INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

SEPSIS IS A CLINICAL SYNDROME distinguished by systemic inflammation, profound cardiovascular alterations and widespread tissue injury. Microcirculatory dysfunction is recognized to be a significant component of the ubiquitous injury in sepsis (19). An increased release of many mediators, including nitric oxide (NO), a short-life effector molecule involved in the regulation of vascular tone and blood flow distribution, has been largely implicated in this process (43). Hypotension complicating excessive vasodilation, or myocardial depression, depresses microcirculatory red blood cell (RBC) flow. Endothelial alterations, interstitial edema, leukocyte activation, and RBC entrapment can worsen these microcirculatory abnormalities. The consequence of these abnormalities is a decrease in the perfused capillary density (19), creating conditions that may limit oxygen availability to the cells. Furthermore, several studies (9, 15, 23, 26) have demonstrated an increased stiffness of RBC in sepsis, which is caused by several mechanisms including oxidation by oxygen free radicals (21, 27), ATP depletion, and increase in intracellular calcium (30), thus rendering the erythrocytes less deformable for penetrating the microcirculation (20, 41).

Hemoglobin solutions were initially developed with the hope of finding an alternative to the problems associated with blood transfusion, including time-consuming and expensive cross-matching, limited shelf life and supplies, and disease transmission. Nevertheless, a new, broader concept of "hemoglobin therapeutics" developed with the realization that hemoglobin solutions are not only RBC substitutes but also have a number of additional properties. One of these is that hemoglobin solutions may penetrate the microcirculation more easily than the RBC due to their small particle size and low viscosity. Their slightly right-shifted O₂ dissociation curve, compared with human blood, may also increase oxygen unloading. Other important properties of hemoglobin solutions are their specific vasopressor effects, mostly related to NO-binding properties (10, 35, 37-39) and their high colloid osmotic pressure (24). Several studies have indicated that resuscitation fluids containing hemoglobin improve oxygen transport to the tissues compared with nonhemoglobin solutions (14, 18). Diaspirin crosslinked hemoglobin (DCLHb), derived from old human erythrocytes, is one of these modified hemoglobin solutions. DCLHb is prepared by cross-linking between the -subunits of hemoglobin within the hemoglobin tetramer by means of a reaction with the diaspirin compound, bis(3,5-dibromosalicyl)fumarate and then purifying the solution by heat pasteurization to inactivate any contaminating viruses.

We sought to define the effects of two different doses of DCLHb in an endotoxic shock model in the anesthetized dog. In the first part of the study, we studied the acute hemodynamic effects of DCLHb during the hyperdynamic phase of endotoxic shock. In the second phase, we tested the hypothesis that the hemoglobin solution would improve oxygen extraction capabilities when blood flow was acutely reduced by cardiac tamponade.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Surgical preparation. Eighteen mongrel dogs (24 ± 5 kg) were anesthetized with pentobarbital sodium at an initial intravenous dose of 30 mg/kg body wt, followed by a continuous infusion of 4 mg · kg¹ · h¹ (Infusomat II, Melsungen, Germany), through the left forepaw vein. After endotracheal intubation with a cuffed endotracheal tube, the dog was mechanically ventilated with room air using a Servo ventilator 900B (Siemens-Elema, Solna, Sweden). Controlled ventilation was facilitated with pancuronium bromide, given as an initial bolus of 0.15 mg/kg, followed by an infusion of 0.075 mg · kg¹ · h¹. Respiratory rate was set at 12 breaths/min, and tidal volume was adjusted to obtain an end tidal CO₂ (PET_CO2) between 28 and 34 Torr. These ventilatory conditions were not changed thereafter. A right femoral arterial catheter was inserted and connected to a pressor transducer for arterial pressure monitoring. The right forepaw vein was cannulated for normal saline and DCLHb infusion. A balloon-tipped pulmonary artery catheter (Swan-Ganz model 93A-131-7-Fr, Baxter; Irvine, CA) was inserted through the right external jugular vein under guidance of pressure waves, as determined from a four-channel monitor (Sirecust 302A, Siemens, Erlangen, Germany). A left thoracotomy between the fourth and the fifth intercostal space was performed with bleeding controlled by electrocautery. A 16-gauge polyethylene catheter (Intracath, Deseret Medical, Sandy, UT) with multiple side holes around the tip was positioned in the pericardial space via a 2- to 3-mm incision in the anterior pericardium, with its tip adjacent to the diaphragmatic surface of the left ventricle. The catheter was secured with purse-string sutures; 30 ml of warm sterile saline was injected into the pericardial cavity to ensure the absence of leakage and then drained before sealing. The thoracic cavity was then carefully closed in three layers, and a chest tube (Argyle, Trocar catheter A75, 28Ch-40 cm, Sherwood, Tullamore, Ireland) was placed through the seventh intercostal space to allow gentle evacuation of the chest. A splenectomy was performed through a midline laparotomy after maximal splenic contraction to 1 mg epinephrine (spread on the surface of the spleen) to prevent autotransfusion of erythrocytes during hypotension. Ultrasonic flow probes were placed around the common hepatic artery (3-4 mm), the portal vein (10-12 mm), and the left renal artery (3-4 mm) for simultaneous measurement of blood flow in these vessels. In each dog, a multipurpose catheter (5-Fr, Cook, Bjaerskov, Denmark) was inserted via the right jugular vein into the superior hepatic vein. Good position was confirmed by direct hand-feeling on the hepatic vein. A 16-gauge, 20-cm intravenous catheter (Argyle, Intramedicut, Sherwood, Tullamore, Ireland) was inserted via the splenic vein into the portal vein.

Hemoglobin solution. DCLHb (10 g/dl) was prepared by Baxter Healthcare. The solution was electrolyte-balanced (135 meq/l Na, 4 meq/l K, and 110 meq/l Cl) and had the following properties: pH 7.40 at 37°C, oncotic pressure of ~43 mmHg, oxygen affinity (P₅₀) of 32 mmHg at 37°C and pH 7.4, viscosity <1.5 centistokes at 37°C, and methemoglobinemia <5%.

Experimental protocol. After surgical preparation, the dog was placed in the supine position and allowed to stabilize for 30 min. The pericardial cavity was emptied using a 5-ml syringe to ensure a slightly negative intrapericardial pressure before control measurements (B1) were obtained. The animals then received a slow intravenous bolus of 2 mg/kg E. coli endotoxin (055:B5, control no. 3120-10-7, Difco, Detroit, MI), and a second set of measurements (B2) was obtained 30 min later. A normal saline infusion was then started and titrated to restore pulmonary occlusion pressure to baseline. A third set of measurements was obtained after 30 min (B3). The saline infusion was then kept at a constant rate of 20 mg · kg¹ · h¹ throughout the study. The dogs were randomly divided into three groups: endotoxin alone (n = 6), endotoxin and DCLHb at a dose of 500 mg/kg (n = 6), and endotoxin and DCLHb at a dose of 1,000 mg/kg (n = 6). In the two latter groups, DCLHb was infused 10 min after the B3 measurements. A fourth set of measurements (B4) was obtained 30 min later. Cardiac tamponade was then induced by repeated injections of normal saline, heated to 37°C, into the pericardial sac. Measurements were repeated every 20 min thereafter in all animals. When mean arterial pressure had declined to 20% of the baseline level, the dog was considered to be in a decompensatory state, the data collection was ended, and the dog was killed. A timeline describing the experiment is shown in Fig. 1.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Schematic representation of the experimental protocol. DCLHb, diaspirin cross-linked hemoglobin; IPP, intrapericardial pressure; B1, baseline; B2, 30 min after endotoxin; B3, 30 min after fluid resuscitation; B4, 30 min after administration of DCLHb in the treated groups.

Measurements and calculations. All pressures were determined from a strip-chart recorder (2600S recorder; Gould, Cleveland, OH) at end expiration. Cardiac index (ml · min¹ · kg¹) was measured by the thermodilution technique (cardiac output computer, COM-2, Baxter) using three to five 5-ml injections of D 5% in iced water. Each injection was started at end inspiration. A temperature probe was used online to control for variations in injectate temperature. Regional blood flow was estimated simultaneously in the common hepatic artery, portal vein, and left renal artery by a previously calibrated blood flowmeter (model T208, Transsonic systems, Ithaca, NY).

Statistics. In each animal, the determination of the whole body and liver critical O₂ (O_2 crit) was obtained from a plot of O₂ vs. O₂ using the method described by Samsel and Schumaker (33). O_2 crit was defined as the point of intersection of two best-fit regression lines, as determined by a least sum of squares technique. Paired sets of linear regressions were calculated for all possible combinations of points separated into low (supply-dependent) and high (supply-independent) O₂ groups. Points were constrained to fall on either regression line but not on both. The pair of regressions with the lowest sum of standard errors of estimate was taken as the set that best fit the data. The values of O₂ and O₂ at the intersection point were then calculated using the two regression equations and were called O_2 crit and O_2 crit, respectively. Because O₂ and O₂ were derived from independent techniques of measurement, ER_O2 at critical point (ER_O2 crit) was calculated by dividing O_2 crit by O_2 crit. An example is shown in Fig. 2. Statistical analysis included a repeated measurements ANOVA followed by Dunett's test. The difference in the slopes of O₂/O₂ were tested by ANCOVA. P < 0.05 was considered statistically significant. All values are expressed as mean ± SD.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Relation between whole body O2 uptake (O2) and whole body O2 delivery (O2) in a control dog (number 6). Lines are dual-regression lines, and the point of intersection of these lines defined the whole body critical oxygen delivery (O2 crit) and the corresponding whole body critical oxygen uptake (O2 crit). ERO2 crit, critical O2 extraction ratio; r2, correlation coefficient.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effects of endotoxin alone. Endotoxin administration resulted in sharp decreases in mean arterial pressure, cardiac filling pressures, cardiac index, and whole body O₂ (Fig. 3). Blood lactate levels increased (Fig. 4). There was no significant difference between the three groups in the total amount of intravenous fluids required (3.9 ± 0.8 liters in the control group vs. 3.5 ± 0.7 and 3.4 ± 0.8 liters in the DCLHb groups).

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Changes in mean arterial pressure (MAP) (A), mean pulmonary arterial pressure (MPAP) (B), cardiac index (CI) (C), and O2 (D) in relation to incremental changes in IPP in 3 groups of animals. Values are means ± SD.

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Changes in arterial lactate concentration in relation to incremental changes in IPP in the 3 groups. Values are means ± SD.

Effects of DCLHb. DCLHb infusion was followed by a dose-dependent increase in mean arterial pressure and mean pulmonary arterial pressure but no significant change in cardiac index or O₂ (Fig. 3). Both systemic and pulmonary vascular resistance indexes increased (Table 1). Relative portal, hepatic artery, and renal blood flows were not affected by the infusion of DCLHb (Fig. 5). DCLHb infusion had no influence on the blood hemoglobin concentration, which somewhat decreased in the three groups throughout the study (Table 2). During cardiac tamponade, the administration of DCLHb was followed by a significant decrease in whole body O_2 crit (12.8 ± 0.7, 8.6 ± 0.9, and 8.2 ± 0.7 ml · kg¹ · min¹ in the control and low- and high-dose DCLHb groups, respectively, P < 0.05 vs. control group for both DCLHb groups, Fig. 6). In the absence of significant differences in whole body O_2 CRIT, whole body ER_O2 crit was significantly higher in the two DCLHb groups (58.7 ± 12.8 and 60.2 ± 9.0% in the low- and high-dose DCLHb-treated groups, respectively) than in the control group (39.1 ± 3.1%; Fig. 6). Liver O_2 crit was slightly lower in the DCLHb treated groups than in the control group, but these differences did not reach statistical significance (Fig. 6). During cardiac tamponade, the DCLHb-treated groups maintained significantly lower arterial lactate levels than the control group (Fig. 4). Mean arterial pressure at whole body O_2 crit was quite similar in the three groups (44 ± 6, 46 ± 8, and 47 ± 4 mmHg, in the control and the two DCLHb-treated groups, respectively).

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Changes in regional blood flow during incremental changes in IPP in the 3 groups of animals. portal, portal vein blood flow; hepart, hepatic artery blood flow; ren, renal artery blood flow. Values are means ± SD.

View larger version (27K): [in this window] [in a new window]   Fig. 6.   Individual and mean whole body (top) and liver (bottom) O2 crit, O2 crit, and ERO2 crit in the three groups of dogs.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main findings of our study are that DCLHb infusion in endotoxic shock can exert a vasopressor effect without affecting cardiac output and regional blood flow and can increase oxygen extraction capabilities.

Many studies have reported a rise in blood pressure after administration of hemoglobin solutions in animals (2, 13, 18, 22, 29, 37-39, 44) and in humans (14, 17, 28). These vasopressor effects are primarily due to the scavenging of NO by hemoglobin (8, 12, 32, 35, 39, 45). Sharma et al. (39) demonstrated, in anesthetized rats, that the administration of the NO precursor L-arginine significantly attenuated the systemic hemodynamic effects of DCLHb. They also showed that DCLHb, when administered in N-nitro-L-arginine methyl ester (L-NAME, a NOS inhibitor)-pretreated rats, accentuated the decrease in blood flow to the gastrointestinal system, spleen, mesentery and pancreas, skin, and musculoskeletal system (9). Furthermore, plasma cGMP concentrations decreased after the administration of both DCLHb and L-NAME. In addition, Rooney et al. (32) showed, in dogs hemodiluted with oxyhemoglobin, that the administration of the NO donor sodium nitroprusside could reverse the vasoconstriction and increase the cardiac output. Ulatowski et al. (45) demonstrated that a cross-linked hemoglobin solution selectively reduced blood flow to the kidneys and intestine in cats, in a manner similar to that seen with NO synthase inhibition. Dietz et al. (6) showed that partial exchange transfusion with - cross-linked hemoglobin in anesthetized dogs resulted in a reduction in the NO-mediated vasodilator responses to acetylcholine and sodium nitroprusside in coronary arteries.

Besides the NO scavenging effect, the vasoconstrictor effects of the hemoglobin solutions may also involve endothelin (ET) (3, 11, 35, 38), a peptide produced by endothelial cells that has direct and potent endogenous vasoconstrictor effects on vascular smooth muscle. Schultz et al. (35) showed that pretreatment with phosphoramidon, a blocker of ET production, attenuated the pressor effect of DCLHb. However, Gulati et al. (11) reported that phosphoramidon did not attenuate DCLHb-induced cardiovascular effects in anesthetized rats, suggesting that the cardiovascular effects of DCLHb are not due to an increase in the conversion of pro-ET to ET. These authors proposed that ET is involved in the vasopressor effects of DCLHb by an effect on its receptor, because an ET -receptor antagonist could attenuate the systemic and regional effects of DCLHb and stroma-reduced hemoglobin (38).

Interestingly, in our study, despite a marked vasopressor effect, the administration of DCLHb did not decrease cardiac output or regional blood flow. This may be related to the concomitant, generous fluid resuscitation. Another explanation could have been the increase in intravascular volume due to the oncotic effects of DCLHb. Nevertheless, the absence of a significant increase in cardiac filling pressures after DCLHb infusion does not support this assumption. We have also extrapolated the intravascular volume changes after DCLHb infusion by comparing the predicted hemoglobin concentration after DCLHb infusion (taking into account the baseline hemoglobin level, the blood volume of the dog, the infused volume of DCLHb, and its hemoglobin concentration) and the measured hemoglobin concentration after DCLHb infusion. Using this approach, we found that the blood volumes of the dogs after DCLHb infusion were virtually unchanged (98 ± 5 and 96 ± 3% of baseline blood volumes for the DCLHb 500 mg/kg- and DCLHb 1,000 mg/kg-treated groups, respectively). The lack of intravascular volume expansion after DCLHb infusion in our model might be the capillary leak phenomenon secondary to the endotoxin infusion which could minimize the oncotic effects of DCLHb by extravascular leak of a certain quantity of DCLHb molecules. Finally, it is not surprising that the hemoglobin level did not increase after DCLHb infusion because the concentration of the infused DCLHb was lower than the hemoglobin blood dog concentration before DCLHb infusion. In nonseptic rats, Sharma et al. (37, 39) and Gulati and colleagues (10) showed that the infusion of DCLHb was followed by an increase in cardiac output and that regional blood flow measured by a radioactive microsphere technique was either increased or not affected. In septic rats, Mourelatos et al. (25) showed that DCLHb immediately increased blood pressure but did not affect cardiac output or regional perfusion; in fact, after 24 h, DCLHb even increased regional perfusion to the brain, heart, and stomach,but not to the liver and the kidney. On the contrary, in septic cats, Ulatowski et al. (45) showed that the infusion of another cross-linked hemoglobin (bovine fumaryl bb-cross-linked hemoglobin) selectively reduced blood flow in the intestines and kidneys by a mechanism consistent with NO scavenging. In a swine model of endotoxic shock, Aranow et al. (2) reported that the infusion of cross-linked human hemoglobin did not impair renal or mesenteric blood flows but did decrease gut mucosal perfusion, when compared with a dextran-treated group. These differences can probably be explained by the different experimental models used. Our data suggest that DCLHb infusion does not alter cardiac output and regional blood flow, provided that generous fluid administration is simultaneously given. The use of pentobarbital sodium as anesthetic agent in our dogs may also have minimized the vasopressor effect of DCLHb because pentobarbital sodium has been demonstrated to attenuate systemic vascular responses to multiple vasopressors (1, 7).

In septic rats, Sielenkämper et al. (40) demonstrated a beneficial effect of DCLHb on tissue oxygen utilization. When oxygen supply dependency was induced by progressive hemodilution, these authors observed that DCLHb infusion increased oxygen uptake and reversed lactic acidosis. In our study, when cardiac tamponade dramatically reduced cardiac output and arterial pressure, DCLHb was also able to improve cellular oxygen availability, as reflected by the lower arterial lactate concentrations found in the DCLHb-treated groups during cardiac tamponade. Because DCLHb did not alter cardiac output and hemoglobin concentration, calculated whole body O₂ was similar in the three groups of dogs throughout the study. Therefore, the improvement in cellular oxygen availability had to be due to an improvement in oxygen extraction capabilities and/or local microvascular perfusion, as indicated by the lower whole body O_2 crit and higher ER_O2 critfound in the DCLHb groups. Hemoglobin solutions may influence O₂ transport in the microcirculation by several mechanisms, including interaction with NO, colloid osmotic effects, and a right-shifted O₂ dissociation curve.

In sepsis (19), systemic hypotension and/or myocardial depression, coupled to local microcirculatory alterations (including interstitial edema, endothelial alterations, and leukocyte or RBC entrapment), depress microcirculatory RBC flow. DCLHb binds NO, an endogenous mediator that contributes to regulating vascular tone (16). DCLHb may have acted primarily on the microcirculation through increasing the tissue perfusion pressure or by scavenging NO. The similar arterial pressure at O_2 crit in the three groups rules out the vasopressor effects as an important mechanism. The NO scavenging effect was also unlikely to play an important role. On the same animal model of endotoxic shock, our laboratory previously demonstrated that the infusion of N-monomethyl-L-arginine, a NO synthase inhibitor, did not improve oxygen extraction capabilities (46). On the contrary, the administration of the NO donor 3-morpholinosydnonimine increased oxygen extraction capabilities. Thus we believe that neither the increase in tissue perfusion pressure nor the NO scavenging effects were implicated in the improvement in oxygen extraction capabilities. Finally, the absence of blood flow redistribution rules out the hypothesis that the improvement in whole body oxygen extraction capabilities could have been due to a possible interorgan blood flow redistribution towards more hypoxic tissues.

The rightward shift of the O₂ dissociation curve of DCLHb (higher P₅₀), compared with human blood (4, 40) may facilitate oxygen offloading in the microcirculation and, therefore, improve tissue oxygenation (40, 42). However, there is no clear evidence, as yet, that right-shifting the O₂ dissociation curve is important to restoring tissue oxygenation. Using a low flow in a pump-perfused left hindlimb dog model, Curtis et al. (5) showed that raising hemoglobin P₅₀ increases tissue skeletal muscle PO₂ but does not affect the ER_O2 crit. Schumaker et al. (36) showed that, in canine steady-state exercise, hemoglobin P₅₀ is not an important determinant of tissue oxygen extraction capacity during normoxia or moderate hypoxia. Finally, we do not think that, in our study, the DCLHb P₅₀ was an important determinant in the improvement in oxygen extraction capablities because the value of dog blood P₅₀, which is between 28.8 and 33.1 mmHg (31, 36), is close to the DCLHb P₅₀, which is ~32 mmHg.

Therefore, we suppose that DCLHb improved tissue oxygenation by its rheological properties. Due to their small particle size and low viscosity, DCLHb molecules penetrate microvessels better than RBC, which must navigate through capillaries with small luminal diameters. These advantageous rheological properties may be particularly important in sepsis, in which both microcirculatory alterations and alterations in RBC stiffness (15) can hinder RBC microcirculatory penetration.

In summary, with its rheological properties, oxygen carrying abilities, and vascular effects, DCLHb may be an interesting option to increase cellular oxygen availability in septic shock.