Intramuscular rocuronium, in doses of 1,000 microg/kg in infants and 1,800 microg/kg in children, produces complete twitch depression in 5-6 min. To determine the rate and extent of absorption of rocuronium after intramuscular administration, blood was sampled at various intervals after rocuronium administration by both intramuscular and intravenous routes to determine plasma concentrations (Cp) of rocuronium.
Twenty-nine pediatric patients ages 3 months to 5 yr were anesthetized with N2O and halothane. The trachea was intubated, ventilation was controlled, and adductor pollicis twitch tension was measured. When anesthetic conditions were stable, rocuronium (1,000 microg/kg for infants and 1,800 microg/kg for children) was injected either intramuscularly (in the deltoid muscle) or intravenously. Four venous plasma samples were obtained from each child 2-240 min after rocuronium administration. A mixed-effects population pharmacokinetic analysis was applied to these values to determine bioavailability, absorption rate constant, and time to peak Cp with intramuscular administration.
With intramuscular administration, rocuronium's bioavailability averaged 82.6% and its absorption rate constant was 0.105 min(-1). Simulation indicated that Cp peaked 13 min after rocuronium was given intramuscularly, and that 30 min after intramuscular administration, less than 4% of the administered dose remained to be absorbed from the intramuscular depot.
After rocuronium is administered into the deltoid muscle, plasma concentrations peak at 13 min, and approximately 80% of the administered drug is absorbed systemically.
Until recently, succinylcholine was the only muscle relaxant that produced effective neuromuscular blockade with intramuscular administration. We recently showed that deltoid (in contrast to quadriceps) injection of rocuronium (1,000 micro gram/kg in infants and 1,800 micro gram/kg in children) produces complete paralysis in 5–6 min in infants and children anesthetized with nitrous oxide and halothane, and also that these doses of intramuscular rocuronium permitted tracheal intubation at 2.5–3.0 min in a few lightly anesthetized patients.  No information exists regarding the rate and extent of absorption of rocuronium from the intramuscular depot. Accordingly, we measured plasma concentrations of rocuronium after intramuscular and intravenous administration to determine bioavailability.
The study was conducted using Organon Inc.'s Investigational New Drug Application, and the protocol was approved by our institutional review board. After obtaining informed consent from parents, we studied 29 pediatric patients who were classified as American Society of Anesthesiologists physical status 1 or 2 and undergoing elective surgery. Patients were divided by age into two groups: infants (ages 3–11 months; 8.4 +/- 1.8 kg [means +/- SD]; n = 13) and children (1–5 yr; 15.5 +/- 4 kg; n = 16). We intended to study 24 patients (12 infants and 12 children) but added five patients because of problems with the assay, which are described later. 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 during the period of the operation.
Eight patients received midazolam orally or nasally before operation; the remaining 21 were not premedicated. Anesthesia was induced with N2O and halothane. After establishing intravenous access, the trachea was intubated and ventilation was controlled to maintain normocapnia. The inspired halothane concentration was adjusted to produce end-tidal halothane concentrations of 0.8–1.0% in children younger than 2.5 yr and of 0.7–0.8% in children older than 2.5 yr. Patients were randomized to receive rocuronium as either an intramuscular (n = 14) or intravenous (n = 15) injection. When end-tidal halothane concentrations and baseline twitch recordings were stable for > 5 min, rocuronium (10 mg/ml) was injected via a 21-gauge needle 1–2 cm into a single deltoid muscle after negative aspiration for blood or as a rapid bolus into a peripheral intravenous catheter in the upper or lower extremity. The rocuronium dose was 1,000 micro gram/kg for infants and 1,800 micro gram/kg for children.
Five venous blood samples were obtained from each patient. The initial sample was obtained after induction of anesthesia but before rocuronium administration; the remaining four samples were obtained between 2 and 240 min after rocuronium administration. For each patient, target times for blood sampling differed, as determined by randomization (Table 1). This “randomized block” approach provides maximal information for a population pharmacokinetic approach when “sparse sampling” is used. 
Supramaximal square-wave train-of-four stimuli were administered at 2 Hz every 12 s to the ulnar nerve via needle electrodes at the wrist. Evoked tension of the adductor pollicis was measured using a Grass FT-03 force transducer (Grass Instrument Company, Quincy, MA) while maintaining preload at 50–100 g. The force signal was amplified (DC bridge signal conditioner; Gould Electronics, Valley View, OH), digitized (NB-MIO-16, National Instruments, Austin, TX) on a Macintosh computer, and displayed (LabView; National Instruments). The ratio of the first component of the train-of-four (T1) to its control value was determined. Each train-of-four was recorded on a strip chart (TA240; Gould).
Respiratory gas was sampled at the y-connector to determine the partial pressure of carbon dioxide by infrared analysis (Capnomac Ultima; Datex, Helsinki, Finland). Arterial oxygen saturation was measured continuously (N200 Oximeter; Nellcor, Hayward, CA). Systolic blood pressure and heart rate (Dinamap; Critikon, Tampa, FL) were measured before and every minute for 5 min after rocuronium administration.
The injection site and the skin of the trunk and face were observed for erythema; signs consistent with histamine release were sought. At the end of surgery, 20 micro gram/kg atropine and 70 micro gram/kg neostigmine were administered to antagonize residual paralysis. In the postanesthetic recovery room, patients were examined for signs of weakness, such as inability to sustain leg lift ; the injection site was examined for signs of tissue inflammation or damage.
Blood samples were iced immediately, and, after centrifugation, plasma was buffered with 0.8 M NaH2PO4and stored at -20 [degree sign] Celsius until analysis using gas-liquid chromatography.  The lower limit of sensitivity of the assay is 10 ng/ml rocuronium, with a coefficient of variation < 12% at that concentration.
Pharmacokinetic analyses were performed using the NONMEM statistical package.* As in all bioavailability analyses, we assumed that all clearances and volumes of distribution were identical with the two different routes of administration (intravenous and intramuscular); in turn, any difference between routes of administration in the plasma concentration (Cp) versus time profile results from differences in bioavailability, rate of absorption, and dosing regimen (if present). Analyses were performed using one-, two-, and three-compartment models. In addition, we tested models in which pharmacokinetic parameters were weight normalized and those in which they were not. Parameters of these pharmacokinetic models included volume of the central compartment (V1), volume of the second compartment (V2), volume of the third compartment (V3), clearance (Cl, elimination clearance equal to V sub 1 [center dot] k10), rapid and slow distribution clearance (Cl sub rapid, equal to V1[center dot] k12; and Clslow, equal to V1[center dot] k13), and volume of distribution at steady state (Vss, equal to V1+ V2+ V3). With intramuscular administration, we assumed first-order absorption with rate constant kafrom the intramuscular depot into the central compartment; bioavailability of intramuscular administration was designated F. Absorption half-life (t1/2 absorption) was determined as ln(2)/ka. For the two-compartment model, distribution and elimination half-lives (t1/2 alpha and t1/2 beta, respectively) were determined using standard equations.  For the three-compartment model, rapid and slow distribution half-lives (t1/2 pi and t1/2 alpha, respectively) and elimination half-life (t sub 1/2 beta) were determined iteratively.**
Intraindividual variability for Cl was assumed to be log-normally distributed, modeled as:Equation 1where Cliis the estimate for Cl for the ithindividual, Cl is the typical value for the population, and etaiis a random variable with mean 0.0 and variance Omega2. Interindividual variability for V1was modeled in a similar manner. Some models permitted interindividual variability for Clrapid, Clslow, V2, V3, ka, and/or F, modeled in a similar manner; however, interindividual variability in Clrapid, was assumed to be the same as for Clslowand that for V2equal to that for V3.
After the “typical” values were determined, we used NONMEM's post-hoc step to determine values for the pharmacokinetic parameters for each patient. Plots of the pharmacokinetic parameters for each patient were plotted against each of the covariates age and weight and examined 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 containing 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 requires a decrease of 6.6 units for one additional parameter or 9.2 units for two additional parameters), and they either improved the pattern of residual differences between measured and predicted values for Cp or decreased the trend between a pharmacokinetic parameter and a covariate.
Because time to peak Cp with intramuscular administration is a function of all the pharmacokinetic parameters, its typical value and confidence limits cannot be determined using Student's t values. Instead, we determined the time to peak Cp using NONMEM's simulation mode. Using the “typical” values for the pharmacokinetic parameters and allowing for the interindividual variability determined in the pharmacokinetic analysis, we simulated the Cp vs. time course of 250 individuals given rocuronium intramuscularly. Time to peak Cp was determined for each of these 250 individuals and the 5th, 50th (median), and 95th percentile values were determined. We also used the typical values for F and kato simulate the typical time course of rocuronium remaining to be absorbed from the intramuscular depot. Finally, we used NONMEM's post-hoc step to estimate the plasma rocuronium concentration in each patient 60 min after rocuronium administration. Mean values for intramuscular and intravenous administration were compared using Student's t test for unpaired data. Peak depression of twitch tension (expressed as a percentage decrease from the control value) and times to 10%(latency), 50%, and peak twitch depression (onset) and time to initial (> 1%), 10%, 25%(clinical duration), and 90% recovery of twitch tension (duration of action) were determined. Values are reported as means +/- SD. Comparisons between infants and children and between intravenous and intramuscular administration were performed using Student's t test for unpaired data; P < 0.05 (corrected, with the Bonferroni inequality, for multiple comparisons) was considered significant.
In six patients (four with intravenous administration, two with intramuscular), values for one to three plasma samples exceeded the upper limit of the standard curve for the rocuronium assay (Table 1). These 11 Cp values exceeding the upper limit of the standard curve for the assay were not used in the pharmacokinetic analysis; however, the standard curve was altered so that no subsequent samples exceeded the upper limit. To compensate for the missing Cp data, we studied an additional five patients, assigned to the same groups as the patients whose data were excluded. Four other patients had one supplemental plasma sample each: the infant (intramuscular group) at 2 min, and the children at 6 min and 27 min (intramuscular group) and at 180 min (intravenous group). In one child (intravenous group), the 190-min sample could not be obtained because venous access was lost. A total of 108 plasma samples (excluding blank samples) were used in the analysis, 57 of them from the intramuscular group (Figure 1).
A two-compartment pharmacokinetic model fit the Cp data better than a one-compartment model (as indicated by an improvement in the objective function of 68 points, and by visual assessment of the residual differences between predicted and observed concentrations;Table 2). Adding a third compartment further improved the quality of the fit (Figure 2). For both the two- and three-compartment models, weight normalization of the pharmacokinetic parameters improved the quality of the fit (model 5 vs. model 6 and model 10 vs. model 11). Permitting interindividual variability in Clrapid, Clslow, V sub 2, V3, kaor F or both did not further improve the quality of the fit. Plots of values for Cl and V1for each patient (determined using NONMEM's post-hoc step) versus covariates age and weight did not reveal a systematic influence of the covariates on the pharmacokinetic parameters.
Thus the “optimal” pharmacokinetic model had three compartments and permitted interindividual variability in Cl and V1. The typical value for Cl was 4.03 ml [center dot] kg sup -1 [center dot] min sup -1 (Table 3) and did not vary with age or weight (Figure 3); the typical value for V1was 105 ml/kg and did not vary with age or weight (Figure 4). With intramuscular administration, the typical value for F was 82.6% and did not vary with age or weight. The typical value of kawas 0.105 min sup -1 (resulting in an absorption half-life of 6.6 min) and did not vary with age or weight. Volume of distribution at steady state was 325 ml/kg (Table 4). Rapid and slow distribution half-lives were 2.5 and 27 min, respectively; elimination half-life was 127 min. These values did not differ between infants and children.
Simulations based on the pharmacokinetic parameters indicated that the median time after intramuscular administration when Cp peaked was 13 min; the 5th and 95th percentiles were 9 and 19.8 min, respectively. Fifteen minutes after intramuscular administration, the quantity of rocuronium simulated to remain in the intramuscular depot was 171 micro gram/kg in infants and 308 micro gram/kg in children (Figure 5); the value is greater in children than in infants because of the larger dose administered. Simulations indicated that 30 min after intramuscular administration, less than 4% of the administered dose remained to be absorbed from the depot. For infants, mean plasma rocuronium concentration 60 min after intramuscular administration was similar to that after intravenous administration (Table 5); however, for children, mean rocuronium concentration was larger after intramuscular administration.
All patients developed complete twitch depression. With both intramuscular and intravenous administration, times to 10% and 90% twitch depression were similar in infants and children (Table 6); however, time to twitch depression was more rapid with intravenous, compared with intramuscular, administration. Time to initial recovery was longer in children given rocuronium intramuscularly than in children given rocuronium intravenously and in infants given rocuronium intramuscularly. Except for these differences in recovery, time to initial recovery of twitch tension, clinical duration, and duration of action were similar after intravenous and intramuscular administration and in infants compared with children. Eleven infants and nine children, eight with intravenous administration and 12 with intramuscular administration, had residual neuromuscular blockade at the end of surgery and received neostigmine.
One child developed mild, transient erythema at the injection site; no other rashes or other signs of histamine release were observed. No patients had clinical signs of weakness during recovery. After administration of rocuronium, heart rate increased more than 20%(median, 26%, range, 21%-58%) in one infant and seven children, four with intramuscular administration and four with intravenous administration. Systolic blood pressure increased more than 20% in two infants and three children (median, 26%; range, 22%-40%), one with intramuscular administration and four with intravenous administration. No patient had more than a 20% decrease in heart rate or systolic blood pressure.
We recently reported that deltoid injection of rocuronium (1,000 micro gram/kg in infants and 1,800 micro gram/kg in children) produces complete paralysis and permits tracheal intubation at 2.5–3.0 min during light halothane anesthesia.  Time to 10% recovery of twitch tension with these doses was 72 +/- 13 min in infants and 86 +/- 27 min in children. These values are larger than reported for clinical doses of rocuronium given intravenously in pediatric patients. For example, Woelfel et al.  reported that time to 25% recovery was 41.9 +/- 13.6 min in infants and 26.7 +/- 6.6 min in children  given 600 micro gram/kg rocuronium; time to 10% recovery was not reported in these studies. Two explanations are possible for rocuronium's prolonged duration of paralysis with intramuscular administration. First, doses of rocuronium administered in our study were larger than those typically given intravenously to infants and children (although the extent to which the large dose is bioavailable after intramuscular administration was unknown), and no data exist regarding duration of intravenous rocuronium doses > 800 micro gram/kg in pediatric patients. Second, duration of paralysis might be increased if absorption from the intramuscular depot were delayed. Although all patients in the initial studies had adequate return of neuromuscular function (as demonstrated by recovery of train-of-four ratio and sustained leg lift ), the lengthy surgical procedures in some of these patients prevented us from assessing the earliest time at which recovery would be adequate. In particular, we were concerned that slow absorption from the intramuscular depot might sustain plasma concentrations of rocuronium and thereby place patients at risk for residual weakness. Thus the goal of the present study was to determine the extent and rate of absorption of rocuronium from the intramuscular depot.
The present study shows that bioavailability of intramuscular rocuronium is 82.6% and the half-life for absorption from the depot is 6.6 min. Based on “typical” values for bioavailability and absorption rate determined in the pharmacokinetic analysis, the quantity of rocuronium remaining to be absorbed from the intramuscular depot 30 min after rocuronium administration is 35 micro gram/kg in infants and 64 micro gram/kg in children (the larger residual dose in children is a function of the larger dose required in those patients). This residual quantity of rocuronium is < 11% of the usual clinical dose (600 micro gram/kg). For infants, plasma rocuronium concentrations are similar 60 min after intramuscular and intravenous administration (Table 5), consistent with the fact that recovery indices are similar for the two routes of administration (Table 6). Thus persistent absorption of rocuronium from the intramuscular depot is not likely to be a common clinical problem in infants. However, for children, plasma rocuronium concentrations are larger 60 min after intramuscular administration than with intravenous administration, consistent with the longer time to initial recovery with intramuscular administration. In our initial study,  the longest time to initial recovery of twitch tension was 121 min in a child (compared with an average value of 70 min in children); this prolonged recovery may have resulted from delayed absorption. These findings reinforce our earlier recommendation that time to 10% recovery of neuromuscular function with these large doses of intramuscular rocuronium is at least 1 h and that we do not recommend its use in procedures of briefer duration.
Pharmacokinetic simulations also indicate that the median time to peak Cp after intramuscular administration is 13 min but ranges from 9.0–19.8 min. This time to peak Cp explains why onset is delayed with intramuscular, compared with intravenous, administration, and suggests that the factor limiting onset of intramuscular rocuronium is the rate of absorption from the intramuscular depot.
The traditional approach to determine bioavailability is to study each patient twice, once with intravenous, administration and once with the alternate route. In addition, most studies obtain many plasma samples from each patient. Our approach differs in that we used an unpaired design with sparse sampling; that is, we studied different patients in the intramuscular and intravenous groups and obtained a small number of samples from each patient. Obtaining paired data from children would require us either to study volunteers or to study those children who require repeated anesthesia and surgery. Neither of these choices seemed appropriate-studying pediatric volunteers raises ethical issues, and studying only those children requiring repeated surgical procedures might preselect children with underlying medical or surgical problems that alter their pharmacologic response. Thus we considered a nontraditional unpaired design. The second deviation from the traditional approach is our use of a sparse sampling regimen. Traditionally, pharmacokinetic studies required > 10–15 samples per patient. In addition, data from each patient were analyzed separately. However, population pharmacokinetic tools such as nonlinear mixed-effects modeling and naive pooled data analysis permit reliable determination of pharmacokinetic characteristics when only a few samples are obtained from each person.
Recently, Wright and Fisher*** simulated whether a sparse sampling unpaired design yielded accurate estimates for bioavailability. Pharmacokinetic data for cisatracurium  were used to simulate data sets of ten persons in a paired study design (i.e., each was given an intravenous and intramuscular dose of the muscle relaxant) or 20 persons in an unpaired study design (in which ten are given an intravenous dose and ten an intramuscular dose). Wright and Fisher permitted interindividual variability in the pharmacokinetic parameters and residual error in the “measured” values for plasma concentration. The magnitude of interindividual variability and residual error was permitted to vary, and the number of samples per patient was either 15 (rich sampling) or three to five (sparse sampling). Despite permitting large amounts of interindividual variability, increasing residual error, a small number of samples, and model misspecification (e.g., using a one-compartment pharmacokinetic model to analyze data simulated from a two-compartment model), the value for bioavailability was estimated accurately. Thus it appears that our nontraditional approach yields reliable estimates for bioavailability.
In our pharmacokinetic analysis, permitting interinvidual variability for kaand F did not improve the quality of the fit. Although there is likely to be interindividual variability in both the rate and extent of absorption from the intramuscular depot, Figure 1shows that the Cp versus time curves for intramuscular administration have the same magnitude of interindividual variability as those for intravenous administration. This suggests that interindividual variability in both kaand F is small.
The present study provides pharmacokinetic parameters for rocuronium for pediatric patients ages 3 months to 5 yr; these values can be compared with values reported previously by Vuksanaj and Fisher  for children ages 4–11 yr. The value for Cl in the present study, 4.03 ml [center dot] kg sup -1 [center dot] min sup -1, is less than the value for the “typical”(50% percentile) 20-kg, 6-yr-old child reported in the earlier study (7.10 ml [center dot] kg sup -1 [center dot] min sup -1). The minimal age overlap in the two studies (only 3 of 20 patients in the previous study and 2 of 15 patients in the intravenous group in the present study were ages 4 or 5 yr) limits comparison of the two studies. However, findings from these two studies suggest that Cl varies minimally during the first few years of life, peaks at 4–6 yr, and then gradually decreases to the value of 2.89 ml [center dot] kg sup -1 [center dot] min sup -1 reported in adults. 
The present study provides additional insight into the potential clinical role of intramuscular administration of rocuronium. First, our analysis suggested that variability in bioavailability and rate of absorption were minimal or absent. Therefore, variability in the time course of intramuscularly administered rocuronium does not result from variability in absorption characteristics. Second, the bioavailability of 82.6% with intramuscular administration partially explains the magnitude of the dose required by this route. Whether intramuscular administration of rocuronium is widely adopted into clinical practice to replace elective intramuscular administration of succinylcholine remains to be determined.
In summary, when rocuronium is administered intramuscularly in the deltoid muscle, the absorption half-life is approximately 6.6 min, plasma concentrations peak at 13 min, and bioavailability exceeds 80%. These findings are consistent with initial neuromuscular recovery from these doses of intramuscular rocuronium occurring at 1–1.5 h, and they reinforce our recommendations that rocuronium should not be administered intramuscularly if duration of surgery is expected to be less than 60–90 min.
The authors thank H. Kenneth Spencer, Ph.D., and Jane Cyran, Ph.D., of Organon Inc., for their continued support of studies examining intramuscular administration of muscle relaxants.
*Beal SL, Sheiner LB: NONMEM's User's Guide. San Francisco, University of California San Francisco, 1992.
**Excel Solver. Microsoft, Redmond, Washington.
***Wright PMC, Fisher DM: Unpublished data.