Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Epidural and intrathecal local anesthetics, opioids, and ₂adrenergic agonists (i.e., clonidine) are commonly used during the postoperative period to relieve pain. Opioids and clonidine act by receptor binding at the spinal level and at the systemic level after absorption in the systemic circulation and subsequent brain distribution. In contrast, local anesthetics act by inhibition of nerve influx transmission only at the spinal level, but their systemic absorption after epidural administration is significant and leads to systemic adverse effects such as cardiac and neurologic toxicities.

After epidural administration, these drugs need to cross the spinal meninges (i.e., dura mater and arachnoid mater) to reach their site of action. However, if the spinal disposition of opioids and clonidine has been studied extensively, the spinal disposition of local anesthetics has been investigated poorly. This has been done only once for local anesthetics (Wilkinson and Lund, 1970), and frequently for opioids (Nordberg et al., 1983, Gourlay et al., 1987, Hansdottir et al., 1991, 1995; Taverne et al., 1992) and clonidine (Eisenach et al., 1993, 1995). The distribution of these agents in cerebrospinal fluid (CSF) has been studied after repeated intrathecal punctures or insertion of indwelling intrathecal catheters to withdraw CSF. These studies demonstrated a rather low CSF bioavailability lower than 4% for pethidine, morphine, and sufentanil and 14% for clonidine. Moreover, a relationship between CSF concentration and analgesic effect has been described for clonidine (Eisenach et al., 1993).

Using experimental ex vivo models, authors have studied extensively the permeability of dura mater (Moore et al., 1982, McEllistrem et al., 1993) or meninges (Bernards and Hill, 1990) to different drugs. They showed that a simple passive diffusion mechanism is likely. Relationships between permeability and different physicochemical properties were studied but the results appeared controversial depending on the models (dura alone or all meninges) and on the nature of meninges (humans or animals). However, these ex vivo models do not take into account the gradient of pressure existing between epidural and intrathecal spaces, the meningeal surface area in contact with solution injected epidurally, uptake by epidural venous blood vessels, and uptake into epidural tissues such as epidural fat.

Given the paucity of data on local anesthetics, excepting ex vivo studies, in vivo investigations thus are strongly required to understand the spinal disposition of these drugs and to study the in vivo drug release from drug delivery systems for local anesthetics currently investigated to improve the clinical and toxicological features of these drugs (Boogaerts et al., 1995, Le Corre et al., 1995, Fréville et al., 1996). We recently validated the use of microdialysis coupled to HPLC to study the intrathecal disposition of bupivacaine (BUP) in rabbits (Clément et al., 1998). This technique, allowing measurement of drug concentrations without removing CSF, also permits access to concentrations of local anesthetics in the epidural space, where it is impossible to withdraw any epidural fluid because of cephalad and caudal spread of the solution injected epidurally (Paul et al., 1989). The aim of this study was to investigate the spinal disposition of lidocaine (LID) and bupivacaine after epidural and intrathecal administration in awake rabbits.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. Local anesthetics (bupivacaine, etidocaine, lidocaine, and ropivacaine) were supplied by Astra (Astra Pain Control, Sweden). A Ringer's solution (8.6 g/liter NaCl, 0.33 g/liter KCl, 0.3/liter CaCl₂, 2H₂O) was used as perfusion fluid. All other reagents were of analytical grade.

Animals. The study was approved by the Committee of Laboratory Investigation and Animal Care of our institution and performed in accordance with French Ministry of Agriculture laws and guidelines for laboratory animal experiments. Experiments were performed on female New Zealand albino rabbits, weighing 3.0 ± 0.3 kg, housed individually in standard cages with free access to food and water. Animals were fasted the night before the experiment.

Animal Preparation. After catheterization of a marginal vein of ear, anesthesia was induced with 25 mg of 2.5% thiopental. Insertion of spinal and vascular catheters was performed under epidural procaine anesthesia and light sedation, maintained with i.v. 1% thiopental (5-10 mg). The technique of epidural anesthesia was described elsewhere (Malinovsky et al., 1997). In the supine position, the femoral area was dissected surgically and an arterial 20-gauge catheter was inserted via the femoral artery. The catheter was heparinized and then closed until blood sampling.

Microdialysis Conditions. Microdialysis sampling was performed using a CMA/102 microinjection pump coupled to a microdialysis probe CMA/120 (membrane length, 10 mm; 0.5-mm outer diameter; molecular mass cut-off, 20 kDa; CMA Microdialysis, Sweden). Microdialysis samples were collected every 2 min during a 1-min interval. The in vitro study of the kinetics of probe response showed that a lag time of 3 min was necessary to obtain the experimental drug concentration (data not shown). Hence, in all the in vivo experiments the first experimental drug concentration considered was that measured at 3 min. Dialysates (sample volume = 1 µl) were collected in vials containing 100 µl of etidocaine (1 µg/ml), and a 50-µl aliquot was injected onto the chromatograph.

Microdialysis Calibration. Retrodialysis, using ropivacaine as internal standard (IS), was applied to calibrate the microdialysis probes. This calibration technique is based on the principle that the relative loss (RL) of a carefully chosen internal standard, added to the perfusate, is related to the relative recovery (RR) of the substance of interest (SI) (Scheller and Kolb, 1991). The K factor values, defined as the ratio RL_IS/RL_SI, were used to determine the extracellular concentration of the compounds of interest according to C = C_dialysate × K/RL_IS (Deleu et al., 1995). The validation of the calibration was assessed by comparing retrodialysis with the zero-net flux method, where the recovery is estimated from dialysate concentrations in a wide range of concentrations while maintaining the extracellular concentration at steady state (Wang et al., 1993).

Evaluation of Unbound Fraction of BUP and LID in CSF. CSF samples (between 1.5 and 2 ml) were withdrawn from the cisterna magna by puncture through the atlanto-occipital membrane in six rabbits. Individual CSF samples from three rabbits were divided in two aliquots for the separate evaluation of BUP and LID binding at a concentration of 100 µg/ml. CSF samples from three rabbits were pooled and then divided in two aliquots for the separate evaluation of BUP and LID binding at the following concentrations: 50, 100, 250, and 500 µg/ml. The binding was studied using microdialysis by spiking a CSF aliquot either with BUP or LID.

Chromatographic Analysis of BUP and LID. The separation and quantification of the local anesthetics in the dialysate (CSF, epidural samples) or in plasma samples were carried out using a HPLC method with UV absorbance detection ( = 205 nm). Dialysate samples were injected immediately onto the chromatographic system. The blood samples were centrifuged and plasma was stored frozen until analysis. BUP and LID in plasma were extracted from plasma before analysis by HPLC according to a previously published method (Le Guevello et al., 1993). The limit of quantification of BUP and LID was 3 µg/ml and 1.5 µg/ml in dialysate and 4 ng/ml and 2 ng/ml in plasma, respectively.

Study Design. CSF disposition of local anesthetics only has been studied in each animal either after intrathecal administration or after epidural administration because meningeal puncture with a needle diameter size larger than 0.4 mm (equivalent to the hole made with the 24-gauge needle) significantly increases the flux of local anesthetics through meninges (Bernards et al., 1994).

Pharmacokinetic Analysis. Intrathecal- and epidural-population pharmacokinetic parameters of BUP and LID were determined using the statistical pharmacokinetic software P-Pharm (version 1.5; Innaphase, Champs sur Marne, France). Intrathecal concentration data after intrathecal administration and epidural concentration data after epidural administration were fitted according to a biexponential model. Intrathecal concentration data after epidural administration were fitted according to a triexponential model. The distribution of the random effect was assumed as normal, and the residual error variance was assumed as heteroscedastic (in proportion to the squared value of the predictions). Initial population parameter estimates were derived from the mean of the individual parameter values obtained by using a stripping algorithm. Individual parameters for each data set (Bayesian estimates) were obtained from the current population parameters and the individual data.

Statistics. All data are presented as mean ± S.D. Student's t test was used to compare individual means. A P value less than .05 was considered as statistically significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Epidural Microdialysis after Epidural Administration. The individual epidural concentration-time profiles of BUP and LID after equimolar epidural administration (low and high doses) are presented in Fig. 1. The values of pharmacokinetic parameters after the administration of the high doses of BUP and LID are listed in Table 1. Pharmacokinetic modeling was achieved only after the administration of the high dose (i.e., the dose used in the intrathecal evaluation after epidural administration) because too few data were obtained in the terminal elimination phase after the administration of the low dose. The epidural concentration-time curves showed a biphasic decline with an apparent terminal elimination phase occurring earlier for BUP.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Measured and population-predicted epidural concentrations of bupivacaine (left) and lidocaine (right) after separate epidural administration of equimolar low dose (bottom, 0.69 µM: 0.2 mg of BUP and .162 mg of LID) and high dose (top, 6.9 µM: 2 mg of BUP and 1.62 mg of LID) in rabbits.

Intrathecal Microdialysis after Intrathecal Administration. Individual CSF concentrations of BUP and LID after equimolar intrathecal administration are illustrated by Fig. 2 and show a biexponential decline. The pharmacokinetic parameters are presented in Table 2. The concentrations of BUP and LID in plasma were beyond the limits of detection during the intrathecal experiment. The AUC_last/area under CSF concentration-time curves from the time of drug administration to infinity (AUC_inf) ratio of intrathecal BUP (87.8%) and LID (91.7%) indicated that the sampling period was long enough to obtain a suitable description of the CSF disposition of these drugs.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Measured and population-predicted intrathecal concentrations of bupivacaine (top) and lidocaine (bottom) after separate intrathecal administration of equimolar dose (0.2 µM: 0.060 mg of BUP and .049 mg of LID) in rabbits.

Intrathecal and Plasma Disposition after Epidural Administration. Figure 3 shows the individual CSF concentrations of BUP and LID after administration of an epidural equimolar dose. The CSF pharmacokinetic parameters are presented in Table 3. The AUC_last/AUC_inf ratio of intrathecal BUP (85.5%) and LID (95.3%) indicated that the sampling period was long enough to obtain a suitable description of the CSF disposition of these drugs. The mean total plasma C_max of BUP was slightly higher than that of LID (2.23 ± 1.25 µM versus 1.24 ± 0.41 µM, P = .13). The T_max of BUP was significantly lower than that of LID (1.6 ± 1.5 min versus 3.8 ± 1.1 min, P < .05).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Measured and population-predicted intrathecal concentrations of bupivacaine (top) and lidocaine (bottom) after separate epidural administration of equimolar dose (6.9 µM: 2 mg of BUP and 1.62 mg of LID) in rabbits.

Unbound Fraction of BUP and LID in CSF. The unbound fractions at the concentrations of 50, 100, 250, and 500 µg/ml were 39.4, 39.5, 39.7, and 43.7% for BUP and 80.0, 76.3, 79.4, and 73.8% for LID, respectively. The mean unbound fractions of BUP and LID, at a concentration of 100 µg/ml, were 39.3 ± 2.3% and 75.8 ± 7.7% (n = 4), respectively.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

At present, the evaluation of CSF disposition of drugs has been achieved only by using classical CSF sampling methods, i.e., repeated punctures or sampling through a catheter. The main methodological drawback of these methods is that sampling of CSF may interfere with CSF dynamics and disturb the disposition of intrathecally administered drugs. To minimize this problem, sampling frequency has been lowered, precluding suitable pharmacokinetic analysis, especially for drugs displaying rapid disposition steps. This point is of interest because numerous drugs used in anesthetic practice have a high lipophilicity and may have rapid kinetics. Hence, for such studies, microdialysis sampling strategy appears of paramount interest. However, the evaluation of the disposition of drugs in CSF might have some limitations because drug diffusion in CSF is not fast and uniform (Rigler et al., 1991) as it can be seen in blood, even though there is slow fluid dynamics in the intrathecal space.

Microdialysis technique, allowing measurement of unbound drug concentrations, leads to the determination of unbound pharmacokinetic parameters of the pharmacologically active fraction of drugs. However, we have established that there is a drug binding in CSF that was linear for BUP and LID in the concentration range studied. The mean unbound fraction of BUP was 2 times lower than that of LID. A lowest unbound fraction of BUP in plasma also has been established compared with LID. However, the plasma protein binding of BUP (Denson et al., 1984) and LID (McNamara et al., 1981) is nonlinear in humans. This difference in the binding pattern may be apparent and may result from the high difference (25- to 100-fold) in drug concentration between plasma and CSF studies and/or from species difference in binding.

The microdialysis technique also permitted us to describe the disposition of BUP and LID in the epidural space. After epidural administration, C_max-epi was lower than C_inj for both drugs and doses. There was also a difference in C_max-epi between BUP and LID (i.e., difference in C_max-epi/C_inj ratio between BUP and LID), which may be explained by a differential diffusion into epidural fat (Rosenberg et al., 1986). Furthermore, the assumption of a higher distribution of BUP within the epidural space is supported by the 4-fold difference in intercompartmental clearance (CL_I) between BUP and LID, which was close to statistical significance (P = .06). The higher epidural elimination clearance (CL_E) of BUP in comparison with LID suggested a more significant uptake of BUP into the systemic circulation and/or into the CSF.

After intrathecal administration of BUP and LID, we showed that the CSF disposition of both drugs displayed a biphasic pattern as described for clonidine in sheep (Castro and Eisenach, 1989) and sufentanil (Hansdottir et al., 1991), morphine, meperidine (Sjöström et al., 1987b), and neostigmine (Shafer et al., 1998) in humans. The biphasic pattern should result from different processes involved in the uptake of intrathecally injected drugs: 1) diffusion along the concentration gradient from CSF through the pia mater into the most superficial portions of the spinal cord (Greene, 1983); 2) access to the deeper areas of the spinal cord through the spaces of Virchow-Robin, which are extensions of the subarachnoid space accompanying the blood vessels penetrating the spinal cord from the pia mater (Greene, 1983); 3) drug diffusion through the arachnoid and dura mater and subsequent diffusion in the epidural space (Cohen, 1968); and 4) uptake into the blood vessels of the pia and dura mater (Vandenabeele et al., 1996).

We found a higher intrathecal CL_E of LID compared with BUP, which is in agreement with the faster systemic uptake of LID after intrathecal administration compared with BUP in humans (Burm et al., 1983). If the CL_E of LID from intrathecal and epidural spaces were close, BUP showed a lower intrathecal CL_E. In contrast to what was observed in the epidural space, CL_I values of BUP and LID in the intrathecal space were very close and lower than the epidural CL_I, highlighting the influence of the epidural fat in epidural drug disposition, especially for BUP. After epidural and intrathecal administrations, the intrathecal apparent elimination half-lives of BUP were close, in contrast to LID, which displayed an unexpected lower intrathecal apparent elimination half-life after epidural administration. It should be noticed that, in contrast to BUP, the CSF concentration profiles of LID were different after intrathecal and epidural administration, the terminal phase occurring later after epidural administration. Moreover, the concentration profiles of LID in CSF after epidural administration and in epidural sites are similar, indicating that the CSF kinetic disposition of LID was influenced by the epidural disposition of this drug, in contrast to BUP.

In the current study, we have estimated the CSF bioavailability of BUP and LID after epidural administration, and our data can be compared to those in the literature. The CSF bioavailability of BUP and LID were higher than those of lipophilic and hydrophilic opioids and close to that reported for clonidine (Table 4). In the current study, the diffusion rate through the meninges (estimated by T_max-csf) was close to that observed for sufentanil in dogs, but much more rapid than the data usually reported (Table 4). The current investigation showed that the CSF bioavailability of LID, the less lipophilic agent, was around three-times higher than that of BUP.

After epidural administration, drug epidural disposition results from three uptake competing processes: 1) through meninges into CSF, 2) through capillary vessels into systemic circulation, and 3) into epidural fat. Using an ex vivo model, Bernards and Hill (1990) suggested that permeability of BUP and LID through spinal monkey meninges was not different. However, the 4-fold difference in absorption rate constant between BUP and LID was not unexpected, considering the high lipophilicity of BUP. Furthermore, this may suggest that in vitro models should not accurately reflect the in vivo situation because of the relative complexity of the processes involved in the drug disposition in the epidural and intrathecal spaces.

Although differences in plasma concentrations between two drugs are dependent not only on diffusion but also on distribution and elimination, our data suggested that BUP diffuses more rapidly than LID into systemic circulation (T_max 2 times lower for BUP) and suggested that the amount of BUP diffusing into systemic circulation was higher than that of LID (C_max 2 times higher for BUP). Indeed, after i.v. administration of LID (4 mg/kg) (Doherty et al., 1995) and BUP (0.6 mg/kg) (Fréville et al., 1996) in rabbits, the mean apparent volume of distribution (Vd) of BUP was larger than that of LID (12.4 liters versus 9.4 liters). This would explain, in part, the difference in CSF bioavailability observed after epidural administration and was supported by the fact that epidural clearance of BUP was higher than that of LID. Moreover, this is supported by the fact that an increase in spinal effects of BUP and, to a lesser extent, of LID is observed when a vasoconstrictor such as epinephrine is added after epidural administration (Covino and Wildsmith, 1998).

The second process, which could explain the difference in CSF bioavailability, is the higher partitioning of BUP into epidural fat compared with LID (Rosenberg et al., 1986).

The CSF bioavailability of LID and BUP found in rabbits can be related to the clinical practice of anesthesia in humans. Indeed, the mean doses used epidurally and intrathecally are 440 mg and 90 mg for LID and 140 mg and 15 mg for BUP (Tucker and Mather, 1998), leading to ratios of epidural doses to intrathecal doses of around 5 for LID and 9 for BUP, i.e., to a clinically based bioavailability of approximately 20% for LID and 10% for BUP.

This first in vivo investigation of the spinal disposition of local anesthetics in animals has several pharmacological implications. Although CSF drug concentrations usually are considered as free drug concentrations, because protein CSF concentrations are very low, we have shown that there is a drug binding in CSF fluid. The investigation of the drug binding for local anesthetics in the human CSF should be performed and may explain the large variability of effect after their intrathecal administration, especially for BUP. Even if the diffusion rate through meninges of BUP was higher than that of LID, our investigation showed a higher CSF bioavailability for LID that should result from a difference in the epidural disposition of these drugs. Furthermore, such a model should be of interest to define the basis of the development of drug delivery systems for local anesthetics currently investigated to improve the clinical and toxicological features of epidural local anesthetics.