To investigate the mechanism by which rare cases of spinal local anesthetic (LA) neurotoxicity occur, we have tested the hypotheses that LAs elevate cytoplasmic calcium (Ca2+(cyt)), that this is associated with a neurotoxic effect, and that lidocaine and bupivacaine differ in their neurotoxicity.
Neurons of the ND7 cell culture line, derived from dorsal root ganglion, were loaded with fura-2 and analyzed by digitized video fluorescence microscopy during 60 min LA exposure, allowing determination of Ca2+(cyt) and time of necrotic cell death (plasma membrane lysis) at the single neuron level.
Lidocaine 0.1% and bupivacaine 0.025% caused minimal changes in Ca. Lidocaine 0.5-5% and bupivacaine 0.125-0.625% caused an early, small (less than threefold), concentration-dependent increase in Ca2+(cyt) that was transient and returned to near baseline within 10 min. Lidocaine 2.5% and 5% then caused a sustained, greater than ten-fold increase in Ca2+(cyt) and death in some neurons during the 60 min exposure period. Pretreatment with thapsigargin eliminated the initial transient increase in Ca2+(cyt), consistent with endoplasmic reticulum (ER) as its source, and increased neuronal death with 5% lidocaine, suggesting that lidocaine neurotoxicity can be increased by failure of ER to take up elevated Ca2+(cyt). The later sustained increase in Ca2+(cyt) seen with 2.5 and 5% lidocaine was prevented in Ca2+ -free medium, and restored when Ca2+ was added back to the buffer in the presence of lidocaine, suggesting that higher concentrations of lidocaine increase influx of Ca2+ through the plasma membrane.
In this model, lidocaine greater than 2.5% elevates Ca2+(cyt) to toxic levels. Bupivacaine and lower concentrations of lidocaine transiently alter Ca2+(cyt) homeostasis for several minutes, but without an immediate neurotoxic effect within 60 min.
CLINICAL, 1,2,in vivo , 3and in vitro 4studies have documented the occurrence of local anesthetic (LA) spinal neurotoxicity, dependent on both concentration and exposure time, but have not determined the mechanism by which LAs cause neurotoxicity. It is difficult to postulate a mechanism of neural damage based on the primary pharmacologic effect of LAs, block of Na+channels. Blockade of Na+channels and resultant electrical inactivity should decrease neuronal metabolism and preserve ATP. Since export of Ca2+cytto the extracellular space is coupled to influx of Na+through the Na+-Ca2+exchanger, block of Na+channels and maintenance of a low cytosolic Na+should act to prevent elevation of Ca2+cyt5Furthermore, LAs not only block Na+channels, but also block Ca2+channels at higher concentrations. 6LAs (at concentrations near therapeutic plasma levels, but much lower than CSF concentrations during spinal anesthesia) limit the increase in Ca2+cytcaused by agents which cause an influx of Ca2+through the plasma membrane, in myocardium, 7airway smooth muscle, 8and secretory cells. 9Blockade of Na+influx is protective during neuronal anoxia. 10LAs are neurotoxic and the chemically dissimilar tetrodotoxin is not when both are given as spinal anesthetics to rats at concentrations that produce similar extents of Na+block. 3
It is therefore most likely that a neurotoxic effect of LAs is mediated by effects other than Na+channel blockade. One potential mechanism is prolonged elevation of Ca2+cyt, in contrast to the physiologic, fleeting elevation of Ca2+cyt, which occurs during response to neurotransmitters. 5,11–13Previous studies have suggested a possible detrimental effect of LAs on Ca2+release from nonmitochondrial intracellular stores. 14–17Those studies were limited by being performed on subcellular fragments, rather than on an intact cell with a native intracellular environment. In a glial cell line, lidocaine caused an increase in Ca2+cytin cells surrounded by Ca2+-free buffer, implying release from an intracellular store. 18In adult dorsal root ganglion neurons, a 30 s pulse of lidocaine transiently increased Ca2+cyt, deriving from both intracellular and extracellular Ca2+stores. 19
In the studies reported here, we tested the hypotheses that LAs alter neuronal Ca2+homeostasis and cause a sustained increase in Ca2+cyt, that this is associated with a neurotoxic effect, and that lidocaine and bupivacaine differ in their neurotoxicity. We have utilized a cell culture line of healthy sensory neurons to allow assays at the single neuron level, to minimize effects of preparative trauma on Ca2+cytand neuronal injury, and to eliminate vascular and other systemic effects of LA. In this model, we have determined the response of neuronal Ca2+cytto lidocaine and bupivacaine during a 60 min exposure to clinically relevant concentrations, ranging from that expected with maldistribution and minimal mixing of high concentrations with CSF, to that expected with complete mixing of lower concentrations. We have also determined the incidence of neuronal death during the 60 min exposure.
Materials and Methods
Chemicals and Buffers
All drugs and other chemicals were obtained from Sigma-Aldrich (St. Louis, Missouri), except where specifically noted, and were of the highest purity available. LAs were obtained as their hydrochloride salts, dissolved in buffer, and adjusted to pH 7.4 prior to use. Expressions of percent LA (g/dl) were calculated from the hydrochloride salt weight, consistent with standard clinical usage. Experimental buffer was HEPES-buffered KRH: 5 mm d-glucose; 25 mm HEPES; 115 mm NaCl; 5 mm KCl; 1.2 mm MgSO4; 1.0 mm KH2PO4; 2.0 mm CaCl2; +NaOH to pH 7.4. 13
All neuronal studies were conducted using the ND7–104 subclone of the ND7 cell line, obtained from Patrick G. Hogan, Ph.D., Investigator, The Center for Blood Research, Boston, Massachusetts. ND7 was derived from rat dorsal root ganglion, immortalized by fusion with mouse neuroblastoma, and has been extensively characterized as a sensory neuron model. 20–23
Neuronal cultures were started with an aliquot grown from the original ND7–104 subclone stock, and used for the lesser of 2 months or 8 passages. Proliferation medium for routine cell growth was L-15 (Gibco BRL Life Technologies, Grand Island, New York) supplemented with 3.3 g/l NaHCO3, 3.3 g/l d-glucose, and 10% fetal calf serum (FCS; Hyclone; Logan, Utah), in T75 flasks. Neuronal suspensions were prepared by gentle trypsinization (0.05% trypsin + 0.53 mm EDTA) for 3 min, followed by FCS to inhibit further proteolysis, centrifugation at 228 × g for 5 min (Beckman GPR centrifuge; Fullerton, California), resuspension in proliferation medium, and plating on glass coverslips.
Round glass coverslips (Fisherbrand 25CIR-1; Fisher Scientific, Pittsburgh, Pennsylvania) were dipped in 75% ethanol and put with sterile forceps into Falcon tissue culture 35 × 10 mm dishes (BD Biosciences, Franklin Lakes, New Jersey), then exposed to ultraviolet light for 20 min. Poly-d-lysine hydrobromide (Sigma P7405, molecular weight > 300,000; Sigma-Aldrich, St. Louis, Missouri) 0.1 mg/ml dH2O was added, 1.5 ml per dish, and incubated 18 h at room temperature. Each dish was then washed twice with 2 ml sterile dH2O, then 1 ml proliferation medium added and the dish and coverslip incubated at 37°C for 30 min. Neurons were then added as a suspension to the coverslip in the tissue culture dish. Neurons were allowed to attach for 6 h, then the medium was changed to differentiation medium, which is identical to proliferation medium except that FCS was decreased to 0.5%, and 1 mm cyclic adenosine monophosphate (cAMP) and 2 ng/ml nerve growth factor (recombinant rat NGF-β) were added. This medium was changed daily, and the cells used 48–72 h after they were initially exposed to differentiation medium. Cells were incubated at 37°C in 6% CO2, remainder room air.
Digitized Video Fluorescence Microscopy
We have previously described 13,24,25the imaging system that was used: an Attofluor RatioVision system (Atto Instruments, Rockville, MD) using a Zeiss (Carl Zeiss MicroImaging, Inc., Thornwood, New York) Axiovert 35 M inverted microscope with a Zeiss 40×, 1.30 NA, oil, Plan, Neofluar lens, equipped with an ICCD camera and a temperature controlled stage. The vendor's software was used for defining regions of interest (i.e. , single cells), background subtraction, pixel by pixel ratioing and calibration, and gray value reporting of the unprocessed image to insure that the 8 bit dynamic range of the video camera was not exceeded. Neurons were completely shielded from the excitation light except during the fraction of a second when an image was acquired. All experiments were conducted at 37.0 ± 0.3°C.
Neurons were loaded with 5 μm fura-2 acetoxymethyl ester (Molecular Probes, Eugene, OR) in KRH + 10% fetal calf serum (FCS) for 20 min at 37°C. This yielded cytoplasmic fura-2: neurons displayed a diffuse, nonpunctate fluorescence throughout the neuron, and lost more than 95% of their fluorescence when exposed to 20 μm digitonin, which lyses plasma membrane but not organellar membranes. 26Excitation was at 334 nm, and 380 nm, with 510 nm dichroic, and 540 ± 25 nm emission. Calculation of Ca2+cytfrom fura-2 fluorescence ratios was performed by calibration with fura-2–free acid solutions containing no Ca2+(10 mm EGTA) and saturating Ca2+(2 mm), in 100 mm KCl, 10 mm NaCl, 10 mm MOPS, pH 7.2. 13Neuronal death was detected by the sudden loss of fura-2 fluorescence from the fluorescent image, indicating loss of plasma membrane integrity, with the lysed neuronal membrane remaining in position on phase contrast view. Dead neurons were excluded from Ca2+cytanalysis after the time of death. At the time of neuronal death, plasma membrane integrity is lost and there is no longer an effective barrier to the influx of Ca2+along its large concentration gradient from outside the neuron. Apparent Ca2+cytvalues at this time provide no indication about the role of Ca2+cytin leading to neuronal death, but are predictably high as a result of neuronal death.
Following loading with fura-2, any planned pretreatment, and establishment of a stable pre-LA baseline for microscopy experiments, LA was added to yield the final concentration indicated for a given experiment. Images were acquired at a fixed interval for the duration of the experiment, which was planned for 60 min for LA exposure without pretreatment. Control experiments with equimolar concentrations of Tris buffer in place of LA were also performed to control for osmotic and other effects. Tris [(tris-hydroxymethyl)aminomethane] is a commonly used laboratory buffer, with a substituted amine pKaof 8.1 (cf. lidocaine pKa= 7.9, bupivacaine pKa= 8.1).
Statistical Analysis of Changes in Ca2+cyt
Because such calibrated values of Ca2+cytare of highest accuracy in comparing successive values in the same neuron, rather than comparing absolute values between neurons, 27our experimental protocols were analyzed using the normalized percent change in Ca2+cytfrom the single neuron's baseline before comparing it with other neurons’ changes. Initial Ca2+cytwas 126 ± 13 (SD) nm; N = 967 neurons. Ca2+cytconcentration at each time point following LA addition for each neuron was normalized by dividing by the average Ca2+cytvalue for that neuron in the 5 min preceding addition of LA. These normalized values were averaged separately for the periods of 0–10 min, and 10–60 min (time periods based on initial inspection of data; see Results). Each value was weighted in inverse proportion to the number of neurons observed in a given experiment, so that each experiment had equal weight. Statistical analysis was then performed on the normalized, weighted 0–10 min, and 10–60 min, averages of each neuron.
For statistical comparisons between LAs, bupivacaine and lidocaine were compared at equipotent concentrations using a ratio of 1:4;e.g. , 0.25% bupivacaine was considered equipotent to, and the same concentration for statistical analysis as, 1% lidocaine. The equimolar Tris buffer controls for bupivacaine and lidocaine were also considered as separate LAs for statistical analysis, and compared with their corresponding equimolar LA concentration;e.g. , lidocaine 1% (37 mm) was compared with Tris buffer equimolar (37 mm) to lidocaine 1%.
To determine the effect of concentration, linear regression for each LA was performed separately for 0–10 min, and 10–60 min, and for −Thapsigargin (Tps) and +Tps. To determine the effect of LA, analysis of variance was performed at each LA concentration, using LA and Tps as independent variables, and 0–10 min and 10–60 min Ca2+cytaverages as repeated measures dependent variables. Post hoc comparisons within significant effects were performed using the Bonferroni correction.
Time of death data from all neurons under a given condition of LA and Tps were combined to produce a single Kaplan-Meier estimator for each condition, and then compared by the Kruskal-Wallis test. All statistical calculations were performed using the algorithms of Systat 7.01 (SPSS Inc., Chicago, Illinois).
Effect of LA on Neuronal Ca2+cyt
Figure 1illustrates representative single neuron Ca2+cytresponses to lidocaine exposures of 60 min. Lidocaine greater than 0.5% caused an initial peak in Ca2+cytto approximately twice baseline, which subsided within 10 min. Lidocaine 0.5 and 1.0% returned to near baseline values of Ca2+cytafter the initial peak. In contrast, lidocaine 2.5% and 5% caused a sustained increase in Ca2+cytto toxic levels, with some instances of neuronal death by plasma membrane lysis (necrosis) observed within the 60 min experimental period. Tris controls equimolar to the lidocaine and bupivacaine concentrations tested caused minimal changes in Ca2+cytand no neuronal death.
Figure 2illustrates representative single neuron Ca2+cytresponses to equipotent bupivacaine exposures of 60 min, using a potency ratio of 1:4. Bupivacaine 1.25% (equipotent to 5% lidocaine) could not be tested because it is not soluble at pH 7.4 at 37°C. Bupivacaine greater than 0.125% caused an initial Ca2+cytpeak similar to that seen with lidocaine, but not a large sustained increase. At the highest bupivacaine concentration that could be tested, 0.625% (19.3 mm), there was generally a slow increase in Ca2+cytas illustrated in figure 2, with a few neurons increasing Ca2+cytseveral-fold over 60 min. No neuronal death was observed for any concentration of bupivacaine during 60 min exposure.
Effect of LA on Neuronal Ca2+cyt
Qualitative inspection of individual neuronal tracings (figs. 1 and 2) indicated a biphasic response to local anesthetic: A first phase comprising a small increase in Ca2+cytand subsequent decline to near baseline, completed within the first 10 min, and a second phase comprising either a more prolonged, steady increase in Ca2+cytfor higher concentrations of lidocaine, or a stable maintenance of near baseline Ca2+cytfor the remainder of the 60 min experimental period. Hence, quantitative analysis of average Ca2+cytwas performed separately for the time periods of 0–10 min, and 10–60 min.
For 0–10 min, summarized in figure 3, lidocaine 0.1–5% and bupivacaine 0.125–0.625% increased Ca2+cytcompared with equimolar Tris controls. Lidocaine 1 and 2.5% caused a greater increase than equipotent bupivacaine. The increase with LA concentration was significant by linear regression for both lidocaine and bupivacaine.
For 10–60 min, summarized in figure 4, the effect of LA was quite different between low and high concentrations. Ca2+cytdid not differ from or was less than Tris controls for both lidocaine 1.0% and below and equipotent bupivacaine, and returned to baseline or lower levels. In contrast, lidocaine 2.5 and 5% and bupivacaine 0.625% showed a large increase in Ca2+cytduring the 10–60 min period, lidocaine 2.5% more than 0.625% bupivacaine. The increase in Ca2+cytwith 5% lidocaine was not greater during 10–60 min than during 0–10 min. However, this may reflect the limits of Ca2+cytmonitoring with fura-2. Because the Ca2+cytcalibration equation is ratiometric, and the Kdof fura-2 for Ca2+is approximately 220 nm, fura-2 begins to saturate and does not accurately report Ca2+cytvalues greater than 1,000 nm. 13
Origin of Increased Ca2+cytCaused by Lidocaine—Effect of Thapsigargin
To test the involvement of the endoplasmic reticulum (ER) in LA-induced Ca2+cytincrease, neurons were pretreated prior to LA addition with 100 nm thapsigargin (Tps), which releases and depletes ER Ca2+by inhibiting the ER's ATP-dependent Ca2+transport. 28Representative single neuron tracings are depicted in figure 5, while statistical comparisons of the Ca2+cytaverages for 0–10 min and 10–60 min after LA addition are depicted in figure 6. As expected, Tps pretreatment caused a transient increase in Ca2+cytwhich returned to a plateau level slightly higher than pre-Tps Ca2+cytas Ca2+cytreleased from the ER was transported out of the cell and sequestered in other organelles. Subsequent addition of LA equipotent to lidocaine 1 or 2.5% caused minimal increase in Ca2+cyt, consistent with the ER being the origin of the initial transient Ca2+cytpeak seen with these LA concentrations in the absence of Tps. In contrast, Tps pretreatment had no effect on the large increase in Ca2+cytcaused by 5% lidocaine for 0–10 min or for 10–60 min, differing from all the other lidocaine and bupivacaine concentrations tested (although drawing quantitative conclusions from such high Ca2+cytvalues is limited by the measurement range of fura-2, as described previously [“Effect of LA on Neuronal Ca2+cyt: Quantitative Analysis,” end of third paragraph]). It is likely that 5% lidocaine still elicited an early Ca2+release from the ER, since an early peak was often visible in the individual cell tracings (fig. 1), but that its magnitude was quantitatively insignificant compared to the small portion of the later Ca2+cytpeak which occurred prior to 10 min. Because the initial and later Ca2+cytpeaks frequently overlapped with 5% lidocaine, and the onset time of the later Ca2+cytpeak was variable, a greater differentiation between the two peaks was not feasible. A small minority (4/65) of neurons treated with 2.5% lidocaine had a Ca2+cytresponse similar to that with 5% lidocaine. These may represent rare neurons with preexisting injury or senescence, so that they are less able to deal with an increase of Ca2+cyt. They may also indicate that the dose–response curve for the Tps-independent increase in Ca2+cytwith lidocaine is fairly steep at 2.5%, with neurons at 2.5% lidocaine very close to responding as they would to 5% lidocaine, and needing only a small impetus from other random variables to push them to that point. Most neurons treated with 2.5% lidocaine had an intermediate response: a slow, delayed increase in Ca2+cytof lesser magnitude, doubling or tripling by 60 min, and were viable at 60 min. This suggests that higher concentrations of lidocaine (2.5 and 5%), unlike bupivacaine and the lower concentration of lidocaine tested (1%), increase Ca2+cytfrom a source other than the ER.
Origin of Increased Ca2+cytCaused by Lidocaine–Effect of Extracellular Calcium
To test the effect of higher dose LA on Ca2+influx from extracellular fluid, extracellular buffer was depleted of Ca2+, neurons exposed to LA, and extracellular buffer then replenished with Ca2+. Representative single neuron tracings are shown in figure 7, and quantitative summaries of all experiments in figure 8. In the absence of extracellular Ca2+, both lidocaine 2.5% and bupivacaine 0.625% caused a small increase in Ca2+cyt, indicating that both LAs release Ca2+from intracellular stores, consistent with the Tps experiments described previously. When Ca2+was added back to the extracellular buffer (at time 0 in the figure), there was no effect in neurons exposed to bupivacaine. For neurons exposed to lidocaine 2.5% and 5%, however, there was an immediate and sustained ten-fold increase in Ca2+cyt, suggesting that lidocaine greater than 2.5% causes a large increase in plasma membrane permeability to Ca2+.
Neuronal Death Caused by Local Anesthetic
The experiments described here are not optimized to study neuronal death, since they follow relatively few numbers of neurons, but to determine quantitative changes in Ca2+cytover time in each neuron. Nevertheless, it is essential to determine the time of death in each neuron in these experiments, because measurement of Ca2+cytis meaningful only when neurons have not undergone necrotic death and still have an intact plasma membrane. 13Survival curves are plotted in figure 9. Lidocaine 5% by itself caused significant neuronal necrosis within the 60 min protocol period, with a mean survival time of 42 min. Pretreatment with Tps increased neuronal death significantly, with a mean survival time of 15 min. Although neuronal death was observed with 2.5% lidocaine ± Tps, the effect of Tps was not statistically significant at 2.5% lidocaine, and the incidence of neuronal death with 2.5% lidocaine did not differ statistically from Tris controls. No death was observed within 60 min in any experiment with lidocaine less than 1%, bupivacaine less than 0.625%, or Tris control.
Our data establish that the highest concentration of lidocaine clinically available (5%) can cause necrotic cell death within 60 min in a neuronal cell culture model. Our data also show that clinical concentrations of lidocaine seen with well-mixed subarachnoid administration, less than 0.5%, 29do not cause necrotic neuronal death within 60 min nor major alterations in calcium homeostasis, consistent with the clinical experience that the vast majority of lidocaine spinal anesthetics do not cause lasting neural injury. Translation of any cell culture model to the clinical situation is always challenging, because the model does not mimic perfectly the clinical situation. One significant difference of our model from clinical practice is that we maintained a constant concentration of LA during the 60 min exposure period, whereas in vivo the CSF concentration would decrease with time as the LA mixed and diffused out of the CSF. We chose this method to maximize reproducibility with a constant LA concentration, and to model the clinical situation of poor LA mixing that appears to increase risk of neurotoxicity. However, our model may overstate the toxicity that would be seen with a given LA concentration that decreased over time.
A new finding reported here is that both lidocaine greater than 0.5% and bupivacaine greater than 0.125% cause an initial, short-lived (approximately 5 min), moderate increase in Ca.2+cytAt lower LA concentrations, Ca2+cytthen returns to normal levels, then decreases below control levels, and death is not observed within our 60 min experimental period. Whether this Ca2+cytincrease or the subsequent decrease causes any clinically significant effect is unknown. However, there are also known examples of major physiologic changes being triggered by a similar, single, short-lived increase in Ca2+cyt;e.g. , the metaphase-anaphase transition. 30Short periods of increased Ca2+cytsimilar to those seen in our experiments (several minutes) can also be associated with synaptic changes affecting neuronal memory and excitability. 31A speculation which is consistent with the observed occurrence of the clinical syndrome of Transient Neurologic Symptoms (TNS) with low concentrations of lidocaine,(≤0.5%) 32and the modulating effects of spinal Ca2+cyton pain processing, 33is that the initial short-lived increase in Ca2+cytmay initiate a period of increased electrical responsiveness by the neuron, causing hyperalgesia. However, we found no differences between bupivacaine and lidocaine in their effects on Ca2+cytat concentrations equipotent to 0.1% and 0.5% lidocaine, and only a small although statistically significant difference at 1% lidocaine, while there is a dramatic difference in clinical incidence of TNS between lidocaine and bupivacaine. 34
The source of the initial short-lived, modest increase in Ca2+cytwith lidocaine greater than 0.5% and bupivacaine greater than 0.125% appears to be the ER. The ER is a major regulator of Ca2+cyt, releasing Ca2+to the cytoplasm or sequestering it in response to multiple stimuli transduced through its IP3and ryanodine receptors. Our data are consistent with a large body of work in muscle showing release of Ca2+from the sarcoplasmic reticulum by LAs, at last partially modulated by the ryanodine receptor, although tissue differences may limit the applicability to neurons. 35,36The subsequent large, sustained increase in Ca2+cytseen with some neurons exposed to 2.5% lidocaine, and all neurons exposed to 5% lidocaine, does not originate in the ER, but represents influx of Ca2+cytfrom the outside of the neuron through the plasma membrane. Our data do not address whether this is an effect on existing ion channels, or a direct effect on the plasma membrane lipid bilayer, although both are plausible given the known interaction of LAs with multiple ion channels and the amphipathic character of LAs which may affect lipid bilayer permeability. 37
It may be of some concern, and requires further study, that both lidocaine and bupivacaine, at low as well as high concentrations, caused an initial release of Ca2+from the ER. Recent data have suggested that depletion of ER Ca2+stores is by itself a severe form of cellular stress, irrespective of Ca2+cytlevels. 38Although with our protocol we did not observe necrosis in neurons exposed to lower concentrations of local anesthetic, we cannot exclude the possibility that delayed neuronal death occurred later than 60 min at lower concentrations.
While our data do not yet establish a change in Ca2+cythomeostasis as the initial, proximate cause of LA neurotoxicity, our data do establish an important role for the later, sustained increase in Ca2+cytin the manifestation of LA neurotoxicity: Ca2+cythomeostasis is altered by LAs, the magnitude of alteration parallels the incidence of neuronal death, the incidence of neuronal death increases when the ER is rendered unable to sequester Ca2+prior to lidocaine treatment, and lidocaine and bupivacaine differ in their effects on Ca2+cyt. Sustained, high magnitude increases in Ca2+cytas seen with 2.5 and 5% lidocaine are generally associated with toxicity. 12,39It is probable that a longer exposure or observation time on larger numbers of cells would reveal significant neuronal death with 2.5% lidocaine as well as with 5%. This can better be determined with a different experimental approach (e.g. , flow cytometry) than the one chosen here (digitized video fluorescence microscopy), which allows continuous monitoring of a small number of cells.
There was a clear, five- to ten-fold difference between equipotent 2.5% lidocaine and 0.625% bupivacaine in terms of the later, sustained Ca2+cytresponse elicited by the LA, consistent with the greater clinical frequency of serious neural injury after lidocaine than bupivacaine. 2There was detectable neuronal death with 2.5% lidocaine and not with equipotent 0.625% bupivacaine in our experiments, although the different survival curves were not statistically different during our 60 min protocol. Our inability to compare 5% lidocaine with equipotent bupivacaine illustrates another factor that may be responsible for part or all of the apparent lesser toxicity of bupivacaine clinically: its decreased solubility compared with lidocaine, such that a bupivacaine preparation equipotent to 5% lidocaine is not available and would not be soluble at physiologic pH. Even commercially available 0.75% bupivacaine must be acidified to pH approximately 4 to stay in solution; we cannot consistently prepare solutions of 0.75% bupivacaine without precipitates forming at pH 7.4 at 37°C.
The effect of Tps on neuronal death suggests hypotheses about both the early and late LA-induced Ca2+cytpeaks. Tps pretreatment eliminated the initial, small, transient Ca2+cytpeak but did not decrease toxicity, suggesting that the early peak is not associated with acute toxicity within 60 min. This is consistent with our finding that the initial Ca2+cytpeak was seen with nontoxic concentrations of both lidocaine and bupivacaine. Tps pretreatment increased neuronal death with 5% lidocaine only, and 5% lidocaine was the only LA concentration tested where Tps pretreatment did not decrease Ca2+cyt(fig. 6), suggesting that the greater than ten-fold late, sustained elevation in Ca2+cytseen with 5% lidocaine is mechanistically involved in its neurotoxicity. Tps is quite specific as a tool to deplete ER Ca2+and determine whether a Ca2+cytpeak originates from the ER, as the early peak with LA does. However, the effect of Tps is not well localized as a cause of cytotoxicity. Tps has several effects on intracellular Ca2+homeostasis which by themselves, independent of Ca2+cyt, can increase toxicity: (1) Tps depletes the ER of Ca2+28,38; (2) By effectively preventing ER uptake of Ca2+cyt, Tps can increase the demand on other intracellular Ca2+cythandling systems, 40particularly (A) mitochondria, with consequent elevation in mitochondrial Ca2+and demands on mitochondrial energy stores, and (B) plasma membrane ion pumps, with increased demand for ATP for export of Ca2+cyt.
The late decrease in Ca2+cytafter the initial increase, seen with 0.5% and 1% lidocaine and equipotent bupivacaine, is consistent with the known Na+and Ca2+channel blocking properties of lidocaine, and with the protective effect of Na+blockade seen in some models of ischemic neuronal injury. 41,42However, a similar decrease was not seen with 0.1% lidocaine, which is still far above the ED50for Na+channel block, suggesting that factors other than Na+and Ca2+channel blockade are responsible.
Our data are generally consistent with those of Gold et al. , 19who used primary cultures of adult rat dorsal root ganglion neurons to assess the neurotoxicity of lidocaine. Their electrophysiologic studies showed that lidocaine greater than approximately 0.25% irreversibly depolarized neurons, consistent with a neurotoxic effect not mediated by Na+channel blockade. They also tested the effect of 30 s pulses of lidocaine on Ca2+cyt, and the toxicity of a 15 min exposure to lidocaine followed by a 60 min recovery without lidocaine. Lidocaine pulses caused an increase in Ca2+cytwith an ED50of 21 mm (approximately 0.5%) lidocaine. Prolonged monitoring of Ca2+cytbeyond the 30 s pulse was not done, nor was the effect of bupivacaine assessed. The amplitude of the Ca2+cytresponse to the lidocaine pulse was diminished, but not eliminated, with nominally Ca2+-free buffer, implicating both intracellular and extracellular Ca2+as sources of the lidocaine-induced increase in Ca2+cyt.
Gold et al. 19reported that both 30 mm (0.8%) and 100 mm (2.7%) lidocaine were more toxic than control after 15 min exposure and 60 min recovery, giving 22% and 32% neuronal death, respectively, which is a greater toxicity than we observed. There are two likely reasons for this discrepancy. First is the difference between neuronal cells used. Their primary cultures of acutely isolated DRG neurons were more recently traumatized by dissection and isolation than our continuous cell line. Roughly 7% of the neurons exposed to control by Gold et al. 19died, versus none of the neurons in our control experiments. Recent studies have documented that axotomy, an inevitable consequence of DRG excision and dissociation, rapidly alters key properties of these neurons. 43Also, continuous cell lines such as we used tend to be more resistant to many types of injury because they are neoplastic, although this introduces another problem in extrapolating to in vivo neurons. 44The differences between our toxicity data and that of Gold et al. 19are consistent with the differences between the neuronal cells used. Although extrapolation from cell culture to in vivo is always difficult, it is reasonable to suggest that the true susceptibility of spinal neurons to lidocaine toxicity in vivo may lie between that of our data and that of Gold et al. 19
Second, Gold et al. 19reported that significant numbers of neurons exposed to lidocaine lifted from the cover slips they were cultured on and were lost to analysis, making quantitation of cell death problematic. While they partially compensated for this by analyzing the number of dead neurons as a percent of remaining adherent neurons, it is unlikely that the population of neurons that lifted had the same characteristics as the neurons that remained adherent. We also assayed individual neurons under the microscope, but used glass coverslips coated with a substrate of high-density poly-d-lysine, which we have found to maintain adhesion of essentially all neurons for at least 60 min during exposure to lidocaine. We were unable to attain cultures stably adhesive during LA exposure using collagen or poly-dl-ornithine with or without laminin or other protein supplementation. This is in itself most likely an indication of LA toxicity, as rounding of cells and detachment from substratum attachment are typical, nonspecific indicators of acute cellular injury. 45,46
In conclusion, lidocaine is clearly neurotoxic and elevates Ca2+cytto toxic levels at clinically available concentrations which might be achieved with poor CSF mixing, in a neuronal cell culture model which eliminates vascular and other systemic effects. Lower concentrations of lidocaine and all concentrations of bupivacaine alter Ca2+cythomeostasis for several minutes, but without an immediate neurotoxic effect within 60 min. Both LAs initially release Ca2+from the ER, but only lidocaine 2.5% or 5% also causes a sustained influx of Ca2+through the plasma membrane.