Pharmacokinetics of Rapacuronium in Infants and Children with Intravenous and Intramuscular Administration

The study was conducted using Organon’s Investigational New Drug Application, (West Orange, New Jersey) and the protocol was approved by the institutional review board of the University of California, San Francisco. After obtaining informed consent from parents, we studied 40 pediatric patients, American Society of Anesthesiologists physical status I or II, undergoing elective peripheral surgery with minimal blood loss and fluid requirements. Patients were stratified into two groups by age: infants (2–11 months, 7.7 ± 1.7 kg [mean ± SD], n = 20) and children (1–3 yr, 13.0 ± 2.6 kg, n = 20). No patient had a history of bleeding disorder, neuromuscular disease, or hepatic or renal insufficiency. Patients were excluded if they received anticonvulsants or aminoglycoside or polypeptide antibiotics perioperatively. Neuromuscular effects of rapacuronium in these patients have been described previously. 2 

Patients were unpremedicated, and anesthesia was induced with nitrous oxide and halothane. After loss of consciousness, nitrous oxide was discontinued and inspired halothane was adjusted to produce end-tidal concentrations of 1.0% in children younger than 2.5 yr and 0.82% in children older than 2.5 yr. After establishing intravenous access, the trachea was intubated and ventilation was controlled to maintain normocapnia. After tracheal intubation, 66% nitrous oxide was added to the inspired gas. Patients were randomized to receive rapacuronium as either an intramuscular (n = 20) or an intravenous (n = 20) injection. If end-tidal halothane concentrations were stable for more than 5 min, rapacuronium (20 mg/ml) was injected either as a rapid bolus into a peripheral venous catheter in an upper or lower extremity or via  a 21-gauge needle 1 to 2 cm into a single deltoid muscle after negative aspiration for blood. The intravenous dose of rapacuronium was 2 mg/kg for infants and 3 mg/kg for children; the intramuscular dose was 2.8 mg/kg for infants and 4.8 mg/kg for children.

Five venous blood samples were obtained from each patient. The initial sample was taken after induction of anesthesia but before rapacuronium administration, the remaining four samples were taken 2–240 min after rapacuronium. For each patient, target times for blood sampling differed, determined by randomization. To prevent degradation of rapacuronium, blood samples were added immediately to vials prefilled with sodium dihydrogen phosphate buffer (0.8 M). Blood was centrifuged within 30 min of sampling, and plasma was stored at −20°C. Concentrations of rapacuronium and its primary metabolite ORG9488 were determined by Corning Hazelton Labs (Hazelton, WI) using a high-performance liquid chromatography–mass spectroscopy technique. The assay is linear for concentrations more than 2 ng/ml for rapacuronium and ORG9488 both and has a coefficient of variation of less than 11% for rapacuronium and less than 20% for ORG9488. The amount of ORG9488 in each vial of rapacuronium was less than 1% of the quantity of rapacuronium (oral communication, January 1999, Viquar Pervaaz, Organon).

Pharmacokinetic analyses were performed using the NONMEM statistical package 4#with the first-order or hybrid approaches. **Analyses were performed using two- and three-compartment models. Parameters of the two-compartment model were weight-normalized volume of the central compartment (V¹), volume of the second compartment (V²), clearance (Cl), and distribution clearance (Cl^distribution). Additional parameters of the three-compartment model were weight-normalized volume of the third compartment (V³) and slow-distribution clearance (Cl^slow); Cl^distributionwas renamed Cl^rapid. Volume of distribution at steady state (V^ss) equaled V¹+ V²or V¹+ V²+ V³. With intramuscular administration, we initially assumed that the absorption process was first-order; that is, the rate of absorption was proportional to the quantity of rapacuronium remaining in the intramuscular depot. This absorption process is described by the equation

in which F is the bioavailable fraction, k  ^absorptionis the absorption rate constant (estimated in the analysis), and time is the interval since drug administration. Additional absorption models were also evaluated. Absorption, distribution, and elimination half-lives were determined using standard formulas.

Intraindividual variability for Cl was assumed to be log-normally distributed, modeled as

in which Clⁱis the estimate for Cl for the ith individual, Cl is the typical value for the population, and ηⁱis a random variable with mean 0.0 and variance Ω². Interindividual variability for each of the pharmacokinetic parameters was modeled in a similar manner. With three-compartment models, interindividual variability in Cl^rapidwas assumed to have the same magnitude as that for Cl^slow; interindividual variability for V²was assumed to have the same magnitude as that for V³. Residual variability between predicted and measured plasma concentrations of rapacuronium initially was assumed to have two components: one proportional to the predicted concentration (“constant coefficient of variation”), and one additive (homoscedastic error). Additional error models with a single component also were evaluated.

After the “typical” values were determined, we used the NONMEM post hoc  step to determine values for the pharmacokinetic parameters for each individual. These values were plotted against each of the covariates age and weight and evaluated for systematic trends (e.g. , a relation between Cl and age) using a smoother (lowess, a nonlinear regression function). If trends appeared, these covariates were incorporated into the pharmacokinetic model. Models with additional pharmacokinetic parameters (either additional compartments, additional parameters to allow for interindividual variability, or the influence of covariates on a particular pharmacokinetic parameter) were accepted if they improved the objective function (P < 0.01 necessitates a decrease of 6.6 units for one additional parameter or 9.2 units for two additional parameters) and either improved the pattern of residual differences between measured and predicted values for plasma concentration of rapacuronium or decreased the trend between a pharmacokinetic parameter and a covariate.

Because the time to peak plasma concentration with intramuscular administration is a function of all the pharmacokinetic parameters, its typical value and confidence limits cannot be determined using Student t  values. Instead, we determined the time to peak plasma concentration using the NONMEM simulation mode. Using the typical values for the pharmacokinetic parameters and allowing for the interindividual variability determined in the pharmacokinetic analysis, we simulated the plasma concentration versus  time course of 250 infants weighing 5 kg and 250 children weighing 15 kg who were administered intramuscular rapacuronium. Time to peak plasma concentration was determined for each of these simulated individuals, and the fifth, fiftieth (median), and ninety-fifth percentile values were determined. We also used the typical values for bioavailability and the absorption rate constants to simulate the typical time course of rapacuronium remaining to be absorbed from the intramuscular depot. Finally, we used log-linear interpolation of the post hoc  estimates for plasma concentration of rapacuronium to estimate the plasma concentration in each subject 30 min and 60 min after administration of rapacuronium.

For ORG9488, the ratio of each plasma concentration to the corresponding plasma concentration of rapacuronium was determined and plotted against time. The influence of age and route of administration on these ratios was assessed visually.

Values are reported as the mean ± SD. Comparisons between infants and children and between intravenous and intramuscular administration were performed using the Student t  test for unpaired data;P < 0.05 was considered statistically significant.

One supplemental plasma sample was obtained from each of three patients: an infant (intravenous group) at 201 min and two children (intravenous group) at 4 and 211 min (fig. 1). The final plasma sample could not be obtained in two patients (one infant, intravenous group at 240 min; one child, intravenous group at 120 min) because of loss of venous access. A total of 161 plasma samples (excluding blank samples) were used in the analysis, 80 of these underwent intramuscular administration.

A three-compartment pharmacokinetic model fit the plasma concentration data better than did a two-compartment model. (The objective function decreased by 127.406 units and the residual differences between predicted and measured concentrations improved. [table 1]). Not permitting interindividual variability in Cl^rapid, Cl^slow, V², and V³increased the objective function by only 5.764 units (model 3 vs.  model 2;P > 0.5 for the deletion of two parameters); therefore, subsequent models did not permit interindividual variability in these parameters. Because subsequent analyses necessitated the use of the NONMEM hybrid approach, model 3 was repeated using the hybrid approach (model 4).

Plots from model 4 showed that with both age groups, measured and population-predicted concentrations after 100 min were biased: The ratios were higher with intramuscular compared with intravenous administration. This could result from incorrect modeling of the absorption process, that is, not allowing for persistent absorption of rapacuronium from the intramuscular depot. To account for this, we tested a model in which the administered dose was absorbed from two separate compartments, each of which had its own rate constant. The sum of the amount absorbed from each of these absorption compartments equals the product of the bioavailability fraction and the administered dose. The fraction of the dose absorbed from each of the two absorption compartments was estimated in the analysis. The objective function for model 5 decreased by 70.429 units compared with model 4, bias was no longer evident in the plots, and the fit of the model to the data was good.

Plots of the post hoc  estimates of Cl versus  weight from model 5 suggested that Cl had both a weight-normalized and an additive component (Cl = slope · weight + intercept; ml/min). Permitting Cl to have weight-normalized and additive components improved the objective function of model 6 by 11.873 units compared with model 5 (P = 0.006 for the addition of one parameter). Plots from model 5 also suggested an alternate explanation: weight-normalized Cl differed between infants and children. Permitting weight-normalized Cl to differ between infants and children (model 7) decreased the objective function by 14.593 units compared with model 5 (P = 0.001 for the addition of one parameter). However, the resulting population-predicted concentrations were biased (e.g. , for infants, measured concentrations were consistently larger than predicted), suggesting that model 7 was not appropriate. Therefore, model 6 was used as the basis for further analyses.

Plots for model 6 suggested that V¹differed as a function of gender. To evaluate this, model 8 permitted V¹to differ between males and females. The objective function decreased by 0.433 units compared with model 6 (P > 0.50 for the addition of one parameter), indicating that V¹did not differ between genders. Covariate plots for model 6 also suggested that the fraction of the intramuscularly administered dose absorbed from the first absorption compartment (as opposed to the second absorption compartment) varied with patient weight or with the volume of drug administered (which was a function of weight and age group). The former was incorporated into model 9 as

where FACT1 is estimated in the analysis, TVCMT1 is the typical value for the fraction of bioavailable drug absorbed from the first absorption compartment [i.e. , with rate constant k  ^absorption(1)], INCMT1 is the value for that fraction for an individual of weight (WT; in kilograms), 10 is approximately the median value for weight in this population, LGTCMT is a logit transformation to ensure that typical and individual values for the fraction of bioavailable drug absorbed from the first compartment are bounded by 0 and 1, and FXCMT1 is the inverse logit transform if LGTCMT. The objective function for model 9 decreased by 7.906 units compared with model 6 (P = 0.005). In model 10, we evaluated whether the volume of drug administered influenced the fraction of the intramuscularly administered dose absorbed from the first absorption compartment. Although the objective function for model 10 decreased by 25.606 units compared with model 6 (P = 0.005), the resulting population-predicted concentrations were biased (e.g. , nearly all initial measured values were markedly less than the corresponding predicted values). Thus, model 9 was used as the basis for further analyses.

The results of model 9 suggested that interindividual variability for the fraction of bioavailable drug absorbed from the first absorption compartment was trivial. In addition, the additive component of the error model was trivial. Therefore, model 11 differed from model 9 by deleting the η term in and by having only the constant coefficient component in the error model. The resulting objective function was identical to that for model 9. Plots of pharmacokinetic parameters from model 11 versus  covariates did not reveal any remaining relation between the covariates and the pharmacokinetic parameters nor bias in the ratio of measured to predicted plasma concentrations (fig. 2).

Thus, the “optimal” pharmacokinetic model (model 11) had three compartments and permitted interindividual variability in Cl, V¹, bioavailability, and each of the absorption rate constants. The typical value for Cl was 4.77 ml · kg⁻¹· min⁻¹+ 8.48 ml/min (fig. 3, table 2). Distributional clearances and volumes of distribution were all weight-normalized. Although Cl varied with weight, half-lives varied minimally (table 3). With intramuscular administration, the typical value for F was 56%. For a 10-kg child, approximately 28% (varying with weight) of bioavailable rapacuronium was absorbed with a rate constant of 0.011 min⁻¹; the remaining 72% was absorbed with the rate constant 0.0491 min⁻¹. The fraction of the bioavailable dose absorbed with a rate constant of 0.011 min⁻¹decreased with increasing weight.

For a simulated 5-kg infant, the median time after intramuscular administration for plasma concentration to peak was 4.0 min; the fifth and ninety-fifth percentiles were 2.5 and 6.3 min, respectively. For the 15-kg child, median time to peak plasma concentration was 5.0 min; the fifth and ninety-fifth percentiles were 3.0 and 7.5 min, respectively. The quantity of rapacuronium remaining in the intramuscular depot initially was larger in children than in infants (fig. 4); values converged approximately 45 min after intramuscular administration. At 30 min and 60 min after administration, mean plasma concentrations of rapacuronium were greater after intramuscular administration than after intravenous administration and greater in children than in infants (table 4).

After intravenous administration of rapacuronium, plasma concentrations of ORG9488 peaked early (fig. 5) and then decreased slowly. With intramuscular administration of rapacuronium, plasma concentrations of ORG9488 peaked 1–2 h after administration and then decreased slowly. During the 30 min after administration of rapacuronium, plasma concentrations of ORG9488 were less than 10% of corresponding rapacuronium concentrations in 71 of 83 measurements. For the remaining measurements, the ratio of ORG9488 to rapacuronium was 10–22%. In samples obtained more than 30 min after rapacuronium administration, the ratio of concentrations of ORG9488 to those of rapacuronium increased progressively (fig. 6). There were no apparent age-related differences in plasma concentrations of ORG9488. However, with intravenous administration of rapacuronium, the ratios of plasma concentrations of ORG9488 to those of rapacuronium were generally greater in infants than in children (fig. 6).

We determined the pharmacokinetic characteristics of rapacuronium with intravenous and intramuscular administration. Our analysis indicates that Cl is approximately 6.5 ml · kg⁻¹· min⁻¹for a 5-kg infant and 5.3 ml · kg⁻¹· min⁻¹for a 15-kg child. These values are slightly less than the Cl values for rapacuronium in adults, with intravenous administration recently reported in two other studies. Szenohrdazsky et al.  5reported that Cl was 9.40 · (1 − 0.00909 ·[Age − 30]) ml · kg⁻¹· min⁻¹in anesthetized volunteers aged 20–42 yr. Fisher et al.  6reported that Cl for rapacuronium was 7.03 ml · kg⁻¹· min⁻¹in adults aged 24–83 yr and did not vary with age. The finding that Cl for rapacuronium is smaller in children than in adults differs from that for other muscle relaxants. For example, the Cls of d -tubocurarine 7and vecuronium 8are similar in children and adults. Because the three studies involving rapacuronium used identical assays, and the pharmacokinetic analyses were performed by the same individual, it is unlikely that the findings of the current study result from experimental error. We are unable to explain why Cl for rapacuronium is smaller in infants and children than in adults. Although infants and children have a smaller Cl compared with adults, the rate at which plasma concentrations decay after bolus doses is more rapid in infants and children (fig. 7). This results from a smaller steady state volume of distribution in infants and children than in young healthy adults (381 and 445 ml/kg in women and men, respectively). 5 

Initially, we modeled absorption from the intramuscular depot using a single first-order rate constant; that is, assuming that the rate of absorption of rapacuronium from muscle was proportional to the quantity remaining in the muscle. This model was biased, as evidenced by plots of the ratio of measured to predicted plasma concentrations versus  time. To address this bias, we tested an alternate absorption model in which bioavailable rapacuronium was absorbed from two different compartments, each of which had different absorption rate constants. The resulting fit of the model to the data improved markedly, suggesting that the more complicated absorption model was appropriate. The traditional model with a single first-order rate constant often is used because it has the correct mathematic form (integration of yields the bioavailable dose). However, there is not necessarily a physiologic basis for this model. In addition, local characteristics, such as the proximity of administered drug to local capillaries and pressure increases resulting from tissue expansion, may influence the rate of absorption. Although our previous studies of intramuscular rocuronium 9did not show the need for a complicated absorption model, our previous success in modeling absorption with a single first-order rate constant may result from a limited quantity or quality of data.

The fraction of rapacuronium absorbed with intramuscular administration was estimated as only 56%. This value is smaller than (although possibly not statistically different than) the bioavailable fraction of 82% estimated for intramuscular administration of rocuronium. 9It is likely that some of the administered dose is lost through the puncture site of the injection; however, this was not quantified. Because weight-normalized volumes of rapacuronium (0.14 ml/kg in infants and 0.24 ml/kg in children) were larger than those for rocuronium (0.1 ml/kg for infants and 0.18 ml/kg for children), it is possible that the quantity of cutaneous loss is larger with a larger administered volume.

Time to peak plasma concentration with intramuscular administration was 4.0 min in a typical infant and 5.0 min in a typical child. These times are shorter than the time to peak plasma concentration with intramuscular administration of rocuronium (13 min for both infants and children). This shorter time to peak plasma concentrations is consistent with the onset of intramuscular rapacuronium being faster than that of intramuscular rocuronium. 2,10 

Thirty to 60 min after intramuscular administration, the simulated percentage of an administered dose that remained to be absorbed from the intramuscular depot was larger with rapacuronium than in our previous study with rocuronium. 9Despite a significant percentage of the bioavailable dose of rapacuronium remaining in the depot, and therefore slowly absorbed, we saw no evidence for residual weakness with intramuscular rapacuronium. It is possible that the differences in the simulated quantities of rapacuronium and rocuronium in the intramuscular depot are artifacts of the analysis, particularly the use of two absorption compartments for rapacuronium and only one for rocuronium. Alternatively, the larger weight-normalized volume of intramuscular rapacuronium compared with that of intramuscular rocuronium may affect the absorption profile. The absence of observations of the quantity of drug remaining in the depot limits our ability to quantify this amount.

Plasma concentrations of the rapacuronium metabolite are less than those of rapacuronium during the first hour after bolus administration of rapacuronium, but exceed the concentration of rapacuronium after approximately 2 h. Because the effects of a bolus dose of rapacuronium dissipate within 15–30 min, and those of an intramuscular dose within 30–60 min, the 2-h concentrations of both the parent compound and the metabolite are not clinically important, despite the metabolite having a potency (defined by the plasma concentration that produces 50% twitch depression) of 50–100% more than that of rapacuronium. 11However, with repeated administration, the concentration of the rapacuronium metabolite increases progressively, presumably leading to a cumulative effect, as has been shown for other nondepolarizing muscle relaxants that have a potent metabolite. 12,13 

Several aspects of our study design warrant comment. We obtained only four samples per patient. In addition, we estimated bioavailability without using the traditional crossover design in which each subject is observed on two or more occasions: once with intravenous administration and another with an alternate route of administration. Recently, Wright and Fisher 14demonstrated that bioavailability could be estimated accurately using mixed-effects modeling (as in the current study) with as few as four samples per patient and a noncrossover design. Similarly, four samples were sufficient to estimate pharmacodynamic parameters accurately in simulations reported by Hashimoto and Sheiner. 15In addition, we used an absorption model that differed from the usual first-order absorption model. Although we have no specific evidence to support this (or the simpler) model, our model with two absorption rate constants fits the data better than does the simpler model.

A second issue regards accuracy of certain of our pharmacokinetic parameters. Because we obtained venous, rather than arterial, samples to determine plasma concentrations of rapacuronium, and relatively few samples were obtained during the early distribution phase, it is possible that estimates for the rapid distribution half-life may be flawed. In addition, use of venous samples probably leads to an inaccurate estimate of V¹.

A third issue regards the estimate for V^ss: 99.1 ml/kg. This value is smaller than the value for extracellular fluid (approximately 20–25% of body weight), 16the body compartment into which muscle relaxants are believed to be distributed. However, the estimate of the V^ssof rapacuronium is not markedly different than that for atracurium in children (129 ml/kg). 17 

Finally, we analyzed data for intramuscular and intravenous administration simultaneously. It is possible that the pharmacokinetic model that we use to explain intramuscular absorption is grossly incorrect. If so, the simultaneous analysis of intramuscular and intravenous plasma concentration data may “contaminate” the results that would be obtained if the intravenous data were analyzed separately.

In summary, Cl for rapacuronium is smaller in infants and children than in adults, but maturational changes in volume of distribution result in a more rapid decay in infants than in children and in children compared with adults. With intramuscular administration, 56% of the administered dose is bioavailable. Rapacuronium is absorbed at two different rates, with approximately 28% having an absorption half-life of 63 min and the remainder absorbed with a half-life of 14 min. Plasma concentrations of rapacuronium’s potent metabolite are less than those of the parent compound during the initial 60 min after bolus rapacuronium administration.