Splanchnic buffering of metabolic acid during early endotoxemia

ORIGINAL INVESTIGATIONS

Splanchnic John A. Kellum,

Buffering of Metabolic Early Endotoxemia Rinaldo

Bellomo,

David J. Kramer,

Acid During and Michael

R. Pinsky

Purpose: We sought to determine the sites of metabolic acid production and clearance during acute endotoxemia. Materials and Methods: In 10 pentobarbital-anesthetized dogs, flow was measured (ultrasonic probes) for the portal vein, hepatic artery, and renal artery. Catheters were inserted into the hepatic vein, pulmonary artery, renal vein, and portal vein. Measurements of blood gases and strong ions were obtained from each site during control conditions and after 30 minutes of intravenous infusion of 1 mg/kg of Escherichia co/i endotoxin. The total metabolic acid flux across each organ was calculated using the standard base excess

formula and the effective strong ion difference method. Pacoz was maintained by controlled ventilation. Results: Mean arterial pH decreased from 7.34 to 7.22 with acute endotoxemia. Although transvisceral pH gradients revealed net acid release, the source of this was purely respiratory (carbon dioxide). During early endotoxemia, the gut significantly increased metabolic acid uptake (36.60 + 6.60 mmol/h, P < ,051. Conclusions: We conclude that during early endotoxemia in the dog, the gut is a major site of metabolic acid removal. Copyright o 7997 by W.B. Saunders Company

M

liver as potential sources of metabolic acid in sepsis and trauma.3,26 Although more recent studies have disputed the role of lactate in this setting,1%4these organs are widely believed to contribute to acidosis in sepsis. It has been shown, for example, that intracellular acidosis of the gut mucosa commonly occurs in sepsis* and may itself be a cause of altered intestinal permeability.22 For these reasons, we focused our investigation on the acid flux across the visceral organs (gut, liver, and kidney).

ETABOLIC ACIDOSIS commonly occurs in patients with the sepsis syndrome.28 Acidosis correlates with increased mortality even in the absence of increased lactate levels and is a more reliable predictor of outcome.24 Although many studies have focused on the pathophysiology, clinical significance, and treatment of lactic acidosis,5,6,10,19,21,25.27 relatively few have investigated other sources of metabolic acidosis in sepsis. This is surprising because lactate ion accounts for less than 50% of the “unmeasured” anions present in patients with sepsis and metabolic acidosis.‘* We have previously shown that during normal physiological conditions the liver takes up these other anions. However, after administration of endotoxin the liver switches to release of unexplained anions.r6 This might account for the appearance of unexplained anions in the circulation of patients with sepsisi and liver disease,r7 and may indeed result in systemic acidosis if these anions accumulate in excess of strong cations. However, further study has revealed that a nonanion gap metabolic acidosis occurs within several minutes of the administration of endotoxin to healthy animals. l5 In addition, patients with sepsis frequently exhibit an acidosis that has a significant nonanion gap component.r2 This acidosis cannot be explained by the release of anions by the liver. Accordingly, we reanalyzed the data from our original study16 and included analysis of net acid flux to determine the sites of metabolic acid production and clearance in acute endotoxemia. Previous studies have implicated the gut and Journa/ofCritica/Care,Vol

12, No 1 (March),

1997: pp 7-12

METHODS This novel analysis was applied data set. t6 The surgical preparation for that study was as follows:

to a previously and experimental

published protocol

Surgical Preparation The study was approved by the Animal Care and Use Committee of the University of Pittsburgh. After an over-night fast, 10 male mongrel dogs were anesthetized with pentobarbital sodium (30 mg/kg intravenously). The trachea was intubated with a 9F cuffed endotracheal tube, and each dog was ventilated (Siemens-Servo 900B, Sulna, Sweden) at a tidal volume of 12 mL/kg. End-tidal carbon dioxide (coZ) was continually monitored (Hewlett-Packard, Palo Alto, CA). Arterial blood gases were periodically sampled, and ventilation was adjusted to maintain an arterial pH between 7.35 and 7.45. A 7F balloon-

From the Department of Anesthesiology/CCM, University of Pittsburgh Medical Centel; Pittsburgh, PA. Received April 20, 1996; accepted September 6, 1996. Address reprint requests to John A. Kellum, MD, University of Pittsburgh Medical Center; Division of Critical Care Medicine, 200 Lothrop Street, Pittsburgh, PA 15213-2582. Copyright 0 1997 by WE. Saunders Company 0883-9441/97/1201-0001$5.00/O

8 tipped pulmonary artery thermodilution catheter with a 15-cm proximal port (American Edwards model 93A-095, Irvine, CA) was advanced into the pulmonary artery through the left external jugular vein. Placement of the proximal port in the right atrium was verified by waveform analysis of the pressure tracing, and placement of the tip in a pulmonary artery was verified by fluoroscopy. A 7F polyethylene catheter was advanced from the right external jugular vein into the left hepatic vein to a distance of approximately 3 cm beyond the inferior vena cava under fluoroscopic guidance. For measurement of arterial pressure, a 5F catheter with multiple side holes was inserted into the high abdominal aorta through the right femoral artery. A 7F polyvinylchloride catheter was also inserted into the right femoral vein for the administration of a continuous infusion of pentobarbital at 2 to 4 mg X kgg’ X h-l. A splenectomy was performed after maximal splenic contraction to 0.5 mL of topical epinephrine (1: 10,000). The splenic vein was cannulated with a 5F polyethylene catheter that was passed into the portal vein to the level of the portal hepatis. The portal vein and common hepatic artery distal to the origin of the gastroduodenal artery were isolated, and ultrasonic flow probes (Transonic Systems, Ithaca, NY) were placed around each vessel. Care was taken to dissect the common hepatic artery as little as possible, so as not to disrupt the nervous sheath. The left renal artery was isolated, and an ultrasonic flow probe was placed around it. All the flow probes were cemented into place using an agar gel mixture, which minimized movement artifact and improved the acoustic signal. A 5F polyethylene catheter was advanced through the left external jugular vein into the left renal vein at the level of the renal pelvis. The infrahepatic vena cava was isolated, and a hydraulic vascular occluder was placed around it for the purpose of a parallel study. The abdomen was loosely closed with interrupted sutures. All the animals were fluid resuscitated with normal saline, as necessary, to maintain the right atria1 pressure between 2 and 5 mm Hg. The animals were then allowed to stabilize. Stability was defined as constant heart rate, arterial pressure, end-tidal COz, and organ flow signals for at least 30 minutes.

Experimental Protocol Each animal served as its own control. Before receiving endotoxin each animal was studied for a control period of at least 30 minutes. During this time, the animals were maintained in a steady state as defined by stable hemodynamics, end-tidal CO*, and arterial blood gas values. Blood was then collected from all the catheters in heparinized syringes for the determination of blood gases, oxygen content, hematoctit, and levels of phosphate and albumin. Blood flows to organs were measured and recorded in real time on a strip-chart recorder (Gould Inc, Cleveland, OH). After obtaining control measurements, Escherichia coli endotoxin (L-2880 lipopolysaccharide, Sigma, St Louis, MO) (1 mglkg) was infused over 5 minutes through the right atria1 port, and the hemodynamic response of the animal was monitored without resuscitation, This was done to avoid any confounding effect of resuscitation fluid on ion concentrations and system acid-base balance. Thirty to 45 minutes after the administration of endotoxin, when the dog was again in a hemodynamic steady state, measurements were repeated as during the control condition.

KELLUM

ET AL

Measurements and Calculations Blood oxygen saturation, oxygen content, and hemoglobin concentration were measured using a cooximeter calibrated for dog blood (Instrumentation Laboratories model 282, Lexington, MA). Blood gases and pH were analyzed with a blood gas analyzer (Radiometer ABL-30, Copenhagen, Denmark). Organ flow data were obtained from ultrasonic flowmeters calibrated ex viva using a standard perfusion circuit as recommended by the manufacturer (Transonic Systems, Ithaca, NY). Sodium, chloride, potassium, calcium, magnesium, phosphate, and albumin were analyzed using an Ektachem 700 analyzer (Kodak, Rochester, NY) and standard reagents.

Current Analysis For this analysis, metabolic acid was determined using two methods. First, the standard base excess was calculated according to the Siggaard-Andersen equations.23 Next, the effective strong ion difference (SIDe) was calculated from the pc02, phosphate, and albumin levels according to the methods described by Figge et al9 Metabolic acid uptake for the gut and kidney was determined by subtracting the arterial concentration from the venous concentration and then multiplying by flow. For the kidney, total flow was estimated as the sum of left renal artery flow + (left renal artery flow X ratio of right to left kidney wet weight). For the liver, metabolic acid uptake was determined as follows: ([hepatic vein] X hepatic vein flow) ([arterial] X hepatic artery flow) - ([portal vein] X portal vein flow).

Statistical Analysis Statistical comparisons were carried out using the Wilcoxon rank sum test, and correlations were tested for using Spearman’s rank correlation test. Error values given in the text are standard deviations. A P < .05 was considered statistically significant.

RESULTS

After the administration of endotoxin, the mean arterial pressure decreased from 113.6 t 19.2 to 57.0 + 14.2 mm Hg (P < .Ol), the cardiac output decreased from 2.1 + 0.4 to 1.5 2 0.4 Wmin (P < .05), and the systemic oxygen consumption was unchanged (131.2 v 125.2 mL/min). During the control period the arterial pH was 7.34 t 0.04 and arterial SIDe was 24.8 ? 2.5. During the endotoxemic condition the arterial pH decreased to 7.22 5 0.05 (P < .OOl) and arterial SIDe decreased to 20.6 k 2.0 (P = .OOl) (Table 1). This was purely a metabolic acidosis, as the arterial pco2 was unchanged (33.2 t- 3.2 v 33.7 + 5.2 mm Hg). Measurements of metabolic acid from the viscera failed to show the site of acid production. Lactate, although increased during endotoxemia (mean difference 0.69 t 0.28 mmol/L), could explain only 16% of the total acidosis. Both methods of metabolic acid measurement

SPLANCHNIC

BUFFERING

DURING

Table Site

Control Art

9

ENDOTOXEMIA

1. Mean

Na+, Cl-,

Nat

Cl-

HcoB Concentrations,

148.7 + 3.7

124.7 i 2.1

PV HV

149.1 + 3.4 146.1 t 8.3

124.1 2 2.6 120.9 -t 7.5

RV

146.2 -t 7.3 149.7 i 3.7

122.0 f 6.7 123.7 i 1.9

Art PV

148.2 f 3.5 147.6 2 2.4

128.4 i- 2.9t 126.5 -t 2.9

HV RV

148.2 f 2.4 148.5 -t 2.9

126.6 + 2.9t 127.5 i 4.3

MV

150.3 rt 3.3

126.5 + 3.5

MV Endotoxemia

Note: All values Abbreviations:

shown Nat,

are mean

sodium;

-t standard

Cl-, chloride;

vein; RV, renal vein; MV, mixed venous. *Difference from arterial concentration tDifference

between

control

deviation. Hco,,

pH, and SlDe for Each Site PC@

33.2 ? 3.2 36.4 i 9.8

7.34 + .04 7.31 + .04

24.8 2 2.5 24.9 + 4.0

18.8 k 2.3

41.6 i 7.8,

7.28 + .09

25.9 t 2.0

18.2 + 2.6 20.5 + 1.7

37.0 -t 6.7 42.1 i 5.5*

7.32 t .05 7.30 i .05

25.6 2 3.2 27.5 2 2.4"

13.6 2 1.8t 15.7 t 1.6*t 15.7 k 2.1t

33.7 ? 5.2

7.22 i- .05t

20.6 2 2.0t

47.9 2 s.o*i 46.0 t 9.6"

7.18 i .04t 7.19 i .05t

24.3 i- 2.4" 23.3 -c 2.6t*

15.3 2 1.9t

41.3 i 8.8"

7.21 i .05t

17.3 t 1.8t

49.0 -t 9.7"

7.18 ? .06t

22.9 i 3.2 24.5 i 2.2t”

Na+, Cl-, and Hco, are expressed

bicarbonate;

Pco~, carbon

dioxide

tension:

in mmol/L,

and Pco, in mm

Art, arterial:

conditions

HV, hepatic

chloride flux changed from control to endotoxemia conditions for any of the organs studied (Table 2). Arteriovenous desaturation across the gut increased from 25% during the control to 38% during endotoxemia conditions. However, this increase was similar for all organs studied and did not correlate with the metabolic acid gradient across the gut. The mean acid gradient across the gut changed from -0.55 mmol/L during control to 2.47 mmol/L during endotoxemic conditions (mean difference 3.01 2 1.48 mmol/L) (Table 3). There was no correlation between metabolic acid flux and regional oxygen consumption (Vo,) or delivery during control conditions. However, during the endotoxemic condition, uptake of metabolic acid by the gut correlated with gut Voz (r = .81, P < .05) but not oxygen delivery (Fig 2).

Chloride,

Lactate

Across

Cl-

lactate metabolic

IO

-0.98 acid

Gut CIlactate Gut

Fig 1. Histograms showing metabolic acid uptake by the visceral organs during control (solid bars) and endotoxemia (hatched bars) conditions. Metabolic acid uptake was determined by the effective strong ion difference multiplied by flow for each organ. Values shown are in mmollh and represent mean data for 10 animals. *P < .05, Wilcoxon rank sum analysis.

metabolic Kidney CIlactate metabolic

acid

acid

and Metabolic

i 2.05

Endotoxemia

-0.09

PValue

+ 0.35

.71

0.03 t 0.04 -0.24 + 1.05 -0.11 i- 0.37

0.03 t 0.06 -0.07 i 0.24 -0.24 ? 0.34

.82 .I0

-0.03

2 0.03

-0.05

i 0.05

.60 .94

-0.02 -0.22

i 0.73 t 0.39

-0.61 -0.04

ir 0.35 2 0.32

.04* .37

-0.01 -0.10

i 0.01 k 0.26

-0.03 -0.22

-t 0.04 i 0.28

.50 .27

Note: All fluxes shown in mmol/min. Metabolic determined by the effective strong ion difference. are mean i standard *Significant at P<

Acid Flux

Each Organ

Control Liver

Liver

Hg.

vein;

(P < .05)

Table 2. Mean

Kidney

PV, portal

(P < .05).

and endotoxic

30 f z E 25 a % ‘L 20 3 u 2 15 P

0

SlDe

PH

18.0 i 1.8 20.2 2 2.2

(standard base excess v SIDe) produced similar results (Y = .98). The direction of the effect and the statistical significance were identical regardless of the method used. Although transvisceral pH gradients revealed net acid release, the source was purely respiratory (COJ for the gut, liver, and kidney. During the control period, the kidney removed metabolic acid from the circulation (6.03 + 4.97 mmol/h). The gut and liver were neutral for metabolic acid. During early endotoxemia, however, only the gut had increased metabolic acid uptake (36.60 t 6.60 mmol/h, P < .05) (Fig 1). Neither lactate nor

2

pco,,

HCO,

deviation. .05.

acid was All values

10

KELLUM

Table 3. Arterial/Venous Excess

Desaturation Gradient

Across

and Standard

Control SVO~

m 1 2

78.6% 68.5%

3

86.7%

4 5

83.4% 70.4%

6 7

84.6% 67.5%

8 9

46.1% 80.7%

10

86.4%

mean

75.3%

Note:

Data

Endotoxemia

SBE Grad

SW2

-5.84

73.5%

-2.22 -2.14

64.4% 45.5%

for

100%);

2.66 2.48 -0.25 3.47

47.8%

0.56 -6.70 2.92

73.1% 54.2% 69.8%

4.67

50.2%

3.20 4.78

1.26 1.09

74.9% 71.3%

1.89 2.43

62.4%

2.47

the

Svo,, venous

ration was always gradient in mmol/L.

SBE Grad

0.94

-0.55

shown

condition. Abbreviations:

Base

the Gut

gut

during

oxygen SBE Grad,

0.49 3.53

the

saturation standard

endotoxemic (arterial base

satuexcess

DISCUSSION

The cause of metabolic acidosis in sepsis is largely unknown. Lactic acidosis appears to account for only part of the acid load, and a large portion remains unexplained. The results of this study suggest that the role played by the visceral organs in early endotoxemia is more complex than previously recognized. Neither the gut nor the liver showed a net release of metabolic acid during early endotoxemia. Thus, despite baseline uptake and later release of anions as reported previously,16 the liver did not show net metabolic acid uptake or release. This can be explained by the movement of other ions balancing the effect on base excess (either exchange of other ions or the concomitant release or uptake of cations). Interestingly, the flux 14

cs 0

0:2

0:4

0:s 018 Fixed Acid Uptake (mmol/L)

i

112

1:4

Fig 2. Scatter plot for metabolic acid uptake by the gut expressed in mmollmin versus gut oxygen consumption. A regression line shown was calculated by the least-squares method; r = .81, P < .05.

ET AL

of neither of the two most abundant extracellular ions (sodium and chloride) changed from control to endotoxemia conditions for any of the organs studied (Table 1). In contrast, the gut actually took up metabolic acid (in addition to unexplained anions) in the endotoxic condition and did so in relation to VoZ. Whether or not this represents an energy-requiring mechanism is unclear. Decreases in metabolic acid across any vascular bed are consistent with known physiological mechanisms. Blood bicarbonate levels increase as blood passes from the arterial to the venous circulation due to the Haldane effect.7 This effect occurs by two mechanisms. First, and most important, is carbamino carriage, which is greater in reduced than in oxygenated hemoglobin. Second, reduced hemoglobin exhibits greater buffering capacity than oxyhemoglobin7 Thus, fully desaturated blood is more basic than saturated blood when co2 is held constant. A large part of this “buffering” can be attributed to the Hamburger shift (chloride movement into the red cell during oxygen release), which increases the SIDe on the venous side. It would be possible, then, to attribute metabolic acid uptake across a vascular bed to the Haldane effect/ Hamburger shift. However, to explain an increase in metabolic acid uptake with endotoxin, there would need to be a major change in arterial/venous (A/V) desaturation and/or a decrease plasma chloride. Although A/V desaturation across the gut increased from 25% during control to 38% during endotoxemia, this increase was similar to that seen with the other organs and did not correlate with the metabolic acid gradient across the gut (Y = .17). Furthermore, this degree of change in desaturation would be expected to increase the base excess by only about 0.375 mmol/L.20 The measured mean increase in base excess was eight times that (3.01 mmol/L). Even if one ignores the baseline desaturation and given that the mean hemoglobin concentration was 14.3 g/dL and assuming an oxygen transformation factor of 0.19,23 the maximum change in base excess would be 1.03 mmol/L. Therefore, only a small fraction of the total metabolic acid uptake by the gut can be attributed to the Haldane effect. As for the Hamburger shift, plasma chloride concentrations did not change across the gut nor would this effect be expected to occur in the red cells only as they crossed the splanchnic circulation and nowhere else, unless a significant

SPLANCHNIC

BUFFERING

DURING

ENDOTOXEMIA

11

change in AN desaturation occurred in this area. No such change occurred. Another mechanism that can produce increased bicarbonate concentrations in venous blood relative to arterial blood is hemoconcentration. In certain situations, such as ischemia, fluid shifts may effect the concentration of bicarbonate. Changes in free water did not appear to play a role in this study. During endotoxemia, the serum sodium concentrations did not change from the artery to the portal vein (148.2 & 3.5 mmol/L v 147.6 2 2.4 mmol/L). However, serum bicarbonate levels increased across the gut from 13.6 5 1.8 mmol/L to 15.7 +- 1.6 mmol/L (P < .05) (Table 1). The results of this study do not reveal the site of metabolic acid production. An increase in metabolic acid as well as chloride levels in arterial blood during endotoxemia suggests that chloride movement into the intravascular space may be an important mechanism. The mean chloride levels in arterial blood increased during early endotoxemia (mean difference 3.7 + 1.1 mmol/L, P < .05). This finding could be a result of Donnan equilibrium in the setting of increased protein transudation from the vascular space. l1 A surprising finding in this study was the small degree to which lactate was responsible for the development of metabolic acidosis (16%). However, this finding may be misleading because we did not measure intracellular lactate. It is possible that intracellular acid (such as lactate) increased significantly during the experiment. Such acid production might explain the marked increase in co2 production observed in our animals. This

would occur presumably as intracellular bicarbonate was titrated by the acid and released from the cells as co2. This acid production might also produce movements in other measured ions (such as chloride) as the cells attempt to normalize their pH by pumping anions out or cations in. This hypothesis warrants further investigation. Another potential mechanism is unreversed adenosine triphosphate (ATP) hydrolysis. It has been proposed that the hypoxic cell will continue to hydrolyze ATP-releasing hydrogen ions into the circulation.13 Such a process could also explain chloride movement into the circulation to preserve electrical neutrality. This explanation is unsatisfactory however, because cellular ATP depletion does not appear to occur in experimental animal models of sepsis.*” In human studies, tissue po2 actually increases with worsening sepsis, further challenging the notion that cellular hypoxia is the major pathophysiological abnormality in sepsis.2 In summary, although anions are released by the liver (which may produce systemic acidemia), neither the liver nor the gut appears to release metabolic acid during early endotoxemia. In fact, the gut becomes a major site of metabolic acid removal and does so in relation to regional Vo2. The kidney also clears metabolic acid during control conditions, but fails to increase its clearance during early endotoxemia. Changes in hemoglobin buffering attributable to increased AN desaturation across the gut (Haldane effect/Hamburger Shift) are of insufficient magnitude to explain the metabolic acid uptake (“splanchnic buffering”) seen.

REFERENCES 1. Bellomo R, Kellum JA, Pinsky MR: Transvisceral lactate fluxes during early endotoxemia. Chest 110: 198-204, 1996 2. Boekstegers P, Weidenhofer muscle partial pressure of oxygen Care Med 22:640-650, 1994

S, Kapsner in patients

T, et al: Skeletal with sepsis. Crit

3. Cain SM, Curtis SE: Systemic and regional oxygen uptake and delivery and lactate flux in endotoxic dogs infused with dopexamine. Crit Care Med 19:1552-1560, 1991 4. Connett RJ, Honig CR, Gayeski TE, et al: Defining hypoxia: A systems view of VOW, glycolysis, energetics, and intracellular Paz. J Appl Physiol68:833-842, 1990 5. Cowan BN, Burns HJ, Boyle P, et al: The relative prognostic value of lactate and haemodynamic measurements in early shock. Anaesthesia 39:750-755, 1984 6. Curtis SE, Cain SM: Regional and systemic oxygen delivery/uptake relations and lactate flux in hyperdynamic endotoxin-treated dogs. Am Rev Respir Dis 145:348-354, 1992 7. Ferguson JKW, Roughton FJW: The direct chemical

estimation of carbamino compounds of Co2 with haemoglobin. J Physiol83:68-77, 1934 8. Fiddian-Green R: Hypotension, splanchnic hypoxia and arterial acidosis in ICU patients. Circulatory Shock 21:326-332, 1987 9. Figge J, Mydosh T, Fencl V: Serum proteins and acid-base equilibria: A follow-up. J Lab Clin Med 120:713-719, 1992 10. Fowler AA, Hamman RF, Zerbe GO, et al: Adult respiratory distress syndrome-Prognosis after onset. Am Rev Respir Dis 132:472-478, 1985 11. Ganong WF: Review of Medical Physiology (ed 15). Norwalk, CT, Lanage, 1991, p 6 12. Gilfix BM, Bique M, Magder S: A physical chemical approach to the analysis of acid-base balance in the clinical setting. J Crit Care 8:187-197, 1993 13. Gores GJ, Flarsheim CE, Dawson TL, et al: Swelling, reductive stress and cell death during chemical hypoxia in hepatocytes. Am J Physiol257:C347-C354, 1989 14. Hotchkiss RS: Rust RS, Dence CS, et al: Evaluation of

12

the role of cellular hypoxia in sepsis by hypoxic marker [18F] fluoromisonidazole. Am J Physiol261:R965-R972, 1991 15. Kellum JA, Bellomo R, Kramer DJ, et al: Etiology of metabolic acidosis during saline resuscitation in endotoxemia. Am J Res Crit Care Med 151:A318,1995 16. Kellum JA, Bellomo R, Kramer DJ, et al: Hepatic anion flux during acute endotoxemia. J Appl Physiol 78:2212-2217, 1995 17. Kellum JA, Kramer DJ, Pinsky MR: Strong ion gap: A methodology for exploring unexplained anions. J Crit Care 10:51-55, 1995 18. Mecher C, Rackow EC, Astiz ME, et al: Unaccounted for anion in metabolic acidosis during severe sepsis in humans. Crit CareMed 19:705-711, 1991 19. Mizock BA, Falk JL: Lactic acidosis in critical illness. Crit Care Med 20:80-93, 1992 20. Nunn JF: Nunn’s Applied Respiratory Physiology (ed 4). Oxford, England, Butterworth-Heinemann, 1993 pp 219-227 21. Peretz DI, Scott HM, Duff J, et al: The significance of lactic acidemia in the shock syndrome. Ann NY Acad Sci 119:1133-1141. 1965

KELLUM

ET AL

22. Salzman AL, Wang H, Wollert PS, et al: Endotoxininduced ileal mucosal hyperpermeability in pigs: Role of tissue acidosis. Am J Physiol266:G633-G646, 1994 23. Siggaard-Andersen 0: The Acid-Base Status of Blood (ed 4). Copenhagen, Munksgaard, 1974, p 97 24. Stacpoole PW, Wright EC, Baumgartner TG, et al: Natural history and course of acquired lactic acidosis in adults. Am J Med 97:47-54, 1994 25. Stacpoole PW, Harman EM, Curry SH, et al: Treatment of lactic acidosis with dichloroacetate. N Engl J Med 309:390396, 1983 26. van Lambalgen AA, Runge HC, van den Bos GC, et al: Regional lactate production in early canine endotoxin shock. Am J Physiol254:E45-E51,1988 27. Vincent JL, Dufaye P, Berre J, et al: Serial lactate determinations during circulatory shock. Crit Care Med 11:449451,1983 28. Young LS: Gram negative sepsis, in Mandell GL, Douglas RG Jr, Bennett JE (eds): Principles and Practice of Infectious Diseases. New York, NY, Churchill Livingstone, 1990, pp 611-632