There is a growing appreciation of the importance of the dorsal root ganglion (DRG) as a potential source of afferent discharge underlying neuropathic pain. This comes from experimental findings in animal models of neuropathy as well as from clinical observations, and includes electrophysiological recordings, provocative maneuvers targeting the DRG, diagnostic blocks, and neuromodulation.27,30,32,33,41,45,50 In patients with lumbar radicular low back pain, for example, tensioning the sciatic nerve by straight-leg lifting mechanically distorts the DRG, triggering radiating leg pain, and afferent discharge in corresponding nerves. Likewise, the DRG is implicated as a driver of pain in herpes zoster and postherpetic neuralgia, trigeminal neuralgia (TN), phantom limb pain, and complex regional pain syndrome.5,8,14,37–40,49
Ablative procedures aimed at the DRG and trigeminal ganglion are used frequently in the treatment of chronic pain, notably TN, chronic headaches, and occipital neuralgia. More recently, electrical counterstimulation directed at the DRG has been introduced as a therapeutic modality in a variety of chronic pain conditions.33 These procedures, however, have important limitations beyond their invasive nature. Resection of ganglia or dorsal rhizotomy, for example, carries a risk of sensory loss and possibly anesthesia dolorosa, and DRG stimulation may be associated with background paresthesias. The efficacy of pharmacotherapy in the management of neuropathic pain probably also depends in large part on the suppression of ectopic neural discharge (ectopia) originating in sensory ganglia and sites of nerve injury. Drugs with proven efficacy include particular members of an idiosyncratic group of therapeutic categories including antidepressants, anticonvulsants, local anesthetics, and antiarrhythmics.21 The most striking common denominator among them is their “membrane stabilizing” property. They all suppress peripheral nervous system (PNS) ectopia.6,12,53 Unfortunately, and not coincidentally, these agents also have in common some cardiac risk and a characteristic set of adverse central side effects, notably somnolence, vertigo, and nausea. This is almost certainly due to membrane stabilization of a shared population of neurons in the central nervous system (CNS).
We report here neurophysiological observations in rats which point to the feasibility of a novel drug-based pain-relief strategy, applicable to the frequently intractable neuropathic pain conditions noted. This approach can be realized with commercially available technologies and will avoid adverse PNS and CNS side effects. Specifically, we show that dilute lidocaine selectively blocks the ectopic generation of impulses within the DRG with relatively modest suppressive effects on the propagation of impulses through the intervertebral foramen. Implementation of this strategy requires delivery of low, nonblocking concentrations of a membrane-stabilizing drug to focal pain drivers. The primary target is the epidural space surrounding the DRG within the foramen. Perineurial delivery to sites of focal nerve injury is also expected to be effective. We also tested effects of applying opiates and gamma aminobutyric acid (GABA) to the DRG surface as an alternative to lidocaine, and found them much less promising.
2. Material and methods
2.1. Animals and surgery
Results derive from experiments on 16 adult male (340 ± 55 g) Sprague-Dawley rats (Harlan/Envigo Labs, Jerusalem, Israel). The animals underwent an initial surgical procedure to induce peripheral nerve injury, followed by an acute terminal electrophysiological protocol to evaluate effects of intraforaminal application of drugs. Experiments were performed with the approval of the Institutional Animal Care and Use Committee of the Life Sciences Institute, the Hebrew University of Jerusalem and in accordance with guidelines of the U.S. Public Health Service's Policy on Humane Care and Use of Laboratory Animals.
Under chloral hydrate anesthesia (400 mg/kg intraperitoneally (i.p.), cat.#15307; Sigma-Aldrich, Rehovot, Israel) using aseptic precautions, we opened the biceps femoris muscle to expose the left sciatic nerve in the lower part of the popliteal fossa and freed it locally from connective tissues. Using 5-0 silk, the nerve was tightly ligated at its point of trifurcation into tibial, common peroneal, and sural tributaries and transected 1 mm distal to the ligature. About 5 mm of the distal nerve stump was removed to suppress regeneration. The exposure was then closed in layers, a topical bacteriostatic powder was applied (bismuth subgallate, Dermatol, Floris, Misgav, Israel), and prophylactic penicillin was given (60 mg intramuscularly, Penibrin, Sandoz, Kundl, Austria). Recovery was uneventful. Preoperatively and postoperatively, the rats were maintained in our Institute's specific pathogen-free facility in transparent plastic cages (23 × 17 × 12 cm) bedded with pine wood shavings. Solid, pelleted food (Teklad product 2918 supplied by Envigo-Harland Ltd., Jerusalem, Israel), and water were available ad libitum. The day–night cycle was 12:12 hours with lights on at 7:00 AM.
Between 3 and 8 days postoperatively (dpo), we prepared the rats for electrophysiological recording using the teased-fiber method.50 The objective was to simultaneously monitor spike activity generated ectopically within the L5 DRG and impulse conduction along axons that traverse the L5 intervertebral foramen. Rats were again anesthetized with chloral hydrate (400 mg/kg i.p. supplemented with 100 mg/kg i.p. as needed), the trachea was cannulated and the rats were mounted prone with the left leg in extension in a Narishige spinal frame. Heart rate and rectal temperature (37°C) were monitored and maintained.
We began by revealing the L4 and L5 spinal nerves just distal to their exit from the corresponding intervertebral foramen and severing the L4 spinal nerve (Fig. 1). Then, using fine Lempert nippers, we exposed the dura over the lower lumbar DRs in a laminectomy that extended from the L3 DRG to the L3 root entry zone. The tip of a PE10 catheter (Intramedic; Clay Adams, Parsippany, NJ), prefilled with sterile saline (0.9% NaCl, 37°C) and cut at an angle, was inserted along the spinal nerve into the L5 foramen until its end rested on the surface of the L5 DRG. The length of the catheter was 16 cm and it had an internal volume of 10 μL. The distal ∼1 cm of the catheter, which had previously been heat shaped into an arc, was anchored to tissue at the exit of the foramen using cyanoacrylate glue and its proximal end was connected to a 10 μL Hamilton syringe. A pool was then formed around the exposures using the skin edges and filled with paraffin oil (34°C).
Next, we re-exposed the sciatic nerve from its cut end to the sciatic notch, covered it with paraffin oil in a second pool (34°C), and opened the perineurium at a point ∼15 mm proximal to the neuroma and ∼50 mm distal to the L5 DRG, using fine scissors. This permitted teasing of fine axon bundles (“microfilaments”) from the main nerve trunk using a pair of specially honed jewelers' forceps. Microfilaments were in-continuity proximally with the DRG, but cut distally. Finally, the spinal dura was opened, the L4-6 DRs were cut near their point of entry into the spinal cord and the cut end of the L5 DR was draped over a pair of Ag/AgCl stimulating hook electrodes (Fig. 1).
2.2. Experimental protocol
2.2.1. Electrophysiological recording
For recording neuronal activity, teased microfilaments were adhered to an Ag/AgCl recording electrode referred to a nearby indifferent. In 3 rats, to increase the yield of active axons, we used 2 independent electrodes and recording channels to simultaneously monitor from 2 microfilaments. Raw signals were fed into the high impedance headstage of a differential amplifier and low-pass filtered at 10 kHz (and sometimes also high-pass filtered at 10 Hz; NeuroLog System, Digitimer, Welwyn Garden City, United Kingdom). Signals were viewed on analog and digital oscilloscopes and digitized at 20 kHz (Digidata 1322a and AxoScope 9.0; Axon Instruments, Union City, CA) for later off-line analysis using Matlab (version R2012a; Mathworks, Natick, MA).
Spontaneous discharge originating within the L5 DRG was passively recorded on teased microfilaments.50 Typically, we began with a fairly large axon bundle containing ongoing activity of many units and then progressively split it into smaller microfilaments. After each split, we retained the most favorable fragment aiming for one that contained at least 2 active units with amplitudes well above the noise level and spike shapes that were easily distinguishable. Units had to fire in a stable, sustained manner to avoid confusion with transient injury discharge that is sometimes caused by the teasing process.
Square-wave stimulus pulses were delivered to the L5 DR at 0.1 or 0.2 Hz through a photoelectric stimulus isolator (2 mA, 0.1 ms, model DS2; Digitimer). Pulses were sufficient to drive all A fibers in the recorded microfilament as documented by saturation of the response evoked, the microfilament's compound action potential (μCAP). μCAPs typically contained 10 to 20 individual afferent A fibers as determined by gradually increasing stimulus frequency and counting the recruitment of individual fibers to saturation34 (Figs. 2A–C). The μCAP served as a monitor of spike propagation through the L5 intervertebral foramen (Fig. 1) and permitted calculation of axonal conduction velocity by dividing propagation distance (∼70 mm) by response latency. All the spontaneously active units in this study were Aβ and Aδ fibers (response latency 0.95-11.0 ms and conduction velocity 6.4-73.7 m/s). In the time window used, 3 to 8 dpo, spontaneous ectopic activity in A fibers is much more prevalent than in C fibers and seems to be a major contributor to ongoing pain and allodynia in many rat models of neuropathy.12,13,17,27,30,34,45,50 Microneurographic recordings in humans also implicate DRG A fibers as a prominent contributor to some neuropathic pain conditions.5,40,41
2.2.2. Source of spontaneous discharge
Nearly, all sciatic nerve afferents in the rat belong to cell somata that reside in the L4 and L5 DRG (97.5%16). In our preparation, recorded activity could not have originated in the L4 DRG, as the L4 spinal nerve had been cut. However, the L6 DRG does contribute a small number of afferents, particularly in rats with a postfixed lumbosacral plexus, and hence, a small fraction of the recorded activity could have originated in the L6 DRG. The L6 spinal nerve is not readily accessible in our preparation and was not cut. However, we used 2 methods to confirm the L5 source and to minimize the risk of unknowingly drawing conclusions based on activity originating in L6 DRG afferents. The first was based on the fact that the pattern of the spontaneous activity in stimulated L5 afferents is “re-set” by the intermittent L5 DR stimulus pulse. That is, their activity becomes transiently time-locked to the pulse (Fig. 3). Activity originating in the L6 DRG is not reset in this way by L5 DR stimulation. The second method, which relates to the pattern of discharge suppression when lidocaine is applied to the L5 DRG, is described below (section 2.2.7).
2.2.3. Drug application to the dorsal root ganglion—lidocaine
Once an appropriate microfilament(s) was chosen, we passively recorded 11 minutes of spontaneous activity, interrupted only by the stimulus pulses that were delivered to the L5 DR as described (section 2.2.1). Data from this “reference period” provided data on the units in the microfilament(s). We then recorded activity during the course of a series of intraforaminal injection trials using saline as a control, followed by lidocaine in stepwise increasing concentration. A “trial” consisted of a period of preinjection “baseline” recording (30 seconds for the first 12 experiments, later extended to 60 seconds), 10 seconds during which saline or lidocaine was injected and then an additional 10 minutes of postinjection recording. An “experiment” on a particular microfilament consisted of the reference period followed by a trial series which began with saline and proceeded to trials of lidocaine in concentrations beginning with 0.002% (or occasionally 0.02%) and ascending to 2%, in decade steps. Overall, this took about 75 minutes.
At the beginning of each experiment, saline filled the catheter. The injection series began with introduction of a 10 μL bolus of 0.002% lidocaine (37°C) into the injection catheter. This caused the extrusion of the 10 μL of saline already present in the catheter onto the surface of the DRG. A tiny air bubble separated consecutive bolus volumes and also served as a visual indicator of the advance of the solution within the catheter. After the saline trial, preparations were made to inject the first bolus of lidocaine (lowest concentration). Specifically, we drew 10 μL of 0.02% lidocaine (37°C) into the Hamilton syringe and attached it to the catheter. This took ∼1 minute. At this point, the next trial began with the preinjection baseline recording period. Then, the 10 μL of 0.02% lidocaine in the syringe was advanced into the catheter causing the 0.002% lidocaine already in the catheter to be extruded onto the surface of the DRG. Effects of this bolus injection were then monitored. For subsequent trials, this same procedure was repeated, with the extrusion of 10 μL of 0.02% lidocaine, then 0.2% lidocaine and finally 2.0% lidocaine. Lidocaine solutions were prepared by dissolving lidocaine HCl (in powder form, cat. # L5647; Sigma-Aldrich) in sterile saline at 2.0% (85 mM). This stock solution was further diluted to form the various working concentrations.
2.2.4. Drug application to the dorsal root ganglion—opiates and gamma aminobutyric acid
Experiments were also performed using 2 μ-opiate receptor agonists, morphine sulfate (0.9, 9, and 90 mM dissolved in saline) and fentanyl citrate (50 μg/mL, 0.15 mM, both from the University Hospital pharmacy) and GABA (100 μM, 1.0, 10, and 100 mM in saline [cat.# A2129; Sigma-Aldrich]). The experimental procedure was as described above for lidocaine, always beginning with the lowest drug concentration in the series and working up. Because none of these drugs reliably silenced spontaneous discharge or blocked foraminal through-conduction, at the end of each of these experiments, we routinely confirmed correct placement of the catheter by documenting suppression of discharge with injected lidocaine.
2.2.5. Data analysis—spontaneous discharge
Drug effects on spontaneous DRG firing were analyzed at the level of the individual active unit. The first step was to sort the multiunit activity recorded on microfilaments into single units. This was performed in Matlab using the Wave_clus toolbox (version 2.542). Briefly, individual spontaneous spike events were first separated from noise and from the DR stimulus artefact and response using window voltage thresholding. Briefly, 25 wavelet coefficients (25 data points) were calculated for each event. The Lilliefors modification of the Kolmogorov–Smirnov normality test was applied and the 10 coefficients with the largest deviation from normality were features used as input to the superparamagnetic clustering algorithm that defined individual units.
In microfilaments with active units of similar waveform, spike sorting was sometimes imperfect. For example, a waveform initially defined by the spike-sorting algorithm as a single unit with an irregular discharge pattern and a relatively high-firing frequency was sometimes later found to consist of 2 similar, but distinct units, each of which fired tonically at lower frequency. We interpret these as cases in which the software at first failed to distinguish the 2 units as independent. Later in the session, a small change in the waveform of 1 or both units due to any of a variety of factors, or lidocaine-induced silencing of one, permitted the units to be separated. By the same token, 2 waveforms initially identified as independent units with tonic discharge were occasionally merged by the software into a single unit with irregular high-frequency discharge. We logged 6 instances of the former type of sorting error and 8 of the latter. Drug trials on such units were treated as defined by the spike-sorting algorithm (ie, 1 or 2 units) up until the moment that the sorting error was revealed. Thereafter, they were treated as independent units. For each sorted unit, we plotted the corresponding train of interspike intervals (isi, Fig. 2D). Firing rate was calculated by counting the number of spikes in a 30-second sliding time window advanced in steps of 1 second.
The primary metric for defining a drug effect was “silencing”, when firing rate fell to zero for a period of at least 30 seconds. Additional metrics describing firing rate were “slowing”, where firing rate fell to ≤50% of the preinjection baseline rate, and “accelerating” where firing rate increased to ≥200% of baseline. We also noted latency of response to applied drugs and response duration. In each trial, the change in firing rate was calculated in comparison with discharge rate during the 30-second (12 rats and 12 filaments) or 60-second (4 rats and 7 filaments) baseline period immediately preceding the drug injection. We note that baseline activity was not necessarily identical to activity during the initial reference period, or to that observed before the previous drug trial. For example, if after injection of a particular lidocaine dose, one unit fell silent, whereas others either failed to respond or recovered after transient slowing, we proceeded with the experimental protocol based on the remaining active units at their current firing rates. This means that particularly sensitive units, in which spontaneous activity was silenced for a long period by a low lidocaine concentration, were not included among the units that were exposed to higher concentrations of lidocaine. For this reason, more units were tested with low concentrations than with high concentrations.
Experiments ended when all units in the microfilament fell silent for ≥30 seconds. Even if 1 or more later resumed firing, they were not exposed to higher lidocaine concentrations. Rather, at this point, 10 μL of 2% pontamine sky blue dye was extruded into the foramen which was then opened to permit visualization of dye spread. In each case, the dye completely covered the DRG epidurally and extended a variable distance beyond, covering as much as 8 mm of the cauda equina central to the L5 DRG. Finally, the animal was killed with an overdose of chloral hydrate followed by cervical fracture.
2.2.6. Data analysis—axonal through-conduction
Assessment of drug effects on through-conduction was based on the μCAP. We began by establishing, in each individual rat, a time window for capturing the μCAP. The “μCAP magnitude” was the area under the curve from 0.5 ms before the rising phase of the μCAP until 0.5 ms after the end of the falling phase of the μCAP. To establish a baseline μCAP magnitude, this value was averaged over all μCAPs evoked during the preinjection baseline recording period before saline or drug injection. Then, during the 10 minutes that followed each injection, averages were made within a running window of 30 seconds that advanced in steps of 1 interstimulus interval (5 or 10 seconds). Thus, each window averaged 6 or 3 μCAPs depending on whether the DR stimulation frequency was 0.2 or 0.1 Hz. Drug-induced changes in the magnitude of μCAPs over the course of the 10-minute observation period were expressed as a percentage of the preinjection baseline value. The degree and timing of maximal decrease in μCAP magnitude was noted for each trial and these values were averaged over all trials that tested a given lidocaine concentration.
The μCAP combines all the conducting units contained in the microfilament. It is important to note, however, that its magnitude is not a linear function of the number of axons recorded, as each unit makes a different contribution to the μCAP's area under the curve. A drop of 50% in μCAP magnitude, eg, does not necessarily indicate that conduction was blocked in 50% of the axons in the microfilament. However, because sensitivity to lidocaine is similar across fiber types, in A fibers at least,29 the decline in μCAP magnitude is a reasonable estimate of through-conduction fidelity. Moreover, it was always possible to visualize the loss and return of individual units in the μCAP over the course of drug trials. Minor drift in μCAP magnitude frequently occurred without loss or gain of all-or-none units. The likely cause was variation in the electrical coupling between the microfilament and the recording electrode.
2.2.7. Diagnosing whether drug action suppresses electrogenesis or blocks spike propagation
In principle, a drug-induced reduction in the number of spontaneous impulses recorded could be due to either of 2 causes: suppression of impulse generation, or block of the forward propagation of impulses already generated. We applied an analytic method, based on observation of firing pattern during the course of discharge suppression, that permitted us to distinguish between these 2 possibilities. The observations also provided confirmation that discharge indeed originated in the L5 DRG and not elsewhere. Specifically, drug-induced suppression of firing takes 1 of 2 forms depending on whether it reflects suppression of spike electrogenesis or conduction block. If lidocaine is applied to axons carrying tonic rhythmic discharge, individual spikes in the impulse train begin to fail as Na+ channels are progressively blocked. This causes the isi of the unit to increase in a discontinuous manner, in integer multiples, until the point of total conduction failure. A radically different pattern is seen if lidocaine acts directly at the locus of spike initiation. Here, the isi increases in a smooth (continuous) manner until firing ceases (Fig. 4). If the pacemaker site at which the impulses arise is near its repetitive firing threshold,12 firing may also cease from 1 spike to the next with no change at all in isi. In such cases, the mode of suppression cannot be diagnosed with certainty. Documentation of these differences, and an explanation of their basis, is given elsewhere.36
2.3. Statistical analysis
Statistical analysis was performed in Excel or Matlab. The proportion of units tested that reached a given criterion change in discharge rate, complete “silencing” for ≥30 seconds, “slowing” or “accelerating”, was tested using the Fisher exact probabilities tests. Mean values were compared using 2-tailed Student t tests. Variability is given ± SD except where SEM is specifically noted. The criterion used for statistical significance was P < 0.05.
Rapid responses to drugs applied within the intervertebral foramen confirmed free penetration through the dural capsule of the DRG in vivo.1 The key observation was that epidural lidocaine suppressed spontaneous discharge originating in the DRG at concentrations more than 2 orders of magnitude lower than those required to fully block the propagation of action potentials in axons that passed through the foramen. That is, dilute solutions of lidocaine suppress spike electrogenesis selectively. Neither morphine, fentanyl, nor GABA substantially suppressed electrogenesis in the DRG and none of the 3 stopped axonal through-conduction, even at very high concentrations.
3.1. Spontaneous discharge in dorsal root ganglia
During the reference period, we identified 52 spontaneously active units (16 rats, 19 microfilaments, 2.7 ± 0.9 active units per microfilament). Firing rates and patterns were much as described in previous studies of spike activity originating in the DRG.17,27,50 Briefly, most fibers fired tonically (ie, with a fixed isi), with either occasional individual spikes dropped from the spike train (“tonic with misses”, 31/52, 59.6%), or with more prolonged silent intervals (“bursting” pattern, 3/52, 5.8%). These patterns are easily recognized in raw records and in isi dot displays such as in Figures 2–4. The characteristic isi in units with missed spikes typically ranged from 80 to 40 ms, translating to about 10 to 20 Hz (not counting the misses). Bursting units had an instantaneous firing frequency of 18 to 36 Hz during bursts (isi = 56-28 ms) with overall firing rate depending on the on–off duty cycle. A minority of units fired with an irregular isi (17/52, 32.7%). Of these, 6 fired in the range of 15 to 52 Hz and 11 fired at <5 Hz. In 1 unit (1.9%), the isi varied in a complex cyclic manner with a period of ∼120 seconds. Units with such firing patterns also occur in experimental neuromas.10
3.1.2. Location of spike electrogenesis
In principle, discharge in centrally connected sciatic nerve microfilaments might originate anywhere from the cut end of the DR to the teased microfilament itself (Fig. 1), although evidence from previous studies prioritizes the DRG.34 In this study, application of the isi pattern method of Matzner and Devor36 established an almost exclusive origin in the L5 DRG. The analysis was applied to 75 lidocaine trials in which spontaneous discharge fell silent after the application of 1 concentration or another of lidocaine to the DRG surface. In about half of these trials (39/75, 52.0%), the unit's firing slowed to a stop by smooth increase in isi (Fig. 4A). This indicates suppression of spike electrogenesis. In the remaining units, firing ceased instantaneously, making the mode of suppression indeterminate. None of the units observed showed clear evidence of axonal conduction block, although in all 75 units the propagating axon was exposed to lidocaine throughout its traverse through the L5 foramen. Smooth increase in isi was also the exclusive pattern in units in which lidocaine slowed firing, but did not stop it completely (Fig. 4B). The integer multiple pattern of isi change, which is diagnostic of axonal propagation block, was not seen in any trial in any unit. In most of the trials in which the unit was silenced (44/75, 58.7%), discharge began to recover during the 10-minute observation period. In these, and in all the units slowed but not stopped, the recovery of firing always took the form of gradually shortening isi (Figs. 4A and B). These observations indicate that the spontaneous discharge included in the present analysis uniformly arose within the DRG and not elsewhere. Indeed, it probably arose within sensory cell somata themselves.12
During the course of the study, we encountered 3 units in which firing remained unchanged, despite application of a blocking concentration of lidocaine (2%) to the L5 DRG. This suggests that the lidocaine did not actually access these afferents. To check, we applied 2% lidocaine (10 μL) distally along the exposed L5 spinal nerve near the junction with the L6 spinal nerve. In 1 case, firing ceased following the isi integer multiplying pattern indicating axonal conduction block rather than suppressed electrogenesis (Fig. 4C). In the other 2, firing stopped suddenly. We conclude that the origin of these impulses was in the L6 DRG, or conceivably in a location in the L5 DRG that was somehow not accessed by the lidocaine when delivered intraforaminally. These 3 units were excluded from the analysis.
3.2. Suppression of discharge as a function of lidocaine concentration
3.2.1. Spontaneous discharge
Irrespective of conduction velocity, or initial firing rate or pattern, lidocaine suppressed spontaneous discharge in a concentration-dependent manner. Concentration dependence was seen in the likelihood that firing would be silenced or slowed after lidocaine application and in the kinetics of suppression. An example of the response of 5 units in a single microfilament to the application of 0.02% lidocaine is shown in Figure 5. Two of the units were silenced (1 recovered) and 3 were slowed. Figure 6 plots the percent of units across all experiments that continued to fire (ie, were not silenced) at each of the lidocaine concentrations tested, where “silenced” means that firing rate fell to zero for at least 30 seconds. The lowest lidocaine concentration tested, 0.002% (10−3 of the concentration normally used for clinical nerve blocks), silenced discharge in only 5 of the 40 units tested (12.5%). An additional 17.5% (7/40) showed transient slowing (to ≤50% of the baseline firing rate) without complete silencing. The majority (70.0%) continued firing largely unchanged. Corresponding values for saline were (4/55 silenced and 13/55 slowed). Proportions for 0.002% lidocaine were not significantly different from saline (combining units silenced and slowed, 12/40 vs 17/55 P = 0.486, Fisher test). Using 0.02% lidocaine, a much larger proportion of units was silenced (28/50, 56%) or slowed (6/50, 12%; combined 34/50, 68%; P < 0.001 vs saline) and likewise with 0.2% lidocaine. Lidocaine of 2% silenced all the units tested. No instances were observed of increased firing, and no previously silent units began to fire in response to applied saline or lidocaine.
3.2.2. Duration and latency
The duration of silencing also increased with increasing lidocaine concentration (Fig. 7). After exposure to 0.002% lidocaine, 4 of the 5 units that were silenced resumed firing within the 10-minute observation time (duration of silence averaged 207 seconds). For comparison, of the 14 units silenced after exposure to 2% lidocaine, only 2 resumed firing within 10 minutes (duration of silence 155 and 462 seconds; 4/5 vs 2/14, P = 0.017). Paradoxically, in units that were silenced by low concentrations of lidocaine, the latency from drug application to silencing was shorter than in units that were silenced only after a high concentration of lidocaine was applied. Thus, latency was 13.0 ± 16.0 seconds for units silenced using 0.002% or 0.02% lidocaine (n = 33) and 46.6 ± 73.8 seconds (n = 42) using 0.2% or 2% lidocaine (P = 0.006). A likely explanation is that the greater resistance to silencing of the latter units reflects a larger diffusion distance to the ectopic pacemaker site and/or greater intrinsic excitability, and hence a longer response latency.
3.2.3. Suppression of transforaminal through-conduction
The sensitivity of axons to conduction block by topical lidocaine was evaluated by tracking the reduction in magnitude of the recorded μCAPs evoked by L5 DR stimulation (Figs. 1 and 2). During the reference period (before injections), evoked μCAPs were stable, with little variation from stimulus to stimulus and at most minor drift over the 10-minute observation period (Figs. 6 and 8A and B). Responses after saline administration were similar. Likewise, application of lidocaine at 0.002% and 0.02% to the DRG caused no change in μCAP magnitude on average (both P > 0.2 compared with saline), although 0.02% lidocaine did markedly suppressed spontaneous firing (Figs. 6 and 8). Increasing the concentration to 0.2% caused a measurable decrease in μCAP magnitude in 11 of 13 microfilaments tested. This occurred in the first half of the observation period. In all but 1, the μCAP substantially recovered before the end of the observation period (Fig. 6). Recovery occurred in a stepwise manner, 1 unit at a time, reflecting asynchronous return of the individual units in the microfilament. The maximal decline in μCAP magnitude was to 51.7 ± 26.5% of preinjection baseline (Figs. 6 and 8A). The decline was statistically significant compared with saline (P = 0.017) and with 0.02% lidocaine (P = 0.036).
By contrast, the application of 2% lidocaine produced consistent and sustained conduction block, with complete elimination of the μCAP (Figs. 6 and 8A and C). Decline began during the course of the injection in 4 of the 6 microfilaments tested, and after 20 and 90 seconds in the remaining 2. Depending on the microfilament, it took between 2 and 8 minutes for the last of the units to be blocked. Block persisted at least until the end of the observation period in all but 1 of the microfilaments. In this case, the μCAP recovered to ∼40% of its original magnitude. Overall, spontaneous electrogenesis proved to be much more sensitive to lidocaine than axonal spike propagation. That is, there is a prominent therapeutic window in the concentration of lidocaine applied, in the range of 0.02% and 0.2%, within which ectopic discharge is suppressed with little or no change in spike through-conduction.
3.3. Topical application of morphine or fentanyl
Response to opiates differed sharply from response to lidocaine. The opiate receptor agonists morphine (0.9, 9.0, 90 mM) and fentanyl (0.15 mM) were applied to the surface of the L5 DRG in 4 rats, morphine in 3 (13 units in 4 microfilaments), and fentanyl in 3 (10 units in 3 microfilaments). In 2 of these rats, both opiates were tested. Preinjection baseline firing was as described above (3.2.1). The initial response to drug application was rated as “increased” if firing frequency increased to at least double the baseline firing rate, “decreased” if it fell to <50% of baseline and “no-change” if it remained within >50% and <200% of baseline. In 26 trials using morphine or fentanyl, 10 (39%) showed initial excitation, 7 (27%) showed initial slowing, and 9 (35%) were largely unaffected (Table 1 and Fig. 9). Changes in firing rate took the form of smoothly decreasing or increasing isi. Responses began within the injection period or shortly thereafter and either began to fade within 1 to 3 minutes or persisted until the end of the observation period. Neither morphine nor fentanyl caused loss of any of the units in the μCAP indicating that intraforaminal application does not block axonal through-conduction, even at the high concentrations used.
3.4. Topical application of gamma aminobutyric acid
Gamma aminobutyric acid (100 μM, 1, 10, 100 mM) was applied to the L5 DRG in 3 rats (5 microfilaments). Altogether, 14 spontaneously active units were exposed to GABA (42 trials overall, Table 1). Like the opiates, and unlike lidocaine, the initial response to topical GABA was predominantly neutral or excitatory. Only 1 instance of suppression was noted, a unit in which firing frequency was reduced using 100 μM, 1, and 10 mM GABA (3 of the 42 trials run, 7%). As with the opiates, initial responses began within or shortly after the injection period and lasted for several minutes. Applied intraforaminally, GABA only minimally affected axonal conduction past the DRG. Propagation was lost in only 2 units in the μCAP of 1 of the microfilaments, and this only when the DRG was challenged with the highest concentration of GABA, 100 mM (Table 1 and Fig. 9).
Epidural application of dilute lidocaine (0.02% or 0.2%) to the surface of the DRG within the intervertebral foramen in rats selectively suppressed the ectopic generation of afferent impulses within the ganglion with comparatively minor block of impulse propagation through the foramen. This difference reveals a therapeutic window for analgesia. Considering that there is further drug dilution after application, the minimally effective concentration for discharge suppression is almost certainly lower than 0.02% (0.85 mM, see 4.11 below). Application of the opiates morphine and fentanyl, and of GABA, agents also expected a priori to be suppressive, had neither effect. The observation of selective (differential) suppression of DRG electrogenesis with dilute lidocaine has direct implications for the control of dysesthesias and spontaneous pain, and of tactile allodynia driven by ectopic impulse generators in the PNS.
4.1. Selective suppression of ongoing discharge originating in the dorsal root ganglion
Impulse propagation along axons is highly secure. The longitudinal current available to sustain the advancing spike is much greater than the minimum required. This high “safety factor” explains why application of high concentrations of Na+ channel blockers over several nodes of Ranvier is required to block nerves. Electrogenesis, by contrast, is a threshold process. Near the threshold for repetitive firing, the typical state of afferents with tonic-with-misses or bursty discharge, a slight reduction of excitation or excitability is enough to silence firing.9,12,26,44 Ectopia originating in DRGs is particularly insecure. Dorsal root ganglion neurons are subject to profound membrane accommodation, their discharge is largely driven by lidocaine-sensitive subthreshold oscillations and without a blood–DRG barrier they are exposed to numerous blood-borne molecules.1,2,18,31,46
The therapeutic window between spike electrogenesis and spike propagation explains why systemic lidocaine at ≤5 mg/kg delivers effective analgesia in patients and animal neuropathy models.4,7,17,18,45,47,51 The resulting plasma concentration, 5 to 20 μM,35 is sufficient to moderate ectopic firing, but far too low to block nerve conduction. Widespread nerve block, of course, would be lethal. But although effective, systemic dosing generates significant adverse side effects because of off-target drug delivery, particularly in the CNS. Thus, optimal exploitation of the therapeutic window also requires focal drug targeting.
As for effect duration, the virtual space between the DRG capsule and the bony foramen was filled by the 10 μL volumes injected, but there was no residual space for a drug reservoir. Moreover, the rich vascularization of the DRG efficiently dilutes applied drugs and washes them out of the ganglion. These factors account for the relatively brief effects observed. Sustained suppression of discharge, and pain relief, requires sustained drug delivery to the locus of electrogenesis.49
4.1.2. Opiates and gamma aminobutyric acid
Endogenous opiates (eg, enkephalins) and GABA are neurotransmitters. In light of the virtual absence of synapses in DRGs,28 their cognate receptors are expected to be located at central and perhaps, peripheral axon ends. However, both receptor types are known to be inserted into the somatic membrane and to be capable of mediating conductance changes in DRG somata.20,43 μ-opiate receptors couple to K+ channels and typically hyperpolarize and inhibit neuronal activity. GABAA-Rs contain Cl− channels which may hyperpolarize or depolarize depending on Cl− gradients, but channel opening always creates a current shunt and hence in practice, GABA is usually inhibitory. Empirically, however, most DRG neurons tested with opiates and GABA was either indifferent or excited.
A priori, opiates and GABAergic agents are not expected to affect nerve conduction (but see Ref. 48). However, the opening of ion channels in DRG somata might shunt into the stem axon enough of the longitudinal current required for forward spike propagation to cause branch-point block.3 Recently, Du et al.19 documented activity-related GABA release within DRG and on this basis predicted that consequent shunting might threaten sensory transmission past the afferent T-junction and into the CNS. If so, this would constitute a novel gate-control mechanism and could provide the elusive explanation of why DRG neuronal somata bear neurotransmitter receptors in the first place.11 Unfortunately, our results revealed no indication of the predicted conduction block (Table 1). The mystery thus remains.
4.2. Clinical implications
4.2.1. Systemic vs focal administration
The clinical efficacy of systemic lidocaine, which is almost certainly due to suppression of neuropathic ectopia, is shared by other membrane-stabilizing drugs with proven analgesic efficacy including tricyclic antidepressants, anticonvulsants, and antiarrhythmics.6,12,53 A major limitation of these drugs for pain relief, however, is the generation of adverse, often intolerable central side effects: somnolence, vertigo, and nausea.21 There is also potential cardiac risk. Our results suggest that these side effects might be avoided by targeting dilute lidocaine or other membrane stabilizers directly to ectopic pacemakers such as the DRG. Conditions in which the DRG is implicated as a pain driver include postherpetic neuralgia, a pain state caused by viral infection of a single ganglion, TN where fifth nerve root compression triggers hyperexcitability within the trigeminal ganglion, phantom pain, complex regional pain syndrome, and radicular low back pain.5,8,14,15,33,37 The principle also applies to pain generated at focal sites of nerve injury such as nerve-end neuromas and nerve entrapment sites.12
The 5% lidocaine patch is an example of such focal targeting. The patch reservoir contains a blocking concentration of drug, but levels in the skin are far lower and resulting systemic concentrations are lower still.22 Distant skin, and even skin under the patch, is not rendered numb. This rules out nerve fiber block. Rather, suppression of ongoing pain felt in the skin (and sometimes beyond) is almost certainly due to selective block of electrogenesis. Another example is amputation stump and phantom limb pain. Using microneurography, Nystrom and Hagbarth41 observed ongoing spike activity in stump nerves. Blocking the neuroma eliminated pain on percussion (Tinel sign). But the ongoing discharge persisted, suggesting an origin in the DRG. This mechanism was verified by Vaso et al.49 who documented elimination of phantom limb sensation and pain using nonblocking concentrations of lidocaine applied intraforaminally. Spontaneous pain was suppressed, but the Tinel sign persisted indicating that afferent through-conduction was not blocked. Bolus dosing yielded a transient analgesic effect, but it could be sustained by repeated intraforaminal injections.
4.2.2. Approaches to clinical implementation
For selective pain suppression, dilute lidocaine and other membrane stabilizers must be delivered to the actual site of electrogenesis. Clinical signs, pain provocation, and diagnostic blocks are needed to identify which DRG, or other pacemaker loci, drive pain in the individual patient. The best current means of sustained drug delivery to the DRGs are implantable pump systems, although a port can serve for short-term trialing.49 The pump would deliver the dilute membrane-stabilizing drug through a permanently implanted catheter inserted into the intervertebral foramen in a manner similar to DRG stimulating electrodes.33 But rather than filling the pump with a dilute suppressive drug, it could be filled with a high concentration of the drug-of-choice which would then be extruded very slowly. Subsequent dilution in the target tissue would provide a steady-state level of nonblocking concentrations for extended periods, requiring reservoir refills only at long intervals, hence easing problems of pump management and patient compliance. In practice, the rate of drug delivery could be titrated to effect.
Local drug concentration within the ganglion (or at the nerve-injury site) presumably needs to be maintained at values similar to those present in the plasma during successful systemic delivery of the same agent. Required lidocaine concentrations are in the 100 μM range.22,35,52 In current medical practice, the candidate drugs are routinely and safely delivered to the systemic circulation over long periods. There should therefore be little concern about toxicity in the DRG or nerve from focal administration of the same drugs at the same concentrations. Systemic concentrations resulting from focal administration to the DRG, in comparison, would be far lower and thus cardiac and CNS toxicity, including the dose-limiting CNS side effects noted, should be virtually absent. These same considerations apply to superfusion of alternative foci of ectopic electrogenesis such as sites of nerve trauma or compression (eg, neuromas and carpal tunnel). The normal barriers to drug penetration at these sites are compromised by the lesion and hence, access should be as at the DRG. Indeed, focal drug delivery to injured nerves is a part of current clinical practice, but the underlying concept is (proximal) nerve block rather than precision delivery to sites of electrogenesis.25
The objective of the treatment is to achieve pain control without central or peripheral side effects and without affecting normal sensory or motor function. Such a modality could be implemented today, with existing commercially available devices. In the future, the mechanical device might be replaced by a cellular or tissue-based “biopump.” In this implementation, injection of engineered cells or microorgans would release a sustained low level of an inhibitory peptide to the surface of a DRG, or site(s) of nerve injury.23,24 Such next-generation technology promises a general solution applicable to many neuropathic pain conditions in which there is an identifiable focal pain driver.
Conflict of interest statement
The authors have no conflict of interest to declare.
The authors thank Anne Minert and Hanoch Meiri for their assistance.This work was supported by the Israel Science Foundation, the National Network of Excellence in Neuroscience (NNE) program initiated by Teva Pharmaceuticals, the Seymour and Cecile Alpert Chair in Pain Research, and the Hebrew University Center for Research of Pain.
. Allen DT, Kiernan JA. Permeation of proteins from the blood into peripheral nerves and ganglia. Neuroscience 1994;59:755–64.
. Amir R, Devor M. Chemically-mediated cross-excitation in rat dorsal root ganglia. J Neurosci 1996;16:4733–41.
. Amir R, Devor M. Electrical excitability of the soma of sensory neurons is required for spike invasion of the soma, but not for through-conduction. Biophys J 2003;84:2181–91.
. Boas RA, Covino BG, Shahnarian A. Analgesic responses to i.v. lignocaine. Br J Anaesth 1982;54:501–5.
. Burchiel KJ, Baumann TK. Pathophysiology of trigeminal neuralgia: new evidence from a trigeminal ganglion intraoperative microneurographic recording. Case report. J Neurosurg 2004;101:872–3.
. Catterall WA. Common modes of drug action on Na+ channels: local anaesthetics, antiarrhythmics and anticonvulsants. Trends Pharmacol Sci 1987;8:57–65.
. Chaplan SR, Bach FW, Shafer SL, Yaksh TL. Prolonged alleviation of tactile allodynia by intravenous lidocaine in neuropathic rats. Anesthesiology 1995;83:775–85.
. Defrin R, Devor M, Brill S. Tactile allodynia in patients with lumbar radicular pain (sciatica). PAIN 2014;155:2551–9.
. de Jong RH. Local anaesthetics. St. Louis: Mosby, 1994.
. Devor M. The pathophysiology of damaged peripheral nerve, 2nd edition. In: Wall PD, Melzack R, editors. Textbook of pain. 2nd ed. London: Churchill-Livingston, 1989. pp. 63–81.
. Devor M. Unexplained peculiarities of the dorsal root ganglion
. PAIN 1999;82(suppl 6):S27–35.
. Devor M. Neuropathic pain
: pathophysiological response of nerves to injury. Chapter 61. In: Wall and Melzack's Textbook of Pain, 6 Edition (McMahon SL, Koltzenburg M, Tracey I, Turk DC, eds), pp 861–888. London, United Kingdom: Churchill Livingstone, 2013.
. Devor M. Ectopic discharge
in Abeta afferents as a source of neuropathic pain
. Exp Brain Res 2009;196:115–28.
. Devor M. Neural basis of pain in herpes zoster and postherpetic neuralgia: the ectopic pacemaker hypothesis. In: Watson CPN, Gershon AA, Oman MN, editors. Herpes zoster: postherpetic neuralgia and other complications. Switzerland: Springer International Publishing, 2017. pp. 157–87.
. Devor M, Amir R, Rappaport ZH. Pathophysiology of trigeminal neuralgia: the ignition hypothesis. Clin J Pain 2002;18:413.
. Devor M, Govrin-Lippmann R, Frank I, Raber P. Proliferation of primary sensory neurons in adult rat dorsal root ganglion
and the kinetics of retrograde cell loss after sciatic nerve section. Somatosensory Res 1985;3:139–67.
. Devor M, Wall PD, Catalan N. Systemic lidocaine silences ectopic neuroma and DRG discharge without blocking nerve conduction. PAIN 1992;48:261–8.
. Dong H, Fan YH, Wang YY, Wang WT, Hu SJ. Lidocaine suppresses subthreshold oscillations by inhibiting persistent Na(+) current in injured dorsal root ganglion
neurons. Physiol Res 2008;57:639–45.
. Du X, Hao H, Yang Y, Huang S, Wang C, Gigout S, Ramli R, Li X, Jaworska E, Edwards I, Deuchars J, Yanagawa Y, Qi J, Guan B, Jaffe DB, Zhang H, Gamper N. Local GABAergic signaling within sensory ganglia controls peripheral nociceptive transmission. J Clin Invest 2017;127:1741–56.
. Dunlap K, Fischbach GD. Neurotransmitters decrease the calcium component of sensory neurone action potentials. Nature (London) 1978;276:837–9.
. Finnerup NB, Attal N, Haroutounian S, McNicol E, Baron R, Dworkin RH, Gilron I, Haanpaa M, Hansson P, Jensen TS, Kamerman PR, Lund K, Moore A, Raja SN, Rice AS, Rowbotham M, Sena E, Siddall P, Smith BH, Wallace M. Pharmacotherapy for neuropathic pain
in adults: a systematic review and meta-analysis. Lancet Neurol 2015;14:162–73.
. Gammaitoni AR, Alvarez NA, Galer BS. Safety and tolerability of the lidocaine patch 5%, a targeted peripheral analgesic: a review of the literature. J Clin Pharmacol 2003;43:111–17.
. Goins WF, Cohen JB, Glorioso JC. Gene therapy for the treatment of chronic peripheral nervous system pain. Neurobiol Dis 2012;48:255–70.
. Hasson E, Slovatizky Y, Shimoni Y, Falk H, Panet A, Mitrani E. Solid tissues can be manipulated ex vivo and used as vehicles for gene therapy. J Gene Med 2005;7:926–35.
. Ilfeld BM. Continuous peripheral nerve blocks: a review of the published evidence. Anesth Analg 2011;113:904–25.
. Jack JJB, Noble D, Tsien RW. Electric current flow in excitable cells. Oxford: Clarendon Press, 1983:518.
. Kajander KC, Wakisaka S, Bennett GJ. Spontaneous discharge originates in the dorsal root ganglion
at the onset of a painful peripheral neuropathy in the rat. Neurosci Lett 1992;138:225–8.
. Kayahara T, Takimoto T, Sakashita S. Synaptic junctions in the cat spinal ganglion. Brain Res 1981;219:277–90.
. Kirillova I, Teliban A, Gorodetskaya N, Grossmann L, Bartsch F, Rausch VH, Struck M, Tode J, Baron R, Janig W. Effect of local and intravenous lidocaine on ongoing activity in injured afferent nerve fibers. PAIN 2011;152:1562–71.
. Kirk E. Impulses in dorsal spinal nerve rootlets in cats and rabbits arising from dorsal root ganglia isolated from the periphery. J Comp Neurol 1974;2:165–76.
. Kovalsky Y, Amir R, Devor M. Subthreshold oscillations facilitate neuropathic spike discharge by overcoming membrane accommodation. Exp Neurol 2008;210:194–206.
. Kuslich S, Ulstro C, Michael C. The tissue origin of low back pain and sciatica. Orth Clin North Amer 1991;22:181–7.
. Liem L, van Dongen E, Huygen FJ, Staats P, Kramer J. The dorsal root ganglion
as a therapeutic target for chronic pain. Reg Anesth Pain Med 2016;41:511–19.
. Liu CN, Wall PD, Ben-Dor E, Michaelis M, Amir R, Devor M. Tactile allodynia in the absence of C-fiber activation: altered firing properties of DRG neurons following spinal nerve injury. PAIN 2000;85:503–21.
. Mao J, Chen LL. Systemic lidocaine for neuropathic pain
relief. PAIN 2000;87:7–17.
. Matzner O, Devor M. Method for distinguishing between drug action on impulse propagation versus impulse generation. J Neurosci Methods 1993;49:23–31.
. McLachlan EM, Janig W, Devor M, Michaelis M. Peripheral nerve injury triggers noradrenergic sprouting within dorsal root ganglia. Nature 1993;363:543–6.
. Ninian Bruce A. Vaso-dilator axon-reflexes. Q J Exp Physiol 1913;6:339–54.
. Nordin M, Balague' F. Biomechanics and ergonimics in disk herniation accompanied by sciatica. In: Weibstein JN, Gordon SL, editors. Low back pain: a scientific and clinical overview. Rosemont: American Academy of Orthopaedic Surgeons, 1996. pp. 23–48.
. Nordin M, Nystrom B, Wallin U, Hagbarth KE. Ectopic sensory discharges and paresthesiae in patients with disorders of peripheral nerves, dorsal roots and dorsal columns. PAIN 1984;20:231–45.
. Nystrom B, Hagbarth K. Microelectrode recordings from transected nerves in amputees with phantom limb pain. Neurosci Lett 1981;27:211–16.
. Quiroga RQ, Nadasdy Z, Ben-Shaul Y. Unsupervised spike detection and sorting with wavelets and superparamagnetic clustering. Neural Comput 2004;16:1661–87.
. Robertson B. Characteristics of GABA-activated chloride channels in mammalian dorsal root ganglion
neurones. J Physiol 1989;411:285–300.
. Strichartz GR, Zhou Z, Sinnott C, Khodorova A. Therapeutic concentrations of local anaesthetics unveil the potential role of sodium channels in neuropathic pain
. Novartis Found Symp 2002;241:189–201; discussion 202–185, 226–132.
. Sukhotinsky I, Ben-Dor E, Raber P, Devor M. Key role of the dorsal root ganglion
in neuropathic tactile hypersensibility. Eur J Pain 2004;8:135–43.
. ten Hoope W, Hollmann MW, de Bruin K, Verberne HJ, Verkerk AO, Tan HL, Verhamme C, Horn J, Rigaud M, Picardi S, Lirk P. Pharmacodynamics and pharmacokinetics of lidocaine in a rodent model of diabetic neuropathy. Anesthesiology 2018;128:609–19.
. Tremont-Lukats IW, Challapalli V, McNicol ED, Lau J, Carr DB. Systemic administration of local anesthetics to relieve neuropathic pain
: a systematic review and meta-analysis. Anesth Analg 2005;101:1738–49.
. Trigo FF, Marty A, Stell BM. Axonal GABAA receptors. Eur J Neurosci 2008;28:841–8.
. Vaso A, Adahan HM, Gjika A, Zahaj S, Zhurda T, Vyshka G, Devor M. Peripheral nervous system origin of phantom limb pain. PAIN 2014;155:1384–91.
. Wall PD, Devor M. Sensory afferent impulses originate from dorsal root ganglia as well as from the periphery in normal and nerve injured rats. PAIN 1983;17:321–39.
. Wallace MS, Dyck JB, Rossi SS, Yaksh TL. Computer-controlled lidocaine infusion for the evaluation of neuropathic pain
after peripheral nerve injury. PAIN 1996;66:69–77.
. Xiao WH, Bennett GJ. C-fiber spontaneous discharge evoked by chronic inflammation is suppressed by a long-term infusion of lidocaine yielding nanogram per milliliter plasma levels. PAIN 2008;137:218–28.
. Yang RH, Xing JL, Duan JH, Hu SJ. Effects of gabapentin on spontaneous discharges and subthreshold membrane potential oscillation of type A neurons in injured DRG. PAIN 2005;116:187–93.
Keywords:© 2018 International Association for the Study of Pain
Dorsal root ganglion; Ectopic discharge; Neuropathic pain; Safety factor; Selective block