Local Myotoxicity from Sustained Release of Bupivacaine from Microparticles

Animals were cared for in accordance with protocols approved by the Animal Care and Use Committee at the Massachusetts Institute of Technology (Cambridge, Massachusetts), and the Guide for the Care and Use of Laboratory Animals  of the US National Research Council.

1,2-Dipalmitoyl-sn -glycero-3-phosphocholine was obtained from Avanti Polar Lipids (Alabaster, AL). Bupivacaine hydrochloride, bovine serum albumin, and α-lactose monohydrate were from Sigma-Aldrich (St. Louis, MO). Poly(lactic-co-glycolic) acid (PLGA; molecular weight 90 kd, G:L = 65:35) was obtained from Medisorb (Alkermes; Cambridge, MA).

PLGA microparticles were prepared as described.9In brief, PLGA dissolved in methylene chloride was homogenized at 3,000 rpm (Silverson L4RT-A; Longmeadow, MA) for 1 min in 1% poly(vinylalcohol) (molecular weight 25 kd; Polysciences, Inc., Warrington, PA) containing 100 mm Tris pH 8.5. The methylene chloride was removed by rotary evaporation. Particles of the desired size range were separated by wet sieving and were then lyophilized to dryness. In particles containing drug, bupivacaine free base was substituted for 50% of the polymeric mass dissolved in methylene chloride.

Blank 1,2-dipalmitoyl-sn -glycero-3-phosphocholine-albumin-lactose (DAL) particles were prepared as described9using a Büchi 190 Mini Spray Dryer (Büchi Co., Flawil, Switzerland). In brief, 1,2-dipalmitoyl-sn -glycero-3-phosphocholine dissolved in 100% ethanol was mixed with an aqueous solution of bovine serum albumin and α-lactose monohydrate, such that the final solute composition was 60% 1,2-dipalmitoyl-sn -glycero-3-phosphocholine, 20% albumin, and 20% lactose in 70:30 (vol/vol) ethanol:water. This solution was spray dried under the following conditions: inlet temperature, 110°–115°C; air flow, 600 l/h; flow rate, 12 ml/min; and aspirator pressure, −20 mbar. The resulting outlet temperature was between 50° and 55°C. In particles containing drug, bupivacaine free base was added to the ethanolic solution, to an amount that equaled 10% of the total solute mass. The total mass of the other three solutes was reduced by an equal amount.

Particle size was determined with a Coulter multisizer (Coulter Electronics Ltd., Luton, United Kingdom).

Fifty milligrams of DAL particles or PLGA microspheres was suspended in 1 ml phosphate-buffered saline, pH 7.4, at 37°C and inserted into the lumen of a Spectra/Por 1.1 Biotech Dispodialyzer (Spectrum Laboratories, Rancho Dominguez, CA) with an 8,000 molecular weight cutoff. The dialysis bag was placed into a test tube with 12 ml phosphate-buffered saline and incubated at 37°C on a tilt table (Ames Aliquot Mixer; Miles Laboratories Inc., Elkhart, IN). At predetermined intervals, the dialysis bag was transferred to a test tube with fresh phosphate-buffered saline. The bupivacaine concentration in the dialysate was quantitated by measuring absorbance at 272 nm and referring to a standard curve. Observation of the entire spectrum and performance of a protein assay (BCA Protein Assay Reagent Kit; Pierce Chemical Co., Rockford, IL) confirmed the absence of albumin from the samples that were measured. Infinite sink conditions were maintained during the release experiments as evidenced by low concentrations of released drug (<0.17 mg/ml).

Adult male Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) weighing 310–420 g were housed in groups, in a 6 am–6 pm light–dark cycle. Under brief isoflurane–oxygen anesthesia, a 23-gauge needle was introduced posteromedial to the greater trochanter, the needle was withdrawn approximately 1 mm, and 0.3 ml drug-containing solution was injected.

Thermal nociception was assessed by a modified hot plate test.1,21Hind paws were exposed in sequence (left, then right) to a 56°C hot plate (model 39D Hot Plate Analgesia Meter; IITC Inc., Woodland Hills, CA). The time (latency) until paw withdrawal was measured by a stopwatch. (Thermal latency in the uninjected leg was a control for systemic effects of the injected agents.) If the animal did not remove its paw from the hot plate within 12 s, it was removed by the experimenter to avoid injury to the animal or the development of hyperalgesia. The experimenter was blinded as to what treatment specific rats were receiving.

The duration of thermal nociceptive block was calculated as the time required for thermal latency to return to a value of 7 s from a higher value. Seven seconds is the midpoint between a baseline thermal latency of approximately 2 s in adult rats and a maximal latency of 12 s.

The sciatic nerve and adjacent tissues were harvested after euthanasia with carbon dioxide and processed to produce hematoxylin and eosin–stained slides.17In all analyses, the pathologist (R.P.) was not aware of the nature of the samples before examination. Samples were scored for myotoxicity (0–6).22The myotoxicity score reflected two separate but related processes that are hallmarks of local anesthetic myotoxicity: nuclear internalization and regeneration. The former is characterized by normal size and cytoplasm chromicity, but with nuclei located away from their normal location at the periphery of the cell. In the latter case, cells are shrunken, with more basophilic cytoplasm. Scoring was as follows: 0 = normal, 1 = perifascicular internalization, 2 = deep internalization (>5 cell layers), 3 = perifascicular regeneration, 4 = deep regeneration, 5 = hemifascicular regeneration, 6 = holofascicular regeneration.

C2C12 mouse myoblasts (American Type Culture Collection CRL-1772; Manassas, VA) were cultured to proliferate in Dulbecco's modified Eagle's medium supplemented with 20% fetal bovine serum and 1% penicillin streptomycin. All cell culture supplies were purchased from Invitrogen (Carlsbad, CA) unless otherwise noted. Cells were then plated in 24-well tissue culture plates with 50,000 cells/ml/well in Dulbecco's modified Eagle's medium supplemented with 2% horse serum and 1% pen-strep, and were left to differentiate into myotubules for 10–14 days. During differentiation media were exchanged every 2–3 days.

Bupivacaine hydrochloride was added to Dulbecco's modified Eagle's medium (with 2% horse serum and 1% Penn strep) at a concentration of 0.125% wt/vol. The medium was then filtered through a 0.22-μm cellulose acetate membrane and then serially diluted to prepare the remaining concentrations. Particles were irradiated with ultraviolet light for 2 h before suspension in media and serial dilution. The medium had a neutral pH at the outset; subsequently, pH was monitored by use of a pH-sensitive dye in the medium.

For time points longer than 4 days, the drug medium was exchanged every 2–3 days. Cells were maintained at 37°C in 5% CO²with the remainder being room air.

To quantitatively assess cell viability after adding drug- or particle-containing media, a colormetric assay (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide [MTT] kit, Promega G4100; Madison, WI) was performed at selected time points. The yellow tetrazolium salt is metabolized in live cells to form insoluble purple formazan crystals. The purple crystals are solubilized by the addition of a detergent. The color can then be quantified by spectrophotometric means. At each time point, 150 μl MTT was added, and then cells were incubated at 37°C for 4 h before 1 ml solubilization solution (detergent) was added. The absorbance was read at 570 nm using the SpectraMax 384 Plus fluorometer (Molecular Devices, Sunnyvale, CA) after being incubated at 37°C in the dark overnight. Cells were also monitored visually to confirm the results of the assay. Each plate had wells that contained medium without cells or other additives whose absorbance was subtracted from the rest of the plate as background. Each plate also had wells that contained medium and cells but no additives; all experimental groups were normalized to those wells.

Neurobehavioral data are presented as median, with 25th and 75th percentiles in parentheses, because they were not normally distributed. They were analyzed with the Mann–Whitney U test, or a Wilcoxon signed rank test when comparing sensory and motor tests in the same animals. Results of cell survival assays were described with parametric measures and tests (mean, SD, t  test, analysis of variance). A P  value less than 0.05 indicated statistical significance. Multiple comparisons were done in a planned manner (i.e. , comparisons were selected individually), and the P  value required for statistical significance (α) was determined by dividing 0.05 by the number of comparisons. Therefore, for myotoxicity scores (10 comparisons), α= 0.05/10 = 0.005, so P < 0.005 was required for statistical significance; for durations of block and cell culture data, there was only one comparison, so the α remained 0.05. For ease of understanding, the α for each comparison is provided with each result. All P  values are two-tailed. Statistical analyses were performed with SPSS 12.0 (SPSS Inc., Chicago, IL).

The DAL particles were produced as a fine white powder, with a median volume-weighted diameter of 4–5 μm. The PLGA particles were also a white powder, with a median volume-weighted diameter of approximately 60 μm. These values are in agreement with our previous reports for similar particles.9 

Animals were injected with 0.5% wt/vol bupivacaine hydrochloride, 10% wt/wt bupivacaine DAL particles, or 50% wt/wt bupivacaine PLGA microspheres (table 1). The sensory and motor blocks from encapsulated bupivacaine were longer than those from free bupivacaine. On gross dissection, tissues injected with free drug appeared normal. In the other groups, particles were found in discrete pockets adjacent to the sciatic nerve, but the tissue appeared otherwise normal. On histologic examination, there was evidence of perifascicular myotoxicity in all animals. The injury was limited to the site of injection, at the surface cell layers of the muscle adjacent to the depot of particles. It was considerably more pronounced in animals injected with encapsulated bupivacaine than with bupivacaine solution, both in extent (fig. 1) and in severity (figs. 1 and 2; P = 0.002 for comparison of 0.5% wt/vol bupivacaine with 50% wt/wt bupivacaine PLGA microspheres; P = 0.004 for comparison with 10% wt/wt bupivacaine DAL particles; both by Mann–Whitney U test with α= 0.005; sample sizes are given in fig. 2).

Many particulate formulations display a rapid initial release of drug referred to as “burst” release. However, comparison of the release kinetics between the particle types reveals that the peak rate of the burst is not the only or perhaps not even the main consideration in myotoxicity. Fifty milligrams bupivacaine-containing DAL particles or PLGA microspheres was suspended in 1 ml saline, placed in dialysis tubes, and the release of drug was measured as described in the Materials and Methods. The release from particles was compared with that from 1 ml bupivacaine hydrochloride, 0.5% wt/vol. Figure 3shows the rate of bupivacaine release from those samples per hour over time. Free 0.5% wt/vol bupivacaine caused a much higher peak level of drug than the other two formulations, yet 0.5% wt/vol bupivacaine caused little myotoxicity in vivo  (fig. 2) and as previously described,22,23whereas the encapsulated formulations do (figs. 1 and 2and Kohane et al.  17). Therefore, the magnitude of the peak level of bupivacaine released is not the sole determinant of myotoxic potential. In contrast, the encapsulated formulations caused severe myotoxicity despite lower peak release rates, suggesting that injury may be a product of a critical concentration applied over time.

This is not to suggest that the magnitude of the burst release is irrelevant. The total amount of drug in 0.5% wt/vol bupivacaine solution is considerably less than that contained in the particles. To demonstrate the potential effects of rapid release of particle contents, six animals were injected with an aqueous solution containing a quantity and volume of bupivacaine equal to the total contained in the DAL formulation, i.e. , 0.6 ml bupivacaine, 1.25% wt/vol. The duration of sensory block was longer than that from 0.5% wt/vol bupivacaine (table 1; P = 0.004 by Mann–Whitney U test; α= 0.05; sample sizes in table). On dissection, the tissues appeared grossly normal. However, severe myotoxicity was seen in all samples (n = 6) on microscopic examination, comparable to that seen with the drug-loaded DAL particles (P > 0.05 by Mann–Whitney U test; α= 0.005; sample sizes in fig. 2), and significantly higher than that seen with 0.5% wt/vol bupivacaine (P = 0.002 by Mann–Whitney U test; α= 0.005; sample sizes in fig. 2). We did not perform an analogous experiment for the PLGA microspheres because the amount of bupivacaine contained in 75 mg of those particles might cause severe systemic toxicity in a 350-g rat (100 mg/kg bupivacaine); the dose of bupivacaine that kills 50% of adult rats (LD⁵⁰) injected at the sciatic nerve is approximately 30 mg/kg.24 

Local anesthetic–induced myotoxicity generally recovers rapidly, often within 2 weeks. However, we have noted that some controlled-release formulations cause myotoxicity at least as far out as 1 month after injection.25This is at a time when bupivacaine release from such microparticles has ended, after approximately 3 weeks of release at a rate that is too low to produce nerve blockade.26One possible explanation of this observation is that local anesthetic myotoxicity is time dependent. To assess this hypothesis, we allowed cells from a myoblast cell line (C2C12 cells) to differentiate into myotubes and then exposed them to a range of concentrations of bupivacaine over a variety of time frames (2 and 6 h; 1, 2, and 4 days; 1, 2, and 3 weeks; fig. 4). Myotoxicity increased with the concentration of bupivacaine, but also markedly with duration of exposure. For example, 62 ± 12% of cells exposed to 0.025% wt/vol bupivacaine survived a 2-h exposure, whereas only 1 ± 2% survived at 3 weeks (p << 0.001 by Student t  test; α= 0.05; n = 12 per group). Concentrations of bupivacaine that would be expected to have little or no anesthetic efficacy and that showed minimal toxicity over short durations of exposure showed decreased viability over the course of days to weeks. For example, 88 ± 12% of cells exposed to 0.0025% bupivacaine survived at 2 h, whereas 52 ± 13% survived after 3 weeks (p << 0.001 by Student t  test; α= 0.05; n = 12 per group).

Blank (no bupivacaine) particles injected at the sciatic nerve caused relatively minimal myotoxicity when injected at the sciatic nerve (fig. 2). This lack of toxicity was also seen in vitro : C2C12 cells exposed to DAL and PLGA particles displayed either minimal or no decrease in cell viability compared with untreated cells over a period of 4 days, even at very high concentrations (fig. 5). (PLGA microspheres at 5,000 μg/ml produced a small but statistically significant decrease in cell viability.) To assess the possibility that the particles increase the toxicity of local anesthetics, C2C12 cells were incubated in local anesthetic solutions with or without particles. PLGA microspheres did not enhance the toxicity of various bupivacaine concentrations (fig. 6). Survival in cells exposed to 0.05% wt/vol bupivacaine for 1 day and 0.01% bupivacaine for 4 days was statistically significantly higher in the groups that received particles as well, although the difference was small. Data showing that DAL particles do not increase bupivacaine myotoxicity are not shown.

Nonetheless, it is possible that particles have an indirect role in the development of myotoxicity. Animals were injected with 38 mg blank (no bupivacaine) microparticles of both types which had been suspended in 0.3 ml bupivacaine, 0.5% wt/vol. On dissection, the tissue appeared mildly inflamed, and particles were located in discrete pockets in the vicinity of the nerve. Microscopy revealed inflammation and myotoxicity in areas adjacent to the pockets of particle. The median myotoxicity score was higher (fig. 2) in these groups than in those injected with 0.5% wt/vol bupivacaine alone, although this was not statistically significant by the stringent criteria used here (e.g. , for PLGA microparticles suspended in bupivacaine solution, P = 0.016 by Mann–Whitney U test; α= 0.005; sample sizes in fig. 2). We note, however, that when the results for both particle types suspended in bupivacaine were pooled and compared with 0.5% bupivacaine, the difference was statistically significant (P = 0.004 by Mann–Whitney U test; α= 0.005; n = 9 when combining the two particle-containing groups).

The initial burst release of bupivacaine seems to cause myotoxicity, but only if the magnitude or some product of magnitude and duration of exposure are above a certain undefined threshold. This contribution to local injury is theoretically amenable to engineering, i.e. , it should be possible to minimize that burst release. Of much greater concern is the observation that very low—even subanesthetic—concentrations of bupivacaine can become myotoxic over extended periods of time. This finding raises the possibility that myotoxicity could be an inevitable concomitant of long-term exposure to conventional (amino-amide and amino-ester) local anesthetics, irrespective of the technology used to deliver them. Myotoxicity is a well-known occurrence in clinical19or investigational27use of conventional local anesthetics. Although it can have severe consequences,19it has not generated much clinical concern. In fact, intramuscular local anesthetic injection is a standard treatment for trigger points in myofascial pain syndromes,28and local anesthetic myotoxicity is generally reversible. The distinction that must be made, however, is that those treatments generally involve a single-shot drug injection with a brief duration, whereas microparticulate systems can result in very high local concentrations and/or weeks of local anesthetic exposure. An even greater concern is that local anesthetics also have considerable local neurotoxicity29–31; the potential for controlled-release devices to injure nerves has not been examined extensively, but injury to muscle suggests that nerve injury might also be possible. Of note, animals that achieved sciatic nerve blocks lasting approximately 9 days after injection with microparticles containing tetrodotoxin, bupivacaine, and dexamethasone frequently had one or more cycles of block recurrence after the initial block wore off,25a pattern potentially attributable to nerve injury. We have also seen this pattern with high concentrations of tricyclic antidepressants used as local anesthetics,23which have been shown to be highly neurotoxic.23,32 

The effect of the particles themselves on myotoxicity is difficult to explain fully. It would seem from our results here and from previous experience that the particles themselves cause little direct myotoxicity. However, the fact that the myotoxicity of bupivacaine solution is increased in the presence of particles in vivo  suggests that they might enhance that toxicity. One possibility is that the particles release some agent (e.g. , lactic or glycolic acids, residual organic solvent, excipients) that potentiates local anesthetic toxicity, but our cell culture data do not support that conclusion. Another possibility is that the presence of discrete pockets of particles allows more reliable identification of sites where the local anesthetic was deposited, thus improving the accuracy of sampling. However, we do not see any sign of such severe toxicity in any animal injected with bupivacaine solution. Furthermore, we did not see such toxicity in an animal model specifically designed to remove sampling bias by injecting very large volumes of local anesthetic solutions (1.5 ml).23It is possible that the inflammation caused by the particles worsens myotoxicity by some unknown mechanism, perhaps by their proinflammatory effects.17,20,25Finally, the macroscopic deposits of particles—as opposed to the individual particles—may slow the decline of the local concentration of drug, thereby increasing the toxicity of bupivacaine solution. The merits of the last two possibilities cannot be evaluated by the methods used in this study. The inflammatory response to particles may prove to be problematic in its own right, irrespective of myotoxicity or neurotoxicity, given the large mass that may have to be injected to achieve clinically relevant nerve blocks in humans.

Although we cannot rule out the possibility that residual organic solvents from the particle production process contributed to the observed myotoxicity, it is unlikely that they play a major role. Particles of both types do not cause myotoxicity in the absence of local anesthetics.17Furthermore, vehicles that do not involve organic solvents (e.g. , cross-linked hyaluronic acid) only cause myotoxicity when they contain local anesthetics.8 

Myotoxicity seems to be related to both the release kinetics of bupivacaine (burst and duration of release), and perhaps the presence of the particles themselves. Even very low concentrations of bupivacaine seem to be myotoxic if the duration of exposure is sufficiently prolonged. One possible implication of these findings is that any type of prolonged duration local anesthesia using drugs of this type will be myotoxic, and potentially neurotoxic.