- Question: Do the degree and duration of the plasma volume expansion after IV infusion of about 200 mL of 20% albumin differ between postoperative patients and healthy volunteers?
- Findings: In both groups, the maximum plasma dilution was about 15% (corresponding to 400 mL), while the intravascular half-life of the infused fluid averaged 10 h in the patients and 7 h in the volunteers.
- Meaning: Twenty percent albumin is an effective plasma volume expander in the postoperative period.
Albumin has become the fluid of choice in the operating room and in intensive care when considering plasma volume expansion with a colloid. Two different formulas of albumin are available: an iso-oncotic (4%–5%) and a hyperoncotic (20%) solution. There are studies on the iso-oncotic formula in the literature,1–3 and a few on the hyperoncotic solution in septic patients.4,5 The only study of 20% albumin in the postoperative period that compared iso- and hyperoncotic preparations reported that the plasma volume expansion is dependent on the amount of infused albumin and not on the fluid volume.6
Hyperoncotic albumin has also been given to reduce peripheral edema in normovolemic patients during the de-escalation phase of fluid therapy, the reason being that because 20% albumin is believed to cause a fluid shift from the interstitial space to the plasma. However, this view has recently been called into question because of the Revised Starling Equation, which holds that there is no oncotic gradient between the plasma and the interstitial fluid.7 Moreover, the so-called nonabsorption rule means that fluid can only be recruited to the plasma via the lymph, the endothelial glycocalyx layer, and from glands.8 The intravascular persistence of infused albumin may also be shorter in inflammatory conditions, such as those associated with major surgery, due to breakdown of the endothelial glycocalyx layer (shedding), which serves as the key barrier to capillary leakage of plasma proteins.9 These findings have been documented in microcirculatory research,7 but they need validation in large biological systems.
The purpose of this study was to obtain data on the pharmacological and volumetric properties of hyperoncotic albumin to better understand the physiological basis for its use. The key outcome measures were degree and half-life of the plasma volume expansion induced by 20% albumin. Our hypothesis was that 20% albumin administered after major abdominal surgery in middle-aged and old patients persists intravascularly for a shorter period of time compared to when the same volume of albumin is given to young, healthy volunteers. Therefore, we assessed the duration and degree of extravasation of albumin, plasma volume expansion, and biomarkers of the endothelial glycocalyx shedding.
This study included a total of 30 subjects: 15 postoperative patients (2 males and 13 females) and 15 healthy volunteers (9 males and 6 females). The data were collected during 2016 as part of a larger project aimed at exploring the clinical efficacy of hyperoncotic albumin, including an arm with infusion of 20% albumin during ongoing abdominal surgery.
The study was approved by the Regional Ethics Committee in Stockholm (Institutional Review Board 2014/2146-31/1; Chairperson Pierre Lafolie) and written informed consent was obtained from all subjects participating in the trial. The trial was registered before patient enrollment at clinicaltrials.gov (NCT02556580, Principal investigator: Robert G. Hahn, Date of registration: September 22, 2015). As the study evaluated a pharmaceutical entity, approval was also obtained from the Medical Products Agency Eu-nr 2015-000996-26 on April 28, 2015.
For all subjects, inclusion criteria were age >18 years, American Society of Anesthesiologists physical status I–II, and blood hemoglobin concentration >10.0 g/dL. The volunteers were not to use daily medication and considered to be completely healthy. For patients, a planned open abdominal surgery with expected operation duration >3 h and a postoperative blood hemoglobin concentration of >9.0 g/dL were also required. The manuscript adheres to the Consolidated Standards of Reporting Trials guidelines.
Patients were recruited from the surgical preoperative ward by one of the research members on the day before the surgery. A total of 32 patients were enrolled, but 17 of these were excluded for various reasons.
The study was performed in the morning of the first day after the surgery when the patients were still in the postoperative ward. The fluid therapy during the operations was managed according to a goal-directed protocol, based on measurements of cardiac stroke volume by an esophageal Doppler system (CardioQ, Deltex Medical, Chichester, UK). In short, a bolus consisting of 200 mL of colloid fluid (3% dextran 60, Plasmodex, Meda AB, Sweden) was infused if a previous fluid bolus was associated with an increase of stroke volume by 10% or more. The duration of the operations was 5.9 ± 1.6 h (mean ± SD) and the perioperative blood loss was 700 mL (interquartile range 363–800 mL). After midnight, no patient received a colloid infusion in any form (ie, no blood, plasma, albumin, or synthetic colloid).
Hemodynamic stability was confirmed before starting the albumin infusion (mean arterial pressure [MAP] >60 mm Hg, heart rate between 50 and 100 beats/min, low plasma lactate, and a stable low dose, if any, of norepinephrine). All patients had a well-functioning epidural block during surgery and a constant-rate infusion of ropivacaine 2 mg/mL plus sufentanil 0.5 µg/mL during the postoperative period. The patients also had given an IV infusion of glucose 25 mg/mL with electrolytes at a constant rate of 40–100 mL/h (mean 71 mL/h) during the postoperative period. Recruited patients were excluded from the infusion study if the operating time had been <3 h.
At 6:30 am, an IV infusion of 3 mL/kg of human 20% albumin (Alburex, CSL Behring, King of Prussia, PA) was started at a constant rate over 30 minutes. Blood samples were collected on 15 occasions from a preexisting radial artery cannula (closed-system, Argon Safedraw Gabarith PMSET, Frisco, TX); at baseline, every 10 minutes for 1 h and then at 75, 90, 120, 150, 180, 210, 240, and 300 minutes after starting the infusion. The baseline samples were withdrawn in duplicate and the mean value of the 2 was used in further calculations.
The systolic pressure, diastolic pressure, and MAP and the heart rate were measured at the time of each blood sampling by an invasive hemodynamic monitor (Datex Ohmeda S/5, Helsinki, Finland).
Urine samples were collected from a catheter bag at 0, 60, and 300 minutes.
Patients were allowed to sip small amounts of water or juice (“comfort hydration,” maximum 100 mL) but not to eat, sit up, or raise their legs during the protocol period.
Control group consisted of volunteers >18 years old and were otherwise healthy. They arrived at the hospital ward between 7:00 and 9:00 am after fasting from midnight. To prevent dehydration, they were instructed to ingest 1 sandwich and to drink 1 glass (2 dL) of clear liquid, 2 h before any blood sampling was started. The experimental procedure was the same as for the surgical patients except that blood samples were drawn from a cannula placed in an antecubital vein of the arm that was not used for infusion.
After resting 30 minutes in the supine position, the volunteers were given 3 mL/kg of 20% albumin (Albunorm, Octapharma, Lachen, Switzerland) intravenously at a constant rate over 30 minutes with an infusion pump. To avoid sample dilution, a small volume of blood was drawn from the cannula and readministered together with 2 mL of 0.9% saline. Arterial pressures were measured using the oscilloscopic method, which correlates well with invasive pressures in young subjects (correlation coefficient 0.78, typical difference invasive-noninvasive 0.1 ± 16.5 mm Hg [mean ± SD] for the systolic pressure and 11.0 ± 12.2 mm Hg for the diastolic pressure10). Volunteers voided just before the infusion protocol started and then as needed during the experiment. A bladder catheter was not used, but the urine was collected and sampled on the same points in time as in the postoperative patients.
Blood for measurement of the hemoglobin concentration was collected in EDTA tubes. In the postoperative group, the samples were analyzed by Sysmex XN (Sysmex Corp, Kobe, Hyogo, Japan) while in the volunteer group, by a Cell-Dyn Sapphire (Abbott Diagnostics, Abbott Park, IL). The coefficient of variation for these analyses, based on the duplicate baseline samples, was 0.8%.
Blood was also sampled in lithium-heparin plasma gel tubes and used for measurement of the plasma albumin, C-reactive protein, amino-terminal pro-hormone of brain natriuretic peptide, and creatinine concentrations on the hospital’s central laboratory instrument (Cobas 8000, Roche Diagnostics, Basel, Switzerland).
The plasma and urinary concentrations of 3 shedding products of the endothelial glycocalyx layer were measured on frozen samples taken at 0, 60, and 300 minutes using enzyme-linked immunosorbent assay kits. These were human CD138/syndecan 1 (Diaclone, Besancon Cedex, France), heparan sulfate (AMS.E-EL-H2364 from Amsbio, Abingdon, UK), and hyaluronic acid (R&D Systems, Inc, Minneapolis, MN). According to the manufacturers, the coefficient of variations of these analyses were 6.2%, <10%, and <7.2%, respectively.
Fresh urine samples taken at 0, 60, and 300 minutes were also used to analyze the concentrations of neutrophil gelatinase–associated lipocalin and creatinine. Neutrophil gelatinase–associated lipocalin is a biomarker of tubulointerstitial inflammation that is regarded to be an early sign of kidney injury.11 Creatinine dilution was taken as evidence of stimulated diuresis because this metabolic waste product is excreted at a relatively stable rate over time.12
The blood volume before the infusion of 20% albumin (BVo) was estimated from height (m), weight (kg), and sex according to the Nadler formula.13
The amount of hemoglobin in the circulation (hemoglobinmass) at baseline was obtained as the product of BVo and the blood-hemoglobino. During the experiments, the sampled hemoglobin was subtracted from the hemoglobinmass. The blood volume at a later time n (BVn) was obtained by dividing the new hemoglobinmass with the blood hemoglobin measured at that later time (blood hemoglobinn).14
The hemoglobin-derived plasma dilution (with correction for blood sampling) was then
where (1 − hematocrito) converts the data on blood volume to plasma volume (PV).
By using a similar mass balance equation, the net capillary leakage of albumin (ie, the reduction of the albumin content of the plasma minus the lymphatic return of albumin) was obtained as the change in albumin mass from time 0 to time n, with correction for the sampled and infused amounts of albumin:2
The half-life (T1/2) of the infused amounts of albumin and fluid was calculated from a simple washout equation:
where X is amount, k is the elimination rate constant, t is the time, and ln 2 is the natural logarithm of 2, which is 0.693.
The recruitment of intra/extracellular fluid to the plasma at 300 minutes in the volunteer group has been described in detail elsewhere.15 The recruitment of intracellular fluid was obtained by a sodium balance method and extravascular recruitment by mass balance, ie, plasma volume expansion plus urinary excretion and sampled blood volume minus the infused volume of 20% albumin.
Data with a normal distribution are presented as the mean ± SD.
The difference between the 2 groups with respect to the plasma volume expansion over time (primary outcome) was evaluated by using repeated-measures ANOVA.
Point-wise comparisons of the plasma volume expansion and plasma content of albumin were studied with 1-way ANOVA.
The duration (half-life) of the volume expansion had a skewed distribution and was presented as the median (interquartile range), and the difference between the groups evaluated by using the Mann-Whitney U test (secondary outcome).
Differences between the groups were also given as the mean difference and 95% CI.
Stepwise multiple regression analysis was used to screen for confounding effects of body mass index, age, sex, and mean MAP on the group-wise comparisons.
P < .05 was considered statistically significant.
The study was powered (90%) to detect a 50% increase in the capillary leakage of fluid (clearance) in the postoperative patients as compared to the volunteers. This calculation is based on a plasma clearance of 5% albumin of 20 (SD = 10) mL/min.2 The standardized difference is then 1.5 and the total number of required subjects 22 (the software used was GPower version 184.108.40.206, Department of Psychology, Düsseldorf, Germany).
Thirty-two patients were enrolled but 17 were excluded before data analysis. Reasons were shorter duration of surgery than anticipated (N = 4), excessively long surgery (N = 3), patient-centered factors (tiredness, pain, vomiting; N = 4), complication during surgery (thrombosis; N = 1), high doses of norepinephrine infusion perioperatively (N = 1), and technical problems (N = 4).
The included 15 patients were operated for colonic resection in combination with additional surgery on at least 1 organ, such as uterus, pancreas, and kidney (N = 9), rectum amputation (N = 5), and intraabdominal lipoma (N = 1).
The patients were older than the volunteers (65 ± 13 vs 31 ± 12 years; P < .001) but the body weights and body mass indices were quite similar (Table 1).
The subsequent text gives descriptive data for all subjects together with the between-group differences. Data for the separate groups are given in Table 2.
The infusions of 20% albumin diluted the plasma volume by almost 15% (Figure 1A), which represented twice the infused fluid volume (Figure 1B). The mean between-group differences were small, –0.9% (95% CI, [–4.7]–2.9) at 30 minutes and 2.1% (95% CI, [–2.3]–6.5) at 300 minutes. Moreover, there was no statistically significant difference in plasma volume expansion over time (repeated-measures ANOVA), and no interaction between volume expansion and time.
The plasma albumin concentration increased by 0.78 ± 0.16 g/dL by the end of the infusions (mean value 0.19 g/dL higher in the patients, 95% CI, 0.14–0.24; P < .001), and remained elevated by 0.54 ± 0.17 g/dL at 300 minutes (mean difference 0.09 g/dL, 95% CI, 0.04–0.22; Figure 1C).
The rates of net elimination of albumin and infused fluid from the intravascular space were calculated from decay curves (Figure 2). Capillary leakage of albumin occurred at a rate of 3.3 ± l.4 g/h (mean group difference 0.9, 95% CI, 0.0–1.9; Figure 1D), which represented 7.7% ± 3.3% of the administered amount.
The half-life of the exogenous albumin in the plasma was 7.7 (5.4–10.7; median and interquartile range) h and was marginally longer in the patients (geometric mean difference 1.2, 95% CI, 0.8–2.0; Mann-Whitney U test, no significant between-group difference).
The induced plasma volume expansion subsided with a half-life of 8.2 (4.2–11.5) h, which tended to be longer in the postoperative patients (geometric mean difference 1.5, 95% CI, 0.8–2.8; no significant between-group difference).
The urine volume was slightly lower in the postoperative group (Figure 1E).
Body mass index was a confounder to several of the group-wise comparisons. Using stepwise multiple regression, body mass index increased with the plasma volume expansion at 30 minutes (correlation coefficient r = 0.43; P < .02), with the rise in serum albumin at 30 minutes (r = 0.52; P < .01), with the half-life of the exogenous albumin (log-transformed, r = 0.42; P < .03), and with the half-life of the plasma volume expansion (log-transformed, r = 0.47; P < .02). Age, sex, or MAP did not serve as statistically significant predictors to the main outcome parameters.
The 16 subjects with body mass index >25 kg/m2 had a plasma volume expansion of 15.6 (5.0)% at 30 minutes, while the others showed 11.9 (4.0)% (P < .03). The half-lives of the volume expansion were 10.3 (6.3–16.3) h and 5.5 (3.6–8.0) h, respectively (P < .04).
The C-reactive protein values were higher in the postoperative patients than in the volunteers (Figure 1F). The plasma heparan sulfate concentration and the urine syndecan-1 concentration were also significantly higher in the postoperative group compared to the volunteers (P < .01 or stronger; Figure 3). The plasma shedding products decreased in the volunteer group after the albumin infusion but remained unchanged in the postoperative patients (Table 2).
Eleven of the 15 patients, but no volunteer, had an amino-terminal pro-hormone of brain natriuretic peptide concentration at baseline that was above the normal range for their age and sex. The concentration increased by 40%–50% during both series of experiments (Table 2).
The urinary neutrophil gelatinase–associated lipocalin concentration remained essentially unaffected, with 4 patients having values above the level of detection before the infusion, 4 at the end of the experiment, and 3 remaining the same.
The urinary creatinine concentration did not change in the patient group (from 9.7 ± 4.0 to 9.0 ± 6.6 mmol/L), whereas it decreased in the volunteer group (from 18.0 ± 11.2 to 9.2 ± 5.6 mmol/L; P < .002). The difference between these changes was 8.1 mmol/L (95% CI, 2.5–13.7) and was due to the higher values among the volunteers before the infusion started (P < .01) while the concentrations were similar at 300 minutes (P = .93).
We found no difference in intravascular pharmacokinetics of 20% albumin in patients after major abdominal surgery as compared to nonoperated controls, despite evidence of increased inflammation in the postsurgical patients.
Our hypothesis of a shorter intravascular persistence of albumin and infused fluid in postoperative patients was refuted. The reason why a higher net capillary escape rate of albumin was expected stems from reports of shedding of the endothelial glycocalyx layer due to inflammation and hypervolemia.8,9 Clear evidence of shedding has been reported in severe trauma16 and after vascular surgery,17 where the plasma concentrations of shedding products increased several fold. The lengthy major abdominal surgeries studied here were associated with a clear rise in the plasma C-reactive protein and amino-terminal pro-hormone of brain natriuretic peptide concentrations, suggesting an inflammatory reaction and fluid-induced cardiac distention. Nevertheless, our measurements of 3 shedding products in plasma and urine showed only minor deviations from normal values and no changes in response to the infusions (Figure 3). Moreover, the net capillary escape rate of albumin and fluid were not increased.
Overall, an effective and long-lasting plasma volume-expanding effect occurred due to infusion of 20% albumin in both the postoperative patients and the volunteers. The volume-expanding effects of iso-oncotic colloids have a half-life of 2–3 h,18 whereas the expansion of 20% albumin lasted 3–4 times longer (Table 2). The longest intravascular persistence times occurred in subjects with overweight, suggesting that caution with repeated infusions of 20% albumin is warranted in them. Age and sex had negligible influence on the key outcome parameters.
The plasma volume expansion, as indicated by the hemodilution, was about twice the infused amount of fluid (Figure 1B), which means that fluid must have been recruited from noncirculating sources. According to the “nonabsorption rule,” fluid can be recruited to the plasma from the lymph and the glycocalyx layer. If the glycocalyx was drained of fluid, some shedding might be anticipated, but this apparently did not occur. Fluid can also be absorbed via fenestrated capillaries in the gastrointestinal tract and glands, but mainly in response to a low hydrostatic pressure.7–9 The MAP was indeed lower in the patients than in the volunteers, and this might explain why allocation of fluid seemed to occur somewhat more rapidly, although the majority of the recruitment should still be ascribed to oncotic forces.
The plasma volume-expanding properties of albumin preparations were studied by Lamke and Liljedahl,6 who infused 50 g of albumin as a 5%, 20%, and 25% solution in postoperative patients. Using iodine-labeled albumin, they found the plasma volume to increase by 500 mL with all 3 concentrations. The fluid-binding capacity of albumin we found is quite similar to that reported by them (10 mL/g albumin).
Margarson and Soni5 used the hemodilution method to compare the volume expansion of 20% albumin in septic patients and volunteers. They found that 200 mL of 20% albumin expanded the plasma volume by 430 mL in septic patients and by 500 mL in volunteers. The net albumin leakage occurred 30% faster in the septic patients.14
The measurements of shedding products suggest that urinary excretion is an important route of elimination for these. Concentrations in plasma and urine were virtually identical for heparan sulfate, slightly lower in urine for hyaluronic acid, and higher in urine for syndecan-1. A previous study held that the excretion of hyaluronic acid, but not of syndecan-1, should be corrected for urine flow.19 A more effective urinary excretion might still explain why the plasma concentrations of syndecan-1 were lower in the patients than in the volunteers (Figure 3).
No correction for hemodilution was made for the plasma concentrations of shedding products, but this might be warranted if their volume of distribution is similar to the plasma volume. Different practices exist for this issue.20,21
The key indication for treatment with albumin is plasma volume support in patients with signs of hypovolemia and organ hypoperfusion, although 5% albumin is more widely used for this purpose. The present subjects were not hypovolemic and albumin was therefore not indicated for this reason. The most common purpose of administration of 20% albumin to normovolemic patients is to recruit edema. More rarely, 20% albumin is used to raise a low plasma albumin concentration, or to make use the scavenger capacity of albumin.22
Our results support that 20% albumin has a dehydrating effect, even in subjects in a physiological state of renal fluid conservation, as the urinary excretion in both groups exceeded the infused fluid volume. Nevertheless, the urinary excretion clearly tended to be larger in the volunteer group, which can probably be explained by the absence of a surgical stress response. It is unclear why they started the study in a state of renal fluid conservation, which is a sign of dehydration in healthy subjects,12 but this might reflect a low daily intake of water. By contrast, the urine flow in the patients was probably normal, as their preinfusion and postinfusion urinary concentration of creatinine was close to that found in the normal population.19
Limitations of the present study include the differences between the 2 groups with regard to age and sex profiles, although these 2 parameters did not stand out in our analysis of confounders. Not only surgery but also health factors differed between the groups, but it is very difficult to recruit volunteers with a mean age of 65 who are perfectly healthy. Therefore, the scope was restricted to the study of 20% albumin in a relevant patient group who most commonly undergo major abdominal surgery, using younger but healthy volunteers as reference.
Plasma albumin was on the low side in the postoperative patients, which is commonly seen after major surgery. This might explain why their increase in plasma albumin was greater, and why the net capillary leakage of albumin tended to be slower than in the volunteers2 (Table 2). Capillary leakage rates and intravascular half-lives of albumin and infused fluid were expressed as the changes in intravascular content, which grossly represents the true leakage minus the returned albumin and fluid by the lymph. The true capillary leakage of albumin would have required the use of an albumin-bound tracer to be measured, but this technique was not applied.
Most calculations have used the hemodilution as the index of relative changes in plasma volume. The hematocrit is known to be low in the capillaries, but the hemoglobin concentration is quite similar when taken from different major vessels in the dog,23 as well as in volunteers.24 Although not a perfect measure, the dilution of the hemoglobin concentration remains the most widely used indicator of fluid-induced plasma volume expansion. Blood was sampled from the arterial line in the patients, but this is expected to have only a marginal influence on the results.24
One should note that 200 mL of 20% albumin raised the plasma amino-terminal pro-hormone of brain natriuretic peptide concentration, and a few subjects did not show adequate elimination of the exogenous albumin and fluid components (rather a steady state) during the study period (Table 2). The plasma volume expansion was modest, and no objective signs of fluid overload (such as dyspnea) were seen, but the excessively long duration of the hypervolemia should be kept in mind and considered when follow-up infusions are planned.
Infusion of 3 mL/kg of 20% albumin caused a long-lasting plasma volume expansion (half-life 7–10 h), amounting to twice the infused amount. No relevant differences were found between postoperative patients and younger healthy volunteers. Shedding of the endothelial glycocalyx layer was slight, or completely absent. Thus, our results suggest that 20% albumin can be used as an effective plasma volume expander in the postoperative setting.
Nurse anesthetist Sandra Månsson assisted in the collection of data from the postoperative patients.
Name: Emma Hasselgren, MD.
Contribution: This author helped collect the patient data and co-write the manuscript.
Name: Markus Zdolsek, MD.
Contribution: This author helped collect the data on the volunteers and perform calculations.
Name: Joachim H. Zdolsek, MD, PhD.
Contribution: This author helped organize and collect the data on the volunteers.
Name: Håkan Björne, MD, PhD.
Contribution: This author helped organize and collect the data on the patients.
Name: Camilla Krizhanovskii, PhD.
Contribution: This author wasresponsible for analyses of the endothelial shedding products.
Name: Stelia Ntika, MSc.
Contribution: This authorhelpedperform the analyses of the endothelial shedding products.
Name: Robert G. Hahn, MD, PhD.
Contribution: This authorhelpedplan the study, write applications, design graphs and tables, co-write the paper, and submitted and revised the manuscript.
This manuscript was handled by: Tong J. Gan, MD.