The choice of the intravenous fluid influences the tolerance of acute normovolemic anemia in anesthetized domestic pigs

Introduction The correction of hypovolemia with acellular fluids results in acute normovolemic anemia. Whether the choice of the infusion fluid has an impact on the maintenance of oxygen (O2) supply during acute normovolemic anemia has not been investigated so far. Methods Thirty-six anesthetized and mechanically ventilated pigs were hemodiluted to their physiological limit of anemia tolerance, reflected by the individual critical hemoglobin concentration (Hbcrit). Hbcrit was defined as the Hb-concentration corresponding with the onset of supply-dependency of total body O2-consumption (VO2). The hemodilution protocol was randomly performed with either tetrastarch (6% HES 130/0.4, TS-group, n = 9), gelatin (3.5% urea-crosslinked polygeline, GEL-group, n = 9), hetastarch (6% HES 450/0.7, HS-group, n = 9) or Ringer's solution (RS-group, n = 9). The primary endpoint was the dimension of Hbcrit, secondary endpoints were parameters of central hemodynamics, O2 transport and tissue oxygenation. Results In each animal, normovolemia was maintained throughout the protocol. Hbcrit was met at 3.7 ± 0.6 g/dl (RS), 3.0 ± 0.6 g/dl (HS P < 0.05 vs. RS), 2.7 ± 0.6 g/dl (GEL, P < 0.05 vs. RS) and 2.1 ± 0.4 g/dl (TS, P < 0.05 vs. GEL, HS and RS). Hemodilution with RS resulted in a significant increase of extravascular lung water index (EVLWI) and a decrease of arterial oxygen partial pressure (paO2), and O2 extraction ratio was increased, when animals of the TS-, GEL- and HS-groups met their individual Hbcrit. Conclusions The choice of the intravenous fluid has an impact on the tolerance of acute normovolemic anemia induced by acellular volume replacement. Third-generation tetrastarch preparations (e.g., HES 130/0.4) appear most advantageous regarding maintenance of tissue oxygenation during progressive anemia. The underlying mechanism includes a lower degree of extravasation and favourable effects on microcirculatory function.


Introduction
The correction of hypovolemia is an essential goal in the treatment of critically ill patients. However, the use of acellular fluids (that is, crystalloid or colloidal solutions) results in the dilution of the circulating cell mass (acute normovolemic anemia) with a corresponding decrease of O 2 transport capacity [1][2][3].
Acute normovolemic anemia is initially compensated by increases of cardiac output and arterio-venous O 2 extraction [4]. Moreover, O 2 supply (DO 2 ) physiologically exceeds O 2 demand (reflected by total body O 2 consumption (VO 2 ) under resting conditions) by the factor three to four. When DO 2 begins to decrease at lower Hb concentrations, the total body O 2 demand is still met, and VO 2 remains constant despite decreasing Hb concentrations (O 2 supply-independency of VO 2 ). When DO 2 falls below a critical value, the amount of O 2 delivered to the tissues becomes insufficient to meet their O 2 demand and VO 2 starts to decline (O 2 supplydependency of VO 2 ) [5]. The corresponding critical hemoglobin concentration (Hb crit ) represents the ultimate limit of the individual anemia tolerance.
Among clinicians all over the world, infusion practice varies considerably and different intravenous (IV) fluids are used for volume replacement [6]. However, currently available IV fluids differ in their pharmacodynamic and kinetic profile (for example, molecular weight, plasma half-life, volume expansion effect) and in their effects on rheology and microcirculatory function.

Materials and methods
After approval by the local governmental review board (Regional Council Darmstadt, department for veterinary affairs/V54), experiments were performed in 36 healthy farm-bred pigs of either sex (body weight 24.0 ± 3.7 kg). All animals received good care in compliance with the Guide for the Care and Use of Laboratory Animals.
Animals were orotracheally intubated and ventilated with ambient air at a rate of 14 cycles/min and a positive end-expiratory pressure of 5 cmH 2 O (Servo 900B, SiemensElema, Solna, Sweden). Tidal volume was individually adjusted to provide arterial normocapnia and was then maintained throughout the entire protocol.

Instrumentation and monitoring
Animals were placed in the supine position and a fivelead electrocardiogram (II, V5) was installed for detection of arrhythmias and ST-segment changes. A double-lumen catheter (Arrow, Reading, PA, USA) was inserted into the cranial vena cava and a Swan-Ganz-Catheter (Baxter, Irvine, CA, USA) was floated into a branch of the pulmonary artery. Each one 6F introducer sheath (Arrow, Reading, PA, USA) was inserted into the right femoral vein and artery, respectively. For continuous measurement of arterial blood pressure and cardiac output, a thermodilution catheter was placed into the left femoral artery (Pulsion Medical Systems, Munich, Germany). Body core temperature was kept constant using a warming pad.
Acute normovolemic anemia was induced by withdrawal of blood (1mL/kg/min) and simultaneous infusion of one of the IV fluids. To maintain normovolemia during the hemodilution protocol, the following infusion rates were chosen with regard to the different plasma expansion effects of the fluids investigated (see Table 1): Ringer's solution 3 mL/kg/min, gelatin 1.2 mL/kg/min, tetrastarch and hetastarch 1 mL/kg/min, respectively. Infusion fluids were warmed in an immersion bath until IV infusion. For precise synchronization of blood withdrawal with the corresponding infusion rates, a bidirectional precision pump (Harvard Apparatus, Holliston, MA, USA) was used.
The target parameter of the hemodilution protocol was the animal's individual critical Hb concentration (Hb crit ), which was defined as the Hb concentration corresponding with a critical limitation of O 2 delivery to the tissues (DO 2 ) and the onset of O 2 supply-dependency of total body O 2 consumption-VO 2 (see below). At the end of the hemodilution protocol, animals were killed by intracardiac injection of saturated potassium chloride solution.

Measurements
Intravascular blood volume was determined at baseline using the "whole-blood" method of the indocyanin green (ICG) indicator dilution technique, which has already been described in detail elsewhere [7]. The pressure transducers of the cardiovascular catheters were connected with a multi-channel recorder and a personal computer (Hugo-Sachs, March-Hugstetten, Germany) for continuous measurement of hemodynamic parameters. Cardiac output was continuously measured using the pulse contour analysis (PICCO classic monitor, Pulsion Medical Systems, Munich, Germany). The volumetric preload and extravasation parameters, intrathoracic blood volume (ITBV) and extravascular lung water (EVLW), were assessed after exchange of each 10% of the circulating blood volume using the transpulmonary thermodilution technique. Arterial and mixed venous blood samples were withdrawn at baseline, after exchange of each 10% of the circulating blood volume and at Hb crit for blood gas analysis and assessment of Hb concentration (GEM-3000 and 682 CO-Oxymeter, Instrumentation Laboratory, Lexington, MA, USA).

Calculated parameters
Body surface area (BSA) was calculated according to Holt [8], from body weight (BW) and a species-dependent constant: where k = 9 for the species pig.
The measured values of cardiac output (CO) and O 2 consumption (VO 2 ) were indexed to BSA: Systemic and pulmonary vascular resistance indices (SVRI and PVRI) were calculated from standard formulae: where MAP = mean arterial pressure, CVP = central venous pressure, CI = cardiac index, MPAP = mean pulmonary arterial pressure and PCWP = pulmonary capillary wedge pressure.

Determination of Hb crit
Hb crit is the correlate of the critical limitation of DO 2 , and marks the onset of total body O 2 supply-dependency. The corresponding decrease of VO 2 was prospectively detected in an automated and investigator-independent manner: VO 2 was measured every 60 sec with a metabolic monitor (Delta-Trac II™ MBM-200, Datex-Engstrom, Helsinki, Finland). VO 2 values were simultaneously recorded and computed with specific software (Delta-Crit-System, DCS) [9]. During the stabilization period, the DCS enters VO 2 values into an online regression analysis and calculates the mean and standard deviation (SD). During the subsequent hemodilution period, the VO 2 value obtained every 60 secis compared to the mean value predicted by the DCS. When three consecutive VO 2 values are outside the predefined range (3-fold SD of the regression line), a significant decrease of VO 2 is assumed (see Figure 1) and the computer alerts visually and acoustically [9]. For determination of Hb crit , the corresponding Hb concentration was measured in the arterial blood sample.

Statistics
Statistical analysis was performed with the SAS 9.1 software package (SAS Institute, Cary, NC, USA). All data are presented as mean ± SD. Distribution of data was assessed with the Shapiro-Wilk test. In the case of normal distribution, the time effect on the different variables, and differences between groups at different time points were tested by repeated analysis of variance (ANOVA). Post hoc analysis of differences detected with ANOVA was performed with the Student Newman Keuls test. In the case of non-normal distribution, the time effect on the parameters, and the between-group differences were tested by analysis of variance on ranks (rANOVA). Post hoc analysis of differences detected with rANOVA was performed with Tukey's test. For all parameters, statistical significance was accepted at P < 0.05.

Baseline characteristics
On analysis of baseline body weight, blood volume index and all investigated parameters of hemodynamics and O 2 transport, there were no significant differences detected between the groups (see Tables 2 and 3).

Secondary endpoints: hemodynamic and O 2 -derived parameters
In Tables 2 and 3, hemodynamic and O 2 -derived parameters are presented at the respective critical Hb concentrations, that is, after hemodilution to 3.7 g/dL, 3.0 g/dL, 2.7 g/dL and 2.1 g/dL, respectively.
Hemodilution to Hb 3.7 g/dL (Hb crit of the RS-group) As compensation for hemodilution, cardiac index (CI) increased by 46 to 60% in all groups (P < 0.05 vs. baseline). Consistently, stroke volume index (SVI) and heart rate (HR) increased by 24 to 40% and 14 to 18%, respectively. While the increases in SVI were significant in the TS-group, GEL-, and RS-groups, the increase in HR was not significant in any group. Mean arterial pressure (MAP) decreased by 11 to 21% in the GEL-, HS-and RS-groups (non-significant) and systemic vascular resistance index (SVRI) decreased in all groups by 35 to 46% (P < 0.05 vs. baseline), while mean pulmonary arterial pressure (MPAP) increased by 22 to 50% (P < 0.05 vs. baseline in the GEL-, HS and RS-group). Stroke volume variation (SVV) remained unchanged in the RS-group and tended to decrease in the TS-and GEL-groups. In the HS-group, the decrease in SVV was statistically significant. Compared with baseline, ITBV index (ITBVI) increased in all groups (P < 0.05), reflecting increased venous return to the heart. The ITBVI was higher in animals hemodiluted with TS or HS than after fluid replacement with RS (P < 0.05). As a consequence of hemodilution, DO 2 I decreased by 14 to 30% in all groups (P < 0.05 vs. baseline). Moreover, the pH-value and base excess (BE) decreased significantly in the GEL-, HS-and RS-groups. As the protocol was terminated at A linear regression analysis including the calculation of SD was performed with VO 2 values collected during the 60-minute stabilization period. During the subsequent hemodilution protocol, a critical limitation of oxygen delivery (DO 2 ) was assumed, when three consecutive VO 2 values fell below the lower 3σ-range.  the onset of the critical DO 2 , lactate concentration was not yet elevated at this time point (see Table 3). The EVLW index (EVLWI) was significantly increased in the RS-group (P < 0.05 vs. baseline and P < 0.05 vs. GEL-, HS-and TS-groups, see Figure 3). Consistently, arterial oxygen partial pressure (paO 2 ) was decreased in the RS-group (P < 0.05 vs. baseline and P < 0.05 vs. GEL-, HS-and TS-groups).
Hemodilution to Hb 3.0 g/dL (Hb crit of the HS-group) As the limit of anemia tolerance was already met at Hb 3.7 g/dL in the RS-group, hemodilution to 3.0 g/dL could only be performed in the HS-, GEL-and TS-groups. A slight increase in CI, HR and SVI and a decrease in MAP and SVRI was observed in these groups, and SVV was significantly reduced compared with baseline. The ITBVI did not change further, but remained elevated above the baseline level (P < 0.05). There were no differences between the HS-, GEL and TS-groups in hemodynamic parameters at Hb 3.0 g/dL.
Consistent with progressive hemodilution, DO 2 I further decreased in the remaining HS-, GEL-and TSgroups. In animals hemodiluted with hetastarch, O 2 extraction was significantly increased as reflected by lower partial venous oxygen pressure (pvO 2 ) (P < 0.05 vs. baseline and vs. GEL-group) and by a higher O 2 extraction ratio (P < 0.05 vs. baseline and vs. GEL-and TS-groups).
Hemodilution to Hb 2.7 g/dL (that is, Hb crit of the GELgroup) Hemodilution could be continued to Hb 2.7 g/dL in the GEL-and TS-groups. CI did not increase further, but was now maintained by an elevation of HR in both groups (P < 0.05 vs. baseline). The MAP and SVRI decreased slightly (non-significant vs. baseline). In both groups, the ITBVI was still higher than at baseline (P < 0.05). However, at Hb 2.7 g/dL, the ITBVI was higher and SVV was lower in the TS-group (P < 0.05 vs. GELgroup). In the GEL-group, the O 2 extraction ratio (O 2 -ER) and EVLWI were elevated when compared with baseline (P < 0.05, see Figure 3). No further differences between the TS-and GEL-groups were observed in hemodynamic or O 2 -derived parameters at Hb 2.7 g/dL.

Discussion
The main findings of the present study are that during acute normovolemic anemia, 1) the choice of the IV fluid has an impact on the maintenance of tissue  oxygenation as reflected by variation in the extent of anemia tolerance; 2) anemia tolerance is higher when using colloids than when using crystalloids, and 3) among the colloids, tetrastarch (6% HES 130/0.4) provided higher anemia tolerance than did gelatin (3.5% urea-crosslinked polygeline) and hetastarch (6% HES 450/0.7). This was reflected by significantly lower values of Hb crit and a higher allowable volume of blood exchanged for tetrastarch. Hb crit is the hemoglobin concentration associated with a critical limitation of O 2 supply and hallmarks the ultimate tolerance limit of acute normovolemic anemia [10,11]. In our previous experimental studies, hemodilution to Hb crit was associated with 100% mortality, if no further treatment (such as elevation of fraction of inspired O 2 , transfusion of red blood cells, or infusion of artificial O 2 carriers) was initiated after institution of critical normovolemic anemia [12][13][14]. Several authors found Hb crit at values between 1.6 and 3.0 g/dL [10,12,[14][15][16][17][18][19][20][21]. In detail, Hb crit was reduced by 1) hypothermia (moderate reduction of body core temperature reduces total body O 2 demand [15]); 2) hyperoxemia (bioavailability of physically dissolved O 2 is excellent in profound anemia [12,16,17]); 3) infusion of norepinephrine (stabilization of coronary perfusion pressure during hemodilution [19]); 4) artificial O 2 carriers (maintenance of arterial oxygen content despite reduced hematocrit [18,22]), and 5) con-tinuous neuromuscular blockade (lowering skeletal muscular O 2 demand [23]).
While different infusion fluids were used for hemodilution in the aforementioned studies, their particular impact on the limit of anemia tolerance has not been fully elucidated. In a similar hemodilution study performed in anesthetized and splenectomised dogs, van der Linden and coworkers found no differences between pentastarch (6% HES 200/0.5) and a 3% gelatin preparation in relation to the value of Hb crit [20]. However, the typical increase of CI in compensation for dilutional anemia was absent in that study, which was explained by cardio-depressant effects of the anesthesia regimen employed [20,21].
In the present study, gelatin provided more extensive anemia tolerance than did hetastarch. While the occasional use of hetastarch is predominantly reported by US physicians [24], European physicians rather avoid this fluid due to its adverse effects on coagulation and renal function [25]. Nevertheless, a recent clinical study found a reduction of mortality in trauma victims resuscitated with hetastarch in addition to the advanced trauma life support (ATLS)-standard of care treatment (that is, crystalloids along with blood products) [26].
Hetastarch is a hyperoncotic infusion fluid with a high viscosity in vitro (see Table 1). However, the effects of these properties in vivo are still not fully understood.
While a decrease of plasma viscosity entails the increase of venous return, thereby enabling the hemodynamic compensation of acute anemia [27], recent experimental data suggest that increased plasma viscosity prevents microvascular collapse (for example, after fluid resuscitation from hemorrhagic shock) [6]. Furthermore, the effect of hetastarch on plasma viscosity is limited by its oncotic properties, as the reabsorption of interstitial fluid involves the dilution of viscogenic materials thereby lowering plasma viscosity in the long term [28].
In the present study, animals hemodiluted with hetastarch featured a significantly higher O 2 -ER at Hb crit when compared with animals of the TS-or GEL-group, indicating that macro-and microhemodynamic compensation of anemia was completely exhausted. In other words, hetastarch failed to prevent microcirculatory collapse at an earlier stage of hemodilution than did tetrastarch or gelatin. Although plasma viscosity was not assessed in the present study, it may be assumed that a dilution-related fading of viscogenic potential might have contributed to this result.
Whether fluid resuscitation should be performed with crystalloids or colloids, has been a matter of controversy for decades, and the discussion is still open [29]. Although most crystalloid infusion fluids are plasma-isotonic, they cross the capillary membranes within 20 to 30 minutes of infusion, and most of the volume infused is shifted into the interstitium. To maintain normovolemia with crystalloids, it has been recom-mended to replace an acute blood loss in a ratio of at least 1:3 [30]. However, the exclusive use of crystalloids for volume replacement results in edema formation and may thereby compromise tissue oxygenation [31]. Actually, excessive tissue hydration due to pure crystalloid volume replacement in the perioperative phase was held responsible for many postoperative complications including increased incidence of anastomotic dehiscence in abdominal surgery, postoperative vomiting and orthostatic dysregulation [32,33]. Moreover, an experimental study in pigs subjected to colon anastomosis surgery found that microcirculatory blood flow and oxygen tension in perianastomotic colon tissue were increased in animals infused with tetrastarch when compared with crystalloid fluid management [34]. This finding was explained by homogenisation of mucosal microcirculatory blood flow after infusion of tetrastarch.
In the present study, with Ringer's solution, the exchange of blood was associated with increased pulmonary edema formation (elevated EVLWI) and an impairment of pulmonary gas exchange (decreased paO 2 ). This phenomenon was not observed in animals hemodiluted with gelatin, hetastarch or tetrastarch. Although EVLWI merely reflects the degree of tissue hydration at the site of pulmonary circulation, it may be supposed that a relevant edema formation also occurred in peripheral O 2 -consuming tissues: while O 2 extraction increased consistently in animals hemodiluted with any of these colloids, O 2 extraction was not increased when animals of the RS-group met their individual Hb crit . As this finding may reflect a microcirculatory disorder related to excessive tissue hydration, we conclude that the comparatively early VO 2 -decrease in the RS-group might, in addition to the anemia-related restriction of the O 2 transport capacity, also be attributable to an edema-related impairment of O 2 uptake at the site of pulmonary and peripheral microcirculation.
A certain transcapillary filtration rate is also characteristic for gelatin preparations [35]. Their low molecular weight (30 to 40 kDa) entails a rapid passage into the interstitial space and a rapid clearance by glomerular filtration, finally reducing volume efficacy to 80% (that is, 20% extravasation rate). Consistently, EVLWI increased much later in the GEL-group than in the RS-group. Moreover, when hemodilution was continued below Hb 3.7 g/dL, O 2 extraction could be augmented, reflecting that microcirculatory function might have been maintained more adequately than in the RS-group.
The strict maintenance of normovolemia is essential for adequate hemodynamic compensation of acute anemia, that is, for the increase in cardiac output. However, one weakness of the present protocol is the verification of normovolemia at time points characterized by extreme anemia. During extreme hemodilution, the kinetic of ICG elimination is significantly altered by dilution of albumin and increased cardiac output. Therefore, a measurement of circulating blood volume at this point would yield results not comparable with the baseline measurement [7], so normovolemia was deduced from clinically assessable parameters, for example, ITBVI and SVV.
While ITBVI represents LV preload, decreases in SVV reflect adequate volume responsiveness. After hemodilution with either fluid, ITBVI increased above the baseline level, indicating that volume replacement was adequate to achieve macro-hemodynamic compensation of acute anemia (increase of venous return to the heart, augmentation of LV preload). However, the increase of ITBVI was more pronounced in the TS-and HS-groups. In the course of hemodilution, SVV decreased in the TS-, GEL-, and HSgroups. In the RS-group, SVV remained unchanged and at Hb crit , it was significantly lower than in the HS-group. Both findings reflect that volume responsiveness was most strongly expressed in the starch groups. In the RS-group, however, volume responsiveness and volume efficacy were limited by partial extravasation despite infusion of 3 mL RS per mL blood withdrawn.
Although this procedure provided adequate LV preload for hemodynamic compensation of acute anemia, it may be argued that the infusion of higher volumes of RS (4 or 5 mL per mL blood withdrawn) might have elevated intravascular volume and might have improved volume responsiveness. On the other hand, extravasation and tissue hydration would have exacerbated and might have further impaired O 2 uptake and tissue oxygenation.
Likewise, the volume effect of gelatin appeared to be limited by its short intravascular half-life and its extravasation tendency. No differences between the colloid groups were observed in LV preload or volume responsiveness until hemodilution to Hb 3.0 g/dL. However, after continuation of hemodilution to 2.7 g/dL, ITBVI tended to decrease and SVV to increase in the GELgroup, and both parameters differed significantly from the TS-group. Nevertheless, animals of the GEL-group still featured sufficient hemodynamic compensation of acute anemia (CI, SVI and ITBVI remained increased vs. baseline), which precludes that they were actually hypovolemic.
These findings indicate that 1) the infusion of 1.2 mL gelatin per mL blood withdrawn was adequate to maintain LV preload and volume responsiveness over a wide range of the hemodilution protocol; 2) in spite of this, the volume effect of gelatin could not be sustained until the end of the protocol, and 3) that extravasation and fading volume responsiveness might have compromised microvascular perfusion, which finally caused animals of the GEL-group to earlier achieve Hbcrit than the TSgroup.
We chose Hb crit as the primary endpoint of the hemodilution protocol. In our previous studies, lactate concentrations began to increase 60 to 90 minutes after institution of Hb crit [12][13][14]19]. Of note, we employed the identical hemodilution protocol in the present study but animals were killed immediately after achievement of Hb crit , so that elevated lactate concentrations could not yet be expected.
In the present experimental model, the effects of the investigated infusion fluids became apparent when the O 2 transport capacity was driven to its critical limit. Whether differences between the groups existed already at less severe degrees of anemia, is difficult to conclude.
In summary, our data suggest that the choice of the IV infusion fluid used for acellular volume replacement has an impact on the maintenance of O 2 supply. Among the fluids investigated in the present study, tetrastarch (HES 130/0.4) provided the most extensive anemia tolerance when compared with gelatin, hetastarch or Ringer's solution. While the exact underlying mechanism remains to be elucidated, extravasation and formation of interstitial edema were associated with decreased anemia tolerance, indicating that the microcirculatory effects of the fluids investigated had a major impact on tissue oxygenation.

Conclusions
During acellular treatment of an acute blood loss in lieu of allogeneic blood transfusions (for example, correction of hypovolemia, bridging an unexpected massive blood loss, or declining of transfusion for religious reasons), modern low-molecular tetrastarch preparations (such as HES 130/ 0.4) appear most suitable to maintain O 2 supply over a wide range of levels of acute normovolemic anemia.

Key messages
• The correction of hypovolemia with acellular fluids implies the dilution of the cell mass remaining in the vasculature (acute normovolemic anemia).
• Sustainment of normovolemia and nutritive microcirculatory blood flow has a major impact on the maintenance of tissue oxygenation and thereby on the tolerance of acute anemia.
• The fluids investigated in the present study (tetrastarch, gelatin, hetastarch and Ringer's Solution) differed in their volume effects and their extravasation rate and might thereby have influenced microcirculatory function.
• In profound anemia, modern low-molecular tetrastarch preparations were most advantageous to maintain normovolemia as well as tissue oxygenation.