Background

Dichloroacetate (DCA) is an effective alternative to bicarbonate to treat lactic acidosis and stabilize acid-base homeostasis during liver transplantation. Although DCA presumably is metabolized by the liver, the impact of end-stage liver disease and liver transplantation on DCA distribution and elimination is unknown. Therefore, the pharmacokinetics of DCA were determined in patients with end-stage liver disease undergoing orthotopic liver transplantation.

Methods

Thirty-three patients undergoing liver transplantation were given DCA as two 40-mg/kg infusions over 60 min, the first after induction of anesthesia, the second 4 h later. Plasma DCA concentrations were determined by gas chromatography-mass spectroscopy. One- and two-compartment pharmacokinetic models were fitted to DCA concentrations versus time data using a mixed-effects population approach. Various models permitted changes in central compartment volume and/or plasma clearance to account for changes in hepatic mass and function and circulatory status during the paleohepatic, anhepatic, and neohepatic periods.

Results

The optimal model had two compartments. DCA clearance was 0.997, 0.0, and 1.69 ml x kg(-1) x min(-1) during the paleohepatic, anhepatic, and neohepatic periods, respectively (P < 0.05). Interindividual variability in central compartment volume differed during the paleohepatic and neohepatic periods. There was no apparent influence of blood or fluid requirements during surgery on DCA clearance or volume of distribution.

Conclusions

Absence of DCA clearance during the anhepatic period indicates that DCA is metabolized exclusively by the liver. Differences in interindividual variability in central compartment volume during the paleohepatic and neohepatic periods possibly result from physiologic changes during surgery. Finally, the results indicate that the newly transplanted liver eliminates DCA better than the native liver.

Key words: Dichloroacetate. Liver: disease; transplantation. Pharmacokinetic modeling: NONMEM; population techniques.

DICHLOROACETATE (DCA) is an effective alternative to sodium bicarbonate for treatment of lactic acidosis. [1]DCA activates the regulatory enzyme pyruvate dehydrogenase, promoting oxidation of pyruvate, thereby decreasing formation of lactic acid from pyruvate. [2]In addition, DCA stimulates hepatic pyruvate dehydrogenase and increases hepatic uptake of circulating lactic acid. [3-5]Because of these effects, DCA has been used to attenuate the marked lactic acidosis that accompanies orthotopic liver transplantation. [6].

Despite the potential utility of DCA, its clinical pharmacology is complex. Conventional one-compartment modeling of DCA pharmacokinetics in human volunteers suggests accumulation during repeated administration, prompting the unproved hypothesis that DCA inhibits its own metabolism. [7,8]However, a one-compartment model may be inadequate to describe the pharmacokinetics of DCA in humans. Thus, we examined whether the pharmacokinetic profile of DCA in humans is adequately described by a one-compartment pharmacokinetic model.

Experimental evidence indicates that the liver may play an important role in removing DCA from plasma. First, less than 1% of intravenously administered DCA is excreted intact in urine. [9]Second, oxalate production by incubated liver is increased in the presence of DCA. [10]Third, urinary excretion of oxalate is increased sevenfold after intravenous DCA administration. [7]These findings have important implications in the setting of liver transplantation, where the native liver has end-stage disease and the capacity of the graft liver to metabolize DCA is unknown. Furthermore, the native liver is removed before the graft liver is anastomosed, causing an approximate 1-h period--the "anhepatic stage"--during which there is no hepatic function. Therefore, we examined two other questions: (1) How important is the liver in the metabolism of DCA in vivo, and (2) what is the effect of the graft liver versus the native liver on DCA clearance during liver transplantation?

The protocol was approved by the U.S. Food and Drug Administration (IND no. 35,790) and institutional review boards of the Oregon Health Sciences University and the Portland (Oregon) Veterans Administration Medical Center. After obtaining written informed consent, we studied 33 patients with end-stage liver disease having a Pugh-Child score [11]ranging from 6 to 14 (median 10) undergoing orthotopic liver transplantation. Details of the clinical course of 28 of these patients have been described previously [6]; the additional five patients had similar diagnoses (Table 1) and clinical courses. Despite the presence of end-stage liver disease, all patients were clinically stable preoperatively, and none had fulminant hepatic failure. Surgery was performed without venovenous bypass. Duration (mean+/-SD) of the paleohepatic period was 207+/-33 min (range 102-460 min); the anhepatic period lasted 73+/-20 min (range 47-136 min).

Table 1. Diagnoses for 33 Patients Undergoing Liver Transplantation

Table 1. Diagnoses for 33 Patients Undergoing Liver Transplantation
Table 1. Diagnoses for 33 Patients Undergoing Liver Transplantation

After induction of anesthesia, patients received 40 mg/kg intravenous DCA in 150 ml of D5W, infused over 60 min. A second, identical, dose of DCA was administered 240 min after the start of the initial infusion. The second DCA infusion was started before the anhepatic stage in 26 patients and during the anhepatic stage in the remaining 7. This infusion was completed before the anhepatic stage in 14 patients, during the anhepatic stage in 11 (at an average of 30 min into the anhepatic stage), and after reperfusion of the graft liver in 8. In one of the eight patients in whom the second DCA infusion continued into the reperfusion stage, the infusion spanned the entire anhepatic stage, because that stage lasted only 47 min.

Arterial blood samples, 10 ml each, were collected into iced heparinized tubes before, at 15, 30, 45, 60, 75, and 90 min, and at 2, 3, and 4 h after the beginning of the first infusion and at 15, 30, 45, 60, 75, and 90 min and at 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 14, 15, 16, 18, and 20 h after the beginning of the second infusion; total sampling interval was 24 h. Blood was centrifuged at 4 degrees C and the plasma immediately stored at -70 degrees C until analysis. DCA plasma concentrations (Cp) were determined by gas chromatography-mass spectroscopy. [6,12]The assay is sensitive to DCA concentrations of 0.4 micro gram/ml with a coefficient of variation of 2.4% at that concentration.

The pharmacokinetics of DCA were determined using a population approach, [13]i.e., values for all subjects were analyzed simultaneously to yield estimates for "typical values" of the pharmacokinetic parameters, standard errors (precision) of these estimates, and interindividual variability. The analysis was performed using the NONMEM program.*

Parameters for the one-compartment model were volume of the central compartment (V) and clearance (Cl, equal to V *symbol* k0). Each parameter was permitted to have one or two components, one proportional to weight, the other constant. Parameters for the two-compartment model were volume of the central compartment (V1), volume of the second compartment (V2), clearance (Cl, equal to V1*symbol* k10), and distribution clearance (Cldistribution, equal to V1*symbol* k12). Volume of distribution at steady-state (Vss) was equal to V1plus V2. Because the one-compartment analysis indicated that all parameters should be weight-proportional only (see results), all two-compartment models were weight-proportional only. For the two-compartment model, distribution and elimination half-lives (t1/2 alpha and t1/2 beta, respectively) were determined using standard equations. [14].

Variability between subjects was modeled by expressing the pharmacokinetic parameters of each individual as a function of the typical value for the population and a factor for that individual. Because inter-individual variability tends to assume a log-normal (i.e., skewed) distribution, interindividual variability for clearance was modeled as: Equation 1where Cliis the estimate for clearance for the ithindividual, Cl is the typical value for the population, and etaiis a random variable with mean 0.0. All models permitted interindividual variability in Cl and V (or Vi); some models permitted interindividual variability in V2or Cldistribution. In some models, interindividual variability was permitted to differ during the times at which the native and the transplanted liver were being perfused (the paleohepatic and neohepatic periods, respectively).

The effect of liver transplantation was evaluated by testing various models in which Cl, V (or V1), Cldistribution, and V2were permitted step changes at the time of test clamping** or actual clamping of the hepatic vasculature. In addition, each of Cl and V (or V sub 1) was permitted to assume a new mean value and interindividual variability during the neohepatic period. Because each of these new models requires additional terms (called thetas in NONMEM parlance), they were used only if they significantly improved both the pattern of residual differences between observed and predicted values and the objective function (P < 0.05, e.g., a model with one additional theta improved the objective function by 3.8). In addition, for each model, we determined the differences between individual parameter estimates (determined using the NONMEM post hoc step) and "typical" values (determined from NONMEM's population fit). These differences were plotted against the number of units of blood, fresh-frozen plasma, and platelets transfused and the volume of Cell Saver (Haemonetics, Braintree, MA) solution and crystalloid administered; trends were sought by plotting a smoother (Supersmoother***) through these data and examining for a systematic deviation of this smoother from the horizontal line with zero elevation.

Recognizing that this model is potentially flawed (see discussion), we evaluated one additional four-compartment model. This model has the same central and peripheral compartments as the two-compartment model described previously. The two additional compartments represent the native and the transplanted liver, respectively. At the time that the hepatic vasculature of the native liver is clamped, drug movement to and from the native liver (compartment 3) decreases to zero so that all DCA within the native liver is irreversibly removed from the pharmacokinetic analysis. Conversely, before perfusion of the transplanted liver (compartment 4), no drug is present in that compartment; at the time of perfusion, drug moves to and from the central compartment with the appropriate rate constants. In addition, irreversible elimination (clearance) is permitted from compartment 3 during the paleohepatic period and from compartment 4 during the neohepatic period. This model was compared to previous models by its effect on the objective function and the pattern of residual differences.

DCA Cp values increased during the initial 1-h infusion, decreased during the next 3 h, then increased during the second 1-h infusion by an amount similar to that during the first infusion (Figure 1). The rate of decrease in Cp values during the next 3 h (during which the anhepatic period usually occurred) was slower than that during the comparable period after the first infusion. Eleven patients had fewer than three Cp values during the anhepatic period (Figure 2). The remaining 22 patients had three to six Cp values during the anhepatic period; in 11 of these, there were Cp values after termination of the second DCA infusion but during the anhepatic period.

Figure 1. Time course of plasma dichloroacetate (DCA) concentration during liver transplantation. The x-axis is time, and the y-axis is (log) DCA concentration, expressed as mean (SE). DCA (40 mg/kg) was infused for 60 min beginning at time 0 and at 240 min (thick lines).

Figure 1. Time course of plasma dichloroacetate (DCA) concentration during liver transplantation. The x-axis is time, and the y-axis is (log) DCA concentration, expressed as mean (SE). DCA (40 mg/kg) was infused for 60 min beginning at time 0 and at 240 min (thick lines).

Close modal

Figure 2. Dichloroacetate (DCA) plasma concentration (Cp) values during the anhepatic period (circles = individual values; lines connect values for individuals). The x-axis is time, and the y-axis is (log) DCA concentration. Eleven patients have fewer than three Cp values. In some of the remaining patients, DCA Cp remained relatively constant after the second DCA infusion.

Figure 2. Dichloroacetate (DCA) plasma concentration (Cp) values during the anhepatic period (circles = individual values; lines connect values for individuals). The x-axis is time, and the y-axis is (log) DCA concentration. Eleven patients have fewer than three Cp values. In some of the remaining patients, DCA Cp remained relatively constant after the second DCA infusion.

Close modal

With the one-compartment model, in which interindividual variability was allowed for both Cl and V, permitting both Cl and V to be weight-proportional improved the objective function markedly compared to models in which Cl and V were constant (Table 2). Permitting both a weight-proportional and constant component for Cl and V did not further improve the fit. Therefore, all additional fitting allowed the pharmacokinetic parameters to be weight-proportional only.

Table 2. One-compartment Pharmacokinetic Models Tested and Their Objective Function

Table 2. One-compartment Pharmacokinetic Models Tested and Their Objective Function
Table 2. One-compartment Pharmacokinetic Models Tested and Their Objective Function

The two-compartment model, in which interindividual variability was allowed for both Cl and V1markedly improved the objective function and quality of the fits compared to the one-compartment model (Table 3). However, this model demonstrated a systematic misfit of the Cp values during the anhepatic period: Postinfusion Cp values were predicted to decrease significantly in all subjects, whereas some subjects demonstrated a slight increase and others a smaller-than-expected decrease (as compared to the decrease observed at a comparable period in relation to the first DCA infusion). This systematic misfit suggested the need to evaluate a model in which one or more pharmacokinetic parameters changed at the onset (and completion) of the anhepatic period.

Table 3. Two-compartment Pharmacokinetic Models Tested and Their Objective Function

Table 3. Two-compartment Pharmacokinetic Models Tested and Their Objective Function
Table 3. Two-compartment Pharmacokinetic Models Tested and Their Objective Function

All models in which step changes in V1or Cl occurred at actual clamping were similar to or better than models in which these changes occurred at test clamping; therefore, no additional models involving test clamping were evaluated. Permitting Cl to change at the time of clamping and return to the baseline value at reperfusion improved the objective function and the fit; with this model, Cl was absent during the neohepatic period.**** Additional models were tested in which V1, Cldistribution, and V2were permitted to change individually or in combination at the time of clamping and return to the baseline value at reperfusion. These values were not justified statistically compared to the model in which only Cl changed during the anhepatic period.

The next model tested permitted Cl to have a different value during each of the paleohepatic, anhepatic, and neohepatic periods. This model indicated that Cl was absent during the neohepatic period and was greater during the neohepatic period than during the paleohepatic period. Permitting V1to assume a new interindividual variation at the time of reperfusion (and allowing Cl to return to the baseline value) further improved this model. There were no effects of the number of units of blood, fresh-frozen plasma, or platelets transfused or the volume of Cell Saver solution and crystalloid administered on the post hoc individual pharmacokinetic parameters (Figure 3).

Figure 3. The influence of transfusion (as an indicator of blood loss during surgery) on dichloroacetate (DCA) clearance (Cl) during liver transplantation. The x-axis is the number of units of erythrocytes transfused during surgery. The y-axis is (log) Cl, determined using NONMEM's post hoc step. Each point is the value from a single individual. The thick line at Cl of 0.997 ml *symbol* kg sup -1 *symbol* min sup -1 is the typical value for the population. The thin line, determined using a smoother (Supersmoother), suggests that there is no relationship between the quantity of blood transfused and Cl.

Figure 3. The influence of transfusion (as an indicator of blood loss during surgery) on dichloroacetate (DCA) clearance (Cl) during liver transplantation. The x-axis is the number of units of erythrocytes transfused during surgery. The y-axis is (log) Cl, determined using NONMEM's post hoc step. Each point is the value from a single individual. The thick line at Cl of 0.997 ml *symbol* kg sup -1 *symbol* min sup -1 is the typical value for the population. The thin line, determined using a smoother (Supersmoother), suggests that there is no relationship between the quantity of blood transfused and Cl.

Close modal

The additional four-compartment model, which accounted for irreversible drug loss from the native liver and perfusion of the transplanted liver at the time of reperfusion, did not further improve the fit.

Thus, the final ("optimal") model had two compartments and permitted Cl to decrease to zero at the time of clamping and to exceed its baseline value at the time that the transplanted liver was perfused (Table 4). With this model, there were no systematic differences between observed and predicted values during any of the three periods of liver transplantation (Figure 4). Although the value of V1remained constant throughout the sampling period, its interindividual variability differed during the paleohepatic and neohepatic periods. During the paleohepatic period, "typical" distribution and elimination half-lives were 20 and 455 min, respectively. During the neohepatic period, "typical" distribution and elimination half-lives were 20 and 279 min, respectively. Vss was 618 ml/kg during all periods.

Table 4. "Typical" Pharmacokinetic Parameters for the Optimal Model

Table 4. "Typical" Pharmacokinetic Parameters for the Optimal Model
Table 4. "Typical" Pharmacokinetic Parameters for the Optimal Model

Figure 4. Relationship between observed plasma dichloroacetate (DCA) concentration and that predicted by NONMEM's post hoc step with the optimal two-compartment model. The x-axis is (log) time, and the y-axis is the ratio of observed to predicted DCA concentration. Each line is from a single individual. Dotted lines = paleohepatic period; solid lines = anhepatic period; dashed lines = neohepatic period. If observed values were identical to predicted values, each line would lie horizontally at 100%.

Figure 4. Relationship between observed plasma dichloroacetate (DCA) concentration and that predicted by NONMEM's post hoc step with the optimal two-compartment model. The x-axis is (log) time, and the y-axis is the ratio of observed to predicted DCA concentration. Each line is from a single individual. Dotted lines = paleohepatic period; solid lines = anhepatic period; dashed lines = neohepatic period. If observed values were identical to predicted values, each line would lie horizontally at 100%.

Close modal

In the current study, we demonstrate that a two-compartment model is needed to fit the plasma DCA concentration versus time data. Using this two-compartment model, DCA clearance is absent during the anhepatic period and then recovers to greater than its baseline value during the neohepatic period. In addition, interindividual variability in central compartment volume during the neohepatic period differs from that observed during the paleohepatic period. Finally, we were unable to demonstrate that DCA pharmacokinetics were influenced by other factors during surgery, such as the number of units of blood or platelets transfused.

Our finding that a two-compartment model is needed to fit the Cp versus time data differs from that of previous studies in which a one-compartment model was adequate. [6,7,15]We speculate that this difference results from the other investigators obtaining their initial sample 30 min after completing a DCA infusion (whereas we obtained our first sample 15 min after beginning an infusion), thereby limiting their ability to identify initial distribution. Recently, Stacpoole and associates reevaluated their DCA pharmacokinetic data in patients with lactic acidosis and found that pharmacokinetics models with two or more compartments were needed in 45 of 111,***** supporting our finding that a one-compartment model is inadequate.

Stacpoole and associates [7,15]also reported that the elimination half-life of DCA increased with its repeated administration. Such a finding can be explained by the longer sampling interval available with larger or repeated doses, revealing that the Cp versus time curve flattens with time. Because our subjects all received the same DCA dose, we have limited information regarding its pharmacokinetic dose-linearity in patients with end-stage liver disease. However, the initial increase and decrease in plasma DCA concentration associated with the second DCA dose were comparable to those of the first dose (Figure 1).

We observed that the pharmacokinetics of DCA changed at several stages of liver transplantation. At the time that the vasculature of the native liver was clamped, Cl decreased acutely, approaching zero. Despite our analysis indicating that Cl was absent during the anhepatic period, two factors limit our knowledge about the magnitude of the change in Cl during this period. First, because one-third of our patients had fewer than three Cp values during the anhepatic period, we have limited information to determine the influence of the anhepatic period on the pharmacokinetics of DCA in some subjects. However, in the remaining subjects, there are sufficient samples to observe differences in the shape of the Cp versus time curve during the paleohepatic and neohepatic periods. Second, the anhepatic period is brief (mean duration 73 min), inadequate to determine a complete pharmacokinetic profile of a drug with DCA's distribution and elimination characteristics. However, similar modeling has been performed by Fiset et al. [16]to determine the influence of a 68-min period of cardiopulmonary bypass on the pharmacokinetics of alfentanil in children.

Examining our Cp values during the anhepatic period suggests the need to permit one or more pharmacokinetic parameters to change at the onset and completion of the anhepatic period: Of 11 subjects in whom adequate Cp values are available during the anhepatic period and after completion of the DCA infusion, 7 demonstrate a slower absolute decrease in Cp compared to the similar period after the first DCA infusion. This supports our belief that initiation of the anhepatic period influences the pharmacokinetics of DCA. Our modeling approach suggested that the largest statistical improvement occurred by permitting Cl rather than Cldistribution, V1, or V2(or various combinations of the latter three) to change at the onset of the anhepatic period. However, it is possible that a combination of physiologic distributional changes during the anhepatic period manifest as a change in Cl. Further insight might be provided by studies in which additional samples are obtained during the anhepatic period and in which the anhepatic period was longer, perhaps in animals. Regardless, the Cp data suggests that the pharmacokinetics of DCA are markedly influenced by the anhepatic period and our analysis suggests that the most likely explanation is an absence of Cl during that period. This absence of Cl during the anhepatic period indicates that DCA is eliminated by the liver exclusively. The importance of the liver to DCA metabolism is consistent with the findings of Curry et al., [7]who observed a sevenfold increase in urinary oxalate excretion after DCA administration.

Our model also indicates that the transplanted liver eliminates DCA better than the native liver, despite a lengthy period of hypothermic ischemia. Unlike the anhepatic period, during which some subjects have few samples, each subject had numerous samples obtained over several hours during each of the paleohepatic and neohepatic periods. This should permit adequate identification of DCA's pharmacokinetics during each of these periods. Few studies have examined drug metabolic function of the newly transplanted liver. Kelley et al. [17]used an isolated perfused rat liver model to compare freshly harvested livers to those preserved hypothermically for 24 h in a medium similar to that used clinically. These investigators found that preserved livers recovered their ability to extract vecuronium, fentanyl, and morphine quickly and completely. [17]Considering the data of Kelley et al. with those from the current study, it appears that the ability of the liver to metabolize many drugs recovers within minutes after transplantation.

Our model also indicated that values for V1varied across the population to a greater magnitude during the neohepatic period than during the paleohepatic period. We speculate that this might result from the many physiologic changes that occur during the anhepatic period and at the onset of the neohepatic period (e.g., rapid changes in circulating blood volume, cardiac output, and body temperature).

Our optimal model does not account for DCA in the native liver irreversibly lost when that liver is removed from the circulation nor does it account for DCA that distributes into the transplanted liver. To explain this modeling error, we tested an additional model accounting for these distributional changes: This model failed to provide a better fit to the data than models not accounting for these changes. We offer two possible explanations. First, the magnitude of physiologic changes related to removal and replacement of the liver may vary enough between individuals to obscure the loss of DCA with removal of the native liver and distribution of DCA to the new liver. Alternately, the quantity of DCA present within the native liver may be sufficiently small that its removal cannot be observed. This is best appreciated by considering that, because the Vss of DCA is 62% of body weight, DCA is functionally distributed in total body water. The liver, which is 75% water by mass and accounts for 2% of body weight, represents about 2.4% of total body water. Thus, if DCA distributed evenly in its Vss, the liver would contain only 1/40th of the total body DCA content at the time of hepatectomy.

One limitation of this study is the lack of comparable pharmacokinetic data in subjects with normal hepatic function. For example, there are no data from anesthetized healthy subjects or patients undergoing nonhepatic surgery. Studies performed in awake volunteers or critically ill patients with lactic acidosis have employed a different study design, precluding meaningful comparison. [7,15]However, all patients enrolled in the current study were clinically stable, and none had fulminant hepatic failure. Additional studies are in progress to determine whether fulminant liver failure affects metabolism of DCA. Many patients with fulminant hepatic failure requiring liver transplantation also have hepatorenal syndrome and are hemodialyzed perioperatively. Hemodialysis augments plasma clearance of DCA, [18]which might further influence the pharmacokinetics of DCA in patients undergoing liver transplantation.

One additional limitation regards our ability to define the value for V1. Several factors influence the ability to define V1correctly. First, immediately after drug administration, venous concentrations are less than corresponding arterial values and tend to provide overestimates of V1; however, all samples in the current study were arterial. Second, frequent sampling immediately after drug administration, particularly with bolus administration, is needed to define rapid initial changes in Cp. Conversely, absence of this rapid initial sampling (as in the current study, in which the first Cp value was obtained 15 min after beginning the initial infusion) might overestimate V1. However, as the intention of our study is to provide pharmacokinetic parameters to guide prolonged nonchanging infusions of DCA, the current study should provide appropriate pharmacokinetic parameters.

Traditional pharmacokinetic modeling assumes that Cl, distributional clearances, and volumes of the various compartments remain constant during the period of sampling. For many drugs and physiologic situations, this assumption may be appropriate. However, in other situations, such as during cardiopulmonary bypass or during removal or replacement of an organ from which the drug is eliminated, the assumption is likely to be flawed. Recently, Szenohradszky et al. [19]could not demonstrate that rocuronium's Cl increased at the time that the transplanted kidney was perfused, suggesting minimal elimination of rocuronium via the kidney. In contrast, in the current study, we demonstrate that Cl is absent during the anhepatic period, strongly indicating a predominant role of the liver in elimination of DCA. Newer pharmacokinetic software tools, such as NONMEM, permit novel modeling of these physiologic changes and may offer insight into the role of physiologic changes in drug elimination or distribution.

In summary, DCA pharmacokinetics are best described by a two-compartment model. Clearance is absent during the anhepatic period, indicating that DCA's major, if not sole, route of elimination is via hepatic metabolism. During the neohepatic period, clearance exceeds its value during the paleohepatic period, indicating that the newly transplanted liver is able to metabolize DCA better than the native liver. Finally, we observed that interindividual variability in V1differed during the paleohepatic and neohepatic periods, possibly the result of the marked physiologic changes during liver transplant surgery.

*Beal SL, Sheiner LB: NONMEM User's Guide. San Francisco, University of California, San Francisco, 1992.

**The suprahepatic inferior vena cava is briefly clamped late in the dissection stage to test for hemodynamic stability in the setting of decreased venous return. This occurs 10-40 min before portal vein clamping (initiation of the anhepatic stage).

***Statistical Sciences, Inc: Modern regression methods, S-plus User's Manual. Version 3.0. Seattle, Statistical Sciences, Inc, 1991, pp 1-46.

****Because NONMEM does not permit clearance (Cl) to equal zero, we tested models in which the lower limit of Cl during the anhepatic period was 0.001 times the value before the anhepatic period. Cl reached that value, indicating there was no clearance during the anhepatic period.

*****Curry SH, Henderson GN: Personal communication. 1995.

1.
Stacpoole PW, Lorenz AC, Thomas RG, Harman EM: Dichloroacetate in the treatment of lactic acidosis. Ann Intern Med 1988; 108:58-63.
2.
Wolfe RR, Jahoor F, Herndon DN, Miyoshi H: Isotopic evaluation of the metabolism of pyruvate and related substrates in normal adult volunteers and severely burned children: Effect of dichloroacetate and glucose infusion. Surgery 1991; 110:54-67.
3.
Shangraw RE, Rabkin JM, Lopaschuk GD: Dichloroacetate activates hepatic pyruvate dehydrogenase in humans undergoing liver transplantation (abstract). ANESTHESIOLOGY 1994; 81(suppl):A277.
4.
Park R, Arieff AI: Treatment of lactic acidosis with dichloroacetate in dogs. J Clin Invest 1982; 70:853-62.
5.
Graf H, Leach W, Arieff AI: Effects of dichloroacetate in the treatment of hypoxic lactic acidosis in dogs. J Clin Invest 1985; 76:919-23.
6.
Shangraw RE, Winter R, Hromco J, Robinson ST, Gallaher EJ: Amelioration of lactic acidosis with dichloroacetate during liver transplantation in humans. ANESTHESIOLOGY 1994; 81:1127-38.
7.
Curry SH, Chu PI, Baumgartner TG, Stacpoole PW: Plasma concentrations and metabolic effects of intravenous sodium dichloroacetate. Clin Pharmacol Ther 1985; 37:89-93.
8.
Stacpoole PW: The pharmacology of dichloroacetate. Metabolism 1989; 38:1124-44.
9.
Lukas G, Vyas KH, Brindle SD, Le Sher AR, Wagner WE Jr: Biological disposition of sodium dichloroacetate in animals and humans after intravenous administration. J Pharm Sci 1980; 69:419-21.
10.
Demaugre F, Cepanec C, Leroux JP: Characterization of oxalate as a catabolite of dichloroacetate responsible for the inhibition of gluconeogenesis and pyruvate carboxylation in rat liver cells. Biochem Biophys Res Commun 1978; 85:1180-5.
11.
Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R: Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973; 60:646-9.
12.
Haessler R, Davis RF, Wolff RA, Kuzume K, Shangraw R, Van Winkle DM: Dichloroacetate reduces plasma lactate levels but does not reduce infarct size in rabbit myocardium. Shock 1996; 5:66-71.
13.
Beal SL, Sheiner LB: Methodology of population pharmacokinetics, Drug Fate and Metabolism: Methods and Techniques. Edited by Garrett E, Hirtz J. New York, Marcel Dekker, 1985, pp 135-83.
14.
Gibaldi M, Perrier D: Pharmacokinetics. 2nd edition. New York, Marcel Dekker, 1982, p 48.
15.
Stacpoole PW, Wright EC, Baumgartner TG, Bersin RM, Buchalter S, Curry SH, Duncan CA, Harman EM, Henderson GN, Jenkinson S, Lachin JM, Lorenz A, Schneider SH, Siegel JH, Summer WR, Thompson D, Wolfe CL, Zorovich B: A controlled clinical trial of dichloroacetate for treatment of lactic acidosis in adults. N Engl J Med 1992; 327:1564-9.
16.
Fiset P, Mathers L, Engstrom R, Fitzgerald D, Brand SC, Hsu F, Shafer SL: Pharmacokinetics of computer-controlled alfentanil administration in children undergoing cardiac surgery. ANESTHESIOLOGY 1995; 83:944-55.
17.
Kelley SD, Cauldwell CB, Fisher DM, Lau M, Sharma ML, Weisiger RA: Recovery of hepatic drug extraction after hypothermic preservation. ANESTHESIOLOGY 1995; 82:251-8.
18.
Curry SH, Lorenz A, Henderson GN, Mars DR, Stacpoole PW: Haemodialysis studies with dichloroacetate. Eur J Clin Pharmacol 1991; 40:613-7.
19.
Szenohradszky J, Fisher DM, Segredo V, Caldwell JE, Bragg P, Sharma ML, Gruenke LD, Miller RD: Pharmacokinetics of rocuronium bromide (ORG 9426) in patients with normal renal function or patients undergoing cadaver renal transplantation. ANESTHESIOLOGY 1992; 77:899-904.