Pharmacokinetics of Thiopental Enantiomers during and following Prolonged High-dose Therapy 

The Human Research and Ethics Committee at this institution approved the study. Twenty patients (13 males; 7 females) were included who received a high-dose thiopental infusion of more than 24-h duration between April 1995 and October 1997. Patients received thiopental for intracranial hypertension complicating severe head injury (nine patients), refractory vasospasm complicating subarachnoid hemorrhage (eight patients), and other diagnoses (three patients: one with sagittal sinus thrombosis, one with cerebral arteriovenous malformation complicated by intracerebral hemorrhage, and one with vasospasm complicating pituitary adenoma resection). Standard therapy was ineffective for all patients before consideration of thiopental. Patients with severe head injury, for example, were administered thiopental for treatment of persistent intracranial hypertension despite administration of mannitol, hyperventilation, sedation with midazolam–morphine, and surgical evacuation when appropriate. The demographic characteristics for the 20 patients are summarized in table 1. No patient had a history of significant hepatic, renal, or cardiac disease before hospital admission. Four patients died during or within 7 days of completion of a thiopental infusion, including two patients in whom postinfusion blood sampling could not be continued for an adequate period to determine V^mand k^mvalues.

Intravenous thiopental (Pentothal; Abbott Australasia, Sydney, Australia) was administered at a loading dose of 5–10 mg/kg, followed by a continuous infusion adjusted to maintain an electroencephalography (EEG) burst suppression pattern of 2–6 bursts/min. All patients were administered an infusion of norepinephrine (2–8 μg/min) to maintain a mean arterial pressure of more than 80 mmHg during thiopental therapy. Prophylactic antibiotic treatment, including second- and third-generation cephalosporins, aminoglycosides, piperacillin or imipenim, or both, was maintained during and after thiopental infusion. The only other comedication routinely administered was oral sucralfate.

All patients underwent hourly clinical neurologic assessments, including reactivity of pupillary responses to light and motor responses to verbal command and painful stimuli, during and after the course of an infusion until discharge from the intensive therapy unit. The postinfusion times to recovery of pupil reactivity to light and times to a Glasgow Coma Score (GCS)13of more than 3 (i.e. , the time when GCS was first ≥ 4) were documented for each patient.

A modified Rankin scale 14outcome at 1 month after the start of thiopental therapy was used to estimate the severity of the patient's neurologic condition. Grade 0 signifies normal neurologic examination for age; Grade 1 is no significant disability despite the presence of neurologic symptoms or signs; Grade 2 is mild disability but independent in activities of daily living; Grade 3 is moderate disability requiring assistance with activities of daily living but able to walk with or without assistance; Grade 4 is severe disability requiring constant nursing care and attention; and Grade 5 is death.

Arterial blood samples were collected from an indwelling cannula during and after completion of an infusion into 5-ml tubes (SST, Becton Dickinson and Co., Franklin Lakes, NJ). All samples were centrifuged at 3,000 rpm within 60 min of collection, and the sera were stored at −20°C until analysis. The mean duration of blood sample collection after commencement of thiopental for the 18 patients in whom Michaelis-Menten elimination kinetics were calculated was 219 h (SD, 66; 95% confidence interval [CI], 186–252). Blood samples were collected once to twice daily after the start of an infusion until its discontinuation, then again at 8–12 h intervals after completion of an infusion until recovery of clinical responsiveness. Thereafter, samples were drawn every 12–24 h for another 48–72 h, the actual times being noted accurately. Additional samples were collected in all surviving patients at the times of completion of a thiopental infusion and times to a GCS score of more than 3. The mean total number of blood samples collected from these patients was 14 (SD, 4; 95% CI, 11–16), including a mean of 10 (SD, 4; 95% CI, 8–12) postinfusion samples.

Thiopental enantiomer serum concentrations were determined in all patients and pentobarbital enantiomer concentrations in five patients by chiral stationary phase high-performance liquid chromatography. 15The limits of quantitation for the R-  and S-  thiopental and R-  and S-  pentobarbital enantiomers were 10 and 100 ng/ml, respectively. Analysis of thiopental and pentobarbital enantiomer binding to serum proteins was performed in 17 and 7 patients, respectively, by equilibrium dialysis. Serial serum samples from each patient were pooled before equilibrium dialysis. Pooled serum samples were adjusted to pH 7.4 and then equilibrated at 37°C for 10 h, as previously described. 10The fractions unbound (f^u) for the separate enantiomers were determined using the differences in postdialysis buffer and serum concentrations. The mean predialysis serum concentrations of pooled samples of R-  and S-  thiopental were 17 mg/l (SD, 12; 95% CI, 11–23) and 19 mg/l (SD, 13; 95% CI, 13–26), respectively. The mean predialysis serum concentrations of pooled samples of R-  and S-  pentobarbital for the seven patients were 4.6 mg/l (SD, 4.6; 95% CI, 3.1–6) and 3.4 mg/l (SD, 3.4; 95% CI, 1.7–5), respectively.

The doses of R-  and S-  thiopental enantiomers were, by definition, one half the dose of thiopental. Each set of thiopental enantiomer serum concentration (C)–time (t) data were fitted by each of two models, using the program PKAnalyst (MicroMath Scientific Software, Salt Lake City, UT): a linear one-compartment model with constant intravenous rate and first-order output and a nonlinear one-compartment model with constant intravenous rate and Michaelis-Menten output. Thiopental shows “multicompartment behavior” after intravenous bolus doses and brief infusions; however, as it approaches a steady state after a prolonged infusion, its behavior is described adequately by a single “well-mixed” compartment. 6For both linear and nonlinear models, Cl^sswas defined as the ratio of average thiopental dose rate for the last 12 h of thiopental therapy to average serum concentration measured during this period, during quasi steady state conditions.

The linear model is based on the elimination constant (K^el) and ratio of the total dose to the volume of distribution (V^d) for the infusion time (T^iv) using equation (1): where C(t) is the serum analyte concentration at time t; T = t − T^ivfor t > T^ivand T = 0 for t < T^iv. In this model, the product of V^dand K^elis equal to the total body clearance (TCl). 7,8 

The nonlinear model is based on the V^m, k^m, and ratio of total dose to V^dfor a given T^ivusing the Michaelis-Menten equation (2): where k^mis defined as the thiopental serum concentration at which the rate of metabolism is equal to one half its theoretical maximum rate (V^m).

The model parameters V^d, V^m, k^m, and elimination t^1/2were estimated separately for total and unbound R-  and S-  thiopental enantiomers, and for thiopental (i.e. , the sum of the enantiomers) so that these could be compared to values reported in the literature. C^maxwas defined as the maximum observed serum thiopental concentration during the course of an infusion. The percentage saturation (%Sat) of the enzyme system at C^maxwas calculated from the following equation:%Sat =[C^max/(k^m+ C^max)]× 100.

At higher concentrations, when C ≥ K^m, the value of t^1/2increases with concentration. At lower concentrations, when C << k^m, the elimination t^1/2will be constant; the values reported, accordingly, were estimated from ln2 × k^m/V^m(= ln2 × the inverse of the first-order rate constant when C << k^m). Total clearance (TCl) for R-  and S-  thiopental was calculated from the ratio of the total dose to the total area under the curve (AUC), as computed by the trapezoid rule using the linear one-compartment model with extrapolation of the terminal phase. The mean residence time for the thiopental enantiomers was estimated from the ratio of the area under the moment curve (AUMC) to the total area under the curve minus one half the duration of infusion. The values of Cl^ss, TCl, and mean residence time, as determined, would have been subject to influence by a number of factors, including sampling frequency, variability of thiopental infusion rate caused by clinical requirements, relation of steady state concentration to dose rate at values exceeding the relevant value of k^m, and clinical condition of the patient. Hence, the primary objective of the data analysis was to provide a comparison between enantiomers because no previous studies have. The limitations of the approach are acknowledged, but the models chosen were the simplest that adequately fitted the data for the relevant time-averaged infusion rate.

Equations describing R- , S- , and (R-  plus S- )-thiopental kinetics were determined by nonlinear least-squares regression analysis using weights of 1/C²and the modified Powell algorithm 16to assess the best fitting parameters. The criteria considered when choosing between models included visual inspection of the model predictions versus  measured values, standard deviation of observed versus  predicted concentration–time data, coefficients of determination, and correlation coefficients between observed and predicted values. The correlation coefficient between observed and predicted values exceeded 0.9 in all data sets when using the nonlinear model with values more than 0.98 for 16 of the 18 patients. A comparison between the models was also made using a modified Akaike Information Criterion 17termed the Model Selection Criterion in the software used. One-way analysis of variance comparison of R²(0.97 ± 0.01 vs.  0.90 ± 0.08, P = 0.0005), correlation coefficient (0.99 ± 0.01 vs.  0.98 ± 0.02, P = 0.01), standard deviation of observed versus  predicted concentration–time data (0.18 ± 0.06 vs.  0.31 ± 0.14, P = 0.001), and Model Selection Criterion (4.0 ± 1.2 vs.  3.0 ± 0.8, P = 0.007) values between nonlinear and linear models consistently indicated that nonlinear interpretation of the data was the preferred approach. The linear model was preferred in only two patients and, in these, the difference in the correlation coefficients between observed and predicted values was small (0.96 and 0.98, respectively) by using a nonlinear model; hence, it was decided to report the nonlinear model parameters for all 18 patients. Figure 1shows two sets of data series from patients 4 and 11 to illustrate the linear and nonlinear model approaches.

Because the data were derived from the same serum samples, the differences between the respective values of C^max, V^m, k^m, %Sat, V^d, Cl^ss, TCl, mean residence time, terminal t^1/2, and f^ufor R-  and S-  thiopental in each patient were tested against a value of zero using the Student t  test for paired data; the respective values of the enantiomeric ratio of the relevant pharmacokinetic parameters were tested against a value of unity using the Student single-sample t  test. 18Multiple linear regression analysis 18was also performed to analyze the relation between clinical variables (total thiopental dosage, duration and dose rate, patient age, gender, and total body weight), and pharmacokinetic parameters (C^max, Cl^ss, %Sat, V^m, k^m, and elimination t^1/2of thiopental) and the duration of postinfusion pupillary and clinical unresponsiveness. A P  value of < 0.05 was significant. The analyses were performed using Statistix for Windows 1.0 (Analytical Software, Tallahassee, FL) on a personal computer.

Demographic data for the patients, including age, weight, total thiopental dose, and infusion duration, are summarized in table 1. The mean times of recovery of pupillary responsiveness and times to a GCS of more than 3 after completion of a thiopental infusion and 1-month Rankin scale outcomes are also summarized in table 1. Examples showing fits of nonlinear models to the data for five patients are given in figure 2. From these cases, it can be seen that that rate of decrease of serum concentrations of both enantiomers after cessation of infusion is initially very slow.

Summary data for Cl^ss, C^max, V^d, k^m, V^m, elimination t^1/2, TCl, and mean residence time for R-  and S-  thiopental based on total serum concentrations from all patients are shown in table 2; corresponding values based on unbound serum concentrations are shown in table 3. Pharmacokinetic differences of approximately 10–20% were found between R-  and S-  thiopental, as indicated by the statistical significance of the enantiomeric ratios and the differences for the various parameters (tables 2 and 3). Although the magnitude of the differences varied among the population of subjects, the overall values were sufficiently similar that it was not possible to relate clinical events selectively to either enantiomer. Pharmacokinetic findings for the sum of enantiomers (i.e. , for thiopental) included the following: Cl^ss: 0.102 l/min (n = 20, SD = 0.05, 95% CI = 0.08–0.12); TCl: 0.094 l/min (n = 18, SD = 0.05, 95% CI = 0.07–0.12); C^max: 73 mg/l (n = 20, SD = 33, 95% CI = 57–89);%Sat: 68 (n = 18, SD = 16, 95% CI = 60–76); k^m: 44 mg/l (n = 18, SD = 41, 95% CI = 23–64); V^m: 1.9 mg · l⁻¹· h⁻¹(n = 18, SD = 1.2, 95% CI = 1.3–2.5); and elimination t^1/2: 14.6 h (n = 18, SD = 7, 95% CI = 11–18). The mean value for the maximal rate of metabolism (V^m× V^d), 0.46 g/h (n = 18, SD = 334, 95% CI = 0.30–0.63), did not differ significantly from the average dose rate for the duration of a thiopental infusion, 0.50 g/h (n = 20, SD = 0.22, 95% CI = 0.39–0.60;P = 0.7).

The relative serum concentration–time curves of thiopental and pentobarbital formed during and after completion of a thiopental infusion is shown in figure 2for five patients. Values of fu of R-  and S-  pentobarbital for the seven patients who underwent serum protein binding analysis was 32%(SD = 5, 95% CI = 27–37) and 34%(SD = 10, 95% CI = 25–44), respectively (P = 0.4).

Postinfusion times of recovery of pupillary responsiveness and times to a GCS of more than 3 were significantly correlated (P = 0.003, n = 16, adjusted R²= 0.4); postinfusion time to GCS of more than 3 also significantly but weakly correlated with the elimination t^1/2of thiopental (P = 0.02, n = 16, adjusted R²= 0.3). Times of onset of pupillary responsiveness or times to GCS or more than 3 after completion of a thiopental infusion were not significantly correlated with other clinical or pharmacokinetic variables, including patient age and weight, total thiopentone dose, duration or dose rate, and Cl^ss, C^max, V^m, and k^mof thiopental.

A multitude of previous studies have described the pharmacokinetics of thiopental in linear terms. During linear conditions, a constant fraction of the thiopental body burden is eliminated per unit-time and the circulating concentration is proportional to the dose. It is well-established that thiopental can show nonlinear (Michaelis-Menten or saturable) kinetics when administered in high doses for prolonged periods. 1–8The progression from linear to nonlinear conditions coincides with increasing body (and thus hepatic mixed-function oxidase) burden of thiopental.

Thiopental elimination becomes nonlinear when serum concentrations exceed approximately 30 mg/l. 2,19During nonlinear conditions, a constant amount of the body burden is eliminated per unit-time and the circulating concentrations are disproportional to dose. As a general rule, if the rate of administration of a substance leads to protracted concentrations greater than the Michaelis constant (k^m) then the resultant concentrations can become substantially greater than those predicted by linear pharmacokinetic models and the elimination t^1/2after cessation of administration increases with decreasing body burden until it reaches its limiting value of linear conditions. In the current study, mean k^mvalues of 20 and 24 mg/l were found for R-  and S-  thiopental, respectively; hence, it is not surprising that the pharmacokinetics of thiopental were better described by a nonlinear model in most patients. In 11 patients, the average dose rate was actually greater than the V^m, thus predisposing these patients to a phenomenon often known as “runaway accumulation,” in which increasing serum thiopental concentrations (and unwanted pharmacodynamic effects) may be seen despite an essentially constant infusion rate administered per clinical requirements. This study has shown that this kind of accumulation often occurs in doses of thiopental that are typically used for neuroprotection of head-injured patients. Moreover, the t^1/2shortly after cessation of infusion (as shown in figs. 1 and 2) was much greater than the late-phase (limiting) values (as given in table 2). Furthermore, with approximately a 10-fold range in thiopental metabolic parameters, it is very difficult to predict serum concentrations from a given dose in a patient without monitoring, despite dosing to a reasonably precise pharmacodynamic end point. Within this patient group, attendant relevant pathophysiologic changes would be expected to contribute to the variability in thiopental pharmacokinetics and any thiopental serum concentration–response relation.

Recent studies have found a higher Cl^ssand V^dof R-  thiopental relative to S-  thiopental after a single intravenous bolus administration of thiopental, 9,10but no previous study has considered the potential for nonlinear pharmacokinetic behavior of the enantiomers from a prolonged high-dose infusion. The mean values of Cl^ss, V^d, V^m, and t^1/2for thiopental (i.e. , the summed enantiomers) in the current study are comparable to those previously reported in high-dose thiopental therapy. 1–4Pseudo steady state serum concentrations of thiopental, however, in the current study (mean value 74 mg/l) were higher than in previously reported series (mean values of 31 mg/l by Russo et al.  4and 49 mg/l by Turcant et al.  2). This most probably reflects a higher average dose-rate requirement (0.50 ± 0.22 g/h) to achieve EEG burst suppression in our patient population. The serum thiopental concentrations at the cessation of infusion exceeded the k^mvalues in all but two patients (respectively, 60 vs.  150 mg/l; and 42 vs.  48 mg/l). In these two patients, concentration–time data were better fitted by a linear than a nonlinear model.

Our findings indicate that R-  and S-  thiopental pharmacokinetic differences, as found previously with anesthetic induction doses, are also apparent with high-dose infusions:R-  thiopental concurrently showed a consistently greater Cl^ssand V^dthan S-  thiopental. Although small, the enantiomeric differences were statistically significant, supported by the enhanced power of the paired analysis. In contrast, S-  thiopental was consistently found to have a greater affinity for serum proteins, so that the enantiomeric differences in Cl^ssand V^dvanished when f^uof the two enantiomers were taken into account. Although thiopental undergoes nonlinear serum protein binding on increasing concentrations through the sub-to-low microgram range, it is essentially constant in the high-concentration range, as was encountered in this investigation. 20–22We also found that S-  thiopental has a higher theoretical maximal rate of metabolism when compared to R-  thiopental, but this difference also vanished when the f^uwere considered. Hence, the pharmacokinetic differences between R-  and S-  thiopental identified in our patient population appear to be strongly influenced by the higher free-fraction of R-  thiopental. Based on mean data, the value of C^max× f^ufor S-  thiopental is not different than that of R-  thiopental. Nevertheless, the contributions of both enantiomers to the pharmacodynamics and pharmacokinetics of a racemic drug need to be considered.

Previous studies indicated that the dose-derived anesthetic potency of S-  thiopental in the mouse is approximately twice that of R-  thiopental. 11,12We performed studies in recombinant γ-aminobutyric acid (GABA) A receptors transfected into Xenopus laevis  oocytes that have shown that the order of potency of GABA facilitation (mean EC⁵⁰for 3 μM GABA) to be S-  thiopental (26.0 μM) more than rac -thiopental (35.9 μM) more than R-  thiopental (52.5 μM) more than rac -pentobarbital (97.0 μM). These findings concur with the differences in potency found in vivo . 23We also found, in the rat, that the distribution coefficients from plasma to central neural tissues were approximately 10–15% higher for R-  thiopental than for S-  thiopental. 24Thus, in apportioning the clinical central nervous system effects between enantiomers, it appears that the higher receptor-intrinsic effectiveness of S-  thiopental is partially offset by pharmacokinetic factors.

In the five patients for whom simultaneous serum pentobarbital and thiopental concentration measurements were taken, the ratio of R-  to S-  pentobarbital concentrations was the reverse of that seen for the respective parent compounds, R-  and S-  thiopental. During the infusion, the concurrent serum concentrations of pentobarbital were 10–30% those of thiopental, which is similar to previous findings. 1,2After cessation of thiopental, the ratio of serum pentobarbital:thiopental concentrations increased with time, most likely because of the longer half-life of pentobarbital and its continued formation by metabolism. The potential clinical significance of the pentobarbital is therefore of interest.

The clinical and EEG effects of pentobarbital at different blood concentrations in humans are not well-defined. Animal and human studies suggest that plasma pentobarbital concentrations necessary for desired clinical and EEG effects are similar to those reported for thiopental. 25–29For example, EEG burst suppression occurs at plasma pentobarbital concentrations ranging from 25 to 75 mg/l in patients with severe head injury, 28,29but plasma f^uof pentobarbital would appear to be approximately twice those of thiopental at the same plasma concentrations. 30,31However, the intrinsic potency for facilitation of GABA response in vitro  indicates rac -pentobarbital to be only 40% that of rac -thiopental. 23Thus, the different relative potencies from in vitro  and in vivo  studies support the likelihood of a similar “clinical” potency for total (bound and unbound) serum concentrations of thiopental and pentobarbital in which the potencies of the R-  enantiomers are significantly less than the respective S-  enantiomers. 11,12,32,33Because the postinfusion pentobarbital serum concentrations in our patients were typically less than 10 mg/l, the amount of pentobarbital formed probably was not making a major contribution to the clinical effects observed.

The mean duration of pupillary unresponsiveness and time to first GCS of more than 3 after completion of a thiopental infusion in the current study were 22 and 91 h, respectively. Our attempt to identify factors that may have influenced the postinfusion duration of clinical and pupillary unresponsiveness was limited by the heterogeneity of the patient population and the relatively small number of patients in the study. Thus, we were unable to successfully predict the duration of clinical effect on the basis of clinical and pharmacokinetic parameters known at the time of completion of an infusion. A more detailed statistical analysis of a larger series of patients undergoing thiopental therapeutic drug monitoring has been attempted separately (Cordato et al.,  unpublished observations). Overall, our findings indicate that an individual patient's inherent capacity to metabolize thiopental is a significant contributing factor to its postinfusion duration of effect.

In conclusion, the pharmacokinetics of both R-  and S-  thiopental became nonlinear at the doses infused in these head injury patient. Although significant, the pharmacokinetic differences between the thiopental enantiomers are smaller than the pharmacodynamic differences found in animal and GABA facilitation studies. The ratio of serum pentobarbital concentration relative to the serum thiopental concentration increases with time after completion of a thiopental infusion but the concentrations of pentobarbital are relatively low and, compared to the infused thiopental, may not be of clinical significance. The pharmacodynamic differences between R-  and S-  thiopental are unknown in humans, and future research in this area may determine whether one or another enantiomer has a clinical advantage over the racemate.

The authors thank Ms. Xiao Qing Gu for technical assistance.