INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS REULTS DISCUSSION REFERENCES

In septic shock, despite an adequacy of cardiac output, an impairment in tissue aerobic metabolism, particularly with respect to the stomach and small intestine, leads to the production of lactate and hydrogen ions in the circulation (1). Lactic acidosis may result as a consequence of a reduction in oxygen delivery either because of a decrease in bulk splanchnic flow, or because microvascular injury leads to a redistribution of perfusion to some areas while other areas remain perfused (4, 5). Alternatively, a defect in the utilization of oxygen at a mitochondrial level may be primarily involved in sepsis (6, 7).

In addition, lactic acidosis in septic shock may reflect abnormal lactate extraction and utilization by the liver and other organs rather than an increase in production by the splanchnic tissues (8, 9). Although, under normal conditions, there is a large metabolic reserve in hepatic lactate metabolism (10), in pathological conditions, such as sepsis, this may not be the case. In a phenformin model of lactic acidosis in dogs, Arieff and coworkers (11) showed that although splanchnic lactate production was only modestly increased, lactate extraction by the liver did not increase to attenuate the acidosis that occurred.

The pathophysiology of lactic acidosis in septic shock is not clear. It has not been determined whether the major factor in lactic acidosis is overproduction by various tissues or an impairment in extraction by the liver. Moreover, in septic shock, vasoactive agents are administered to improve systemic blood pressure (Psa) (12), and these agents may lead to changes in splanchnic and hepatic blood flow that in turn may affect splanchnic lactate production and hepatic lactate extraction in sepsis (12). Agents such as phenylephrine and norepinephrine have -adrenergic effects, and these agents may cause splanchnic ischemia by decreasing splanchnic blood flow (13). Dopamine may increase splanchnic blood flow by means of dopaminergic receptors, while dobutamine has predominantly -adrenergic properties that may lead to an increase in splanchnic blood flow (12). Combinations of vasoactive agents, such as dobutamine and norepinephrine, may also be administered in sepsis, during which tissue oxygen delivery is maximized by dobutamine, while Psa is maintained by norepinephrine.

Because the pathophysiology of lactic acidosis in sepsis is not well defined, we examined splanchnic and hepatic hemodynamics and lactate metabolism in Escherichia coli sepsis in dogs. We further determined the effects of various vasoactive treatments on these parameters in this model. The objective was to define which agent best preserved splanchnic blood flow and improved lactate metabolism in sepsis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS REULTS DISCUSSION REFERENCES

Animal Preparation

These studies were performed in accordance with protocols approved by the University Animal Care Committee. Mongrel dogs (25 to 30 kg) were anesthetized with thiopental sodium (20 mg/kg, intravenously) and were given a loading dose of sufentanil (10 µg/kg) followed by maintenance doses of sufentanil (1 µg/min) and midazolam (5 µg/kg/ min) (16). The trachea was intubated with an endotracheal tube and the lungs were mechanically ventilated (Harvard Apparatus, Hollison, MA). The initial ventilator settings (approximate rate of 10 breaths/min; tidal volume, 15 ml/kg) were adjusted to maintain initial arterial PCO₂ at ~ 35 mm Hg and pH at ~ 7.35. During the course of the study, as metabolic acidosis ensued, the ventilator rate was increased as necessary to try to maintain the pH between 7.25 and 7.35. The inspired oxygen concentration was titrated to maintain arterial PO₂ > 150 mm Hg, so that arterial hypoxemia did not contribute to increased lactate production found over the course of the study.

With the animal in the supine position, vascular catheters were inserted percutaneously under sterile conditions. An arterial catheter was placed into the left femoral artery and advanced into the midabdominal aorta. This catheter was used to obtain arterial lactate concentration (Lac_Art), oxygen content (Ca_O2) (see below), and arterial blood gases. Into the left jugular vein, a thermistor-tipped catheter was advanced into the pulmonary artery to measure mean pulmonary arterial pressure (), pulmonary wedge pressure (Pwe), right atrial pressure (Prat), and cardiac output () by the thermodilution technique (Columbus Instruments, Columbus, OH). The average of three determinations was used in the reporting of for each condition. Samples for mixed venous lactate (Lac) and oxygen content (C_O2) were obtained from the distal port of the catheter.

The abdomen was opened by a midline incision, the preparation of which was similar to that described by Kellum and coworkers (17). The spleen was removed. A catheter was advanced into the splenic vein and positioned in the portal vein at the point at which it enters the liver. The portal vein catheter measured 0.2 cm in external diameter and contained one end hole and two side holes, the latter within 1.0 cm of the tip.

Another catheter of similar characteristics was advanced from the right internal jugular vein into the hepatic vein. This hepatic vein catheter was usually placed into the right hepatic vein because of easier placement, although hepatic lactate concentrations and oxygen contents were similar when determinations from the right and left veins were compared (between right and left hepatic vein measurements, the coefficient of variation [CV] of lactate measured over a range of 1 to 10 mM averaged 3.8 ± 1.2%, while the CV of hepatic oxygen content determined over the course of the experiment averaged 4.9 ± 2% [this included 12 pairs of determinations that were obtained in 2 preliminary experiments; see below]).

The hepatic vein catheter was positioned into the hepatic vein by palpation, and proper positioning of the catheter was verified at autopsy. The hepatic catheter was also used to measure hepatic venous pressure (see below). The pressure wave form was monitored to ensure that pressure oscillations were not overdamped, and in that case, it was unlikely that the catheter was pushed too far into the parenchyma to cause stagnation of blood flow. Moreover, it is important to note that in the present study calculation of total hepatic flow was determined from the sum of hepatic artery flow and portal vein flow (see DATA ANALYSIS, below). This calculation would not be affected by the position of the hepatic vein catheter, because the analysis was dependent on knowing the flow into the liver rather the amount of flow that exited the right and left hepatic veins.

The portal vein catheter was used to obtain blood samples for portal vein lactate (Lac_pov) and oxygen content (Cpov_O2). The hepatic vein catheter was used to obtain hepatic vein lactate (Lac_hv) and oxygen content (Chv_O2). In addition, a flow probe (Transonic, Ithaca, NY) was placed around the proximal portion of the portal vein to measure portal blood flow (pov). Another probe (Transonic) was placed around the hepatic artery to measure hepatic artery blood flow (ha). We have found that these probes are stable with little zero drift over the course of the experiment. Nevertheless, we did check zero flow by the snare technique, in which each vessel was transiently occluded; the zero flow measured by occlusion was compared with that found when electrical zero was obtained. The snare technique was performed at the beginning of the experiment and just after the last measurement was determined in order to make sure there was no zero drift between the initial and final measurements. Moreover, manual calibration of the probes, in which pulsatile flow was driven by a pump through a dissected artery and vein, was also performed periodically to check the electrical calibration (approximately every 3 to 4 mo), and these determinations were found to be stable and accurate over this period.

All of the fluid-filled catheters were connected to transducers (Cobe, Lakewood, CO) that were referenced relative to the left atrium. Measurements were obtained with the animal in the supine position and were taken at end expiration, during which the ventilator was turned off for approximately 3 to 5 s. All signals were displayed on an eight-channel recorder (Astra Med, West Warwick, RI).

Protocols

Baseline measurements (presepsis; see below) were obtained 1 h after completion of the preparation, when parameters were stable. Sepsis was then induced as previously described by intravenous infusion of 10¹⁰ colony-forming units of live E. coli (serotype 0111:B4) given over a 30-min interval (18). A constant infusion of approximately 5 × 10⁹ colony-forming units of E. coli per hour was then maintained for the remainder of the study. After 4 h of infusion, measurements were again made and this condition was termed the septic condition. In randomized design, one of the treatments (see below) was then administered for 1 h while E. coli was continuously infused. A final set of measurements was obtained at the end of this 1-h period (treatment condition). The duration of the experiment was approximately 6 h (i.e., 1-h baseline waiting period, 4 h of sepsis, and 1 hr of treatment). By design, shock was defined as a drop in Psa of 25% compared with the baseline value, and the goal of drug treatment was to restore Psa to approximately presepsis values. Furthermore, measurements were obtained at similar Pwe between conditions. Volume infusion with Pentaspan (DuPont Pharma, Mississauga, ON, Canada) was given as necessary to maintain the filling pressure at similar levels during the three conditions.

Five treatment regimens (n = 8 for each group) and a nonseptic time-control treatment (n = 7) were studied in respective groups of dogs in which the treatments were administered in randomized design. The rationale for the different treatments was based on common usage of the various vasoactive agents in clinical medicine (see DISCUSSION). These treatments included a sham treatment sepsis group, in which normal saline solution was infused as control treatment, a dopamine treatment sepsis group (dopamine), a phenylephrine treatment sepsis group (phenyl), a norepinephrine treatment sepsis group (norepi), and a combination treatment group in which dobutamine and norepinephrine (norepi/dobut) were infused. In the latter group, the dosage of dobutamine chosen was 5 µg/kg/min and norepinephrine was titrated to restore Psa to the presepsis value. In all treatments, the drugs were infused by pump and were maintained at a constant rate once the required blood pressure was reached. To control for the effect of time on our results, we also included a nonseptic time-control group, in which normal saline solution was infused without E. coli over the interval of the experiment.

In each condition (i.e., presepsis, sepsis, and treatment), a complete set of hemodynamic measurements was obtained. These measurements included Psa, , Pwe, Prat, , pov, ha, and systemic vascular resistance (Rsva). Rsva was calculated from (Psa  Prat)/. Blood samples for lactate concentrations (Lac_Art, Lac, Lac_pov, and Lac_hv), oxygen contents (Ca_O2, C_O2, Cpov_O2, and Chv_O2), and hematocrits (Hct_Art, Hct, Hct_pov, and Hct_hv) were also measured. Oxygen contents were directly measured by the carbon monoxide scrubbing technique (19). Lactate determination was made by an automated lactate dehydrogenase-based assay. In preliminary experiments, we determined that the coefficients of intraassay (n = 18 pairs) and interassay variation (n = 10 pairs) of lactate measurements were 0.57 ± 0.21 and 1.2 ± 0.33%, respectively. Because of the small variability in the lactate measurements, only one lactate measurement was obtained for each determination.

In a subset of each treatment group, moreover (approximately two or three in each treatment group), liver function tests were performed during baseline, sepsis, and treatment. These tests included serum bilirubin, alanine transaminase (ALT), aspartate transaminase (AST), and -glutamyltransferase (GGT). These determinations were performed by Clinical Chemistry (Health Sciences Centre, Winnipeg, MB, Canada). The objective was to determine whether biochemical evidence of hepatic dysfunction developed over the course of sepsis in our model.

Supplemental Studies

In addition to the groups described above, a supplementary protocol was performed in separate groups of septic (n = 12) and nonseptic (n = 11) dogs, in which the effect of bolus infusion of lactic acid (5 mmol/kg) on hepatic lactate extraction was examined. This part of the experiment was performed because under nonseptic conditions, the delivery of lactate to the liver was small, so that it was difficult to compare hepatic extraction between septic and nonseptic conditions. An identical preparation was used in which lactic acid (0.4 M) was infused through the lower abdominal aorta over 30 min. Measurements of lactate extraction were obtained immediately after termination of the infusion.

Data Analysis

Splanchnic lactate production (SLP) was calculated from (Lac_Art  Lac_pov) × pov(1  Hct_pov) (20). Total lactate presented to the liver (Lac_tot) was calculated from [ha × Lac_Art × (1  Hct_Art) + pov × Lac_pov × (1  Hct_pov)]. Percent liver lactate extraction (Lac Ex%) was calculated by [Lac_tot  Lac_hv × (1  Hct_hv)]/Lac_tot × 100. Splanchnic oxygen consumption (sp_O2) was calculated from (Ca_O2  Cpov_O2) × pov while hepatic oxygen consumption (h_O2) was calculated from [(ha × Ca_O2 + pov × Cpov_O2)  (ha + pov) Chv_O2].

In the preceding analyses, we used uptake of lactate from plasma based on the work of Naylor and colleagues (20). These investigators showed that erythrocyte lactate equilibrates only slowly with plasma, and therefore little net erythrocyte lactate is likely to be removed from the cell during flow through the splanchnic and liver areas. However, we also compared the plasma results with those obtained when whole blood lactate concentrations were used as performed by Arieff and coworkers (11) (see DISCUSSION).

Statistics

Between-group analyses were determined by two-way repeated measures ANOVA split plot design (factor A, different treatment groups; factor B, different time periods), in which the interaction between the two factors was also assessed. If a significant interaction was present, then the treatment effects were different. In that case, a Student- Newman-Keuls (SNK) multiple range test was used to determine where differences occurred. Within-group analyses were determined by one-way repeated measures ANOVA and SNK multiple comparison test. When two conditions were compared in nonrepeated analyses, paired and unpaired t tests were used in the appropriate conditions. The p value for accepting significance was < 0.05. Results are reported as means ± 1 SE.

	REULTS

TOP ABSTRACT INTRODUCTION METHODS REULTS DISCUSSION REFERENCES

Sepsis Effects

After 4 h of E. coli bacteremia, Psa decreased approximately 50% from baseline in all groups (see Figure 1, upper left), while there was no change in the nonsepsis time control group (Table 1). By design, Pwe (see Figure 1, lower left) was unchanged between presepsis and sepsis. There were also no changes in between baseline and sepsis conditions (see Figure 1, upper right). Compared with the presepsis value, Rsva decreased during sepsis in all groups (p < 0.05) (see Figure 1, lower right).

View larger version (29K): [in this window] [in a new window]   Figure 1.   Mean arterial blood pressure, cardiac output, pulmonary wedge pressure (Pwe), and systemic vascular resistance during the three conditions are shown. By one-way repeated measures ANOVA and Student-Newman-Keuls (SNK) multiple comparison test, *p < 0.05 versus sepsis. By two-way repeated measures ANOVA and SNK, &p < 0.05 versus all other treatments; #p < 0.05 versus phenylephrine; +p < 0.05 versus sham; !p < 0.05 versus nonseptic controls.

When all of the sepsis groups were combined, the changes in pov, ha, and total hepatic blood flow observed between presepsis and sepsis were 0.47 ± 0.05 versus 0.34 ± 0.03* L/min, 0.20 ± 0.02 versus 0.30 ± 0.03* L/min, and 0.69 ± 0.05 versus 0.63 ± 0.04 L/min (*p < 0.05 pre- versus postsepsis by paired t test; p < 0.05 versus nonseptic time-control group by ANOVA). Portal blood flow decreased during sepsis by about 25%, while hepatic artery flow reciprocally increased. The net result was that total hepatic flow was slightly, but not significantly, decreased in sepsis (see presepsis and sepsis conditions in Figure 2).

View larger version (18K): [in this window] [in a new window]   Figure 2.   The effects of the various treatments on portal vein flow, hepatic artery flow, and total hepatic flows are shown. By one-way repeated-measures ANOVA and SNK, *p < 0.05 versus sepsis. By two-way repeated-measures ANOVA and SNK, !p < 0.05 versus other treatments except for norepi; #p < 0.05 dobut/norepi versus phenylephrine and nonsepsis; +p < 0.05 dopamine versus phenylephrine and nonsepsis.

As shown in Table 2, progressive metabolic acidosis developed in the sepsis groups over the course of the experiment; this was not observed in the nonseptic time-control group (see Table 1). There were no differences in arterial blood gas parameters observed under any of the conditions among the treatment groups. Hct decreased over the course of the experiment between baseline and sepsis (see Table 2).

The arterial lactate concentrations observed in the various groups are shown in Figure 3. During sepsis, Lac_Art increased as compared with baseline (1.4 ± 0.08 versus 3.8 ± 0.4 mM; p < 0.05). In the nonseptic time-control group, there were no changes in arterial lactate detected over the course of the study (see Table 1).

View larger version (14K): [in this window] [in a new window]   Figure 3.   Arterial lactate concentrations are shown during presepsis, sepsis, and treatment conditions. There was a significant increase in arterial lactate concentrations in pre- versus postsepsis (*p < 0.05 by one-way ANOVA and SNK). None of the treatments caused an increase in arterial lactate compared with sepsis values.

The relationship between the total hepatic plasma lactate delivery and percent lactate extraction for the individual dogs in the presepsis condition is shown in Figure 4 (top). At baseline, when all groups were combined, total hepatic plasma lactate delivery, lactate extraction, and percent lactate extraction were 0.52 ± 0.05 mmol/min, 0.002 ± 0.011 mmol/min, and 0.2 ± 2.6%, respectively (when whole blood lactate was considered, the results were 0.90 ± 0.08 mmol/min, 0.003 ± 0.019 mmol/min, and 0.8 ± 0.026%, respectively).

View larger version (18K): [in this window] [in a new window]   Figure 4.   Total hepatic plasma lactate delivery for the individual dogs is shown on the ordinate against percent lactate extraction on the abscissa. (Top panel ) Presepsis condition. (Middle panel ) Septic condition. (Bottom panel ) Treatment condition. Triangles represent data obtained during lactic acid infusion (see text).

In the sepsis condition, the relationship between total hepatic plasma lactate delivery and percent extraction for the individual dogs is shown in Figure 4 (middle). During sepsis, total hepatic plasma lactate delivery, lactate extraction, and percent lactate extraction averaged 1.74 mmol/min, 0.12 ± 0.02 mmol/ min, and 8 ± 1%, respectively, which were all significantly increased (p < 0.05) as compared with presepsis. The results obtained with whole blood lactate were 2.4 ± 0.08 mmol/min, 0.16 ± 0.03 mmol/min, and 7.8 ± 1.1%, respectively, and were also increased as compared with presepsis (p < 0.05).

Under conditions of nonsepsis, because lactate delivery to the liver was so small, we compared hepatic lactate extraction in septic and nonseptic groups when a bolus amount of lactic acid was administered (see SUPPLEMENTAL STUDIES, above). In the lactate infusion experiments (Figure 4, top triangles; n = 11; some dogs had two measurements, in which the second was 15 min postinfusion), the results showed that over a range of plasma lactate delivery of 1.6 to 26.5 mmol/min (mean ± SE, 6.3 ± 1.3 mmol/min) plasma lactate extraction and percent extraction were 0.94 ± 0.1 mmol/min and 14.9 ± 4.5%, respectively. These results were significantly higher than values in septic dogs, in which measurements were compared over a similar range of total hepatic plasma lactate concentrations (range, 1.55 to 17.0 mmol/min; mean, 5.0 ± 0.6 mmol/min), where plasma lactate extraction was 0.18 ± 0.05 mmol/min and percent extraction was 3.6 ± 1% (p < 0.02 versus nonsepsis).

Splanchnic lactate production was determined under each condition. At baseline, splanchnic tissues (on average) consumed lactate, since the mean values for plasma and whole blood lactate were negative and averaged 0.017 ± 0.004 and 0.03 ± 0.006 mmol/min, respectively. During sepsis, splanchnic tissues produced lactate and mean production increased to 0.07 ± 0.012 mmol/min for plasma and to 0.097 ± 0.017 mmol/ min for whole blood (p < 0.05 versus baseline; see Table 3).

Hepatic and splanchnic oxygen consumption were also measured under each condition (see Table 4). For all sepsis groups (n = 40), splanchnic oxygen consumption averaged 13.6 ± 11 ml/min at baseline and 11.8 ± 13 ml/min during sepsis. Hepatic oxygen consumption was also unchanged between baseline and sepsis (17 ± 2 versus 13.3 ± 2 ml/min).

Liver function tests (LFTs) (n = 12) were performed to detect biochemical evidence of hepatic dysfunction in sepsis. LFT results increased slightly, but not significantly between baseline and sepsis. Bilirubin measured 1.9 ± 0.5 versus 2.1 ± 0.4 µmol/ L; ALT measured 43 ± 10 versus 48 ± 9 U/L; AST measured 39 ± 8 versus 79 ± 21 U/L; LDH measured 128 ± 13 versus 224 ± 53 U/L; and GGT measured 4 ± 1 versus 6.3 ± 1 U/L.

Treatment Effects

In the various treatment groups, the respective (mean ± SE) doses of the vasoactive drugs used were 15 ± 1.2 ug/kg/min dopamine, 6.6 ± 0.7 µg/kg/min phenylephrine, 4.1 ± 0.03 µg/ kg/min norepinephrine, and in the dobut/norepi treatment group, 5 µg/kg/min dobutamine with 4 ± 0.4 µg/kg/min norepinephrine.

Systemic hemodynamics measured during the treatment condition are shown in Figure 1. During treatment, Psa returned to the presepsis value by design. In the sham treatment group, Psa did not increase over the treatment interval. Pwe value were not different among the treatments (see Figure 1, lower left), and similar amounts of intravenous fluids were given in the five treatment groups (dopamine, 15 ± 7 ml/kg; phenylephrine, 13 ± 10 ml/kg; norepinephrine, 13 ± 10 ml/kg; dobut/norepi, 14 ± 7 ml/kg; and sham treatment, 14 ± 8 ml/kg).

During treatment, increased in the dopamine, norepinephrine, and dobut/norepi groups, particularly as compared with the phenylephrine group, in which decreased with treatment. In the nonseptic time-control group, remained unchanged under the three conditions (see Table 1). Phenylephrine treatment caused a significant increase in Rsva, which was not found with the other treatments.

The effects of the various treatments on pov, ha, and total hepatic flow (h,tot) are shown in Figure 2. In the phenylephrine group, pov significantly decreased during treatment as compared with sepsis. pov decreased slightly, but not significantly, with norepinephrine. Dopamine treatment resulted in a nonsignificant increase in pov as compared with sepsis, while dobut/norepi produced a negligible effect (pre- versus posttreatment).

The changes in ha and h,tot observed with the different treatments are also shown in Figure 2. Treatments with dopamine, norepinephrine, and dobut/norepi tended to increase ha as compared with pretreatment, while phenylephrine treatment significantly decreased ha. There was a marked reduction in h,tot in the phenylephrine group that was significantly different from that found in the dopamine group. In the nonseptic time-control group, the changes in ha, pov, and h,tot are shown in Table 1 and were unchanged during the time of of the study.

Arterial lactate concentrations measured during treatment are shown in Figure 3. There were no significant changes in arterial lactate when any of the treatments were administered, although lactate concentrations in the phenylephrine group were nonsignificantly higher than those measured in the other groups. For all dogs during treatment, lactate averaged 4.2 ± 0.6 mM; p < 0.05 versus baseline.

None of the treatments altered lactate extraction and percent lactate extraction as compared with sepsis. The individual data are shown in Figure 4 (bottom), while the group treatment data are shown in Table 3. When results for all dogs were combined, total plasma lactate hepatic delivery, lactate extraction, and percent lactate extraction measured during treatment averaged 1.6 ± 0.16 mmol/min, 0.08 ± 0.02 mmol/ min and 7 ± 2%, respectively (all p < 0.05 versus baseline); for whole blood calculations, the results averaged 2.2 ± 0.3 mmol/min, 0.11 ± 0.03 mmol/min, and 6.9 ± 1.8%, respectively (all p < 0.05 versus baseline).

There were no effects of any of the treatments on splanchnic lactate production. The treatment data are shown in Table 3. When results for all dogs were combined, splanchnic lactate production measured during treatment averaged 0.055 ± 0.01 mmol/min for plasma and 0.075 ± 0.016 mmol/min for whole blood (p < 0.05 versus baseline).

In the treatment groups, there were also no changes in splanchnic or hepatic oxygen consumption observed under the three conditions. Splanchnic and hepatic oxygen consumption did not change when any of the treatments were administered (see Table 4).

In Table 4, the arterial minus portal oxygen content differences are shown. When results for all dogs were combined, there was no difference in mean values at baseline versus sepsis (3.6 ± 0.19 versus 4.4 ± 0.5 ml of O₂ per 100 cm³ of blood). As shown in Table 4, the arterial minus portal content oxygen difference decreased during dopamine infusion. The results obtained in the dopamine group were significantly different from those observed during sham treatment and phenylephrine treatment.

Liver function tests showed no change with the various treatments and the data for all of the treatment groups were pooled. Bilirubin measured 1.8 ± 2 µmol/L, ALT measured 63 ± 13 U/L, AST measured 87 ± 27 U/L, LDH measured 238 ± 60 U/L, and GGT measured 8 ± 2 U/L.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS REULTS DISCUSSION REFERENCES

The main points of this study are that the elevation in plasma lactate found in this model of sepsis could be attributed to an increase in SLP as well as an impairment in hepatic lactate extraction (HLE). Both processes would serve to promote lactic acidosis in sepsis. Furthermore, none of the vasopressors used in this experiment altered SLP or HLE as compared with no treatment.

Although SLP increased in sepsis, we would argue that the mechanism was not due to a reduction in tissue oxygen delivery. The results showed that when pov decreased in sepsis and further decreased during phenylephrine treatment, the appropriate response to the reduction in oxygen delivery was observed. Under the various conditions, the splanchnic arterial to portal oxygen content difference [C(art-pov)O₂] widened appropriately as pov decreased, such that splanchnic oxygen consumption did not decrease. Accordingly, because splanchnic tissues were able to extract more oxygen as blood flow decreased, and because oxygen consumption was unchanged between conditions, there is no evidence to suggest that the increase in SLP observed during sepsis was due to tissue hypoxia. We recognize, however, that because tissue measurements such as intramucosal pH were not performed, this still remains an assumption.

Some authors have suggested that in clinical sepsis, microvascular oxygen delivery may be deranged such that some capillaries in the gut are grossly underperfused, which would lead to anaerobic glycolysis and the release of lactate, while others are still perfused (4). In that case, a decrease in oxygen splanchnic oxygen consumption between nonseptic and septic conditions should still be detectable, because anaerobic metabolism would be expected to accelerate as pov decreased further. However, it also important to recognize that small local changes in anaerobic metabolism may be outside the sensitivity of our measurement technique.

The present findings would therefore support the view that under conditions of adequate cardiac output, tissue hypoxia may not be the mechanism of the lactic acidosis found in sepsis. This conclusion would support that of Hurtado and co-workers (7) and Curtis and Cain (6), who found in acute models of endotoxemia that sepsis had a direct effect on cellular metabolism and that an increase in arterial lactate in sepsis may occur independently of tissue hypoxia.

In the present study, total hepatic plasma lactate delivery at baseline averaged 0.52 mmol/min and lactate extraction was 0.002 mmol/min. During sepsis, total plasma lactate delivery approximately tripled to 1.74 mmol/min, while hepatic lactate extraction increased to 0.12 mmol/min. This represented an increase in lactate extraction to 8%. During sepsis, therefore, hepatic lactate extraction increased as compared with baseline, but in terms of studies performed in nonseptic preparations, this increase in lactate extraction appeared modest in amount. Lupo and colleagues (21) examined the kinetics of lactate transport in rat liver in vivo. According to these investigators, it was estimated that under nonseptic conditions, the normal canine liver (20- to 25-kg dogs) would be able to metabolize lactate at approximately 2 mmol/min. In a study of sheep, in which lactic acid loading was performed, Naylor and coworkers (20) reported that maximal hepatic plasma extraction was 5.72 mmol kg^0.75/h. With respect to our canine model, this value would correspond to ~ 1.2 mmol/min when the weights of the dogs were considered. When bolus lactic acid infusion was compared between septic and nonseptic dogs, the results showed that hepatic clearance of lactic was decreased in sepsis (see Figure 4 and Results).

Uptake of lactate by the liver is thought to be regulated by an active transport process as well as by diffusion (21). At blood lactate concentrations of 1 to 5 mM, hepatic lactate uptake appears to be mediated primarily by an active transport process. Lactate may be cotransported with a hydrogen ion, so that when lactate anion is taken up by the liver, the concomitant uptake of hydrogen ion results in the formation of bicarbonate ion in the plasma. At high lactate concentrations, diffusion of the nondissociated anion as well as other nonspecific active transport processes appear important in lactate metabolism (21).

The factors that regulate hepatic uptake of lactate by the lactate-hydrogen cotransporter are still being worked out (21, 25- 27). Blood pH may be an important factor in the regulation of hepatic lactate uptake. When the pH external to the hepatocyte is less than the pH inside the cell, lactate uptake is increased as compared with the opposite situation (26, 27). During lactic acid loading, we did not find a difference in portal pH between septic (6.86 ± 0.05) and nonseptic (6.9 ± 0.04) groups, although intracellular pH was not measured. In addition, the lactate anion transport system can also be inhibited by other ions. Pyruvate anion is a potent inhibitor of lactate anion uptake, while D-lactate anion is another inhibitor of L-lactate anion (the naturally occurring isomer) (25). In clinical sepsis, therefore, the presence of cellular acidosis and circulating ions may play a role in limiting the maximal hepatic uptake of lactate.

In our analysis of lactate metabolism, we used the approach of Naylor and coworkers (20), in that plasma lactate, rather than whole blood lactate concentrations, was primarily used in the calculations. The latter investigators showed that during lactate loading, it took 1 to 2 h for plasma lactate to equilibrate within the red cell. In the present study, the percent hepatic lactate extraction and splanchnic lactate production were determined from calculations which involved (1  Hct), because it was unclear that steady state equilibrium of lactate between plasma and the red cell was ever achieved in our model. Nevertheless, the conclusions were essentially the same when whole blood calculations were used.

As part of the present study, we also examined the effects of commonly used vasopressors on hepatic lactate metabolism and splanchnic blood flow in E. coli sepsis (14). We recognize, however, that other treatments may be used, such as epinephrine, amrinone, prostacyclin, and dopexamine, and that the latter treatments may give results different from those found in the present study. We chose to study those treatments with which we were most familiar. As shown in Figure 4 and Table 3, none of the vasopressor strategies altered hepatic lactate use in our model. However, the five treatments had different effects on portal and hepatic blood flows.

As compared with pretreatment, phenylephrine caused a decline in pov and a severe reduction in ha, such that total hepatic flow was significantly reduced. Norepinephrine and particularly the combination of dobut/norepi tended to maintain total blood flow as compared with pretreatment. The dose of norepinephrine was quite high, but was required to maintain the predetermined Psa. There was no difference in the norepinephrine doses used with and without dobutamine. We think that this resulted because any systemic vasodilation caused by dobutamine was offset by the greater increase in contractility that occurred with the combined treatment versus norepinephrine alone ( nearly doubled with dobut/norepi as compared with an approximately 40% increase with norepinephrine alone).

Dopamine, through its dopaminergic and -adrenergic effects, appeared to have the most positive effects on splanchnic and hepatic arterial blood flows. As compared with pretreatment, both portal blood flow and hepatic artery blood flow increased with dopamine treatment. However, despite the different effects of the preceding treatments on splanchnic and hepatic blood flows, SLP as well as hepatic and splanchnic oxygen consumption were unchanged in the various groups.

Our results also show that as compared with baseline, pov decreased in sepsis, while ha increased, such that h,tot remained unchanged. The hemostatic control of total liver blood flow has been ascribed to the hepatic arterial buffer response (28). Under physiological and hemorrhagic shock conditions, a reduction in pov elicits an increase in ha. The mechanism may be due to a reduction in the washout of the local vasodilator adenosine. In animal models of sepsis, variable changes in hepatic blood flow have been reported (11, 29). Our results would support the contention that, when pov decreases in sepsis, there is a concomitant increase in ha such that total hepatic flow remains unchanged.

In the present study, we did not measure lactate production and extraction across all organ beds in sepsis to determine the contribution of each to the acidosis observed, because our primary objective was to examine SLP and hepatic lactate extraction. It is further recognized that other organs have been reported to contribute to altered metabolism in sepsis (32), such as the lung, which was found by Bellomo and coworkers (32) to be a contributor in an acute canine model of E. coli endotoxemia, but was not a contributor in this experiment. In the present study, it is also noteworthy that the increase in SLP observed between baseline and sepsis (i.e., 0.09 mmol/min [plasma]) closely approximated the increase found in hepatic lactate extraction (i.e., 0.11 mmol/min). This close coupling between production and extraction would reinforce the notion that splanchnic and hepatic lactate metabolism are the important contributors to the acidosis found in sepsis, at least as determined by this experiment.

In terms of the present study, it is important to recognize that the model used is restrictive in that it involves the use of large inocula of bacteria, which may yield results different from those of other models using, for instance, compartmentalized infection, antibiotic cotreatment, or sepsis monitored over a longer period of evolution, which may better reflect the human condition (33). This represents a limitation of the present study. Furthermore, the methodology was similar to that described by Kellum and colleagues (17). Although removal of the spleen is invasive, this procedure is rather straightforward and seemed to be the best way of obtaining the measurements required to answer the questions addressed in this experiment.

The present study emphasizes that although the liver was able to increase lactate extraction in sepsis, this increase was modest and, importantly, less than that found during lactic acid loading in nonseptic animals. In terms of hepatic dysfunction in sepsis, we are not sure whether abnormal lactate metabolism is due to a separate component of liver dysfunction or to a generalized process that develops over the course of the condition. Biochemical evidence of hepatic dysfunction was minimal in this model to explain the impairment in hepatic lactate metabolism as part of global hepatic injury. To the extent that animal models may reflect the human condition, our results indicate that both increased SLP and impaired HLE would contribute to the lactic acidosis found in clinical sepsis.