Lidocaine may be useful in the treatment of neuropathic pain states. The authors hypothesized that lidocaine would reduce tactile allodynia observed in a rat nerve injury model. In an effort to determine the site of drug action, effects after intravenous, intrathecal, and regional administration were compared.
Rats underwent ligation of the left fifth and sixth lumbar spinal nerves. The 50% thresholds (g) for left hind paw withdrawal of awake rats to von Frey hairs were documented before, during, and after intravenous administration of lidocaine at programmed/documented pseudo-steady-state plasma concentrations, and correlated with measured plasma concentrations. Responses to lidocaine application intrathecally and regionally to the injured nerves were also recorded.
In rats with tactile allodynia, intravenous lidocaine yielded 66 +/- 11% of the maximal possible effect on thresholds (100% = normal threshold), versus -1.3 +/- 2.7% for saline infusion. Twenty-one days after lidocaine infusion, 30-40% of the maximal possible effect persisted. Threshold increases depended on plasma concentration, rather than quantity of drug administered: rats receiving 15 mg/kg with higher plasma concentrations (1.2 +/- 0.1 micrograms/ml) showed significant allodynia suppression throughout 7 days of follow-up, whereas rats receiving 15 mg/kg at a slower rate with lower plasma concentrations (0.6 +/- 0.1 microgram/ml) did not. The EC50 for acute allodynia suppression was 0.75 microgram/ml. No such allodynia suppression was seen after intrathecal or regional administration of lidocaine despite transient neural blockade.
Intravenous, but not intrathecal or regionally applied, lidocaine produces dose-dependent suppression of allodynia associated with nerve injury. The effects far outlast plasma concentrations of lidocaine. The mechanism of these prolonged effects is unknown.
Methods: Rats underwent ligation of the left fifth and sixth lumbar spinal nerves. The 50% thresholds (g) for left hind paw withdrawal of awake rats to von Frey hairs were documented before, during, and after intravenous administration of lidocaine at programmed/documented pseudo-steady-state plasma concentrations, and correlated with measured plasma concentrations. Responses to lidocaine application intrathecally and regionally to the injured nerves were also recorded.
Results: In rats with tactile allodynia, intravenous lidocaine yielded 66 plus/minus 11% of the maximal possible effect on thresholds (100% = normal threshold), versus -1.3 plus/minus 2.7% for saline infusion. Twenty-one days after lidocaine infusion, 30-40% of the maximal possible effect persisted. Threshold increases depended on plasma concentration, rather than quantity of drug administered: rats receiving 15 mg/kg with higher plasma concentrations (1.2 plus/minus 0.1 micro gram/ml) showed significant allodynia suppression throughout 7 days of follow-up, whereas rats receiving 15 mg/kg at a slower rate with lower plasma concentrations (0.6 plus/minus 0.1 micro gram/ml) did not. The EC sub 50 for acute allodynia suppression was 0.75 micro gram/ml. No such allodynia suppression was seen after intrathecal or regional administration of lidocaine despite transient neural blockade.
Conclusions: Intravenous, but not intrathecal or regionally applied, lidocaine produces dose-dependent suppression of allodynia associated with nerve injury. The effects far outlast plasma concentrations of lidocaine. The mechanism of these prolonged effects is unknown.
Key words: Anesthetics, local: lidocaine; nerve. Injury: peripheral neuropathy; rat; von Frey hairs. Pain, allodynia: mechanical; tactile. Pain, hyperalgesia: mechanical; tactile.
A number of carefully executed clinical studies have shown that systemically administered local anesthetics may have analgesic properties specific to pain states resulting from damage to nerve tissue. [1-7]Analgesia in such neuropathic pain states is attained in the absence of motor or sensory conduction blockade, and at doses without toxic effects. Because neuropathic pain is otherwise difficult to treat and is typically refractory to conventional analgesic agents, these studies have generated considerable interest.
Patients with painful diabetic neuropathy have been shown to benefit with reduced pain scores for several days from the intravenous administration of lidocaine, without alteration in thermal thresholds. [8,4]Patients with pain due to peripheral nerve injury likewise reported decreases in spontaneous pain, albeit of short duration, after receiving intravenous lidocaine. Orally administered available congeners of lidocaine also may be effective, as demonstrated by the analgesic effects of mexiletine in painful diabetic neuropathy and peripheral nerve injury, and tocainide in trigeminal neuralgia. 
After peripheral nerve injury, humans may report both thermal hyperalgesia and tactile allodynia (i.e., pain evoked by light touch or brushing of the skin). Importantly, human psychophysical studies have documented that the predominant evoked pain complaint in peripheral nerve injury sufferers is allodynia. [10,11]Kim and Chung have reported that after ligation of the L5 and L6 spinal nerves in the rat, reliable, long-lasting allodynia develops. In particular, animals so prepared display readily quantifiable paw withdrawal reactions to light mechanical stimuli, comparable in character to the clinical syndrome of tactile allodynia evoked by light touch. We employed this model to study the efficacy and site of action of systemically administered lidocaine in the neuropathic state.
Materials and Methods
All procedures followed protocols approved by the Institutional Animal Use and Care Committee of the University of California, San Diego. Male Harlan Sprague-Dawley rats (weighing 100-200 g) were maintained in a standard facility with a 12-h day (0600-1800) and 12-h night cycle. Water and food pellets were supplied ad libitum. After surgery, rats were housed two or three to a cage with corncob bedding. All surgeries were performed on separate occasions, allowing a minimum of 5 days between procedures for adequate recovery, except as specified later in the case of regional nerve catheter placement at the time of nerve ligation.
Rats were anesthetized with halothane, 1-3%, in a 34% oxygen/air mixture. The left fifth and sixth lumbar spinal nerves were exposed via a dorsal midline incision and ligated tightly with 6-0 silk suture according to the method described by Kim and Chung. Rats were studied between 5 and 60 days after nerve ligation.
Intrathecal PE-10 catheters were implanted under halothane/oxygen anesthesia. The catheters were 9 cm in intrathecal length and terminated near the lumbar enlargement. A rostral 3-cm extension of catheter was tunneled under the skin and externalized between the ears for injection. Rats with neurologic deficits were not used. Catheters were flushed with 10 micro liter preservative-free physiologic saline after insertion and after use for drug delivery.
Catheters made from PE-50 tubing were inserted into an external jugular vein under halothane/oxygen anesthesia. Similar catheters with the addition of a small bouton 1 cm from the intravascular tip, for suture retention, were inserted into a carotid artery, taking care to spare the adjacent vagus nerve. All catheters were tunneled subcutaneously to emerge at the base of the neck, flushed before and after each use with heparinized saline (10 units/ml), and capped when not in use.
Indwelling Nerve Blockade Catheters
To allow drug delivery at the nerve injury site in the six rats described later, catheters were fashioned from PE-10 gently bent into a U shape and fused to a PE-50 extension, as previously described. During creation of the nerve ligation, the PE-10 loop was tied down to the newly lesioned nerve bundle, with the catheter lumen directed proximally toward the dorsal root ganglion. Catheters were externalized through the wound closure to the subcutaneous level, secured to the fascia, then tunneled subcutaneously to exit at the neck. These catheters were periodically flushed with sterile normal saline to ensure patency.
A computer program (Stanpump) designed to rapidly achieve and maintain constant plasma drug concentrations (pseudo-steady-state concentrations) was used to drive a syringe infusion pump (Harvard Apparatus 22, South Natick, MA). A modification of Rowland's kinetics set using experimentally derived rat parameters (Chaplan, Bach, Shafer, and Yaksh: unpublished data) enabled prediction of plasma concentrations and stable maintenance of pseudo-steady-state concentrations in plasma (Figure 2).
Rats were placed in a plastic cage with an open wire mesh bottom and allowed to accommodate for approximately 15 min. A series of eight von Frey-type filaments with exponentially incremental stiffness (0.41, 0.70, 1.20, 2.00, 3.63, 5.50, 8.50, and 15.10 g; Stoelting, Wood Dale, IL) was employed to determine the 50% threshold for paw withdrawal to light mechanical stimuli, using a previously described up-and-down paradigm summarized later. Only the left hind paw was tested, because (1) the ability or willingness of the rat to bear weight on the left hind paw may have altered any results of testing applied to the "normal" right hind paw; and (2) other investigators have shown alterations in both left and right spinal cord dorsal horn neurochemical environments after unilateral lesions, calling into question the validity of the contralateral paw as a control. [17-20]Instead, parallel groups of rats were used as controls. In brief, von Frey hairs were gently applied to the mid-plantar paw, with sufficient force to cause slight filament buckling. A positive response was noted whenever the paw was sharply withdrawn, and a negative response was recorded if no paw withdrawal occurred. The presentation of filaments was always in sequential order of stimulus intensity. The direction of presentation, however, (ascending or descending intensity) was reversed each time the threshold of response was crossed, so as to first locate the approximate minimum stimulus necessary for paw withdrawal, and then refine the measurement by oscillating further around the threshold. Paw withdrawal thresholds were calculated by noting the stimulus level at which the first change in response occurred, indicating preliminary identification of the threshold range, collecting four additional responses, and mathematically interpolating the 50% response threshold using the method described by Dixon. In cases where response thresholds fell outside the range of detection, i.e., continuous positive or negative responses were observed to the limit of the range of stimuli, values of 0.25 g (maximally neuropathic) or 15.00 g (normal, based on our previous observations) were respectively assigned. Because the results thus computed do not yield a mathematical continuum (not all possible values can be generated), nonparametric statistical methods were used to compare such "raw" thresholds. For some comparisons, raw thresholds were converted to percent of maximum possible effect (%MPE), designating pretreatment paw withdrawal thresholds (baseline values) as 0% effect, and assigning a cutoff value of 100% effect to thresholds greater or equal to 15 g: therefore, %MPE values near 100 indicate normal mechanical thresholds (i.e., at or near 15 g), whereas values near 0 indicate allodynia. The following equation was used to compute %MPE: Equation 1.
Experimental Design: Effect of Lidocaine on Normal Rats
Normal (i.e., nonlesioned) rats prepared with both intrathecal and intravenous catheters (N = 12), were divided into two groups. The first group (N = 6) received an infusion of lidocaine designed to rapidly achieve a plasma concentration of 2.5 micro gram/ml, and maintain that plasma concentration for 30 min. The second group (N = 6) received the same volume of infusate containing only saline. Paw withdrawal thresholds were measured after 30 min of infusion and compared. These animals were not further used in the present studies.
Acute/Long-term Effect in Rats with Neuropathy: Lidocaine Versus Saline
Four weeks after nerve ligation, rats were assigned to groups for either lidocaine or saline infusion, using a random number table. The investigator assessing mechanical thresholds was blinded to the treatment. Lidocaine (Abbott, Abbott Park, IL, 2%, diluted in normal saline to 5 mg/ml; N = 6), or an identical volume of saline (N = 5), was continuously infused intravenously at an incremental rate to an endpoint of complete allodynia suppression, or a maximum infusion volume corresponding to a lidocaine dose of 30 mg/kg, over a period of approximately 60 min. At the conclusion of this acute phase of the experiment, intravascular catheters were trimmed and capped, and rats were returned to the animal facility. Testing was performed daily on days 1-14, and then on days 17, 19, and 21.
Acute Time Effect
To examine the relationship between plasma concentration and time, neuropathic rats with intrajugular catheters (N = 6) received lidocaine at rates calculated to maintain a pseudo-steady-state plasma concentration (plateau) close to 2 micro gram/ml, to a cumulative dose of 15 mg/kg. Paw withdrawal thresholds were tested 5, 10, 20, 30, and 40 min into the plateau phase of the infusion, after completion of the approximately 9-min rapid bolus infusion phase required to achieve the targeted concentration. A parallel group of noninjured rats with both intrajugular and intraarterial catheters received identical infusions (N = 4), and arterial blood samples were drawn at all time points corresponding to threshold measurements in the neuropathic rats.
High Versus Low Plasma Concentration Lidocaine
To examine the importance of plasma concentration achieved versus total dose of lidocaine administered, rats with both intrajugular and intraarterial catheters were randomly assigned to receive a dose of lidocaine in such a fashion as to achieve a plasma target concentration previously found to suppress mechanical allodynia as above (high target: group A), or the same total dose in mg/kg, but delivered in such a way that plasma concentrations remained at approximately half the demonstrated effective target concentration (low target: group B). Intravenous infusions of lidocaine designed to achieve and maintain plasma concentrations in the vicinity of 2 micro gram/ml were administered to group A to a cumulative dose of 15 mg/kg (n = 11). Group B received an infusion designed to maintain a plateau concentration of half the concentration in group A, administered over 85 min, for a cumulative dose again of 15 mg/kg (n = 9). Paw withdrawal thresholds were tested every 10 min, and arterial blood samples were drawn when allodynia suppression was achieved, or at the end of the infusion. At the conclusion of the experiment, rats were returned to the animal facility, and paw withdrawal thresholds were tested thereafter on posttreatment days 1, 5, and 7 (n = 8, group A, and n = 6, group B).
Spinal Cord. To determine whether lidocaine has an antiallodynic action when directly applied to the spinal cord, a dose of 500 micro gram lidocaine was administered intrathecally in a volume of 10 micro liter, followed by 10 micro liter saline flush (N = 6). This dose was chosen to grossly exceed any cerebrospinal fluid concentration of lidocaine that might be achieved by systemic infusion, and to produce obvious transient motor dysfunction as a marker of effective drug placement. Paw withdrawal thresholds were tested at baseline before dosing (0 min) and at time points beginning when motor function completely returned to normal after lidocaine instillation (30, 45, 60 min). The effects on mechanical thresholds, expressed as %MPE, were compared to baseline values.
Spinal Nerve. To explore the possibility of a site of action at the injured axon, rats prepared as described earlier with indwelling nerve block catheters were dosed with 1 mg lidocaine regionally applied to the nerve ligation site, in a volume of 50 micro liter (20 mg/ml; N = 6). Paw withdrawal thresholds were tested at 5, 10, 15, 30, and 60 min after drug treatment, and compared to pretreatment baseline values.
Samples were obtained by withdrawing and discarding approximately 0.3 ml of blood from the arterial cannulae, and then withdrawing samples of approximately 0.5 ml volume. These samples were centrifuged and the plasma supernatant was frozen at -20 degrees Celsius until analysis. Lidocaine was extracted from the thawed samples by solid-phase extraction chromatography. 
Gas Chromatography Instrumentation and Chromatographic Conditions
Lidocaine was quantitated by capillary gas chromatography with nitrogen-phosphorus detection. A Hewlett-Packard 5890 II gas chromatograph (Palo Alto, CA) was equipped with a methyl-silicone gum (HP-1) capillary column (25 M x 0.2 mm x 0.33 micro Meter), programmed with injector and detector temperatures of 265 degrees Celsius. Split injections (1.5:1) were performed with a Hewlett-Packard 7673A automatic sampler, and the chromatograms were recorded and analyzed with H-P Chemstation (MSDOS) software. The helium carrier gas flow rate was 0.9 ml/min (32 psi). Hydrogen and air flow rates in the detector were 3 and 120 ml/min, respectively. The oven temperature was programmed at 240 degrees Celsius for 1 min, raised over 1 min to a final temperature of 270 degrees Celsius, and held at 270 degrees Celsius for 4 min. Total run time was 5 min; lidocaine and bupivacaine (used as an internal standard) eluted at 2.4 and 4.0 min, respectively. The absolute limit of detection by this method was 0.05 ng lidocaine/micro liter serum. The interassay and intraassay coefficients of variation were 4.2% and 2.5%, respectively, for serum lidocaine concentrations between 0.1 and 1.0 ng/micro liter.
Results are expressed as mean plus/minus SE, or median (95% confidence intervals). Median paw withdrawal thresholds before and after infusion were compared using the Wilcoxon signed-rank test if paired and the Mann-Whitney U test if unpaired. Changes in %MPE, designating pretreatment paw withdrawal thresholds as 0% effect and threshold values of greater or equal to 15 g as 100% effect (see above), were compared using repeated measures analysis of variance. When P values were less than 0.05, we identified significant differences between pairs by contrast calculations (means comparisons). Nonparametric repeated-measures analyses were performed using the Friedman test. Lidocaine blood concentrations were compared using a two-tailed t test.
Effect of Lidocaine in Normal Rats
No effect of intravenous lidocaine infusion was seen in normal rats: no difference was seen between paw thresholds in the lidocaine and saline groups (15 plus/minus 0 g, both groups; data not shown). No differences in behavior were noted between groups during the infusions.
Acute/Long-term Effect: Lidocaine Versus Saline
Rats with neuropathy, infused with lidocaine, showed a clear suppression of allodynia. The median paw withdrawal threshold was 2.18 g (1.69-2.62 g) before infusion (baseline), whereas after infusion thresholds rose significantly to a median of 11.17 g (7.14-14.10; P = 0.028, Wilcoxon signed-rank). These postinfusion values corresponded to a %MPE of 65.8 plus/minus 10.8. The thresholds in rats receiving saline did not change significantly during the same time interval, with a baseline median of 2.20 g (1.08-2.61 g) and a postinfusion median of 1.56 g (0.00-2.58 g; P = 0.89, Wilcoxon signed rank), corresponding to a %MPE of -1.3 plus/minus 2.7. Comparison of treatment effects on mechanical thresholds (%MPE) over the subsequent 21 days revealed a sustained difference between the two groups (Figure 1), which was significantly different by repeated measures analysis of variance at P < 0.0002. No evidence of toxicity was observed in any rats either during the infusion or the follow-up period. Rats appeared well groomed and had smooth coats, were active and alert, without chromodacryorhinorrhea or other overt signs of stress, and did not autotomize. Though not systematically recorded, informal observation suggested that rats with normalized mechanical thresholds after lidocaine also displayed increased weight-bearing on the affected paw, with decreased guarding behaviors.
Acute Time Effect
Paw withdrawal was inhibited in a cumulative fashion by the continuous administration of intravenous lidocaine at a pseudo-steady-state plasma concentration (in the parallel group) of 1.34 plus/minus 0.07 micro gram/ml for a period of 40 min. Median paw withdrawal threshold increased gradually during the 40-min infusion period to a final value of 12.68 g (7.07-14.87 g), compared to a baseline value of 1.56 g (1.00-2.80 g). Figure 2shows both the stable plasma concentrations generated by the software-driven infusion and the steady increase in paw thresholds, expressed as %MPE, as a function of time. Although plasma concentrations achieved were slightly lower than the targeted levels, more importantly, concentrations were steady over time and correlated in a linear fashion with targeted values. Repeated measures analysis of variance yielded a P value of < 0.001 for %MPE during lidocaine infusion; using contrasts (means comparison), the 30- and 40-min time points were shown to be the points that significantly differed from baseline.
High Versus Low Plasma Concentration Lidocaine
Rats receiving the higher target concentration of lidocaine (group A) achieved allodynia suppression. The median baseline value for paw withdrawal threshold in this group was 1.56 g (1.19-2.19 g), which was significantly below that recorded at the end of the infusion period, 15.00 g (12.41-15.76 g; P = 0.0033, Wilcoxon signed-rank test). Rats receiving the lower target concentration (group B), receiving the same cumulative dose (15 mg/kg) administered over a longer period (85 min), did not achieve significant allodynia suppression (baseline thresholds of 2.80 g (1.64-3.15 g) versus postinfusion, 3.31 g (2.22-4.39 g; P = 0.067, Wilcoxon signed-rank test). Paw withdrawal thresholds in group A were significantly higher than in group B immediately after infusion (P = 0.002, Mann-Whitney). At the end of the infusion period, blood concentrations in the higher target infusion group (A) were 1.21 micro gram/ml plus/minus 0.09 and in the lower target infusion group (B), 0.61 micro gram/ml plus/minus 0.10 (P = 0.0007, two-tailed t test). Rats in group A continued to show significant allodynia suppression during the 7 days of follow-up (P = 0.0006, Friedman test). No suppression of allodynia appeared during the follow-up period for rats in group B (P = 0.67, Friedman test; Figure 3).
Effect Site: Intrathecal and Regional Nerve Application
Rats receiving 500 micro gram of intrathecal lidocaine exhibited transient motor block (flaccidity) of 10-15 min maximum duration affecting the hind quarters but not the forequarters. During this time paw withdrawal could not be assessed. Testing of paw withdrawal thresholds after resolution of motor block (as evidenced by normal stance, gait, and righting reflexes) revealed no differences from baseline at 30 and 60 min; a mean threshold elevation at 45 min representing a 13.7 plus/minus 4% MPE was statistically significant (P = 0.02, repeated-measures analysis of variance with contrasts). Application of lidocaine directly to the site of nerve injury by means of indwelling catheters caused conduction blockade of a transient nature, manifested as transient paw flaccidity with decreased withdrawal responses to pinch as well as to light tactile stimuli. After resolution of acute blockade, no further effect on allodynia was detectable with regional administration. Figure 4shows the mechanical thresholds after both intrathecal and regional lidocaine administration.
After determination of the delayed onset of allodynia blockade as demonstrated in Figure 2, all plasma lidocaine concentrations obtained after at least 30 min of intravenous lidocaine administration were correlated with simultaneously determined paw withdrawal thresholds. This analysis resulted in a sigmoidal dose-response curve. Logit analysis (log[y/(100 - y]) was performed to linearize the sigmoidal curve (Figure 5). The plasma lidocaine concentration corresponding to 50% suppression of allodynia was calculated as 0.75 micro gram/ml using this analysis.
The major findings of this study are that (1) intravenously administered lidocaine produces plasma-concentration-dependent reversal of tactile allodynia associated with nerve injury; (2) the reversal occurs at a dose that has no detectable effect on normal motor function; (3) the effects are unexpectedly persistent; and (4) the effects achieved by systemic administration are not reproduced by application directly to the spinal cord or the affected peripheral nerve. In this model, lidocaine did not suppress allodynia immediately: there was an approximate 30-min latency during pseudo-steady-state drug administration before significant effects were seen. Our results exclude the possibility that the delay in onset relates to slow distribution to either the cerebrospinal compartment or the site of nerve lesion. A plasma concentration-response relationship was evident for allodynia suppression as assessed immediately after infusion. A similar relationship emerged for long-term suppression of allodynia as well, because the group of rats with mean plasma concentrations of 1.2 micro gram/ml immediately after infusion continued to have persistently increased mechanical thresholds, whereas the group with mean concentrations of 0.6 micro gram/ml did not.
Mechanism of Action
Widespread Consequences of Peripheral Nerve Injury. After injury to a peripheral nerve, substantial functional alterations occur in both the peripheral and central nervous systems. Sustained, low-level ectopic spontaneous activity originates at the site of neuroma formation in large peripheral axons as well as in dorsal root ganglion cells. While the basis of this spontaneous electrical activity is not known, abnormalities of axonal sodium channel distribution have been described in association with demyelination after peripheral nerve injury [25,26]that may lead to spontaneous fiber activity. In addition, the appearance after nerve injury of an unusual type of "modified rapidly adapting" cutaneous mechanoreceptor has been identified, which although not spontaneously active, shows abnormally prolonged, weak, irregular discharges to light mechanical stimuli. Prominent increases are also seen in the evoked activity of dorsal horn neurons, which project supraspinally. [29,30]Similar sustained central activity is demonstrable after the application of N-methyl-D-aspartate type glutamate agonists, which leads not only to electrophysiologic facilitation of neuronal responses but also to the behavioral manifestation of tactile allodynia. Thus, continued afferent pathway activity is linked to behavioral states wherein modest stimuli may evoke pronounced responses. These changes in electrical activity are associated with peripheral and central changes including alterations in receptor expression [33,34]second messenger function, [35,36]neurotransmitter production, likely neuronal dropout, and possibly altered balance of inhibitory/excitatory neurotransmitters. The sum total of these mechanisms may provide a scenario whereby nerve injury leads to an anomalous pain state.
Effect of Lidocaine on Neuronal Activity. Owing to its well known properties of conduction blockade, lidocaine has been assayed in afferent systems primarily using electrophysiologic assessments. A number of investigations have examined the effects of lidocaine on evoked or spontaneous neural activity such as described earlier. A systematic examination in patients with painful diabetic neuropathy has suggested a spinal or supraspinal effect site because of suppression of the centrally organized nociceptive flexion response. In addition, considerable evidence from the preclinical literature supports a spinal cord or supraspinal site of action of intravenously administered lidocaine in facilitated pain states. [40-44]Lidocaine, with an octanol: water distribution coefficient of 110 at 36 degrees Celsius, pH 7.4, distributes promptly to central nervous system structures after systemic administration. The effects of a systemically delivered dose appear more potent in central than in peripheral nervous structures. Although peripheral terminals clearly respond to lidocaine, they appear to do so only at a relatively high concentration. A single study has derived in vitro dose-response curves for the suppressant effect of lidocaine on spontaneous activity in acutely injured peripheral terminals. The reported ED50of 5.7 micro gram/ml however, reflects drug in an artificial, protein-free system ; a substantially higher plasma concentration would in all likelihood be required for a comparable investigation in vivo, considering that lidocaine is extensively protein-bound in circulation. In whole animals, the ED50of intravenous lidocaine for discharge suppression in neuromata has been reported to be 6 mg/kg, whereas that for the dorsal root ganglion is lower at 1 mg/kg, a dose that also yields suppression of polysynaptic (spinal cord) sural nerve evoked afterdischarges. Dose-related suppression of neurons in Rexed lamina V to high threshold mechanical and noxious thermal stimuli is seen in decerebrate cats (plasma concentration = 3-10 micro gram/ml). Intravenous lidocaine (1-5 mg/kg) suppresses polysynaptic C-fiber evoked flexor responses to mustard oil and noxious heat, without evidence of conduction block at the peripheral terminal. Intravenous lidocaine (3-4 mg/kg) suppresses noxious-evoked activity in wide dynamic range neurons in the rat, and, in addition, selectively suppresses the increased wide dynamic range neuronal activity seen ipsilateral to chronic peripheral nerve injury. [42,43]To date, no studies have specifically examined the effects of systemically administered lidocaine on supraspinal structures or descending pathways in the context of hyperalgesia or increased evoked responses.
Effect Site of Lidocaine. These potent central effects strongly suggest that lidocaine can exert a central action. Although our local spinal delivery work demonstrates that the prolonged effects cannot be accounted for by a simple direct local action of lidocaine, the spinal cord should not be excluded from further consideration based solely on the current results. Our observations, including the delay in onset of the antiallodynic effect of intravenous lidocaine and the need for systemic delivery, raise the possibility, among other hypotheses, that biotransformation of the administered drug may be a prerequisite. Several metabolites with local anesthetic/antiarrhythmic activity, and half-lives as long as or longer than the parent compound, are produced in the liver, including monoethylglycinexylidide and glycinexylidide. It should be noted that the activity of intravenous lidocaine in a dorsal horn recording preparation exceeds that of iontophoresed lidocaine, suggesting that the species active in producing effects at the spinal cord level could in fact be a metabolite. 
We have considered the possibility that lidocaine might be acting via sympathetic nervous system blockade in this model, known to be responsive to sympathectomy. [51,52]The minimal and transient sympatholytic effects of systemically administered lidocaine recently reported in a rabbit mesentery model do not support this hypothesis, although we did not directly test sympathetic activity in this study.
In addition, we cannot exclude a supraspinal effect of lidocaine based on our results. To date, no investigations have addressed this issue. It is open to conjecture whether lidocaine might exert a direct brain effect resulting in analgesia, or influence descending modulatory pathways.
Persistence of the Allodynia Suppression. The persistence of the effect of brief systemic exposure to lidocaine on allodynia behavior in our study was particularly unexpected, and is not readily explainable according to currently understood mechanisms of action of local anesthetics. The plasma half-life of lidocaine is no more than 30 min in the rat, whereas the duration of allodynia suppression was clearly greater than several hundred half-lives. Long-lasting suppression of allodynia was reproduced in two separate experiments and thus was a reliable finding of this investigation. Though controversial, there is precedent in both the clinical and animal literature for prolonged consequences of lidocaine treatment. Controlled clinical trials [8,4,5]as well as anecdotal reports [55,56]have noted persistent effects. In recent work, we observed decreases in thermal hyperalgesia that persisted for 24 h after intravenous lidocaine treatment in the Bennett chronic nerve constriction (rat) model. The present study, however, is the first to show persistent effects of lidocaine treatment of this magnitude and duration in an animal model of neuropathic pain.
Clearance of lidocaine from a tissue site (as in a peripheral nerve block) depends on local tissue blood flow, and may be prolonged. Unexpectedly, however, we found that the effects after delivery into the systemic circulation far outlasted those of regional application. This finding argues against allodynia suppression occurring as an accompaniment to neurotoxicity, because greater local concentrations achieved by direct application of the drug to peripheral nerve and spinal cord had far less effect than exposures to lower concentrations by intravenous administration.
Prolonged effects of several weeks might appear to point to some form of persistence of the drug at an active tissue site. We attempted to address the possibility of a depot drug action by administering an identical total dose in mg/kg to groups of rats using two different plasma concentration profiles, subjecting one group to a shorter infusion with a higher plasma concentration and the second to a longer infusion attaining half the plasma concentration. The observation that persistent allodynia suppression was only seen in the group receiving the higher plasma concentration appears to argue against any hypothesis whereby a depot of accumulated drug might explain persistent drug action.
Reversible sodium channel blockade is the most often described major pharmacologic property of lidocaine. It is possible that the action of lidocaine may be different in excitable tissues after injury, in a way that remains to be elucidated. For example, after injury, the expression of different isoforms of both acetylcholine receptors and sodium channels at the neuromuscular junction [58,59]leads to a distinct pharmacology of the postinjury state. Functionally differing subpopulations of sodium channels have already been well described within the normal dorsal root ganglion. Channels associated with the smaller dorsal root ganglion cells, which likely give rise to small unmyelinated fibers (C-fibers), display the particular characteristic of use-dependent blockade with lidocaine. Although the role of C-fibers per se has been questioned in both the induction and maintenance of tactile allodynia, [61-63]it is speculated that a channel isoform, hypothetically expressed in afferent pathways after neuronal damage, could be subject to exaggerated use-dependent blockade by lidocaine.
We raise the additional hypothesis that lidocaine may be acting at a novel locus, one other than the sodium channel, perhaps by a mechanism of action that leads somehow to long-term "switching" of neuronal function. Several reports have suggested that lidocaine has second messenger blocking effects, [64-68]however, the concentrations employed in those studies far exceeded the physiologic range used in our study. Of note, a recent report delineates a mechanism whereby the metabotropic glutamate receptor functions as an "on-off" switch through the activation of protein kinases/phosphatases to regulate the inducibility of long-term potentiation in the hippocampus. Such regulation, with durable consequences for the facilitation/inhibition of postsynaptic impulses, may provide one example of a theoretical construct as to how an isolated pharmacologic intervention can exert a persistent effect.
In summary, we have described suppression of allodynia in a rat surgical neuropathy model during acute administration of lidocaine, accompanied by very prolonged continued effects. The site of effect and means of action remain to be elucidated. Further studies are necessary to clarify the mechanisms involved in this potentially very clinically useful effect.