INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

SEPSIS DISTURBS METABOLIC O₂ reserve of the circulation because the capacities to augment cardiac output, to appropriately distribute blood flows between organs, and to extract O₂ are all diminished in this syndrome (4). In peripheral tissues, this circulatory dysfunction has been proposed to explain the occurrence of supply dependency of systemic O₂ uptake at abnormally high values of systemic O₂ delivery in sepsis (32, 33). We have previously studied the generalizability of sepsis-associated circulatory dysfunction to the heart. In sheep rendered septic by peritoneal contamination, Bloos et al. (6) found the capacity to augment both coronary blood flow and myocardial O₂ extraction was depressed when acute hypoxia was used to stress the metabolic O₂ reserve of the heart. As the O₂ needs of the heart are increased by the hypermetabolic milieu imposed by sepsis, this depressed metabolic O₂ reserve in combination with changes in myocardial O₂ delivery, which are normally inconsequential, could lead to ischemia (40). In this circumstance, myocardial dysfunction complicating the consequences of local ischemia would further limit cardiac output reserve, thereby promoting injury to noncardiac circulations.

Isovolemic hemodilution is normally accompanied by a generalized circulatory compensation to maintain tissue oxygenation, including a redistribution of blood flows to the heart and brain (30, 38, 42). This redirected O₂ delivery originates from nonvital organs, for example, from the splanchnic circulation where oxygenation is then supported by increasing tissue O₂ extraction (28). In another study, Morisaki et al. (30) demonstrated that isolvolemic hemodilution limited appropriate increases in myocardial O₂ delivery at hemoglobin levels in sepsis which, in contrast, were well tolerated in nonseptic sham animals. These data may have been the consequence of the impact of sepsis on O₂ extraction reserve (33), because the ability to redistribute O₂ delivery away from the gut in septic animals was also depressed during modest anemia. Therefore, data from this experiment could not support a clinical suggestion that the red blood cell (RBC) transfusion trigger could be reduced in critically ill patients (2).

The current experiment was designed to extend previous work by Bloos et al. (6) and Morisaki et al. (30) characterizing the effects of anemia on the circulatory reserve of the septic heart in specific and the circulation in general. We hypothesized that isovolemic hemodilution to create modest anemia in mature sheep rendered septic by cecal ligation and perforation (CLP) would depress the ability of the heart to support circulatory compensation in this syndrome. We randomized study animals to either isolvolemic hemodilution which created modest anemia or RBC transfusion to maintain high-normal hemoglobin levels. We then determined both systemic and myocardial circulatory compensation to acute hypoxia, which depressed both convective O₂ delivery and increased myocardial O₂ needs. Novel findings of this experiment included 1) the ability to increase cardiac output was sufficient to maintain convective O₂ delivery during severe reductions in arterial O₂ content (Ca_O2), despite the circulatory dysfunction imposed by sepsis, and 2) anemia in this septic model increased coronary blood flow additionally to the septic insult thereby reducing coronary flow reserve and causing an intramyocardial maldistribution of blood flow, whereas RBC transfusion to maintain normal hemoglobin levels supported a superior ability to extract O₂ in the coronary circulation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Fifteen mature, male Suffolk sheep weighing 39-76 kg (average 56.7 kg) underwent instrumentation following 1 wk of acclimatization in our laboratory. On the first day of study, the study animal was anesthetized with halothane and 100% O₂ (5-6 l/min) via mask, after which the trachea was intubated and the sheep was ventilated with 100% oxygen. Through a left posterolateral thoracotomy and with the use of a sterile technique, a saline-filled Silastic catheter (0.125 in. OD, Dow Corning, Midland, MI) was inserted into the left atrium, secured, and exteriorized. The coronary sinus was retrogradely cannulated via the hemiazygos vein with similar grade tubing as previously described (6). Correct placement of this coronary sinus catheter was confirmed during surgery from the demonstration that its O₂ saturation was <30%. Using a direct cutdown technique, we then placed saline-filled Silastic catheters (0.125 in. OD) into the left femoral and carotid arteries, while the left external jugular vein was cannulated with a 8-Fr introducer (Cordis, Miami, FL).

After recovery, the sheep were placed in metabolic cages and allowed free access to food and water. Ringer lactate (4 ml · kg¹ · min¹) was administered postoperatively to maintain adequate hydration. Analgesia was provided with meperidine, 100 mg admixed with each 1,000 ml of Ringer lactate. Catheter patency was maintained by intermittent flushes with heparinized saline (1,000 U heparin/500 ml saline).

Experimental protocol. Three days after the initial surgery, a 7-Fr Swan-Ganz catheter (model 93-131; American Edwards, Santa Ana, CA) was flow directed into the pulmonary artery through the jugular vein introducer (Fig. 1). A baseline, nonseptic study was then performed with the sheep in a conscious state. Systemic arterial and pulmonary artery pressures and cardiac output were measured. Blood was simultaneously obtained from the carotid artery, central vein, and the coronary sinus to measure blood gases, hemoglobin, and lactate. After this nonseptic study, a partial omentectomy was performed under general halothane anesthesia. The cecum was then located, devascularized, and ligated, and the tip was incised. The abdominal wound was closed in two layers with 2-0 coated vicryl ties, and a sterile tracheostomy was then performed. A 39-Fr low-pressure cuffed tracheostomy tube (Shiley) was inserted into the trachea and connected to a system providing humidified and warmed air.

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Study protocol. CLP, cecal ligation and perforation; hb, hemoglobin.

Sheep were randomly allocated to either an RBC transfusion group (group T) or a hemodilution group (group H). In group H, blood was drawn from the carotid artery, and pentastarch was infused isovolemically into the external jugular vein to reach a hemoglobin level of 70 g/l. Sheep in group T received fresh packed RBCs taken from a donor sheep on the transfusion day titrated to a hemoglobin of 120 g/l. The transfusion and hemodilution interventions were completed during the first 24 h after peritoneal contamination to allow circulatory compensation to be adequately expressed before the hypoxia intervention. A pilot study demonstrated that the interval between randomization and the hypoxia study was, however, short enough to maintain the group differentiation according to different hemoglobin levels.

As previously described (5, 6, 29), CLP leads to a panperitonitis and polymicrobial bacteremia. Throughout the time between the laparotomy and the study 48 h later, pentastarch was infused to maintain left atrial pressures (LAP) at nonseptic baseline levels. Analgesia was continued as previously detailed and increased if the sheep showed discomfort.

Forty-eight hours after the CLP, sheep were connected to a semiopen system to lower the inspired O₂ concentration (FI_O2) by mixing room air with nitrogen (Bird O₂ blender, Palm Springs, CA). This system was connected to a metabolic monitor (DeltaTrac II, Datex Intrumentation, Helsinki, Finland) to directly measure systemic O₂ consumption. We repeated all measurements described for the nonseptic study. A radioactive microsphere was then injected to allow later calculation of organ blood flows, while coronary sinus blood was obtained to calculate myocardial O₂ uptake and lactate extraction. Using a polarographic O₂ monitor (model 5570, Ventronics), we subsequently reduced the FI_O2 through up to four successive stages, adjusting to achieve a similar decrease in Ca_O2 between stages. The previous study by Bloos et al. (6) demonstrated the lowest FI_O2 that could be tolerated in this unanesthetized model approximated 0.1. Hemoglobin and arterial O₂ saturations were measured at and in between each stage to calculate the Ca_O2. Twenty minutes of equilibration time was allowed at each experimental level before measurements were repeated, as described in the 48-h baseline study, including the infusion of another set of radioactive microspheres. After the final stage of hypoxia, the study animal was euthanized with pentobarbital, and organs of the animal were then harvested for gamma counting to calculate organ blood flows.

The study protocol was approved by the University Council on Animal Care in accordance with the guidelines set down by the Canadian Council on Animal Care. During the experiment, all animal care was provided by a physician and qualified animal health technicians. All surgery was performed in a surgical suite certified for animal surgery.

Specific measurements and calculations. Systemic and pulmonary pressures were recorded with a two-channel monitor (model 78353A, Hewlett-Packard) and were referenced to the sheep's left atrium. Cardiac outputs were measured in triplicate by the thermodilution technique using a cardiac output computer (model 9570A, Hewlett-Packard). The cardiac output was indexed to the body surface area, and heart work was estimated as the product of cardiac index and mean aortic pressure (1). Blood gas samples were stored on ice before analysis with an ABL-3 blood gas analyzer (Radiometer, Copenhagen, Denmark).

During the experiment, the hemoglobin and the O₂ saturations were measured by a co-oximeter (OSM-II Hemoximeter, Radiometer, Copenhagen, Denmark). Subsequently, hemoglobin levels were confirmed by a Coulter Cell Counter (model 5, Burlington, Ontario, Canada). Lactate was measured by a Greiner G-400 Chemistry Analyzer (Switzerland).

Measurement of organ blood flows. As previously described in this sheep model (5, 6, 29), the microsphere technique was used to quantitate organ blood flows through the different stages of the experimental protocol. We used latex spheres with a diameter of 15 µm labeled with either ⁴⁶Sc, ⁵⁹Zn, ⁸⁵Sr, ⁹⁵Nb, or ¹⁴¹Ce obtained from New England Nuclear (DuPont Canada, Mississauga, Ontario, Canada). After the spheres were mixed for 5 min with a Vortex Mixer (model 58223, Scientific Products, Evanston, IL), an amount equivalent to ~25-30 mCi was injected into the left atrium. With the use of an infusion/withdrawal pump (Harvard Apparatus), sampling of the reference blood from the carotid artery and the femoral artery (10 ml/min) was started during injection of the spheres and was continued for 90 s after the injection.

After the animal was killed, the heart was obtained; the left and right atrium were removed from the heart and discarded. The ventricles were counted together to represent coronary blood flow. Random samples were taken from the liver, diaphragm, small and large gut, while the brain, kidneys, gallbladder, and pancreas were processed as whole organs. In this study, liver blood flow represents hepatic artery blood flow. The gastrocnemius muscle was obtained to represent skeletal muscle blood flow. All tissues were cut in 1-cm long pieces and placed on a petri dish. After drying for 72 h in a heater (Biological Safety Cabinet, Nuaire, Plymouth, MA), the tissue was placed into plastic tubes. Tissue and the reference blood samples were then counted in a multichannel Automatic Gamma Counter System, Series 1185 (Scarle Analytic, Des Plaines, IL). Radioactivity of each isotope in each organ was determined by the stripping technique (25). The counts of the two blood reference samples were averaged, and organ blood flow was calculated (expressed as ml · min¹ · 100 g wet wt tissue¹) by the equation: organ blood flow (ml/min) = 10 ml/min × organ counts/reference blood counts. The adequacy of microsphere mixing was assessed by linear regression between the blood flow to the left and right kidneys. The correlation coefficient (r²) was 0.98 in the hemodilution group versus 0.97 in the transfusion group. The regression lines were not different between the groups, and the slopes were not statistically different from 1.

Analytic approach. All data are expressed as means ± SE. The data were analyzed by analysis of variance using a two-factor design as it is supported by SPSS 6.0. Therefore, effects are shown as group effect (hemodilution vs. transfusion), hypoxia effect, and interaction. The Tukey test was used as a post hoc test to correct for multiple comparisons. A P value of <0.05 was considered to be statistically significant. Dependent physiological parameters were analyzed by regression analysis using the least-squares method. A dummy variable model was used to compare regression lines between the two groups (23).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effects of hemodilution and RBC transfusion. All animals completed the study protocol. Postmortem examination confirmed purulent ascitic fluid with an exudative reaction around a necrotic cecum. Forty-eight hours following CLP, the hemoglobin level was 77 ± 3 g/l in the hemodiluted group and 117 ± 4 g/l in the transfused group. To establish these end points, 745 ± 65 ml of whole blood was withdrawn from group H sheep and replaced with pentastarch, and the group T sheep received 314 ± 127 ml of packed RBCs. Exclusive of the pentastarch infused to isovolemically replace blood withdrawn in the hemodiluted group, group H sheep received 39.8 ± 3.9 ml · kg¹ · 24 h¹ and group T sheep received 35.0 ± 2.1 ml · kg¹ · 24 h¹ (P was not significant) during the 48 h after CLP. Table 1 summarizes the effect of these interventions on circulatory and systemic O₂ metabolism values in the two study groups, just before the hypoxia intervention was begun.

The primary interventions of the study, hemodilution or transfusion, achieved the desired end points regarding the hemoglobin concentration 48 h after CLP (Table 1, Fig. 2). Compared with pre-CLP evaluation, hemodynamic consequences of peritoneal contamination included an unchanged mean arterial blood pressure and an increase in cardiac index, heart rate, and LAP in both study groups (Table 1). Systemic O₂ delivery increased between the baseline and 48-h study in group T but not in group H. Simultaneously, systemic O₂ extraction fell during the 48 h following CLP in group T, whereas calculated systemic O₂ consumption and arterial lactate levels (group H: 0.3 ± 0.05 to 0.3 ± 0.08 mmol/l; group T: 0.5 ± 0.14 to 0.4 ± 0.12 mmol/l; P = not significant) remained unchanged.

View larger version (11K): [in this window] [in a new window]   Fig. 2.   Hemoglobin levels (A) and arterial O2 content (CaO2, B) before CLP and during hypoxia trial. Before CLP, hemoglobin levels were similar between two groups and differed according to hemodilution () and transfusion (), respectively, 48 h after CLP.

Effects of hypoxia and hemoglobin level on systemic hemodynamics. Table 2 compares the baseline study to the final hypoxic stage of selected hemodynamic and O₂ metabolism values. The final FI_O2 approximated 0.11 in both study groups by the end of the fourth study stage; thus the arterial PO₂ fell significantly in both groups. By the final hypoxic stage, the overall depression in Ca_O2 was significantly greater in group T compared with group H (Fig. 2). Neither mean blood pressure nor mean pulmonary artery pressures were affected by the hypoxic intervention in either study group. The cardiac index was greater at baseline in group H compared with group T. Coincident with an increase in the heart rate, the cardiac index increased in both groups between the baseline and final hypoxic study stages and remained higher throughout all hypoxic stages in group H versus group T.

Table 2 also records the depression in systemic O₂ delivery that occurred between baseline and the final hypoxic study stages. During this study period, the systemic O₂ uptake remained unchanged throughout all four stages of hypoxia in group H (Fig. 3). Reducing the FI_O2 was accompanied by an increase in systemic O₂ extraction in both study groups (group H, P < 0.01; group T, P < 0.01). A significant group effect during hypoxia in systemic O₂ extraction (P < 0.01) was likely explained by the lower baseline value in group T because the maximal value was similar in both study groups. In both study groups, arterial lactate rose modestly but significantly by the final stage of study.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Changes of systemic O2 extraction (O2E) and O2 consumption (O2) measured by DeltaTrac over change of CaO2 during hypoxia trial. Analysis of variance for O2E: hypoxia effect, P < 0.01; group effect, P < 0.01. Analysis of variance for O2: hypoxia effect, not siginificant (NS); group effect, NS.

Effects of hypoxia and hemoglobin level on myocardial O₂ metabolism variables. The effect of hypoxia on myocardial O₂ metabolism is demonstrated in Figs. 4-8 and Table 3. Figure 4 shows the effect of study interventions on both myocardial O₂ uptake and heart work, the latter estimated as the mean blood pressure times cardiac index product. This figure demonstrates that the progressive reduction in FI_O2 was accompanied by significant and similar increases in myocardial work and myocardial O₂ uptake in both study groups. Accordingly, coronary blood flow increased in both groups with progressive hypoxia but was significantly greater in group H throughout the entire hypoxia intervention (P < 0.01). The left ventricular endocardial-to-epicardial flow ratios were not affected by the hypoxic intervention in either study group, although they were significantly lower in group H than in group T during all study stages (Fig. 5). Similar to the increase in arterial lactate levels, coronary sinus lactate levels increased in both study groups during severe hypoxia (Table 3).

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Changes of heart work (BPM, beats/min; CI, cardiac index; A) and myocardial O2 (B) over change of CaO2 during hypoxia trial. Analysis of variance for heart work estimated by product of mean blood pressure and cardiac index: hypoxia effect, P < 0.01; group effect, NS. Analysis of variance for myocardial O2: hypoxia effect, P < 0.05; group effect, NS. , Hemodilution; , transfusion.

View larger version (25K): [in this window] [in a new window]   Fig. 5.   Left ventricular endocardial-to-epicardial blood flow ratios. Because there is no interaction or time effect, bars represent flow ratios of all stages for each group. Flow ratios differ with P = 0.026.

View larger version (23K): [in this window] [in a new window]   Fig. 6.   Changes of myocardial O2 delivery (DO2) and myocardial O2E over change of CaO2 during hypoxia trial. Analysis of variance for myocardial O2 delivery: hypoxia effect, P < 0.01; group effect, NS. Analysis of variance for myocardial O2E: hypoxia effect, NS; group effect, P < 0.05.

View larger version (19K): [in this window] [in a new window]   Fig. 7.   Linear regression between arterial and coronary sinus PO2. Hemodilution group: r2 = 0.28, P < 0.01; transfusion group: r2 = 0.51, P < 0.001 showed that at a given arterial PO2, hemodiluted sheep had a greater coronary sinus PO2 than transfused sheep (P < 0.01).

1

1

1

1

View larger version (20K): [in this window] [in a new window]   Fig. 8.   Myocardial O2 vs. coronary blood flow. Regression lines are heart = 25 × myocardial O2  11 (hemodilution group, r2 = 0.82) and heart= 15 × myocardial O2  58 (transfusion group, r2 = 0.79). Two regression lines differ in slope as well as intercept with P < 0.01.

Effects of hypoxia and hemoglobin level on regional O₂ delivery relationships. Figure 9 demonstrates that the hypoxic intervention was accompanied by an increase in the regional O₂ delivery (QO₂) to the heart in both groups. QO₂ to the brain was not affected by the hypoxic trial. Both groups reduced QO₂ to the gallbladder, kidneys, and spleen. There was a statistically significant reduction in regional QO₂ to the liver, pancreas, rumen, and small and large gut in the transfusion group only. The hemodilution group only showed a tendency to reduce QO₂ to these organs without being statistically significant.

View larger version (26K): [in this window] [in a new window]   Fig. 9.   Percent change of organ O2 delivery (organ blood flow × CaO2) from measurement at room air to last hypoxic stage. * P < 0.05, ** P < 0.01 change different from 0. § P < 0.05 group difference.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Critical appraisal has shown considerable variation in both practice and opinion regarding the appropriate RBC transfusion trigger in sepsis (1, 42). Because this may be related to a paucity of basic research, we designed an experiment to measure and compare the effect of two clinically relevant RBC transfusion strategies on the metabolic O₂ reserve of circulation in septic sheep. We found that isovolemic hemodilution to maintain hemoglobin levels between 75 and 80 g/l during sepsis imposed changes on the adaptation of circulation to acute hypoxia, which have not been previously reported in healthy models. Novel information from this experiment is the demonstration that maintaining normal hemoglobin levels by RBC transfusion was accompanied by a greater coronary flow reserve, a better intramyocardial blood flow distribution, and a greater capacity to extract O₂ in the septic coronary circulation.

Background. Sepsis is characterized by a quantifiable tissue injury, which may lead to the multiple organ dysfunction syndrome. Sepsis is also a hypermetabolic state, where tissue O₂ needs are markedly increased. Where research has linked outcomes from sepsis with an inability to match O₂ delivery to these elevated tissue O₂ needs (43), treatment strategies have emphasized optimizing increasing systemic O₂ delivery by 1) increasing the cardiac output with intravascular volume expansion and/or inotropic therapy, and/or 2) increasing Ca_O2 with RBC transfusion and measures that improve arterial oxygenation (i.e., supplemental O₂ and positive end-expiratory pressure). Because it is not uncommon to find modest anemia in septic patients, the appropriateness of transfusing RBCs to increase tissue O₂ availability has been frequently discussed (2, 9, 43) yet not explicitly examined.

A finely regulated control system distributes O₂ delivery to match the metabolic O₂ need of the tissue. Because some organs have a limited ability to increase O₂ extraction, isovolemic hemodilution in health is accompanied by an increase in convective O₂ delivery to the heart and brain (22, 30, 38, 43). Such compensation is facilitated by a redistribution of O₂ delivery away from organs with a greater O₂ extraction reserve, such as the splanchnic circulation (22, 28). The effectiveness of this compensation is evident in recent guidelines which propose that anemia in previously healthy patients can be tolerated to hemoglobin levels as low as 60-70 g/l (2).

In contrast to health, an argument may be made that the RBC transfusion trigger should be higher in sepsis. For example, sepsis is a hypermetabolic process, and increasing the metabolic rate in healthy animal models elevates the optimal hematocrit of the gut (22). Second, sepsis is characterized by circulatory abnormalities that impact on tissue O₂ delivery, including a limited cardiac output reserve (4), an impaired redistribution capacity of O₂ delivery from the splanchnic organs to the heart and brain (5), and microcirculatory dysfunction which limits O₂ extraction capacity (24). We have assessed adequacy of the circulatory reserve when anemia complicates sepsis in different approaches. We found that 1) the hemoglobin concentration exerted independent and negative effects on regional O₂ delivery in septic sheep (17); 2) sepsis in sheep depressed the capacity to increase both coronary blood flow and myocardial O₂ extraction during acute hypoxia, compared with control sheep (6); and 3) the ability to appropriately increase myocardial O₂ delivery during acute hypoxia in septic rats occurred only in animals transfused to maintain hematocrit levels >45% (30). In contrast and confirming previous work (10, 20), nonseptic rats maintained myocardial O₂ delivery reserve to hypoxia with hematocrits <30%. 4) The hemoglobin dissociation curve was shifted left in septic sheep compared with that in nonseptic study conditions (7). Taken together, these data are indirect evidence that sepsis may elevate the "optimal" hemoglobin concentration, which would be clinically acknowledged as a need to transfuse RBCs earlier in sepsis than in health.

The current experiment was therefore designed to test the hypothesis that isovolemic hemodilution to create modest anemia in mature sheep rendered septic by CLP would depress the ability of the heart to support circulatory circulation in this syndrome. We used a large animal model of sepsis complicating peritoneal contamination as previously described by our laboratory (5, 6, 29). This model reproduces the circulatory lesion, which has been reported at both the regional (5) and microregional (29) levels. To obviate potential confounding effects of anesthetic agents on both regional and microregional circulations, this study was carried out with the experimental animal awake (10). After baseline studies, we then exposed the animals to acute hypoxia to determine whether the usual metabolic O₂ reserve of the circulation was altered by hemoglobin status (3, 6, 40). Because the hypoxic intervention was accompanied by a modest increase in the arterial lactate concentration, it is probably reasonable to conclude that acute compensation to maintain O₂ availability in this animal model was exceeded during the final hypoxic study stage (31).

Animals allocated to group T had a mean study hemoglobin that approximated normal hemoglobin levels in sheep, whereas the mean study hemoglobin in group H animals was similar to values proposed as target values above which no transfusion is necessary in recent clinical guidelines (2). Because we have recently found that sepsis lengthens the time required to ensure steady-state conditions in the central and regional circulations (18), we completed interventions to distinguish anemic versus nonanemic sheep 24 h before the hypoxic intervention. We used RBCs stored in CPDA-1 (citrate, phosphate, dextrose, adenine), taken from a donor sheep on the same day as RBC transfusion, because Marik and Sibbald (26) noted that reduced RBC deformability complicating storage may limit tissue O₂ availability in sepsis.

Hemoglobin levels and systemic circulation's metabolic O₂ reserve. Before the hypoxic intervention, both study groups demonstrated a circulatory profile that is typical of sepsis. Demonstrating that anemia imposed an added stress on the central circulation, group H sheep had a higher cardiac index than group T sheep, and this difference was maintained across all subsequent study stages. As O₂ delivery was sequentially reduced during acute hypoxia, an expected increase in systemic O₂ extraction was similar in both study groups. In a previous study by Bloos et al. (6), it was confirmed that sepsis in this animal model depressed the capacity to extract O₂. Data from the current experiment therefore demonstrate that the hemoglobin status of the animal does not influence the progression of this lesion. Because measurement of systemic O₂ extraction reflects the algebraic sum of changes occurring within individual organ circulations, it is possible that changes according to hemoglobin status might have occurred in individual organ circulations. Whereas hemodilution is known not to reduce regional O₂ delivery by increase of organ blood flows (38), severe hypoxia reduced regional O₂ transport to all noncardiac organs observed during this study regardless of the hemoglobin level. In a previous hypoxia study (6), Bloos et al. found that sham animals redistributed QO₂ from the gut to the heart, whereas septic sheep did not. Bloos et al. suggested that septic sheep were unable to increase O₂ extraction sufficiently to give up blood flow. In this study, we found a significant reduction in regional QO₂ to the gut in the transfusion group only. This might reflect a normalization of the O₂ extraction reserve of the gut in the transfusion group. However, this remains speculation because we did not measure the regional O₂ extraction of the gut.

Hemoglobin levels and coronary circulation's metabolic O₂ reserve. Reducing O₂ content was accompanied by an increase in the cardiac index times blood pressure product, a surrogate end point for heart work (1). A parallel increase in myocardial O₂ uptake was the consequence of an increase in flow work and confirms that the metabolic coupling of O₂ availability to changing O₂ needs remains intact in hyperdynamic sepsis. However, the mechanism by which myocardial O₂ uptake was supported differed according to the study subject's hemoglobin status. Thus 1) coronary blood flow was greater at all levels of myocardial O₂ uptake in group H compared with group T animals, and 2) the maximum myocardial O₂ extraction achieved was greater in group T compared with group H animals.