There is a clinical requirement for longer-acting local anesthetics, particularly for the management of post-operative and chronic pain. In this regard, liposomes have been suggested to represent a potentially useful vehicle for sustained drug release after local administration. In the current study, the authors used a transmembrane pH gradient to efficiently encapsulate bupivacaine within large unilamellar vesicles. They report on the kinetics of drug uptake and release and the duration of nerve blockade.
The rate and extent of bupivacaine uptake into large unilamellar vesicles that exhibit a pH gradient (interior acidic) were determined and compared to drug association with control liposomes that did not exhibit a proton gradient. In subsequent studies, researchers examined the kinetics of bupivacaine release from these liposome systems in vitro. Using the guinea pig cutaneous wheal model, the rate of clearance of the liposome carrier was monitored after intradermal administration, using a radiolabelled lipid marker, and the duration of nerve blockade produced by free and liposomal bupivacaine was compared.
Bupivacaine was rapidly and efficiently accumulated within liposomes that exhibited a pH gradient (interior acidic) with trapping efficiencies of 64-82% of total drug, depending on the initial bupivacaine:phospholipid ratio. Little uptake was seen, however, for control vesicles that did not exhibit a transmembrane proton gradient. Using an in vitro model of drug clearance, liposomally encapsulated bupivacaine was found to be slowly released for a longer period of time compared with either the free drug or bupivacaine associated with control (no pH gradient liposomes). In the guinea pig cutaneous wheal model, more than 85% of the liposomal carrier was found to remain at the site of administration for 2 days. The sustained drug release afforded by liposomes that exhibited a pH gradient resulted in an increase in the duration of nerve blockade of as much as threefold compared with either the free drug or bupivacaine in the presence of control (no pH gradient) liposomes. Recovery of half maximal response (R2.5) after administration of 0.75% free bupivacaine, for example, was approximately 2 h, whereas the same dose of bupivacaine in pH gradient liposomes exhibited a R2.5 of approximately 6.5 h.
Large unilamellar vesicles that exhibit a pH gradient can efficiently encapsulate bupivacaine and subsequently provide a sustained-release system that greatly increases the duration of neural blockade.
Key words: Anesthetics, local: bupivacaine. Fats, liposomes.
EXCELLENT pain control can be achieved using local anesthetics to block specific peripheral nerves; however, these agents are limited by a relatively short duration of action. Even bupivacaine and etidiocaine, the longest-acting local anesthetics currently in clinical use, are only effective for 2-12 h, depending on the type of block and the dosage administered. Placement of an indwelling catheter allows repeated drug administration; however, this procedure is associated with potential complications, including blockage or breakage of the catheter, systemic toxicity as the results of elevated plasma drug levels, migration of the catheter subdurally (in the case of epidural catheterization), and infection. [2,3]A local anesthetic that could maintain neural blockade for 24 h or longer after a single administration would be of considerable clinical use. This need prompted the development of several slow-release formulations based on microspheres, [5-8]microdroplets, microcrystals, biodegradable polymers, and liposomes. [12-14]
Liposomes are microscopic spheres that consist of a phospholipid bilayer that encapsulates an aqueous core. They have been used as systemic delivery systems for a number of antineoplastics, antifungal agents, and antibiotics, and also have been found to enhance the therapeutic properties of local anesthetics. This later application takes advantage of the slow rate of clearance of liposomes after intradermal or subcutaneous injection [16,17]to provide a depot from which the local anesthetic can be released slowly, thereby prolonging nerve blockade. In particular, administration of lidocaine or bupivacaine in liposomes has been shown to prolong the duration of epidural anesthesia [18,13]and also, in the case of bupivacaine, to prolong conductance block in a murine model. Comparisons of plasma concentrations over time after administration of lidocaine or bupivacaine are consistent with more sustained release of local anesthetic from liposomes and correlate with a reduction in acute central nervous system and cardiac toxicities. These earlier reports are encouraging, particularly given that relatively unsophisticated liposomal systems were used. Generally, the liposomes that were used consisted of large multilamellar vesicles (MLVs; Figure 1) or small unilamellar vesicles, in which the local anesthetic was either partially encapsulated in the aqueous core or associated hydrophobically, as the free base, with the liposomal bilayer. As such, no mechanism existed whereby drug release rates, and, therefore, duration of action, could be controlled.
In the current work, we took advantage of two important developments in liposome technology. First we used liposomes of well defined size, large unilamellar vesicles (LUVs), produced by an extrusion procedure (Figure 1). These systems, which exhibit relatively large trapped volumes, allow greater quantities of drug to be encapsulated. In addition, LUVs would be expected to show improved drug retention properties because of their greater osmotic stability compared with MLVs. [22,23]In addition, we used a transmembrane pH gradient to encapsulate bupivacaine efficiently within the liposomal carrier. This technique has been termed "remote drug loading" and reflects redistribution of drugs that are weak bases or weak acids in accordance with the Hendersen Hasselbach equation (Figure 2). The remote loading procedure not only allows efficient drug accumulation within liposomes, but should also provide a mechanism whereby local anesthetic release rates can be controlled.
Materials and Methods
1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC) was purchased from Avanti Polar Lipids (Alabaster, AL) and was > 99% pure. Sodium chloride and sodium diphosphate (monobasic) were obtained from Fisher Scientific (Vancouver, B.C., Canada), and cholesterol, N-(2-hydroxyethyl)-piperazine-N-2 ethane-sulfonic acid, and bupivacaine were obtained from Sigma (St. Louis, MO), all with more than 99% purity. sup 14 Carbon-dipalmitoyl-phosphatidylcholine (115 mCi *symbol* mmol sup -1) was chosen as a lipid marker for the in vitro experiments, whereas the nonexchangeable, nonmetabolizable marker,  14Carbon-cholesteryl hexadecyl ether (55.7 mCi *symbol* mmol sup -1) was selected for the guinea pig cutaneous wheal experiments. Both of these radiolabels were purchased from New England Nuclear (Ontario, Canada).
Preparation of Large Unilamellar Vesicles that Exhibit a pH Gradient
Mixtures of DOPC and cholesterol (55:45 molar ratio), including14Carbon-dipalmitoyl-phosphatidylcholine or14Carbon-cholesteryl hexadecyl ether, were weighed and dissolved in benzene:methanol (95:5 v/v) at approximately 100 mg *symbol* ml sup -1. The solution was lyophilized by rapidly freezing the sample in liquid nitrogen and transferring it to a vacuum pump at < 50 mTorr for 5 h, with a liquid nitrogen trap attached. Multilamellar liposomes (100 mM phospholipid) were prepared by hydration of the lyophilized lipid mixture with 300 mM citrate pH 4.0 or 150 mM NaCl, 50 mM N-(2-hydroxyethyl)-piperazine-N-2 ethane-sulfonic acid pH 7.4 (HBS). The resulting MLVs were subjected to five freeze-thaw cycles by freezing in liquid nitrogen for 5 min, followed by incubation at 37 degrees Celsius for 5 min. This procedure was used to maximize the liposome trapped volume. Large unilamellar vesicles were prepared by passing the frozen and thawed MLVs ten times through an Extruder (Lipex Biomembranes, Vancouver, British Columbia, Canada) at 37 degrees Celsius, using two stacked 600 nm polycarbonate filters (Nuclepore Corporation, Pleasanton CA).
A transmembrane pH gradient was formed by dialyzing (Spectra Por 2 dialysis tubing, molecular weight cutoff 12-14,000) LUVs formed in citrate buffer against 100 volumes of HBS at 4 degrees Celsius for 48 h with one change of external buffer after 24 h.
Determination of Trapped Volume
Large unilamellar vesicles were prepared as described earlier, except that the lyophilized lipid mixture was hydrated in 300 mM citrate, 1 mM sucrose pH 4.0 that contained14Carbon-sucrose (0.3 micro Ci *symbol* ml sup -1). After dialysis against HBS, as described earlier, the trapped volume was determined and expressed as micro liter *symbol* micro mole sup -1 DOPC.
Bupivacaine Uptake into Large Unilamellar Vesicles that Exhibit a pH Gradient
Bupivacaine uptake studies were performed using LUVs prepared with 100-nm pore-size polycarbonate filters. Bupivacaine was added to LUVs (5 mM phospholipid) to give the final drug concentrations shown in Figure 3, and the samples were then incubated at 37 degrees Celsius. At various times up to 2 h, 100-micro liter aliquots were removed and unencapsulated drug separated from the liposomes by the use of 1-ml minicolumns. The columns were centrifuged at 2,500 revolutions/min for 3 min, and the eluent then assayed for drug and phospholipid, as described later in the Analytical Procedures section. Controls were also run that consisted of liposomes that did not exhibit a pH gradient (pH 7.4 in and pH 7.4 out), to quantify nonspecific bupivacaine binding.
Equilibrium Bupivacaine Association with Large Unilamellar Vesicles Determined by Ultrafiltration
This technique was used to determine bupivacaine association with liposomes at high drug and phospholipid concentrations. Samples (200 micro liter) were placed in microcentrifuge filtration tubes, molecular weight cut off 30,000 (Costar Scientific, Cambridge, MA), and centrifuged for 1 h at 15,100 gavin a Sorval MC 12V microcentrifuge (DuPont, Newton, CT). The filtrate (15 micro liter) and original sample were then analyzed for bupivacaine by high-pressure liquid chromatography (HPLC). Control experiments confirmed that no liposomal lipid was present in the filtrate.
Release of Liposomal Bupivacaine In Vitro
These experiments were conducted at lipid and drug concentrations similar to those later used in the animal studies. Bupivacaine (26.6 mM) was incubated at 37 degrees Celsius for 1 h with DOPC:Cholesterol (100 mM phospholipid) LUVs that exhibited a pH gradient (pH 4.0 in/pH 7.4 out) or with similar LUVs that did not exhibit a proton gradient (pH 7.4 in/pH 7.4 out). The two samples were then placed in dialysis bags and dialyzed against 500 volumes of HBS, with buffer changes after 4 and 8 h. At various times up to 24 h, samples were removed from inside the dialysis bags and assayed for bupivacaine and phospholipid, as described below.
Clearance Rate of Large Unilamellar Vesicles after Intradermal Injection into Guinea Pigs
Clearance of liposomes from the site of injection was followed, using14Carbon-cholesteryl hexadecyl ether. At various time points, 100-micro liter intradermal injections of DOPC:Cholesterol LUVs (100 mM phospholipid) were made into the guinea pig back. A zero time point injection also was made immediately before killing the animal by carbon dioxide asphyxiation. The skin was then removed and injection sites dissected. To monitor the possibility of radioisotope diffusion between wheal sites, muscle tissue below and the dermal layers surrounding the wheals also were dissected. Tissue samples were weighed and digested in 4 ml Solvable (New England Nuclear, Boston, MA) overnight at 65 degrees Celsius. Samples were then cooled to 4 degrees Celsius, and two 250-micro liter aliquots of H2O2(35%) were added to decolorize, with the second peroxide aliquot added only after foaming subsided. The sample was then divided into four 1-ml aliquots, and 15 ml UltimaGold liquid scintillation cocktail (Packard, Meriden, CT) was added to each vial. Samples were vortexed and counted the following day on a Beckman LS 3801 liquid scintillation counter (Beckman Instruments, Mississauga, Ontario, Canada) after overnight storage in the dark.
All animal experiments were conducted in accordance with guidelines issued by the Canadian Council for Animal Care and were approved by the University of British Columbia Animal Care Committee.
Guinea Pig Cutaneous Wheal Model of Neural Blockade
The duration of nerve blockade produced by free and liposomal bupivacaine was compared using the cutaneous wheal model in Dunkin Hartley guinea pigs from the University of British Columbia Animal Care Facility. To reduce stress, the animals' backs were shaved the day before the experiment was conducted. Wheals were formed by injecting 100 micro liter of the sample into the intradermal layer of the guinea pig's back. As a control, a saline or empty liposome sample also was injected into each animal. Wheals were marked and pricked five times, at intervals, using a 18.5-gauge needle attached to a 5-ml plastic syringe. The syringe was placed inside a plastic sheath so that the needle was exposed at one end but could move freely up and down within the sheath. This system ensured that, when the needle was touched to the guinea pig's back, a constant pressure was applied (equal to the weight of the needle and syringe). After application of the needle to the wheal site, the observer (blinded to the samples injected) scored a positive response if the back muscles twitched. Interanimal variation was reduced by testing all four samples on the same animal in each experiment. Each animal was used three times, with a minimum 48-h interval between experiments. Wheals were never generated in the same location twice. A random number generator was used to assign the formulations to various regions on the back and to determine the order of testing each wheal. In addition, the person testing the response to probe stimuli was blind to the contents of the sample injected at each site.
Bupivacaine was quantified by HPLC, using mepivacaine as an internal standard. To the sample (300 micro liter) we added mepivacaine (20 micro liter, containing 83 *symbol* micro gram *symbol* ml sup -1 in water) followed by acetonitrile (300 micro liter). The mixture was then incubated at 60 degrees Celsius for 5 min, to ensure rupture of the liposomes and release of the entrapped drug. The sample was then centrifuged at 2,500 revolutions/min for 5 min, to pellet the lipid and the supernatant taken for HPLC analysis. Using a 10-cm octadecyl-silica column, bupivacaine and mepivacaine were eluted under isocratic conditions with 35% sodium diphosphate, pH 5.0; 65% acetonitrile containing 20 mM triethylamine at 1.0 ml min sup -1 and detected at 210 nm. All of the solvents used were of HPLC grade.
In some experiments, phospholipid concentrations were determined by phosphate assay after perchloric acid digestion. 
Vesicle size distributions were determined by quasi-elastic light scattering, using a Nicomp 370 submicron particle sizer (Nicomp Instuments, Goleta, CA), as described previously. 
The durations of nerve blockade produced by free bupivacaine, bupivacaine in control liposomes, and bupivacaine in pH gradient liposomes in the guinea pig cutaneous wheal model were compared using an analysis of variance model. Using a block design, these formulations were compared at dosages of 0.75% and 2.0% bupivacaine. Interanimal and day-to-day variance within groups also was evaluated, to confirm the validity of the model (squared multiple r values for 0.75% and 2.0% bupivacaine formulations were 0.762 and 0.961, respectively).
As shown in Table 1, when DOPC:Cholesterol MLVs are extruded through 600-nm pore-size filters, the resulting vesicles have a mean diameter of approximately 350 nm. Although within this population some of the vesicles will contain more than one bilayer, for simplicity, we refer to these as LUVs. In agreement with previous reports, we determined a trapped volume of 3.0 micro liter micro mole sup -1 phospholipid for these LUVs. Also shown in Table 1are the size distribution data for similar liposomes extruded through 100-nm pore filters.
In Figure 3, the rates and extent of bupivacaine uptake by LUVs that exhibit a pH gradient are shown for three different drug-to-lipid ratios. Bupivacaine uptake by vesicles incubated with 1 mM drug (drug-to-lipid ratio 0.2) is rapid and remains stable for the 2-h timecourse. This level of accumulation (150 micro moles bupivacaine/mmole phospholipid) represents 75% of the total local anesthetic and corresponds to an intravesicular bupivacaine concentration of approximately 100 mM and an internal:external concentration gradient of approximately 400:1. At higher drug-to-lipid ratios, rapid accumulation is also observed, but this uptake is less stable (Figure 3). As would be expected, uptake efficiency is dependent on the initial drug concentration. For 2 mM bupivacaine (molar ratio drug-to-lipid 0.4), maximum uptake, seen at 5 min, represents approximately 58% encapsulation, whereas for 3 mM drug (molar ratio drug-to-lipid 0.6), maximum uptake corresponds to approximately 43% encapsulation. To confirm that bupivacaine uptake reflects drug redistribution in response to the imposed pH gradient, control LUVs that did not exhibit a pH gradient (pH 7.4 in and pH 7.4 out) were incubated with 3.0 mM bupivacaine under similar conditions. As shown in Figure 3, these control vesicles exhibited only low levels of drug association.
In the uptake study described earlier, drug-loaded liposomes were separated from free bupivacaine using size exclusion chromatography on 1-ml minicolumns. During passage down the column, it is possible that some initially entrapped bupivacaine would have been lost from the liposomes. In addition, for technical reasons, this kinetic study was undertaken at relatively low drug and lipid concentrations. To determine bupivacaine entrapment under equilibrium conditions and at the high bupivacaine and phospholipid concentrations used for the in vivo studies, an ultrafiltration technique was used. The concentrations of free and liposomally entrapped bupivacaine determined using this technique for samples that contained either 0.75% or 2.0% local anesthetic are shown in Table 2. These data are in good agreement with the results obtained using size exclusion chromatography. In the case of liposomal samples that contained 0.75% bupivacaine, approximately 82% of the drug is encapsulated, whereas at 2.0% bupivacaine, the trapping efficiency is approximately 64%.
We next examined the rate of release of bupivacaine from liposomal systems with and without a pH gradient. As shown in Figure 4, bupivacaine efflux from the sample containing control (no pH gradient) vesicles appears to follow essentially exponential kinetics, with less than 50% of the initial drug concentration remaining at 5 h and only approximately 10% at 24 h. As indicated by Figure 3, bupivacaine will not be entrapped within vesicles that do not exhibit a pH gradient. This control, therefore, is designed to ensure that any reduction in bupivacaine efflux rate seen for pH gradient liposomes result from drug encapsulation and is not simply a consequence of nonspecific hydrophobic binding to the liposome surface. Release of local anesthetic from LUVs that exhibit a pH gradient appears to follow two distinct phases. There is an initial, fairly rapid drug efflux for the first hour, during which approximately 20% of the bupivacaine is lost. This observation is consistent with the uptake data shown in Table 2, which indicate that approximately 18% of the drug is not encapsulated. The subsequent rate of drug release, however, is very much slower, such that even at 24 h, more than 60% of the initial concentration remains within the dialysis bag (Figure 4). It should be stressed that we are not implying that the release rates observed in this in vitro experiment will correspond to those occurring in vivo. Protein binding, for example, will tend to shift the equilibrium in favor of bupivacaine release in vivo. This study is simply intended to illustrate the influence of a transmembrane pH gradient on bupivacaine efflux kinetics.
An important consideration in the development of a sustained release system for local anesthetics is that the carrier should remain at the site of administration for an extended period. Clearly, if the liposomes used in the current study are cleared rapidly from the injection site, they cannot provide extended nerve blockade. We observed that greater than 85% of the liposome dose remained at the wheal site as long as 48 h after intradermal injection (results not shown).
We next compared the duration of nerve blockade produced by free bupivacaine, bupivacaine with pH gradient liposomes, or bupivacaine with control liposomes. It must be emphasized that, in these experiments, the same total amount of local anesthetic was administered for each formulation. The duration of nerve blockade produced by 0.75% bupivacaine is shown in Figure 5. As would be expected, a fairly rapid onset of local anesthetic action was seen for all three samples, although the block produced by bupivacaine with pH gradient liposomes was not complete at the first timepoint, 15 min. In the case of free drug, tactile sensation began to recover after 1 h, with half maximal response (R2.5) seen at approximately 2 h. When bupivacaine was administered with control LUVs (no pH gradient), the duration of action was longer than for free drug, with half maximal response not reached until approximately 3.5 h. Administration of local anesthetic entrapped within LUVs that exhibited a pH gradient resulted in a further increase in the duration of drug action (Figure 5). For this sample, recovery of tactile sensation began at approximately 4 h, with the R2.5 value only being attained by approximately 6.5 h. This represents a greater than threefold increase in the duration of nerve block compared with free drug.
When these three formulations were compared at 2% bupivacaine, similar trends were seen, but with a relative shift in R sub 2.5 values to longer times (Figure 6). Both free drug and bupivacaine with control liposomes exhibited similar recoveries from nerve blockade, with R2.5 values of approximately 4-4.5 h. Again, bupivacaine administered in liposomes that exhibited a pH gradient showed a much longer duration of action, with the half maximal response occurring approximately 11 h after injection. The delayed onset of nerve block seen for pH gradient liposomes at 0.75% bupivacaine was not observed with 2.0% drug (Figure 6). As shown in Table 3, when the duration of nerve blockade produced by the three local anesthetic formulations were compared using an analysis of variance model, free drug and bupivacaine in control liposomes were not statistically different. In contrast, the differences in duration of nerve blockade between these two formulations and bupivacaine in pH gradient liposomes were highly significant at both 0.75% and 2.0% local anesthetic (Table 3).
In the development of a sustained release vehicle for local anesthetics, three important objectives can be identified. First, high drug concentrations must be efficiently encapsulated within the carrier to provide a depot sufficient to maintain therapeutic concentrations of free local anesthetic at the site of nerve block over a prolonged period of time. Second, clearance of the carrier from the site of administration should be slow, so that the full benefit of local drug release is realized. Third, a mechanism allowing control over the rate of local anesthetic release is required to ensure that the duration of drug action can be maximized. As discussed later, the application of a transmembrane pH gradient to encapsulate bupivacaine within large unilamellar vesicles provides a system that meets these three objectives.
We showed previously that a large number of drugs that are lipophilic amines can be rapidly and efficiently encapsulated into liposomes, using a pH gradient. This technique is termed "remote loading," because drug uptake occurs into preformed liposomes and can be performed immediately before clinical administration. In the case of LUVs (100 nm diameter) incubated with bupivacaine at a drug-to-lipid ratio of 0.2, the level of drug uptake observed corresponds to a concentration gradient (inside/outside) of approximately 400:1, based on a trapped volume of 1.5 micro liter/micro mole phospholipid. Given that this accumulation will result in a decrease in the pH gradient from an initial value of 3.4 pH units (pH 4.0 in/pH 7.4 out) to a residual value of between 2.4 to 2.8 pH units, [24,31]the bupivacaine concentration gradient (2.6 log units) is essentially that predicted by the Hendersen-Hasselbach equation. An indication of the efficiency of the remote loading technique is provided by the fact that even at a higher bupivacaine concentration than is used clinically, 2%, greater than 60% of the total drug present can be liposomally encapsulated (100 mM phospholipid).
In addition to showing that high concentrations of bupivacaine can be efficiently loaded into large unilamellar vesicles exhibiting a pH gradient, we also confirmed that these carriers exhibit extended residency times after intradermal injection. Previous reports indicated that the rate of liposome clearance after subcutaneous administration is critically dependent on vesicle size. [16,17]Allen and coworkers, for example, found that, whereas small liposomes (80-90 nm) drain into the blood via the lymphatic system after subcutaneous injection, vesicles of diameter greater than approximately 120 nm did not appear in the blood to any significant extent. Liposomes of mean diameter 350 nm were used in the present research and, consistent with these earlier reports, we observed only low levels of clearance for 48 h after intradermal injection. Given that these LUVs would be expected to show an approximately Gaussian size distribution, it is possible that the lipid loss observed (primarily during the first few hours) reflects clearance of smaller vesicles within the population. Clearly, however, the requirement for extended residency at the site of administration is met by the present liposome formulation.
The importance of providing a mechanism whereby local anesthetic release from the liposome depot can be controlled is also well illustrated by the current research. As discussed earlier, researchers in previous studies using liposomal bupivacaine relied on passive entrapment of the drug or hydrophobic association with the phospholipid bilayer. [13,18-20]These systems are similar to the control (no pH gradient) liposomes used in the current work. In agreement with these earlier studies, we observed a longer duration of action for 0.75% bupivacaine when administered with control liposomes compared with free drug alone. When bupivacaine release rates are controlled using a pH gradient, however, much greater enhancement of analgesic duration is achieved. In contrast to control liposomes, which increase the time of nerve block by less than twofold, pH gradient liposomes exhibit a greater than threefold extension in duration. In addition, at higher bupivacaine doses (2%), there is no significant difference in the time course of neural blockade between free drug and bupivacaine with control liposomes. This difference in behavior compared with that seen at 0.75% bupivacaine may result from the lower ratio of hydrophobic binding sites to local anesthetic at this higher dose (the lipid concentration is maintained constant). In contrast to control liposomes, systems that exhibited a pH gradient retained the ability to extend the duration of action of 2% bupivacaine. Again, almost a threefold increase in duration of neural blockade was observed relative to the free drug.
In conclusion, large unilamellar vesicles that exhibit a pH gradient can be used to greatly extend the duration of action of the local anesthetic bupivacaine.