Ketamine Distribution Described by a Recirculatory Pharmacokinetic Model Is Not Stereoselective 

Five male dogs, weighing 27.5–33 kg (30.2 ± 2.4 kg, table 1), were studied in this Institutional Animal Care and Use Committee–approved study. After an overnight fast, during which water was allowed ad libitum,  the dog was brought to the laboratory. Anesthesia was induced with methohexital (10–15 mg/kg intravenous)via  a foreleg vein, the trachea was intubated with a 9-mm endotracheal tube, and the animal was placed in the left lateral decubitus position. Mechanical ventilation was instituted at a tidal volume of 20–25 ml/kg and at a rate sufficient to maintain end-tidal carbon dioxide tension at 30 ± 5 mmHg. Anesthesia was maintained with 2.0% halothane in oxygen. End-tidal halothane concentrations were monitored with an infrared multiple gas analyzer (PPG Biomedical Systems, Lenexa, KS) after its calibration with known standards.

Using a modified Seldinger technique, an 8-French percutaneous sheath introducer was inserted into an external jugular vein. Arterial access for blood sampling by roller pump or syringe was achieved by inserting a 16-gauge (Angiocath; Becton Dickinson, Sandy, UT) catheter percutaneously into the femoral artery; this also allowed systemic arterial blood pressure to be monitored via  a solid-state pressure transducer (Baxter-Edwards, Irvine, CA). A flow-directed thermal dilution pulmonary artery catheter (Baxter-Edwards 93A-140–7F, with a 20-cm proximal port) was inserted through the sheath introducer, positioned with the tip in the pulmonary artery, and secured for determination of pulmonary arterial pressure and thermal dilution cardiac output (CO), and was placed to facilitate right atrial administration of the ketamine and physiologic markers. The side arm of the sheath introducer was used for maintenance fluid administration and the readministration of autologous blood. Hydration was maintained throughout the study by an infusion of 0.9% saline at a rate of 5–10 ml · kg⁻¹· h⁻¹to maintain a constant pulmonary artery diastolic pressure (± 2 mmHg).

After anesthetic induction and catheter placement, 150 ml whole blood was removed from the dog through the arterial catheter and anticoagulated with 1,000 U heparin. This blood was immediately replaced with 600 ml saline solution, 0.9%, administered intravenously over 30 min. During the first 10 min of the study (from time t = 0 min to t = 10 min), this autologous blood was reinfused intravenously to replace the blood removed during this period of frequent blood sampling.

The study was not begun until the dog was hemodynamically stable. This was defined as less than 10% variation of CO and pulmonary and systemic arterial blood pressures over a 30-min period when heart rate and blood pressures were measured continuously and CO was determined at least every 15 min. The dogs were hemodynamically stable approximately 1 h after removal and saline replacement of the 150 ml blood.

At the onset of the study (time t = 0 min), ICG (Hynson, Westcott, and Dunning, Baltimore, MD), 5 mg in 1 ml diluent, antipyrine (Sigma, St. Louis, MO), 25 mg in 1 ml of diluent, and racemic ketamine (Parke Davis, Morris Plains, NJ), 100 mg in 1 ml, were placed sequentially in a 76-cm intravenous tubing (4.25 ml priming volume). The 3-ml drug volume was flushed through the proximal pulmonary artery catheter port within 4 s using 10 ml dextrose in water, 5%, allowing the simultaneous determination of dye and thermal dilution CO. Arterial blood samples were collected every 0.05 min for the first min and every 0.1 min for the next minute using a computer-controlled roller pump (Cole-Parmer, Chicago, IL), set at a withdrawal rate of 1 ml/s, and a chromatography fraction collector (Model 203; Gilson, Middleton, WI). Subsequently, 30, 3-ml arterial blood samples were drawn manually at 0.5 min intervals to 4 min, at 5 and 6 min, every 2 min to 20 min, at 25 and 30 min, every 10 min to 60 min, every 15 min to 120 min, and every 30 min to 360 min.

Plasma ICG concentrations of all samples obtained up to 20 min were measured on the study day by the high-performance liquid chromatography (HPLC) technique of Grasela et al.,  18as modified in our laboratory to provide sensitivity of 0.2–20.00 μg/ml with coefficients of variation of 5% or less. 19 

Plasma antipyrine concentrations were measured in all samples using a modification of a high-performance liquid chromatography technique developed in our laboratory. 10,20The antipyrine method is linear for plasma concentrations of 0.10–10.00 μg/ml, with coefficients of variation of 5% or less.

Plasma [S -(−)] and [R -(+)] ketamine enantiomer concentrations were measured in all samples using a modification of the stereospecific high-performance liquid chromatography technique developed by Geisslinger et al.  21Plasma samples were extracted in duplicate with C18 Empore high-performance extraction disk cartridges (3M Filtration Products, St. Paul, MN) using prilocaine as the internal standard. Samples were eluted isocratically at 1.0 ml/min from acid glycoprotein (AGP) guard and analytical columns (Advanced Separation Technologies, Inc., Whippany, NJ) using a mobile phase consisting of pH 6.98, 0.02 M phosphate buffer, and isopropyl alcohol in a 97.5:2.5 ratio. Absorbance was monitored at 214 nm. The stereoselective ketamine assay is linear from concentrations of 0.05 to 10.00 μg/ml, with coefficients of variation of 10% or less.

Plasma ICG concentrations were converted to blood concentrations by multiplying them by one minus the hematocrit because ICG does not partition into erythrocytes. To interpret intercompartmental clearances in relation to blood flow, plasma antipyrine and ketamine concentrations were corrected for partitioning into the erythrocyte by calculating the apparent dose of the respective drug. The ratio of the actual dose to the apparent or calculated dose is then used as an in vivo  estimate of erythrocyte partitioning to convert plasma concentrations to blood concentrations. Because the central blood flow of the antipyrine and ketamine models was constrained to that of ICG (i.e.,  CO), the apparent antipyrine and ketamine doses were estimated using blood ICG and plasma antipyrine or ketamine stereoisomer concentrations, respectively, obtained before the first evidence of recirculation according to the following relationship:10 

A delay (central volume [V^C] and nondistributive volume [V^ND];fig. 1) is a mathematical description of the distribution of drug transit times that can be described by the mean transit time (MTT). The apparent volume of a delay element can be determined as the product of the flow through the delay and its MTT. The delay elements of the SAAM II kinetics analysis software (SAAM Institute, Seattle, WA) are composed of a linear chain (or tanks-in-series) of n identical compartments connected by identical rate constants k such that n/k is equal to the MTT of the delay. The solution for a linear chain obtained by successive integration for the exit rate from compartment n is given by the Erlang density function, which is a special case of the τ distribution function, 22 

Indocyanine green, antipyrine, and ketamine concentration versus  time data before evidence of recirculation (i.e. , first-pass data) were weighted uniformly and fitted to the sum of two right-skewed τ(Erlang) distribution functions using TableCurve2D (version 3.0; Jandel Scientific, San Rafael, CA) on a desktop personal computer (Gateway 2000, North Sioux City, SD). 22The application of classic criteria for defining first-pass concentration to our data analysis using the Erlang distribution function has been described in detail by Krejcie et al.  22Because paired first-pass S -(+)- and R -(−)-ketamine concentrations and the respective pulmonary MTT for the enantiomers were not different (paired t  test), S -(+)- and R -(−)-ketamine concentrations were modeled simultaneously; this improved the confidence in the model parameters of ketamine pulmonary uptake. Erlang functions were not used to determine V^ND, instead single delay elements in SAAM II were used in which the optimum number of tanks-in-series were determined iteratively.

In subsequent pharmacokinetic analysis, the descriptions of the central circulation (the blood and apparent tissue volume between the right atrial injection site and the femoral arterial sampling site) were incorporated as parallel linear chains, or delay elements, into independent recirculatory models for the individual markers using SAAM II implemented on a desktop personal computer (Gateway 2000). 9,22The first-pass data were excluded from further data fitting; the results of the Erlang model of the central circulation were placed as fixed parameters in the recirculatory model, thereby reducing the number of parameters to be optimized. The concentration–time data were weighted, assuming a proportional variance model, in proportion to the inverse of the square of the predicted value. Possible systematic deviations of the observed data from the calculated values were sought 23using the two-tailed one-sample runs test, with P < 0.05 corrected for multiple applications of the runs test, as the criterion for rejection of the null hypothesis. 24Possible model misspecification was sought by evaluating the ratio of the measured to the predicted drug concentrations versus  time relation.

The pharmacokinetic modeling methodology has been previously described in detail. 11It is based on the approach described by Jacquez 25for obtaining information from outflow concentration histories, the so-called inverse problem. Antipyrine and ketamine distributions were analyzed as the convolution of their intravascular behavior, determined by the pharmacokinetics of concomitantly administered ICG, and tissue distribution kinetics. 10 

For ICG, the two lumped, parallel pathways of the central circulation, described by the sum of two Erlang distribution functions, were incorporated in a full recirculatory model (fig. 1), including lumped parallel fast and slow peripheral intravascular circuits and Cl^Eusing SAAM II. The antipyrine and ketamine pulmonary tissue volumes (V^T-P) are the difference between their respective apparent central volumes (MTT^{antipyrine or ketamine}· CO) and the central intravascular volume determined by ICG (MTT^ICG· CO).

Peripheral drug distribution can be lumped into identifiable volumes and clearances: a fast nondistributive peripheral pathway (V^ND-Fand Cl^ND-F); a slow nondistributive peripheral pathway (V^ND-Sand Cl^ND-S); rapidly (fast) equilibrating tissues (V^T-Fand Cl^T-F); and slowly equilibrating tissues (V^T-Sand Cl^T-S). The nondistributive peripheral pathways (delay elements) represent intravascular circuits in the ICG model. 10,19The single nondistributive peripheral pathway in the antipyrine and ketamine models, determined by the respective recirculation peaks, represents blood flow that quickly returns the lipophilic drugs to the central circulation after minimal tissue distribution (i.e.,  a pharmacokinetic shunt). 10,11In the antipyrine and ketamine models, the parallel rapidly and slowly equilibrating tissues are the fast and slow compartments of traditional three-compartment pharmacokinetic models; therefore, the central circulation and nondistributive peripheral pathways are detailed representations of the ideal central volume of the three-compartment model. 20Because of the direct correspondence between the recirculatory model and three-compartment models, Cl^Ewas modeled from the arterial (sampling) compartment to enable comparison of these results with previous ones.

As noted previously, preliminary analyses of the first-pass S -(+)- and R -(−)-ketamine arterial plasma concentration data indicated that there were no differences in pulmonary tissue distribution for these enantiomers, suggesting that their distribution to other tissues may also be indistinguishable from one another;i.e.,  intercompartmental clearances and volumes of distribution are the same for each enantiomer. To test this hypothesis, ketamine data were fit to models in which the number of parameters describing peripheral drug distribution were successively decreased. To determine statistically which was the simplest model, the Akaike information criterion (AIC), 29the mean absolute performance error (MAPE) and visual inspection of the plots of the ratios of the measured to the predicted drug concentrations versus  time relations were used as bases for model selection.

For model 1, Cl^E, intercompartmental clearances (Cl^ND, Cl^T-F, and Cl^T-S), and distribution volumes (V^ND, V^T-F, and V^T-S) for the two enantiomers were independent of one another (12 parameters). For model 2, Cl^Eand the distribution volumes remained independent, but all of the intercompartmental clearances were the same for the enantiomers, except Cl^ND, which was allowed to vary reciprocally with the different Cl^Es because both of these clearances are nondistributive and their sum was assumed to be the same for both enantiomers (10 parameters). For model 3, Cl^E, Cl^ND, and V^NDwere allowed to vary, but tissue volumes and their respective intercompartmental clearances were set to be equal (eight parameters).

Other investigators reported differences in the tissue distribution models for ketamine enantiomers. 4,5The apparent differences in tissue distribution of the ketamine enantiomers may be derived from an artifact of modeling the different Cl^Es of the stereoisomers from V^C. Modeling Cl^Efrom a peripheral compartment rather than V^Cwill affect the estimate of the volume of the peripheral compartment from which Cl^Eis modeled, and thus the steady state volume of distribution (V^SS). 26Normally, there is no information in typical input–output pharmacokinetic data to select among elimination models, 27but the simultaneous modeling of enantiomers with different Cl^Es and similar tissue distribution should enable optimal model selection. Therefore, a fourth model was considered (model 4), with elimination modeled from V^T-Finstead of V^Cand only Cl^Ewas allowed to vary for the two enantiomers (six parameters). V^T-Fwas chosen because this compartment includes the major portion of the splanchnic circulation and tissue. 10,28 

Blood ICG, antipyrine, and ketamine enantiomer concentration versus  time relations were well-characterized by the models from the moment of drug administration (time t = 0;figs. 2–4). The arterial blood drug concentration versus  time relations were always characterized by a brief delay between the moment of drug administration and their appearance at the arterial blood sampling site. The one-sample runs test indicated that there were no systematic deviations of the observed data from the calculated values.

The recirculatory pharmacokinetic model parameters of ICG are presented in table 2. A little more than one third of the blood volume (0.95 of 2.68 l) was in the central circuits. The fast peripheral circuit contained 0.20 l, which is an average of 11.7% of the peripheral blood volume (1.73 l) and 7.7% of the total blood volume, and received 48.3%(1.77 l/min) of CO (3.58 l/min, as determined by dye dilution, table 2). The slow circuit contained the majority of the peripheral blood volume (88.3%, 1.53 l, which is 56.9% of the total blood volume) but received only 41.9%(1.49 l/min) of CO ICG Cl^Erepresented only 9.9%(0.30 l/min) of CO.

The recirculatory antipyrine pharmacokinetic model (table 2) always contained a small but readily identifiable pulmonary tissue volume, which averaged 0.16 l or 0.61% of the total antipyrine distribution volume (26.6 l). The rapidly and slowly equilibrating peripheral antipyrine tissue volumes contained an average of 19.0%(5.02 l) and 75.5%(20.09 l) of the total antipyrine distribution volume, respectively, and fast and slow distribution clearances averaged 44.0%(1.64 l/min) and 21.5%(0.82 l/min) of CO, respectively. Antipyrine Cl^Eaccounted for only 2.9%(0.10 l/min) of CO. The balance of CO in the antipyrine model (31.5%, 1.02 l/min) was assigned to the nondistributive peripheral pathway, which represented only 1.3%(0.33 l) of the total antipyrine distribution volume.

The AIC for the various recirculatory ketamine pharmacokinetic models in table 3shows that there was an improvement in AIC (lowering) from model 1 to model 2 in which the number of independent parameters was reduced by two by setting the intercompartmental clearances of the racemates to be equal. However, the AIC increased in model 3, in which there was a further reduction in the number of parameters by setting the peripheral tissue volumes to be equal. Therefore, we rejected the model in which the ketamine racemates distributed to similar tissue volumes when elimination was modeled from the central compartment. However, the lowest (i.e. , most negative) AIC in all dogs was observed in a peripheral elimination model: model 4. The reduction in the number of adjustable parameters afforded by this model (six fewer than model 1, four fewer than model 2, and two fewer than model 3) did not result in a marked increase in mean absolute performance error over that of the models (1, 2, and 3) with more adjustable parameters. Only the central elimination model with the largest number of degrees of freedom, model 1 with independent intercompartmental clearances and peripheral volumes, had slightly better mean absolute performance errors than model 4 for most dogs. Therefore, the ketamine pharmacokinetic results reported and discussed herein are those of model 4.

The recirculatory ketamine pharmacokinetic model (table 2) also contained a pulmonary tissue volume, which averaged 0.31 l and was 0.75% of the total ketamine distribution volume (46.4 l). The rapidly (6.17 l) and slowly (38.48 l) equilibrating peripheral ketamine tissue volumes contained an average of 13.9 and 82.0% of the total ketamine distribution volume (46.4 l), respectively. Fast (1.54 l/min) and slow (0.86 l/min) distribution clearances averaged 44.3 and 21.9% of CO, respectively. Ketamine Cl^Ewas greater for the S -(+)-enantiomer (0.66 l/min) than for the R -(−)-enantiomer (0.49 l/min). The intercompartmental clearance to the nondistributive peripheral pathway (1.17 l/min) was 33.9% of CO, whereas the nondistributive volume (0.48 l) represented only 1.0% of the total ketamine distribution volume.

The intensity and timing of the onset of drug effect for rapidly acting intravenous anesthetics, such a ketamine, are influenced by factors affecting the early arterial drug concentration versus  time profile. 8These factors include intravascular mixing, pulmonary uptake, and distribution to highly perfused tissues. 9,30It has been suggested recently that distribution of drugs to tissue may be affected more by active transport across the endothelium than by factors such as blood flow and diffusion. 12Recently, we showed that the pulmonary uptake of fentanyl involves, not only passive diffusion, but also carrier-mediated endothelial transport. 13Ketamine is a racemic (50:50) mixture of two enantiomers that have been shown to interact with proteins in a stereospecific fashion. The S -(+)-enantiomer binds to NMDA receptors more avidly than the does the R -(−)-enantiomer, 31producing a more potent hypnotic effect. 6Likewise, human hepatic microsomes N -demethylate the S -(+)-enantiomer at a higher rate than R -(−)-ketamine. 32We wished to measure the in vivo  tissue distribution of ketamine and to determine whether there may be differences in the distribution of the stereoisomers, which would be suggestive of a specific protein-mediated transport across the vascular endothelium. In addition, we wished to evaluate ketamine pharmacokinetics using a high-resolution recirculatory model to examine carefully any possible differences in disposition that could reasonably be attributed to differential handling of the two stereoisomers.

The estimate of pulmonary extravascular antipyrine distribution (V^T-P, 0.16 ± 0.08 l) corresponds well with estimates of others for pulmonary extravascular water volume in dogs. 33Ketamine V^T-Pwas twice that of antipyrine (table 2). There was no difference in the V^T-Pof the R -(−)- and S -(+)-ketamine enantiomers. We looked for systematic differences between stereoisomer concentrations in each sample collected during first-pass and found none. In addition, there was no difference in the first-pass MTT, the principal measure of pulmonary uptake. 22,30,34The lack of a stereoisomeric effect in the pulmonary tissue distribution of ketamine suggests that physicochemical differences other than stereoisomerism account for its increased partitioning in the lung compared with antipyrine.

The pulmonary uptake of ketamine is quite small compared with fentanyl (60 times tissue water), 13lidocaine (9 times tissue water)9or sufentanil (84 times tissue water). 30It is closer to that of alfentanil (2.4 times tissue water)22,34or diazepam (approximately 5 times tissue water). 34Thus, interindividual differences in the pulmonary uptake of ketamine would not be expected to have much of an effect on onset of action.

The lack of stereoisomeric effect for the distribution of ketamine in pulmonary tissue suggests that the distribution of the S -(+) and R -(−)-enantiomers in other tissues might also be identical. Because the pharmacokinetics of tissue distribution are described by intercompartmental clearance (Cl^T-Fand Cl^T-S) and volume of distribution (V^T-Fand V^T-S), we systematically explored the modeling implications (i.e. , goodness of fit vs.  number of adjustable parameters) of treating the enantiomers as individual drugs or as identical drugs with respect to these parameters. If these parameters can be made to be common to the two enantiomers without substantially affecting goodness of fit, the AIC should become progressively smaller as more distribution parameters are made common. Table 3shows that in the case of a central elimination model this is true if Cl^T-Fand Cl^T-Sare made common, but does not hold up if V^T-Fand V^T-Sare also made common. Because the S -(+)-enantiomer has the greater Cl^E, we would expect its peripheral distribution volume to be less than that of the R -(−)-enantiomer during this condition. With Cl^Emodeled from V^T-F, we obtained similar fits with fewer adjustable parameters and an improved AIC (fig. 5and table 3); that is, there was a consistent improvement in the AIC when both enantiomers were “forced” to share the same V^T-F, V^T-S, Cl^T-F, and Cl^T-S. Thus, it is reasonable to conclude that there are no significant differences in the pharmacokinetics of tissue distribution between the two enantiomers of ketamine. This exercise provides evidence that the choice of the site from which Cl^Eis modeled (i.e.,  central or peripheral) will affect the peripheral distribution volume and thus V^SS.

Using a central elimination model with all parameters independent for the two enantiomers (model 1) resulted in a V^SSfor R -(−)-ketamine of 55.8 l, compared with 42.9 l for S -(+)-ketamine. This is in contrast to the peripheral elimination model (model 4) in which the V^SSfor either enantiomer was 46.4 l. Using model 1, the Cl^Es were 0.49 and 0.66 l/min, respectively, the same as reported for model 4 in table 2because elimination clearance is independent of the model used.

The ketamine V^SS(model 4) in these five dogs averaged 2.11 l/kg when referenced to plasma and compared with 2.5 l/kg reported by Kaka and Hayton 35and with 1.37 l/kg reported by Schwieger et al.  36The total (i.e.,  racemic) ketamine Cl^Evalues reported in these studies were 32.2 and 18.1 ml · min⁻¹· kg⁻¹, respectively, compared with 22.2 ml · min⁻¹· kg⁻¹for R -(−)-ketamine and 29.9 ml · min⁻¹· kg⁻¹for S -(+)-ketamine (table 1) in the current study.

There was a marked difference in Cl^Efor the two enantiomers of ketamine. On average, the Cl^Ewas 35% greater for the S -(+)-enantiomer compared with the R -(−)-enantiomer and was greater in all five dogs (table 2). This is consistent with the results of White et al.  3and Geisslinger et al.  4,5in humans (15 and 22% greater, respectively) and previous findings in rodents. 1,2In human liver microsomes, Kharasch and Labroo 32showed that the rate of N -demethylation was greater for S -(+)-ketamine than for the R -(−)-enantiomer.

The ICG and antipyrine kinetics compare very well with those from previous studies during similar conditions, 9–11,19,20,22suggesting that the 100-mg dose of ketamine had little if any effect on the factors governing the pharmacokinetics of these markers and drugs. Only in dog 2 was there a brief period of systematically poor fitting. This incident corresponds to the large positive deflection just before 1 min in the measure to predicted drug concentration ratio versus  time plots (fig. 5). This may be the result of a brief hemodynamic perturbation in this dog caused by the ketamine bolus. Because of the need for near-continuous arterial sampling, no hemodynamic measurements could be made during this period. However, at t = 10 min, all such values were within 10% of baseline.

Although antipyrine and ketamine were modeled independently, their mean intercompartmental clearances (Cl^T-Fand Cl^T-S) and the nondistributive intercompartmental clearances (Cl^ND) were remarkably similar. A linear regression for these parameters is shown in figure 6and shows that the correlations were highly significant for Cl^T-Fand Cl^T-S; the correlation for Cl^NDwas not significant because of an outlying datum. This finding suggests that a further model-reduction step, whereby the enantiomers of ketamine share common distribution clearance terms with antipyrine, could be made. It also indicates that factors determining the transcapillary exchange of these two compounds are similar. Antipyrine has been used as a marker of perfusion-limited tissue distribution in many tissues 17because it is minimally plasma protein bound, freely diffuses across lipid membranes, and equilibrates very rapidly with tissues. It would appear that the transcapillary exchange of ketamine is similar to that of antipyrine, despite ketamine's more extensive plasma protein binding (53%), 35indicating that the protein binding of ketamine does not influence its transcapillary exchange.

We described the pulmonary uptake and recirculatory pharmacokinetics of the individual stereoisomers of ketamine based on a multiple-indicator dilution technique with frequent arterial blood sample collection. Although the S -(+)-ketamine enantiomer had a Cl^Ethat averaged 35% more than the R -(−)-enantiomer, we found no difference in the pulmonary uptake of these two stereoisomers. Our conclusion that the peripheral distribution kinetics of the ketamine stereoisomers do not differ is predicated on the assumption that, if the pulmonary uptake of ketamine is not stereospecific, peripheral tissue distribution is unlikely to be stereospecific. Modeling the two different stereoisomer Cl^Es from V^T-Frather than from V^Cyielded a model of ketamine stereoisomer tissue distribution that was consistent with this assumption.