03). The between-subject variability in dilution was most pronounced for autologous plasma. Transport of protein from blood to the interstitial space occurred faster when the infused fluid contained the protein in question. The rate was highest at 60 min, and the process was still in progress at 8 h when approx. 60% of the infused albumin, 45% of the fibrinogen and 75% of the infused antithrombin had been translocated to the interstitial fluid space. In contrast with the proteins, excess plasma water was removed by urinary excretion. It is concluded that the volume expansion is equivalent for the two colloid fluids, although it is more predictable for 5% albumin. The transport of protein outlasted the volume expansion.

  • albumin
  • antithrombin
  • fibrinogen
  • pharmacokinetics
  • transcapillary leak


Albumin (5%) and human plasma are colloid infusion fluids given to replace plasma volume and to raise its oncotic pressure. Their physiological effects are complex and have not been fully investigated. The plasma concentrations of infused proteins increase, whereas the dilution resulting from the plasma volume expansion reduces the concentrations of other proteins. Moreover, the transport of albumin from blood to the interstitial fluid increases [1]. The sum of these alterations may be relevant to postoperative morbidity and to critical illness, which includes organ dysfunction associated with a high transcapillary escape rate for proteins [2,3] and coagulation disorders [4].

In the present study, we compared the duration and degree of plasma volume expansion and also the transcapillary transport of three plasma proteins (albumin, fibrinogen and antithrombin) during and after intravenous infusion of 5% albumin and autologous plasma in volunteers. The aim was to examine the efficacy of autologous plasma as a volume expander and the combined effects of plasma protein clearance and plasma dilution.


After Ethics Committee approval and providing informed consent, 15 healthy male volunteers between the ages 18 and 36 years (mean, 32 years) and weighing 70–94 kg (mean, 80 kg) agreed to participate in two intravenous infusion experiments separated by at least 2 weeks. They received, in random order, 10 ml/kg of body weight of 5% human albumin solution (Baxter Healthcare, Deerfield, IL, U.S.A.) and the same amount of autologous plasma.


Autologous plasma

The volunteers donated plasma by aphaeresis on two occasions in accordance with the routines of the Department of Transfusion Medicine at our hospital. The interval between all donations and any experiments always exceeded 1 week to allow restitution of intravascular volume and plasma protein content. The frozen plasma was thawed to reach room temperature on the morning of the plasma infusion.

Infusion protocol

All experiments started at 07:00 hours. Subjects were allowed a standardized breakfast, not including tea or coffee, but otherwise fasted throughout the experiment. They rested in the supine position for 30 min before baseline measurements were done. A cannula was placed in the antecubital vein of each arm, one for infusion and the other for blood sampling. The albumin solution and plasma were infused at a constant rate over 30 min by means of an infusion pump (Flo-Gard 6201; Baxter Healthcare). Venous blood (7 ml) was withdrawn every 5 min during the first hour, every 10 min during the second hour and, thereafter, every 15 min throughout the 8 h experiment. The first sample in the series was removed in duplicate and the mean value of the two was used in the calculations. To avoid sample dilution, 3 ml of blood was withdrawn from the venous catheter before sampling. This blood was then reinfused and the cannula was flushed with 5 ml of saline. An additional 10 ml of blood was withdrawn at baseline and at the end of the infusion (30 min) for analysis of ionized calcium.

Monitoring included in arterial blood pressure, heart rate and oxygenation by pulse oximetry after each blood sampling (Datex Light; Datex Oy, Helsinki, Finland).

Blood chemistry

The Hb (haemoglobin) concentration, RBC (red blood cell) count and MCV (mean corpuscular volume) were measured on a Technicon H2 (Bayer, Tarrytown, NY, U.S.A.) using colorimetry at 546 nm (Hb) and light dispersion using a helium neon laser (RBC and MCV). The serum albumin concentration was analysed using the Bromcresol Green method, followed by reflection spectrophotometry (Ektachem 250/950 IRC; Johnson & Johnson, Rochester, MN, U.S.A.). Ionized calcium was measured on an ABL 505 (Radiometer, Copenhagen, Denmark).

Coagulation status was tested by measuring the platelet count on the Technicon H2 and also by determining the plasma fibrinogen and antithrombin concentrations, aPTT (activated partial thromboplastin time) and prothrombin time [expressed as the INR (international normalized ratio)] using a Thrombolyser (Behnk, Nordenstedt, Germany). Antithrombin was reported as activity as a percentage of ‘normal’, which represents a plasma concentration of 150 mg/l.

Our own duplicate samples ensured a coefficient of variation of 0.9% for Hb, 0.7% for RBC, 1.3% for MCV, 2.2% for thrombocytes, 1.3% for albumin, 1.1% for INR and aPTT, 2.5% for fibrinogen and 1.9% for antithrombin.

Mass balance calculations

Calculations of plasma volume expansion and protein leak using mass balance, which are based on the repeated blood sampling of Hb and plasma proteins, provides a high time resolution without the use of radioactive tracers. To ensure accuracy, the protein and electrolyte content of all administered albumin and plasma units were measured and the results used in the subsequent calculations.

Plasma dilution

The BV (blood volume) at baseline (t=0) was estimated according to the height and weight of the subjects as described by Nadler et al. [5]. Embedded Image The MHb0 (total Hb mass) was first obtained as: Embedded Image from which sample losses up to a later time t were subtracted: Embedded Image The expanded blood volume at the later time t was then obtained as described previously [6]: Embedded Image This expression was converted from BV into PV (plasma volume) data: Embedded Image where Hct0 is the baseline haematocrit. The accuracy of PVt is improved by performing the same calculations for the RBC and entering the mean value for the Hb and RBC dilution instead of Hbt/Hb0, and Hct0 is finally multiplied by the change in red cell size (MCVt/MCV0). The Hb-derived plasma dilution was then obtained as: Embedded Image

Protein clearance

The disappearance of protein mass from the plasma volume (capillary leak) was obtained as the difference between the protein content of plasma at time 0 and time t: Embedded Image where Proteint is the plasma protein concentration at time t.


The t1/2 (half-life) of the infused fluid volume and protein mass was calculated from a simple washout equation based on the amount (Xd) that was still circulating in the plasma at 60 min and up to the end of the study: Embedded Image where k is the elimination rate constant and ln 2 is the natural logarithm of 2, which is 0.693. This equation is free from any influence of dilution. The time after 60 min was used to avoid any influence of distribution effects of fluid and albumin on the result.

Fluid and electrolyte balance

The volunteers voided urine just before the infusions started and then whenever necessary. The urinary volume was recorded and sampled for measurements of albumin, sodium and potassium excretion, and of the osmolality. All infusions were analysed for albumin, fibrinogen, antithrombin, sodium and potassium content.

The distribution of infused fluid between intra- and extra-cellular fluid spaces at 8 h was estimated by applying mass balance to the whole-body sodium and volume changes [7]. For this purpose, the serum sodium concentration was measured in the plasma samples, and the baseline extracellular volume and total body water were estimated by bioimpedance using a Xitron 4000B (Xitron Technologies, San Diego, CA, U.S.A.) [8,9]. Each reported value is the mean of three successive recordings, each based on 50 frequencies and calculated by the software delivered together with the apparatus.


Data are presented as the means (S.D.), and statistical comparisons were made using repeated-measures ANOVA. When there was a skewed distribution, the data are given as the median (interquartile range), and the Wilcoxon matched-pair test was used for statistics. P<0.05 was considered statistically significant.


Volume expansion

The plasma volume increased by 17 and 21% during the infusions of albumin and plasma respectively, which corresponded to 550 and 600 ml when multiplied by the estimated plasma volume of 3.23 litres at baseline (Table 1). The decay was monoexponential and occurred with a t1/2 of 2.5 (1.9–3.1) and 2.9 (2.5–5.4) h respectively (P<0.03). The between-subject variation for the dilution was smaller for 5% albumin than for the plasma infusions (Figure 1).

View this table:
Table 1 Baseline laboratory data for the 15 studied volunteers

Data are means (S.D.).

Figure 1 Hb-derived plasma dilution during and after infusion of albumin (left) and autologous plasma (right)

Individual dilution-time curves (thin lines) are superimposed on the computer-generated best fit (thick line).

Protein concentrations

The albumin concentration increased by 10% during the albumin experiments and by 5% when plasma was administered (P<0.001 for the difference; Figure 2a).

Figure 2 Serum concentrations of albumin, antithrombin and fibrinogen (upper panels) and the blood platelet count, plasma thromboplastin time and prothrombin time (lower panels) during and after a 30 min intravenous infusion of 10 ml/kg of body weight of 5% human albumin and autologous plasma

Values are expressed as a percentage relative to baseline.

Albumin infusion decreased the fibrinogen and antithrombin concentrations by 12 and 16% respectively, and they were restored slowly (Figures 2b and 2c).

Infusion of plasma raised the concentration of fibrinogen (6%) significantly more (P<0.04) than the concentration of antithrombin (3%). At 8 h, the fibrinogen concentration was still increased by 10%, whereas antithrombin remained at 3%.

For both colloids, the platelet count closely followed the Hb dilution (Figure 2d), whereas the aPTT was prolonged (Figure 2e) and the prothrombin time increased (Figure 2f) only when 5% albumin was infused. Ionized calcium decreased most after autologous plasma (−15% compared with −4%; P<0.0001).

Protein clearance

Albumin disappeared from the plasma during both infusions, and this process continued throughout the follow-up period (Figure 3a). The net flux of fibrinogen and antithrombin across the capillary membrane was highly dependent on whether the infusion contained the protein measured (Table 1, and Figures 3b and 3c).

Figure 3 Extravascular losses of albumin, antithrombin and fibrinogen during and after a 30 min intravenous infusion of 10 ml/kg of 5% human albumin solution and autologous plasma

The losses expressed in weight units (upper panels) and in percentage of the infused amount (lower panels), if any.

The transcapillary escape of protein was also expressed as a fraction of the administered amount. Then approx. 60% of the albumin, 45% of the fibrinogen and 75% of the antithrombin had escaped from the plasma volume at 8 h. These differences were established early on and then remained fairly constant (Figures 3d–3f).

The clearance of plasma proteins was also expressed as the t1/2. When plasma was infused, the t1/2 of the administered albumin was 3.0 (2.6–4.9) h and was 4.8 (3.0–5.4) and 3.3 (2.3–4.5) h for fibrinogen and antithrombin respectively. The t1/2 of albumin when 5% albumin was infused was 4.1 (3.4–5.8) h.

Fluid and electrolyte balance

The urinary excretion exceeded the infused fluid volume in both series of experiments, but it was even higher after infusion of 5% albumin (mean, 1081 ml) than after autologous plasma (892 ml; P<0.0001). Also, the infused sodium was, on average, excreted within the follow-up period (Table 2). However, larger urine volumes were more diluted with respect to sodium (Figure 4, left-hand panel) and had a lower osmolality (Figure 4, right-hand panel). The urinary albumin excretion was measured during ten of the experiments, but was undetectable in seven and below a total of 6 mg in the others.

View this table:
Table 2 Composition of the infused fluid and of the urine excreted from the start of the infusion to 8 h later

Data are means (S.D.). NS, not significant.

Figure 4 Relationships between volume, sodium concentration and osmolality of all urine collected during the entire experimental period of 8 h

The calculations of the distribution of fluid between intra- and extra-cellular fluid spaces indicated that the former increased by 220 (50–320) and 127 (61–278) ml during the albumin and plasma infusions respectively (non-significant difference).

The serum sodium level, which is a key marker in this calculation, decreased by 0.4 (1.4) mmol/l during the experiments, with no difference between the infusions, and the concentration changes over time were small (Figure 5a). The serum chloride concentration was higher (P<0.03) and the bicarbonate concentration was lower (P<0.05) during the experiments with 5% albumin compared with those with autologous plasma (Figures 5b and 5c).

Figure 5 Serum concentrations of sodium (a), chloride (b) and bicarbonate (c) during the experiments


There are many aspects to consider when studying how the human body handles a volume load containing a protein colloid solution. First, there is plasma volume expansion which supports the haemodynamic system in the case of absolute or relative hypovolaemia. In the present study, the volume-expanding properties of the two protein colloid solutions, 5% albumin and autologous plasma, appeared to be quite similar, although the between-subject variability was higher for plasma. With both solutions, the maximum plasma volume expansion was about 75% of the infused volume, the t1/2 was 2.5 h, and the duration was 8 h.

Secondly, the dilution associated with volume expansion modifies the plasma protein concentrations. Such dilution makes it more difficult to raise the plasma concentration of any specific protein, which might be a therapeutic goal when choosing to infuse a protein colloid instead of a synthetic colloid. For example, the increase in the plasma albumin level after infusing nearly 1 litre of fluid never exceeded 10%, although the albumin solution had a 37% higher albumin concentration than plasma (Figure 2a).

A third aspect is transcapillary leakage of plasma components. The disappearance rate of albumin was dependent on the relationship between the concentrations of the protein in the infused colloid fluid and in plasma. Hence the albumin was eliminated faster when it was administered as 5% albumin compared with autologous plasma (Figure 3a). On the other hand, the elimination occurred at a fairly similar rate for the two fluids when expressed as a percentage of the administered amount. Between 40 and 50% of the infused albumin had been translocated from plasma to the interstitial fluid space 4 h after the infusions (Figure 3d). The disappearance of fibrinogen and antithrombin was also governed by the concentration of the protein in the infused colloid fluid, although the leakage occurred more slowly with fibrinogen than with the other two proteins. This difference is probably due to the larger size of the fibrinogen molecule (340 kDa) compared with albumin (67 kDa) and antithrombin (57 kDa) [10].

There was proportionality between the disappearance of fluid and protein from the bloodstream, but the plasma volume decreased by as much as 10 ml for each percent of infused albumin that escaped the vascular system (compare Figures 1 and 3d). Therefore the raised plasma albumin level was maintained and protein leakage still occurred when the volume expansion had subsided 8 h after the administration of fluid.

The trace amounts of albumin in the urine suggests that nearly all capillary leak of protein represented translocation to interstitial tissue. In contrast, the infused fluid volume was excreted as urine, as the volunteers voided a larger fluid volume than they received. The sodium and water balance did indicate some intracellular accumulation of fluid, but this fluid shift is cancelled out if we correct the data for sodium-free evaporation losses during the 8 h of the study (600 ml/day) [11]. The limited net transport of fluid across the cell and plasma membranes implies that no interstitial oedema developed in these volunteers, which could be expected to occur as the interstitial oncotic pressure increases on addition of intravascular proteins. However, the increase in protein content is known to far outweigh the interstitial fluid accumulation in high-permeability oedema [12].

The increased blood–tissue transport of albumin during volume expansion with a protein colloid solution has been studied by other authors. Renkin et al. [1] reported the increased elimination of 125I-labelled albumin 30 and 60 min after letting iso-oncotic albumin expand the plasma volume by 80–90% in rats. Albumin uptake was found in the viscera, heart and skin. They concluded that the increased transport was dissipative and not convective, which implies that diffusion, rather than the increased volume flow, would be the cause. Bent-Hansen [13] also considered that diffusion explains transcapillary leak of albumin, whereas the slower interstitial transport is merely convective. He also pointed out that a selective efflux of albumin from plasma to tissue occurs during the first minutes after intravascular administration, which acts to reduce of the plasma volume expansion.

Infused fluid does not have to contain protein to induce protein leakage. Volume expansion with normal saline doubled the clearance of 125I-labelled albumin in rabbits [14]. Similarly, transcapillary leak was indicated by a progressively larger difference in dilution between Hb and albumin in volunteers receiving Ringer's acetate, dextran 70 and hypertonic saline [15]. The albumin clearance seems to be is increased by ANP (atrial natriuretic peptide) [10,14], which is released during volume expansion.

In the present study, blood–tissue transport of proteins other than albumin, primarily antithrombin, occurred during administration of the antithrombin-free albumin solution. The alteration of the plasma concentration was slight, but long-lived. At 8 h, the concentration of this key inhibitor of the coagulation cascade was still lowered by 3%, whereas the principal coagulation substrate (fibrinogen) had returned to baseline. Such differences might explain why infusion of crystalloid fluid enhances coagulability [16], which can be reversed in vitro by the addition of antithrombin [17]. The balance between promoters and inhibitors of the coagulation cascade seems to be disturbed more when the infused fluid contains coagulation proteins. In the present study, the difference in concentration between fibrinogen and antithrombin was twice as large at 8 h after infusion of autologous plasma compared with 5% albumin.

Protein colloid fluids are known to influence haemodynamics and coagulation. Apart from the beneficial substitution effects of infused components in hypovolaemic subjects without and with coagulation disorders, both albumin [18] and plasma [19], as well as whole blood [20], decrease the ionized calcium level, which reduces the pumping capacity of the heart. In the present study, a reduction of the ionized calcium level was three times larger for plasma than for albumin, the difference being attributable to the content of citrate in stored plasma.

Infusion of 5% albumin prolongs the bleeding time, which is probably due to inhibition of platelet aggregation [4]. Our data show that 5% albumin also increases the aPTT and induces a more long-lasting increase in the prothrombin time. These coagulation tests were hardly affected at all by autologous plasma, and the larger difference in concentration between antithrombin and fibrinogen raises the possibility that autologous plasma promotes further coagulation. All the inhibitors of the coagulation cascade in plasma are small proteins having a molecular masses similar or lower than that of albumin, whereas several promoters (such as fibrinogen) are much larger and, therefore, have a higher tendency to remain intravascular when infused.

The use of plasma as a protein colloid fluid has increased since a meta-analysis showed, in 1998, that albumin solution increases mortality [21], although this conclusion would appear to be controversial today [22]. Plasma is recommended when there is a need for coagulation factors, but autologous plasma is used as a volume expander as well. Little is known about the time course of volume expansion following infusion of heterologous plasma, which has probably increased in clinical practice after the albumin debate. A study in volunteers from 1960 by Hutchison et al. [23] showed that autologous plasma and 6% dextran 70 increase the blood volume by 83 and 82% of the infused amount respectively, whereas heterologous plasma is often a poorer volume expander, particularly in the presence of any transfusion reaction.

Our measurements of the content of the infusions revealed that the 5% albumin solution had a higher albumin concentration than the autologous plasma. The addition of preservatives to the solutions made their electrolyte content (Table 2) deviate from that of plasma (Table 1). Autologous plasma was even hypertonic due to the relatively high sodium content of the CPD (citrate/phosphate/dextrose) anticoagulant solution, which increases volume expansion and prolongs the excretion of infused fluid [24]. The preservatives caused minor alterations in the electrolyte balance during the infusions, whereas larger amounts of these colloid solutions are apparently required to cause marked disturbances (Figure 5).

Limitations of the present study include that the volunteers did not suffer from hypoproteinaemia or hypovolaemia, which are indications for colloid therapy. Rather, they were in a state of slow continuous bleeding as the frequent blood sampling removed almost 300 ml of blood during each experiment. Corrections to account for sampled blood was applied in all calculations. Moreover, the mass balance calculations describe the net transcapillary transport of proteins, which is slower than the true rates, as proteins are returned from the interstitial fluid space to the blood by the lymphatic system. Finally, the plasma volume at baseline was not measured but derived from a mathematical expression based on the height and weight of the volunteers. This might have introduced errors, which we believe are quite small due to the fact that the further presentation is based on summary data for 15 individuals.

In conclusion, 5% albumin and autologous plasma are equally effective plasma volume expanders. Both induce a long-lasting capillary leak of protein (>8 h) which varies in proportion to molecular size and the difference in concentration of the protein in question between the infused fluid and the plasma.

Abbreviations: aPTT, activated partial thromboplastin time; Hb, haemoglobin; INR, international normalized ratio; MCV, mean corpuscular volume; RBC, red blood cell; t1/2, half-life


View Abstract