Effect of Renal Failure and Cirrhosis on the Pharmacokinetics and Neuromuscular Effects of Rapacuronium Administered by Bolus Followed by Infusion

The study was conducted at two sites: the University of California San Francisco and the University of Liverpool, United Kingdom. With local institutional review board approval and informed consent, 59 nonpregnant patients aged 18–65 yr were enrolled at the University of California San Francisco (n = 32) and at University of Liverpool (n = 27). Most patients underwent peripheral procedures; however, some patients with renal failure underwent abdominal procedures, such as placement of catheters for peritoneal dialysis or repair of incisional hernia. Patients were divided into three groups: healthy controls (normal renal and hepatic function, n = 25), chronic renal failure (serum creatinine > 4.5 mg/dl, n = 28), and cirrhosis (Pugh-Child class A or B, n = 6). Patient weight was 76 ± 15 kg (mean ± SD) and did not differ between groups. Patients were not receiving any other drugs expected to influence the neuromuscular response to rapacuronium.

After an 8-h fast, patients were premedicated with 1 to 2 mg lorazepam orally or with 1 to 2 mg midazolam intravenously. After administration of 1–3 μg/kg fentanyl, anesthesia was induced with 3–5 mg/kg thiopental via  an intravenous catheter placed in an upper extremity; anesthesia was then maintained with isoflurane (end-tidal concentration of 0.6%). An intravenous catheter was placed in the contralateral arm or in the external jugular vein (e.g. , if a vascular shunt was present in one arm) to sample blood. Ventilation was controlled to maintain normocapnia (end-tidal partial pressure of carbon dioxide [Pco²] of 30–35 mmHg). Esophageal temperature was maintained more than 36°C.

After loss of consciousness, supramaximal train-of-four (TOF) stimuli were applied to the ulnar nerve every 12 s. Mechanical twitch response of the adductor pollicis was measured with a calibrated force displacement transducer, amplified, and recorded on a strip chart. The first twitch response of each TOF (T1) was stable for more than 5 min before rapacuronium administration. The ratio of the fourth component to the first component of each TOF was determined.

Rapacuronium, 1.5 mg/kg, was administered over 5 s into a rapidly flowing infusion. When T1 recovered to 5% of control, rapacuronium (diluted to a concentration of 2.0 mg/ml) was infused for 30 min, starting at a rate of 50 μg · kg⁻¹· min⁻¹. If twitch depression was outside of the range of 90–95%, the rapacuronium infusion rate was changed in increments no smaller than 8.3 μg · kg⁻¹· min⁻¹. Each infusion rate was maintained for a minimum of 3 min. After the infusion was discontinued, twitch was permitted to recover spontaneously to 25% of control, at which time patients received 50 μg/kg neostigmine preceded by 10 μg/kg glycopyrrolate. In the first 11 patients at the University of California San Francisco (four healthy controls, five patients with renal failure, and two with cirrhosis), twitch was permitted to recover as long as clinically possible; if necessary, neostigmine and glycopyrrolate were administered to ensure complete neuromuscular recovery.

Venous blood, 5 ml, was sampled before and 2, 4, 7, and 10 min after the bolus dose and immediately before the start of the infusion. Additional samples were taken immediately before the end of the 30-min infusion and 2, 4, 7, 10, 20, 30, 45, 60, 75, 90, 120, and 150 min and 3, 4, 5, 6, 7, and 8 h after the infusion. At the University of California San Francisco, additional samples were taken at 10 and 20 min after the start of the infusion. Sodium dihydrogen phosphate was added to blood samples immediately to prevent degradation of rapacuronium. Blood was centrifuged within 30 min of sampling and plasma was stored at −20°C. Concentrations of rapacuronium and its primary 3-OH metabolite, ORG9488, were determined by Covance Laboratories (Madison, WI) using a high-performance liquid chromatography–mass spectometry (HPLC-MS) technique. The assay is linear for concentrations more than 2 ng/ml for both rapacuronium and ORG9488 and has a coefficient of variation of less than 11% for rapacuronium and less than 22% for ORG9488.

The pharmacokinetic characteristics of rapacuronium were determined using a population approach: mixed-effects modeling (NONMEM). Plasma concentration values for all subjects were analyzed simultaneously to determine typical values for the pharmacokinetic parameters, standard errors of these estimates, and interindividual variability. In addition, we determined the influence of renal function, cirrhosis, and other covariates (e.g. , demographic characteristics and preoperative laboratory values) on the pharmacokinetic parameters. 6 

Two-compartment models had the parameters clearance (Cl), distributional clearance (CL^distribution), and volumes of the central and peripheral compartments (V¹and V², respectively). Three-compartment models had, in addition, a slow distributional clearance (CL^slow) and a volume of the deep peripheral compartment (V³); CL^distributionwas renamed CL^rapid. Interindividual variability was permitted for each of these parameters and was assumed to have a log-normal distribution. Residual error between measured and predicted concentrations was initially assumed to have two components, one proportional to the predicted concentration (constant coefficient of variation), one additive; additional error models were tested.

A model-building approach was used. Initially, patients with renal failure and those with cirrhosis were assumed to have the same pharmacokinetic parameters as healthy controls. Two- and three-compartment models, both weight-normalized and non–weight-normalized, were compared to determine the appropriate structural model. Appropriateness of the error model was determined by visual inspection of the residual errors. After the population analysis was performed for each model, the NONMEM post hoc  step was performed. This Bayesian step determines the parameter estimates for each individual in comparison with the population estimates. These differences are quantified through the NONMEM η (eta ) terms. The resulting values for η were plotted against the covariates age, weight, height, gender, group (renal failure vs.  cirrhosis vs.  healthy controls), and preoperative values for hematocrit, hemoglobin, and serum concentrations of creatinine, creatinine clearance (estimated using the Cockcroft-Gault nomogram 7), bilirubin, aspartate transaminase (AST), and alanine transaminase (ALT). After a smoother (lowess, a local regression) was added to each plot, trends were sought by visual inspection. If a relation between a pharmacokinetic parameter and a covariate was observed, this relation was tested in the model. Additional parameters were accepted in the model if they improved the NONMEM objective function statistically (for P < 0.01, 6.6 units for one additional parameter, 9.1 units for two). Half-lives were calculated using standard formulas. The effect of renal failure and cirrhosis on plasma concentrations of ORG9488 was assessed visually.

Maximal twitch depression (expressed as percentage depression from the predrug control value), time from administration of rapacuronium to maximum twitch depression, and (for those patients who received rapacuronium by infusion) time to 5% recovery of T1 after the initial bolus dose were calculated. Total rapacuronium dose administered during the infusion was calculated. For those subjects who received neostigmine at 25% recovery after the infusion, time from end of the infusion to 25% recovery of T1 and time from neostigmine to recovery of the TOF ratio to 0.7 were determined. Values for the three groups were compared using analysis of variance or the Kruskal-Wallis test, followed by the Student-Newman-Keuls test. Values are reported as mean ± SD. P < 0.05 was considered statistically significant.

Two patients with renal failure received only the bolus dose of rapacuronium and blood sampling was discontinued at 10 min because of technical difficulties; these patients are included in the pharmacokinetic analysis but not in the calculation of infusion rates and neuromuscular effects. In an additional nine patients (four healthy controls and five patients with renal failure), no blood was sampled; these patients are not included in the pharmacokinetic analysis.

In one healthy patient, maximum twitch depression was 81%; the remaining patients all developed 98% or more twitch depression. Time to peak twitch depression was 600 s in the patient who developed 81% twitch depression and 275 s in one renal failure patient. In the remaining patients, onset time was similar in the three groups (table 1). Time to 5% recovery of T1 (at which time the infusion was started) was similar in the three groups. The dose of rapacuronium infused over 30 min to maintain target twitch depression was 22% smaller in patients with renal failure than in healthy controls (fig. 1). After the end of the infusion, time to 25% recovery and time from neostigmine to recovery of a TOF ratio of 0.7 were similar for the three groups.

Plasma concentration data were obtained in 50 patients (of whom two received only the bolus dose). After the bolus dose, plasma concentrations of rapacuronium initially decreased rapidly in all groups (fig. 2). During the infusion, rapacuronium concentrations increased slightly and were similar in the three groups. After the infusion, plasma concentrations of rapacuronium initially decreased at a similar rate in the three groups. However, 8 h after the end of the infusion, rapacuronium concentrations were slightly larger in patients with renal failure than in the other groups (fig. 3), despite patients in renal failure receiving smaller doses.

For rapacuronium, weight normalization improved the quality of fit of the two-compartment model (model 1 vs . model 2;table 2); therefore, weight normalization was adopted for subsequent analyses. A three-compartment model markedly improved the quality of fit (model 3 vs . model 1). Permitting interindividual variability in CL^slowto differ from that for CL^rapidand for V³to differ from that for V²further improved the quality of fit (model 4 vs . model 3). An error model with only a single component, a constant coefficient of variation, fit as well as one that also permitted an additive component (model 5 vs . model 4). Therefore, all subsequent models used weight-normalized pharmacokinetic parameters and permitted interindividual variability in each of these parameters; the error model contained only a single component.

With model 5, plots of post hoc ηversus  covariates suggested that patients with renal failure had a smaller clearance. Permitting the typical value of Cl to differ in patients with renal failure compared with the other groups improved the quality of the fit markedly (model 6 vs.  model 5). With model 6, plots of post hoc η suggested that Cl decreased with age. Therefore, model 7 evaluated the effects of age on Cl using the following approach:

CL = THETA(1) · (1 + AGEFACTOR · (AGE − 45)) (1)

CL = THETA(7) · (1 + AGEFACTOR · (AGE − 45)) (2)

where CL is the typical value for a patient of a particular age, THETA(1) is the typical value for a normal or cirrhotic patient aged 45 yr, THETA(7) is the typical value for a renal failure patient aged 45 yr, 45 yr is the median age for the patients studied, and AGEFACTOR is estimated in the analysis. This model was statistically justified compared with the model in which age did not affect Cl (model 6); plots of post hoc η suggested that CL^rapiddiffered between genders. Permitting CL^rapidto differ between genders improved the quality of the fit (model 8 vs.  model 7). Post hoc  plots now suggested that V²was larger in cirrhotic patients, that CL^slowhad both a weight-normalized and additive component, or that serum albumin concentration affected V¹. However, incorporating each of these into the pharmacokinetic model failed to improve the quality of the fit further (models 9–11 compared with model 8). There was no evidence that incorporating serum creatinine or creatinine clearance into the model further improved the quality of fit.

Thus, the optimal model (table 3, fig. 4) for rapacuronium had three compartments and all pharmacokinetic parameters were weight-normalized. Clearance was 24% less in patients with renal failure than in the other groups. In all groups, Cl decreased 0.52%/yr of age compared with the value at age 45 yr (table 4). The coefficient of variation of the parameter estimate for the effect of age on Cl was 69%, suggesting that age might not influence Cl. However, a likelihood profile 8(not shown) demonstrated that the decrease in Cl/yr of age was at least 0.3%. CL^rapidwas 51% smaller in men than in women. Rapid and slow distribution half-lives were shorter in women than in men. Elimination half-life varied minimally as a function of gender, age, and renal function. There were no relation between the pharmacokinetic parameters and other covariates.

Plasma concentrations of ORG9488 were largest immediately after the bolus dose of rapacuronium in all groups, decreased rapidly, then increased during the rapacuronium infusion (fig. 5). After the infusion, plasma concentrations of ORG9488 decreased monotonically at a similar rate in healthy controls and in cirrhotic patients. In patients with renal failure, plasma concentrations of ORG9488 decreased minimally during the 8 h after the infusion (fig. 6). In one patient with cirrhosis, the time course of ORG9488 was similar to that of patients with renal failure; this patient’s serum creatinine was 2.3 mg/dl.

We found that the clearance of rapacuronium was 24% less in patients with renal failure than in normal patients and those with cirrhosis. As a result, doses of rapacuronium required to maintain a target range of neuromuscular depression were smaller in patients with renal failure. The magnitude of decrease in the clearance in renal failure in the current study is similar to that reported previously by Szenohradszky et al.  1(32%) in a similar group of patients aged 18–45 yr. Patients with renal failure studied by Szenohradszky et al.  1did not have a longer median time to 25% recovery of T1 or to a TOF ratio more than 0.7. However, a few of their nine patients with renal failure had prolonged recovery, a finding that would be accentuated if supplemental doses of rapacuronium were administered. Thus, our finding that rapacuronium infusion requirements are decreased by renal failure is consistent with the findings reported by Szenohradszky et al.  1The combined results of these studies suggests that rapacuronium’s recovery profile after a single dose or a bolus dose followed by a 30-min infusion adjusted to maintain target twitch depression is usually not affected by renal function.

In cirrhotic patients, rapacuronium’s clearance did not differ from that in healthy adults. This finding conflicts with that of Duvaldestin et al.  2who reported that rapacuronium’s clearance was larger in cirrhotic patients (median: 6.9 ml · kg⁻¹· min⁻¹; range: 6.1–8.9 ml · kg⁻¹· min⁻¹) than in healthy controls (median: 5.3 ml · kg⁻¹· min⁻¹; range: 4.2–8.4 ml · kg⁻¹· min⁻¹). Duvaldestin et al.  2also reported that rapacuronium’s steady state distribution volume was 50% larger in cirrhotic patients than in healthy controls, whereas we observed no effect of cirrhosis on distribution volume. As in the current study, neuromuscular recovery was similar in patients with cirrhosis compared with healthy controls. The cirrhotic patients in the two studies appear to be similar: the diagnosis of cirrhosis was typically confirmed by liver biopsy and patients had similar Pugh-Child scores (5–9 in the current study, 5–10 in the Duvaldestin et al.  2study). Regardless, these studies suggest that cirrhosis does not prolong the duration of action of rapacuronium either as a single bolus dose or as a bolus followed by a 30-min infusion and that rapacuronium dosing in cirrhotic patients should probably be similar to that in healthy patients. However, the number of cirrhotic patients investigated in both studies is small so that additional pharmacokinetic studies may be needed in this patient population.

We also observed that rapacuronium’s clearance decreased with age. This finding is similar to that of Szenohradszky et al.  1who reported that Cl decreased 0.9%/yr (compared with Cl at age 30 yr) in a group of volunteers and patients aged 18–45 yr. In contrast, Fisher et al.  9reported that age did not affect Cl in a group of patients aged 24–83 yr. The major difference between the single study that did not demonstrate an effect of age and those that did is the duration and intensity of blood sampling: the two studies that demonstrate an effect of age sampled for 8 h or more and obtained more than 15 samples/patient; the study that did not demonstrate an effect of age sampled for less than 4 h and obtained only 3 to 4 samples/patient. It is likely that the limited duration of sampling in the one study prevented the investigators from identifying an age-related effect that was not apparent during the initial several hours of sampling.

The other major finding of the current study is that elimination of rapacuronium’s active metabolite ORG9488 is markedly delayed in patients with renal failure. When Szenohradszky et al.  1gave a single 1.5-mg/kg dose of rapacuronium to patients with renal failure, plasma concentrations of ORG9488 persisted in the range of 200 ng/ml for 8 h. In the current study, in which the total rapacuronium dose averaged 2.66 mg/kg in patients with renal failure, plasma concentrations of ORG9488 also persisted for 8 h after the final administration of rapacuronium and were proportionally larger than those in the previous study. This finding in patients with renal failure contrasts to that in healthy patients and in all but one of those with cirrhosis in whom plasma concentrations of ORG9488 decrease consistently when rapacuronium is no longer administered. The cirrhotic patient who had persistent concentrations of ORG9488 had an abnormal serum creatinine but did not meet our criteria for renal failure; this emphasizes the impact of renal dysfunction on the elimination of ORG9488.

The importance of persistent concentrations of ORG9488 results from its potency—a study by Schiere et al.  4suggests that, based on steady state plasma concentrations, ORG9488 is 2 to 3 times as potent as rapacuronium. The typical concentration of ORG9488 attained in renal failure patients in the current study, approximately 500 ng/ml, is approximately one fourth of the concentration that Schiere et al.  4calculated would produce 50% twitch depression at steady state. Thus, persistence of ORG9488 for many hours after rapacuronium administration might place a patient with renal dysfunction at risk for prolonged weakness in clinical practice. The mode of rapacuronium administration in the current study—adjustment of the infusion rate to maintain twitch tension at 5–10% of control—resulted in patients with renal failure receiving smaller rapacuronium doses than in the other groups and permitted rapid facilitated recovery. Had we dosed rapacuronium at the same rate in all groups, plasma concentrations of both rapacuronium and ORG9488 would have been larger in renal failure patients than in the other groups. In turn, paralysis may have been prolonged by renal failure. However, adjusting the rapacuronium dose minimized the impact of renal failure on accumulation of ORG9488.

Finally, we observed that spontaneous recovery of neuromuscular function to 25% of control and facilitated recovery to a TOF ratio of 0.7 were similar in the three groups. This finding is expected in patients with cirrhosis in whom plasma concentrations of rapacuronium and ORG9488 typically follow a time course similar to that in normal patients. However, we speculate that the decreased Cl of rapacuronium and the markedly prolonged elimination of ORG9488 in patients with renal failure might impair the rate of both spontaneous and facilitated recovery if supplemental rapacuronium is administered without maintaining target twitch depression or if rapacuronium is dosed for more than 30 min.

In the current study, we did not model the pharmacodynamics of rapacuronium. Such modeling depends on accurate data regarding the relative potency and the equilibration rate constant for rapacuronium’s active metabolite compared with rapacuronium. Although Schiere et al.  4reported the pharmacokinetic characteristics of rapacuronium and its active metabolite, their large variability in the potency of both compounds (e.g. , a fourfold range of potency values for ORG9488) limits applicability of simulations based on average potency values. Had Schiere et al.  4performed a crossover study in which volunteers received rapacuronium on one occasion and ORG9488 on another, we might have had sufficient information to model the impact of renal failure and cirrhosis on dose requirements of rapacuronium. However, the results of the current analysis are consistent with our clinical finding that renal failure decreases the infusion requirement for rapacuronium.

A second limitation of the current study is that the number of cirrhotic patients was small, a result of difficulty in recruiting sufficient candidates at the two sites. However, the findings in cirrhotic patients are similar to those in a previous study.

In summary, we confirm that rapacuronium’s Cl is decreased in patients with renal failure. In addition, we confirm our finding that rapacuronium’s Cl decreases with age. Unlike Duvaldestin et al. , 2we found no effect of cirrhosis on rapacuronium’s Cl or volume of distribution in a small cohort of patients. Finally, as in previous studies, we observed that elimination of rapacuronium’s active metabolite ORG9488 is markedly delayed in patients with renal failure. Our findings indicate that rapacuronium dosing is not affected by cirrhosis but that its maintenance requirement is decreased in patients with renal failure.

The authors thank Scott Kelley, M.D., Department of Anesthesia, University of California San Francisco, for recruiting patients and James E. Caldwell, M.B., Department of Anesthesia, University of California San Francisco, for assisting with the study.