Physicians and their patients are greatly concerned about perioperative blood administration. Although isovolemic hemodilution is utilized to decrease the incidence of transfusion, it is unclear at what degree of hemodilution hepatoenteric ischemia and injury occurs. The authors hypothesized that hepatic ischemia, systemic ischemia, and tissue injury would occur during hemodilution in rabbits, and that the severity of ischemia and injury may be dependent on the fluid administered.


Rabbits anesthetized with isoflurane were assigned randomly to a sham-operated group (n = 8) or groups that underwent four isovolemic hemodilutions (25% of the blood volume removed at hourly intervals), with blood replaced with one of three solutions: balanced electrolyte solutions containing 6% pentastarch (n = 8), 6% hetastarch (n = 9), or 5% human albumin in normal saline (n = 8). Arterial ketone body ratio and plasma lactate, respectively, served as measures of hepatic and systemic ischemia. Gastric, duodenal, and hepatic histologic injury was assessed post mortem.


Hemodilution from a baseline hematocrit of about 33% to about 8% (third hemodilution) with all three colloids did not result in a significant increase in plasma lactate concentration or decrease in arterial ketone body ratio. At a hematocrit of about 5% (fourth hemodilution), the hetastarch group had a significantly (P < 0.05) greater plasma lactate concentration than the sham-operated and 5% human albumin groups. There were no significant differences in arterial ketone body ratio or histologic injury between the groups.


Isovolemic hemodilution (approximately 5% hematocrit) with albumin, pentastarch, or hetastarch solutions does not result in significant hepatic ischemia or injury assessed by histology.

PHYSICIANS and their patients are greatly concerned about perioperative allogeneic blood administration. Although the blood supply of the Unites States is now safer than ever before, [1,2] the risk of blood-borne infection will never be zero, because of finite serologic test sensitivity and specificity and human error. [3] Consequently, several effective blood-conserving strategies have been utilized clinically, such as red-cell scavenging, actively warming patients, and isovolemic hemodilution. [4] Weiskopf [5] recently presented an elegant mathematical analysis arguing that isovolemic hemodilution to a hematocrit of 15% could obviate the need for allogeneic transfusion with surgical blood losses between 5 and 10 l. It also has been demonstrated that isovolemic reduction of blood hemoglobin concentration to 5 g/dl in conscious, healthy humans at rest does not result in inadequate systemic oxygen delivery as assessed by oxygen consumption and plasma lactate concentration. [6] Altogether, these investigations strongly support the clinical efficacy and safety of isovolemic hemodilution as a blood conservation strategy.

Although it may seem clear that isovolemic hemodilution is a valuable clinical technique, it is unclear whether the asanguineous solutions used to replace shed blood adversely affect either systemic or regional perfusion-oxygenation supply-demand relationships. Spahn et al. [7] recently noted that the replacement fluid did not adversely affect the compensatory increase in cardiac output as long as isovolemic filling of the heart was maintained. Although guidelines regarding red-cell transfusion exist, [8] the authors are unaware of any guidelines with regard to what particular intravenous fluid or combination of fluids may be administered to obtain a desired hematocrit without incurring systemic or regional ischemia.

The present study was designed to determine what degree of hemodilution is required to cause hepatoenteric ischemia and tissue injury in rabbits. The hypothesis that the severity of ischemia and injury is fluid-dependent also was tested.

The study was approved by the Animal Review Committee of the University of Alabama at Birmingham. All animals received humane care in compliance with the Principles of Laboratory Care formulated by the National Society for Medical Research and with the Guide for the Care and Use of Laboratory Animals prepared by the National Research Council and the National Academy Press (revised October 1996, Washington, DC).

Surgical Protocol 

Male New Zealand white rabbits (Myrtle's Rabbits, Thompson Station, TN), weighing 1.8–2.8 kg (n = 33), were made to fast for 14–16 h before experimentation but allowed free access to water. All rabbits were premedicated with 2 mg/kg intravenous famotidine (Merck, West Point, PA) via a marginal ear vein. Rabbits were anesthetized with 10 mg/kg intravenous ketamine (Parke-Davis, Morris Plains, NJ) via a marginal ear vein. Animals were subsequently administered inhaled 1% isoflurane (Abbott Laboratories, North Chicago, IL) in 99% oxygen (3 l/min flow). Isoflurane administration (inspired concentration) was monitored with an anesthetic agent monitor (model 8100, BCI International, Waukesha, WI) that was calibrated daily. After tracheostomy, mechanical ventilation with a Harvard Apparatus ventilator (model 683, Millis, MA) was performed with PaCO(2) maintained at 32–45 torr (4.2–6.0 kPa). Pancuronium bromide (Elkins-Sinn, Cherry Hill, NJ) was administered 0.1 mg [middle dot] kg-1[middle dot] h-1intravenously to facilitate mechanical ventilation. Mean arterial pressure was monitored by placement of an 18-gauge right femoral arterial catheter. Central venous access was obtained via the right internal jugular vein for pressure monitoring and fluid administration. All pressures were recorded on a Grass Model 7D polygraph (Grass Instruments, Quincy, MA). All rabbits received a maintenance infusion of lactated Ringer's solution at 4 ml [middle dot] kg-1[middle dot] h-1, and esophageal temperatures were maintained at 38–39 [degree sign]C with a heating pad. A 30-min equilibration period followed completion of the surgical preparation.

Isovolemic Hemodilution Protocol 

Rabbits were assigned randomly to groups administered 5% human albumin in normal saline (Abbott Laboratories; n = 8), 6% hetastarch (n = 9), or 6% pentastarch (n = 8) during isovolemic hemodilution. The hydroxyethyl starch solutions administered were Hextend and PentaLyte (BioTime, Berkeley, CA), which are, respectively, hetastarch and pentastarch solutions containing balanced electrolytes (Na+= 143 mM, Cl-= 124 mM, Ca++= 2.5 mM, Mg++= 0.45 mM, K+= 3 mM), glucose (5 mM), and lactate buffer (28 mM). For simplicity, Hextend and PentaLyte are referred to subsequently as hetastarch and pentastarch solutions. A fourth group (n = 8) underwent sham operation, receiving maintenance and replacement fluids (lactated Ringer's solution) for all blood sampling. All fluids were warmed to 39 [degree sign]C in a water bath. Isovolemic hemodilution was performed by an exchange transfusion of 25% of the estimated circulating blood volume, which was calculated as follows:([weight in kilograms x 0.07 l/kg] x 0.25) x 1000 ml = 25% of the blood volume in milliliters. A quantity of one of the test fluids equal to 25% of the blood volume was infused intravenously, and simultaneously an equal volume of blood was removed from the femoral arterial catheter over a 2-min period. This procedure was performed three more times at 1-h intervals for a total of four exchange transfusions. Animals underwent 1:1 (one volume replacement fluid:one volume of blood removed) isovolemic hemodilutions. Before each exchange transfusion, 4.5 ml arterial blood was removed from animals undergoing isovolemic hemodilution for biochemical and physiochemical analysis described subsequently. This volume then was replaced immediately with an equal volume of the experimental fluid assigned to the animal. Sham-operated animals had 3 ml arterial blood removed at matched time intervals, which was replaced 3:1 with lactated Ringer's solution.

Hemodynamics, Arterial Blood Gas Analyses, Hematocrit, Plasma Colloid Oncotic Pressure, and Plasma Total Protein Concentration 

The mean arterial pressure (MAP), central venous pressure, and heart rate were recorded after 30 min of equilibration and 1 h after each isovolemic hemodilution. The pH, PaCO(2), PaO(2), and HCO3-of arterial blood samples were determined after 30 min of equilibration and 1 h after each isovolemic hemodilution. Arterial blood gases were analyzed at 37 [degree sign]C using a blood-gas analyzer (model 1306, Instrumentation Laboratory, Lexington, MA). Hematocrit was assessed by centrifuging blood samples contained in a microhematocrit tube for 3 min in a microcapillary centrifuge (model MB, International Equipment Company, Needham HTS, MA) and then manually determining the hematocrit on a microcapillary reader (International Equipment Company). The remaining volume of all blood samples was then centrifuged at 1000g for 5 min at 4 [degree sign]C. Plasma colloid oncotic pressure was measured after 30 min of equilibration and 1 h after each isovolemic hemodilution with a colloid osmometer (model 4420, Wescor, Logan, UT) using a membrane with a 30,000-d molecular-weight cut-off. The colloid osmometer was calibrated twice daily with Osmocoll II standard (Wescor). The colloid oncotic pressure of each of the experimental fluids and generic hetastarch solution (Abbott Laboratories) also was determined. Plasma total protein concentration was measured after 30 min of equilibration and 1 h after each isovolemic hemodilution. Plasma total protein concentration was determined with a spectrophotometric method. [9] Hematocrit and plasma protein concentrations were used to assess the extent of hemodilution among the groups.

Plasma Lactate, Acetoacetate, and [Greek small letter beta]-Hydroxybutyrate Analyses 

Plasma samples collected after 30 min of equilibration and 1 h after each isovolemic hemodilution were assayed for lactate, acetoacetate, and [Greek small letter beta]-hydroxybutyrate. Plasma lactate concentrations were determined with a GM-7 analyzer (Analox Instruments, Luxenburg, MA), which uses a modification of an amperometric measurement of the oxidation of reduced nicotinamide adenine dinucleotide with a Clark oxygen electrode. [10–12] Plasma was stored at -85 [degree sign]C before measurement of acetoacetate and [Greek small letter beta]-hydroxybutyrate with a GM-7 analyzer using a similar methodology. [10–12] Arterial ketone body ratio (AKBR) was defined as the plasma acetoacetate concentration divided by the [Greek small letter beta]-hydroxybutyrate concentration. Plasma lactate concentration served as a measure of systemic ischemia, whereas AKBR served as a measure of hepatic ischemia.

Tissue Analyses 

Tissue biopsy specimens were obtained from the anterior wall of the stomach, duodenum, and middle lobe of the liver post mortem. Tissue sections immediately were placed in 10% buffered formalin and subjected to routine processing, embedded in paraffin, and stained with hematoxylin and eosin. Morphometric injury was graded by one of the authors, who was blinded to the identity of the resuscitative fluid or experimental condition (e.g., sham vs. hemodilution). The grading scales of tissue injury were as follows. The stomach biopsy specimens were evaluated with the following criteria of tissue injury:(1) mucosal edema, (2) mucosal hemorrhage and ulceration, (3) mineralization, and (4) parietal/chief cell degeneration and vacuolation in areas other than surface ulceration. Specimens from duodenum were evaluated with these criteria:(1) lacteal dilation, (2) individual cell necrosis, (3) submucosal edema, and (4) inflammatory cells in lamina propria. Finally, evaluation of liver biopsy specimens used the following criteria:(1) dilation of sinusoids, (2) inflammatory cells in the capillaries, and (3) hepatocellular vacuolation. Each injury criterion had a possible score of 0–4, with 0 = no abnormality, 1 = minimal, 2 = mild, 3 = moderate, and 4 = marked. Additional tissue biopsy specimens were used to obtain the tissue wet-to-dry weight ratio. After recording the wet weight of the tissue sample, it was placed in a drying oven at 70 [degree sign]C for 2 weeks and reweighed. An increase in the tissue wet-to-dry weight ratio served as a gross measure of tissue edema.

Statistical Analysis 

All parametric variables are expressed as mean +/- SD. Analyses of the effects of isovolemic hemodilution and the intravenous fluid administered on all parametric variables were conducted by one-way measures of analysis of variance at each time point. The Tukey test was used for post hoc comparisons between the groups. Tissue-injury scores are expressed as the median and the first and third quartiles. The effects of isovolemic hemodilution and the intravenous fluid administered on stomach, duodenum, and liver injury scores were analyzed with the Kruskal-Wallis test. Analysis of the oncotic pressure of each undiluted intravenous fluid in vitro was conducted by analysis of variance, with the Tukey test used for post hoc comparisons. An alpha error of < 0.05 was considered significant. As values of plasma lactate concentration indicative of ischemia have not been defined in the anesthetized rabbit, the following a priori definition was formulated. Systemic ischemia was defined as a lactate value greater than the mean plus 2 SDs observed at the 30-min equilibration (derived from all observations in all groups at 30-min equilibration).

General Observations 

All animals survived the experimental protocol. The in vitro colloid oncotic pressures of the experimental solutions administered and the generic hetastarch solution were measured (six measurements per fluid). The values were as follows: 5% human albumin in normal saline, 20.4 +/- 0.4 mmHg; generic hetastarch, 31.5 +/- 0.6; hetastarch solution, 31.3 +/- 0.6; and pentastarch solution, 32.2 +/- 1.0. All three hydroxyethyl starch solutions had a significantly greater colloid oncotic pressure than 5% human albumin in normal saline solution.

Hemodynamics, Hematocrit, Plasma Protein Concentration, and Plasma Colloid Oncotic Pressure 

The MAP of the sham group was significantly greater than that of the albumin group after the third hemodilution and greater than that of the pentastarch group after the fourth hemodilution (Table 1). There were no significant differences in central venous pressure or heart rate among the groups throughout experimentation. The hematocrit values of all three isovolemic hemodilution groups were significantly lower than the sham group after each hemodilution. However, there were no significant differences in hematocrit among the isovolemic hemodilution groups throughout experimentation. The plasma protein values of the pentastarch and hetastarch groups were not significantly different each other during experimentation, but their protein values were significantly different from the albumin and sham groups after each hemodilution. The plasma colloid oncotic pressure of the albumin and hetastarch groups were significantly greater than the sham group after the second, third, and fourth hemodilutions However, the plasma colloid oncotic pressure of the pentastarch group was not different from the other groups throughout experimentation.

Arterial Blood-Gas Analyses 

There were no significant differences in arterial pH, PaCO(2), or PaO(2) among the groups throughout experimentation (Table 2). After the fourth hemodilution, the albumin group had a significantly greater arterial HCO3-concentration than the hetastarch group.

Plasma Lactate and AKBR 

The value of plasma lactate concentration indicative of ischemia was determined to be at least 3.7 mM, based on the 30-min equilibration values of all four groups (Figure 1). Hepatic ischemia was defined as an AKBR not more than 0.25, which has been associated with hepatic mitochondrial dysfunction in rabbits [13] and cats [14] and with severe hepatic dysfunction in humans. [15] These ischemic thresholds are depicted as the dashed horizontal line in both panels of Figure 1. There were no significant differences in circulating lactate between the groups after achieving a hematocrit value of approximately 8% in the isovolemic hemodilution groups after the third hemodilution. After the fourth hemodilution (hematocrit approximately 5%), the hetastarch group had a plasma lactate concentration that was significantly greater than those of the sham and albumin groups, and the hetastarch and pentastarch groups' mean values were greater than or equal to the ischemic threshold. Hemodilution to a hematocrit of approximately 5% did not result in any significant differences in AKBR between the groups. Values of the plasma concentrations of acetoacetate and [Greek small letter beta]-hydroxybutyrate from all groups are depicted in Table 3. The plasma concentrations of acetoacetate and [Greek small letter beta]-hydroxybutyrate of the sham group were significantly greater than those of the pentastarch and hetastarch groups after the fourth hemodilution.

Histopathology and Tissue Wet-to-dry Weight Ratio 

There were no significant differences in stomach, duodenum, or liver histologic injury scores among the groups (Table 4). There were also no significant differences in stomach, duodenum, or liver wet-to-dry weight ratios among the groups.

The findings of the present study are in agreement with other animal [16,17] and human [6] investigations of hepatoenteric and systemic ischemia in the setting of acute isovolemic hemodilution. Noldge et al. [16,17] could not demonstrate a decrease in hepatic lactate uptake, an increase in hepatocellular enzyme release, [16] or histologic liver injury [16] after reducing the hematocrit from about 30% to about 14% in a porcine model wherein 6% hetastarch dissolved in normal saline was used for hemodilution. Similarly, Weiskopf et al. [6] found that isovolemic hemodilution with 5% human albumin solution in awake, resting humans (decrease of hematocrit from approximately 39% to approximately 15%) did not increase circulating lactate concentrations. The present study extended these observations, establishing that at a hematocrit of approximately 8%(24% of the original value of approximately 33%), neither circulating lactate concentrations nor AKBR values of the animals administered any of the colloids studied were different from those of the sham-operated animals. Further, at a hematocrit of approximately 5%(15% of the original value). neither the wet-to-dry weight ratios nor histologic injury scores were different among the groups in the present investigation. Taken as a whole, our data demonstrate that under the conditions of the present study, hepatoenteric ischemia and injury do not occur at a hematocrit less than half that reported previously. [16,17]

Another finding revealed by the present study is that the systemic and hepatic ischemia that occurs during acute isovolemic hemodilution is influenced by the specific asanguineous solution that is administered. After the fourth hemodilution (hematocrit about 5%), rabbits administered hetastarch solution exhibited more systemic ischemia than the sham or albumin groups, as assessed by increases in circulating lactate concentration. However, there was no significant difference in AKBR among the four groups. Although the plasma lactate concentration and AKBR values of the three colloid groups were not always statistically different from one another, an apparent hierarchy of the severity of ischemia was observed at a hematocrit of approximately 5%. It appeared that the severity of ischemia after hemodilution was least with albumin, greater with pentastarch, and greatest with hetastarch administration. This pattern occurred despite the fact that there were no significant differences in MAP, plasma colloid oncotic pressure, or hematocrit values among the colloid groups. One possible explanation for this phenomenon is that the colloids administered possess different viscosities. It is not unreasonable to assume that hemodilution with albumin results in plasma viscosity similar to that before hemodilution. On the other hand, substitution of endogenous plasma proteins with large, branched hydroxyethyl starch molecules likely increase plasma viscosity. Indeed, hetastarch solutions have a greater viscosity than pentastarch solutions in vitro. [18] Consequently, we speculate that at the extremes of hemodilution, either total microcirculatory flow or the heterogeneity of flow is impeded by increased plasma viscosity and that this effect is greatest with hetastarch, less with pentastarch and least with albumin administration.

It is of great interest that at extremely low hematocrit values the groups administered either hydroxyethyl starch solution had as little systemic and hepatoenteric ischemia as the human albumin and sham groups. Unlike hydroxyethyl starch solutions, which are easily manufactured, the supply of human albumin solution is dependent on the availability of human plasma, which could result in occasional albumin shortages. Further, volume for volume, human albumin solution is generally more expensive than the hydroxyethyl starches. Last, a recent metaanalysis revealed that the administration of human albumin to critically ill patients (e.g., hypovolemia, burns) was associated with an increased mortality rate. [19] Although further prospective, randomized clinical investigations involving human albumin solutions are required to either substantiate or disprove the conclusions of this study, [19] it is reassuring that other, equally tolerated colloid solutions are available for isovolemic hemodilution.

In summary, isovolemic hemodilution with human albumin, pentastarch, or hetastarch solutions did not result in significant increases in circulating lactate concentrations (approximately 8% hematocrit), decreases in AKBR (approximately 5% hematocrit), or histologic injury (approximately 5% hematocrit). These observations were obtained in isoflurane-anesthetized rabbits after neuromuscular relaxation with pancuronium, experimental conditions known to decrease metabolic demands. It is conceivable that similar, parallel clinical studies involving anesthetized patients may reveal that hematocrits lower than 15% may be tolerated intraoperatively after isovolemic hemodilution. If lower hematocrits can be safely tolerated, it is likely that the incidence of allogeneic blood transfusion (and blood-borne infection) can be decreased for elective operative procedures.

Donahue JG, Munoz A, Ness PM, Brown DE Jr, Yawn DH, McAllister HA Jr, Reitz BA, Nelson KE: The declining risk of post-transfusion hepatitis C virus infection. N Engl J Med 1992; 327:369-73
Aach RD, Lander JJ, Sherman LA, Miller WV, Kahn RA, Ginick GL, Hollinger FB, Werch J, Szmuness W, Stevens VE, Kellner A, Weiner JM, Mosley JW: Transfusion-transmitted viruses: Interim analysis of hepatitis among transfused and nontransfused patients, Viral Hepatitis. Edited by Vyas GN, Cohen SN, Schmid R. Philadelphia, Franklin Institute Press, 1978, pp 383-96
Zuck TF: Greetings: A final look back with comments about a policy of zero-risk blood supply (editorial). Transfusion 1987; 27:447-8
Schmied H, Schiferer A, Sessler DI, Meznik C: The effects of red-cell scavenging, hemodilution, and active warming on allogenic blood requirements in patients undergoing hip or knee arthroplasty. Anesth Analg 1998; 86:387-91
Weiskopf RB: Mathematical analysis of isovolemic hemodilution indicates that it can decrease the need for allogeneic blood transfusion. Transfusion 1995; 35:37-41
Weiskopf RB, Viele MK, Feiner J, Kelley S, Lieberman J, Noorani M, Leung JM, Fisher DM, Murray WR, Toy P, Moore MA: Human cardiovascular and metabolic response to acute, severe isovolemic anemia. JAMA 1998; 279:217-21
Spahn DR, Leone BJ, Reves JG, Pasch T: Cardiovascular and coronary physiology of acute isovolemic hemodilution: A review of nonoxygen-carrying and oxygen-carrying solutions. Anesth Analg 1994; 78:1000-21
American Society of Anesthesiologists Task Force on Blood Component Therapy: Practice guidelines for blood component therapy. Anesthesiology 1996; 84:732-47
Smith PK, Krohn RI, Hermanson GT, Mallia AK, Gartner FH, Provenzano MD, Fujimoto EK, Goeke NM, Olson BJ, Klenk DC: Measurement of protein using bicinchoninic acid. Anal Biochem 1985; 150:76-85
Cheng FS, Christian GD: Amperometric measurement of enzyme reactions with an oxygen electrode using air oxidation of reduced nicotinamide adenine dinucleotide. Anal Chem 1977; 49:1785-8
Thomas LC, Christian GD: The use of amperometric oxygen electrodes for measurements of enzyme reactions. Anal Chem Acta 1977; 89:83-91
Cheng FS, Christian GD: A coupled enzymatic method to measure blood lactate by amperometric monitoring of the rate of oxygen depletion with a Clark oxygen electrode. Clin Chem Acta 1979; 91:295-301
Yamamoto M, Tanaka J, Ozawa K, Tobe T: Significance of acetoacetate/[Greek small letter beta]-hydroxybutyrate ratio in arterial blood as an indicator of the severity of hemorrhagic shock. J Surg Res 1980; 28:124-31
Kobelt F, Schreck U, Henrich HA: Involvement of liver in the decompensation of hemorrhagic shock. Shock 1994; 2:281-8
Ozawa K, Aoyama H, Yasuda K, Shimahara Y, Nakatani T, Tanaka J, Yamamoto M, Kamiyama Y, Tobe T: Metabolic abnormalities associated with postoperative organ failure: A redox theory. Arch Surg 1983; 118:1245-51
Noldge GFE, Priebe H, Bohle W, Buttler KJ, Geiger K: Effects of acute normovolemic hemodilution on splanchnic oxygenation and on hepatic histology and metabolism in anesthetized pigs. Anesthesiology 1991; 74:908-18
Noldge GFE, Priebe H, Geiger K: Splanchnic hemodynamics and oxygen supply during acute normovolemic hemodilution alone and with isoflurane-induced hypotension in the anesthetized pig. Anesth Analg 1992; 75:660-74
Corry WD, Jackson LJ, Seaman GVF: Action of hydroxyethyl starch on the flow properties of human erythrocyte suspensions. Biorheology 1983; 20:705-17
Cochran Injuries Group Albumin Reviewers: Human albumin administration in critically ill patients: Systematic review of randomized controlled trials. BMJ 1998; 317:235-40