Differential Nerve Block: Direct Measurements on Individual Myelinated and Unmyelinated Dorsal Root Axons

This study protocol was approved by the Stanford University Administrative Panel on Laboratory Animal Care.

Adult male Sprague-Dawley rats (weight 200–350 g, n = 23) were anesthetized using 1–2% enflurane and 70% N²O in oxygen. After induction, the trachea was exposed, cannulated, and connected to a semiclosed circle anesthesia circuit for continued maintenance of anesthesia. End-tidal carbon dioxide and respiratory rates were continuously monitored. With a surgical microscope, a laminectomy extending from the T12 thoracic vertebra to the L6 lumbar vertebra followed by a long dural incision was used to expose the dorsal surface of the spinal cord and the dorsal roots. Individual lumbar dorsal roots were dissected from adjacent roots and cut proximally near their point of entry into the spinal cord and distally near their exit from the spinal canal. Each isolated root was transferred immediately to an artificial cerebrospinal fluid (aCSF) solution with the following composition (mM): NaCL 123, KCl 5, CaCl²2, MgSO⁴1.3, NaHCO³26, NaH²PO⁴1.2, and glucose 10, bubbled continuously with 95% O²/5% CO sub 2. As measured in the perfusion chamber at 37 degrees C +/-0.3 degree C, aCSF pH was in the range of 7.35–7.40 and P^CO2was 35–40 mmHg.

Each root was placed in a Teflon perfusion chamber (internal volume 1.5 ml) for stimulation and recording Figure 1. The ends of each root were trimmed, and the perineurium was left intact. The proximal end of the root (2–3 mm) was led through a slotted partition into a recording compartment, while a suction-type stimulating electrode was attached to the distal end within the perfusion chamber. The slot in the partition was sealed with silicone-based vacuum grease, and the aCSF in that compartment was covered with mineral oil. Using standard single-fiber microdissection techniques, the proximal end of the root was progressively divided into small fascicles, each typically containing one to three electrophysiologically distinguishable axons. Action potentials were recorded using chlorided silver wire electrodes.

Supramaximal (1.5 x threshold) constant voltage stimuli (0.2 ms in duration at 0.3 Hz) were delivered to the intact distal end of the isolated root while single-fiber action potentials in a proximal fascicle were amplified, displayed on a digital storage oscilloscope, and recorded for computer analysis (Figure 2). The stimulus parameters were selected to minimize activity-dependent changes in axon properties. The length of the root from the tip of the stimulating electrode to the recording electrode was measured to calculate conduction velocity from single axon conduction latency measurements. Latencies were measured with an adjusted resolution ranging from 0.1 micro second for the shortest latencies to 5 micro second for the longest. The length of the root exposed to the perfusate was measured between the tip of the stimulating electrode and the partition separating the recording and perfusion compartments (average length+/-SD 19+/-3.8 mm).

The exposed portion of the root was continuously superfused with aCSF at 37 degrees+/-0.3 degree C and a flow rate of approximately 8 ml/min. In pilot experiments, single axon recordings in this preparation were stable (< 5% change in conduction velocity or action potential amplitude) for as long as 6 h.

Each single fiber preparation was tested for stability under control conditions for 15–30 min. After control measurements, the aCSF perfusate was switched to one containing a low concentration of lidocaine (150 or 260 micro Meter; Astra, Westboro, MA), selected from pilot study data to be at or below the EC⁵⁰level. Latency was measured at 1-min intervals until the unit failed or for at least 15 min to ensure steady-state conditions. Maximum conduction velocity slowing typically occurred within 5–10 min of lidocaine exposure. Axons unblocked at the lowest concentration of lidocaine were tested at higher lidocaine concentrations (e.g., 260 then 540 micro Meter) for additional 15-min intervals. Each dorsal root was exposed to this sequence of lidocaine concentrations only once. Data were accepted from those axons demonstrating complete recovery (latency and amplitude within 5% of control values) after return to control aCSF.

In a separate series of experiments, cervical portions of vagus nerves were removed from 14 adult male rats using the anesthetic technique previously described (see methods, dorsal root isolation). Similarly, in vitro perfusion, stimulation, single-fiber recording, and drug exposure were accomplished in the manner previously described for the dorsal root experiments.

The vagus nerve was selected because of the availability of comparable data in other species and because it has been used in most previous single fiber and whole nerve studies.

Fisher's exact test was used to determine the statistical significance of differences in the incidence of conduction block between dorsal root and vagal axons, between myelinated (CV > 3 m/s) and unmyelinated (CV < 1.4 m/s) axons, and between long (> 20 mm) and short (< 15 mm) exposure lengths. Multiple comparisons were evaluated using analysis of variance and Bonferroni's t test. Estimates of EC⁵⁰were calculated from a least-squares fit to a sigmoid function bounded by 0% and 100%. EC⁵⁰95% confidence intervals were estimated from this model.

Data were obtained from 77 dorsal root axons in 34 dorsal roots and 41 vagal axons in 21 vagus nerves. Conduction velocities for dorsal root axons ranged between 25.3 and 0.53 m/s, whereas conduction velocities for vagal axons ranged between 31.4 and 0.76 m/s. Axons were divided into three categories on the basis of conduction velocity. Axons with conduction velocities greater than 3 m/s were considered to be myelinated, whereas those with conduction velocities less than 1.4 m/s were considered to be unmyelinated. A third group of axons was comprised of those with intermediate conduction velocities and probably consisted of both large unmyelinated and small myelinated fibers.

To ensure that all measurements were made under steady-state conditions, the times to achieve a stable reduction in conduction velocity (increase in conduction latency) during continuous exposure to sub-blocking concentrations of lidocaine (150 and 260 micro Meter) were measured. Data were obtained from 18 dorsal root axons and 26 vagal axons, representing both myelinated and unmyelinated fibers. Steady-state effects were uniformly obtained within the first 10 min of drug exposure at either concentration (Figure 3). This observation is consistent with the absence of significant diffusion barriers in this preparation.

Susceptibility to local anesthetic-induced conduction block was measured at three concentrations of lidocaine selected, on the basis of a pilot study, to bracket the concentration of lidocaine estimated to produce a 50% incidence of conduction block (EC⁵⁰). As shown in Figure 4, at the highest concentration of lidocaine (520 micro Meter), 88% of unmyelinated and 100% of myelinated dorsal root axons were blocked, whereas at the lowest concentration of lidocaine (150 micro Meter), only 15% of unmyelinated and 29% of myelinated dorsal root axons were blocked. EC⁵⁰concentrations of lidocaine for each of the previously described conduction velocity groups were estimated from a nonlinear regression model least-squares curve. By this method, the EC⁵⁰lidocaine concentrations for both myelinated and unmyelinated axons were similar: 232 and 228 micro Meter, respectively. Axons in the intermediate conduction velocity group appeared to be slightly more sensitive to lidocaine with an estimated EC⁵⁰of 192 micro Meter, although the 95% confidence intervals for all three groups overlapped. EC⁵⁰concentrations were estimated by simple interpolation between the lower lidocaine concentrations; however, this method did not produce results significantly different from the linear regression model. Within axon types, there was no obvious correlation between conduction velocity (axon diameter) and susceptibility to conduction block. For this analysis, the myelinated and unmyelinated axon groups were subdivided by conduction velocity into quartiles. For myelinated axons, the incidence of conduction block at 260 micro Meter lidocaine in the fastest quartile (75% block; CV 20.29+/-4.08 m/s; n = 8) was not significantly different from that in the slowest quartile (62.5% block; CV 3.59+/-0.25 m/s; n = 8). A similar analysis of conduction block in unmyelinated fibers yielded comparable results with 66.7% of the fastest axons blocked (CV 1.21+/-0.08 m/s; n = 6) and 83.3% of the slowest axons blocked (CV 0.75+/-0.13 m/s; n = 6).

To assess the relative lidocaine sensitivities of dorsal root and peripheral nerve axons, vagus nerve axons were studied in a separate series of experiments. At a lidocaine concentration of 260 micro Meter, a smaller proportion of vagal axons was blocked in each conduction velocity group (Figure 5). However, this difference was statistically significant only for axons in the myelinated axon category (P < 0.05; Fisher's exact test). The EC⁵⁰blocking concentration of lidocaine for unmyelinated vagal axons was approximated by simple interpolation between the two lidocaine concentrations studied (260 and 520 micro Meter). The EC⁵⁰estimated by this technique was 285 micro Meter. The estimate of EC⁵⁰for myelinated axons (less or equal to 345 micro Meter) could not be determined with the same degree of certainty because all 12 axons studied were blocked at 520 micro Meter lidocaine.

For those axons in which 260 micro Meter lidocaine produced conduction block, the mean time interval until block occurred was 3.16 min for dorsal root axons (n = 33) and 6.52 min for vagal axons (n = 14). This difference was not statistically significant. Time to conduction block in myelinated dorsal root axons (2.98+/-0.83 min; n = 12) did not differ significantly from that measured in unmyelinated dorsal root axons (2.81+/-0.74 min; n = 14). Similarly, at 520 micro Meter lidocaine, times to conduction block in myelinated vagal axons (2.80 +/-0.47 min; n = 10) and unmyelinated vagal axons (2.28+/- 0.45 min; n = 9) were not significantly different.

Sensitivity to the conduction blocking effects of local anesthetics can be estimated from measurements of steady-state conduction velocity during drug exposure, because conduction velocity decreases with decreasing numbers of active sodium channels. Conduction velocity was calculated from measurements of nerve length between stimulating and recording electrodes and the time required to conduct an action potential over that distance (latency). Thus, increases in conduction latency directly correspond to decreases in conduction velocity. Exposure to sub-blocking concentrations of lidocaine increased latency (decreased conduction velocity) in both myelinated and unmyelinated dorsal root axons (Figure 6). Linear regression analysis across all conduction velocity groups revealed a significant direct correlation between control conduction velocity (axon diameter) and sensitivity to the conduction velocity slowing effect of lidocaine at 260 micro Meter (r²= 0.40; P < 0.01) and at 150 micro Meter (r²= 0.26; P < 0.01). At 520 micro Meter lidocaine, only 3 of 64 axons (4.7%) remained unblocked. All three of these lidocaine-resistant axons had conduction velocities less than 1.3 m/s, consistent with unmyelinated fibers. As a group, unmyelinated dorsal root axons were significantly less sensitive to the conduction velocity slowing effects of lidocaine when compared to myelinated dorsal root axons (Figure 7). Within the myelinated and unmyelinated dorsal root axon groups, however, there was no significant correlation between axon diameter and sensitivity to lidocaine as measured by changes in conduction velocity.

In a similar manner, vagal axons were tested at 260 and 520 micro Meter lidocaine. Regression analysis did not reveal any significant correlation between axon diameter and changes in conduction velocity at either lidocaine concentration (r²= 0.04 and 0.05, respectively). At 260 micro Meter lidocaine, vagus nerve axons in the intermediate and myelinated conduction velocity groups were slightly less sensitive to the sodium channel blocking effects of lidocaine than were their dorsal root counterparts (Figure 7). This difference was statistically significant only for the intermediate conduction velocity group. At 520 micro Meter lidocaine, only 4 of 41 vagal axons (9.8%) remained unblocked. Two of these axons had conduction velocities less than 1.4 m/s, whereas the remaining two had conduction velocities of 1.46 and 2.72 m/s. As a group, unmyelinated vagal axons appeared to be less sensitive to the conduction velocity slowing effects of lidocaine when compared with myelinated vagal axons; however, this difference was not statistically significant.

The length of dorsal root exposed to lidocaine ranged from 11 to 25 mm. This variation was the result of surgical exposure at the time of harvest and the extent of subsequent trimming of the cut nerve ends. It has been demonstrated in studies of peripheral nerve myelinated axons that susceptibility to local anesthetic-induced conduction block is dependent on the length of nerve exposed to local anesthetic. [10]To examine this effect in dorsal roots, axons with exposed lengths of 15 mm or less were compared with axons having exposed lengths of 20 mm or more. These short and long axon groups had average exposure lengths of 13.5 and 22.4 mm, respectively. A higher percentage of short axons were blocked (Figure 8) at each of the three lidocaine concentrations tested, although these differences were not statistically significant. The effects of exposure length within myelinated or unmyelinated axon groups were similar (Figure 8). Vagal nerve lengths only ranged between 17 and 25 mm (average 20 mm), thus separation into categories by length was not practical.

Although the "size principle" is a clinically appealing explanation for differential nerve block, experimental studies on single axon preparations of mammalian peripheral nerve have failed to demonstrate any correlations between axon diameter (conduction velocity) and susceptibility to local anesthetic conduction block. [5,6]Earlier studies supporting the size principle were almost all based on measurements of CAP amplitude. [8]Because reductions in CAP amplitude produced by conduction block could not be differentiated from those produced by temporal dispersion as a result of altered conduction velocities, conclusions from these studies regarding differential block may be invalid.

Staiman and Seeman [4]studied nine axons in frog sciatic nerve and reported an inverse correlation between axon diameter and lidocaine sensitivity. These data are in conflict with the results of the current study and with the earlier studies of Fink and Cairns. [5,6]This discrepancy may be explained by the major differences between the experimental preparations including species and axon size differences. [5]In the current study, sensitivity of individual dorsal root axons to lidocaine was assessed in two ways. First, the incidence of steady-state conduction block at three lidocaine concentrations was determined. At lidocaine concentrations of 150 and 520 micro Meter, unmyelinated dorsal root axons appeared to be slightly less susceptible than myelinated dorsal root axons to conduction block, although the estimated EC⁵⁰lidocaine concentrations were nearly identical (228 vs. 232 micro Meter). Within axon types, the frequency of conduction block did not correlate with axon diameter (conduction velocity). This is consistent with the results in peripheral nerve studies of Fink and Cairns, [5]who reported that there was no correlation between fiber size and lidocaine blocking concentration within either myelinated or unmyelinated rabbit vagal axon groups.

The second method was based on the principle that the axonal propagation of action potentials depends on the activation of an adequate number of sodium channels to depolarize the next segment of nerve membrane to threshold. Thus, partial blockade of sodium channels will result in both an increase in threshold and a decrease in the magnitude of depolarizing current. These changes decrease the effective spread of depolarization and, as a result, action potential propagation is slowed (conduction velocity decreases) as the number of blocked sodium channels increases. Eventually, conduction fails when the depolarizing current preceding the propagated action potential is insufficient to reach threshold in the adjacent patch of nerve membrane. By comparing the conduction velocity slowing effects of lidocaine in all dorsal root axons, there appeared to be a direct correlation between axon diameter and lidocaine sensitivity. However, the myelinated axon group was significantly more sensitive than the unmyelinated axon group to the conduction velocity slowing effects of lidocaine, which probably accounts for the strength of the overall correlation. The within-group correlations between conduction velocity and lidocaine sensitivity were not significant. However, interpretation of these results is somewhat confounded by the uncertainty in separating myelinated axons from unmyelinated axons on the basis of conduction velocity. For example, large unmyelinated axons may exist with conduction velocities well above the 1.4-m/s limit used in the current study, and conversely, small myelinated fibers are likely to exist with conduction velocities below the 3-m/s limit. By necessity, axons representing these two extremes (< 3 and > 1.4 m/s) could be categorized only in the intermediate conduction velocity group. Not surprisingly, given the small sample size, elimination of the slowest members of the myelinated and fastest members of the unmyelinated axon groups will diminish the ability to detect a correlation between conduction velocity within these groups and any other variable.

The observation that unmyelinated fibers are less sensitive to lidocaine is consistent with the predictions of a mathematical model of nerve conduction in the presence of a local anesthetic. [13]In that model, decremental conduction was found to be greater in myelinated axons than in unmyelinated axons. The authors speculate that, in myelinated axons, "the interposition of the passive internodal segment leads to a loss of energy which aids in the development of decremental conduction." As a result of decremental conduction, action potential amplitude decreases as a function of propagation distance. Eventually, conduction fails when the depolarizing current becomes insufficient to reach threshold in an adjacent membrane segment. Thus, their model predicts that unmyelinated axons will be less susceptible to conduction velocity slowing and conduction block.

In the peripheral nerve studies of Fink and Cairns, [6,14]unmyelinated axons were shown to be substantially less sensitive to lidocaine than myelinated axons. For example, they reported an average lidocaine blocking concentration of 630 micro Meter for unmyelinated axons and 430 micro Meter for myelinated axons in rabbit vagus nerve. [5]These laboratory observations, similar to the results of the current study, are contrary to clinical observations of differential sensory nerve block, whereby sensory modalities subserved by unmyelinated axons (e.g., temperature) are first to be blocked, are associated with the greatest extent of block, and are the last to recover. To understand this apparent discrepancy between clinical and laboratory observations, it may be useful to separate the temporal aspects of differential block, which depend on differences in diffusion path length and diffusion barriers, from those aspects dependent on true differences in axon sensitivity. Although favorable access by local anesthetics to unmyelinated axons may result in the appearance of increased sensitivity to local anesthetic effects (early onset and spread of block), actual differences in nerve membrane sensitivity are not necessary to explain these observations. In clinical practice, the temporal aspects of nerve block are the most easily detected and quantified. With the exception of continuous spinal and epidural anesthetic administration, steady-state conditions during regional nerve blockade are rarely achieved. Thus, the temporal aspects of differential nerve block tend to dominate our clinical perceptions. In addition, because local anesthetics prefer to bind to sodium channels in the open and inactive states, [15,16]as opposed to resting state channels, the appearance of differential block may depend in part on intrinsic differences in background nerve activity. In the presence of lidocaine, the addition of frequency-dependent (phasic) block to the resting, tonic block of sodium channels may result in the appearance of differential block in the active axon groups. For example, the sodium channels in large and small axons may be equally susceptible to block by lidocaine. However, if the magnitude of spontaneous background activity is substantially greater in a group of small axons, these axons would incorrectly appear to possess sodium channels with a higher sensitivity to lidocaine.

As proposed by Fink, [7]the clinical appearance of differential small fiber conduction block may be strongly dependent on the length of axon exposed to local anesthetic. Thus, in spinal anesthesia, axons in the long intrathecal segments of lumbar roots are more susceptible to conduction block than those in the significantly shorter cervical and thoracic roots. This increased sensitivity is likely the result of decremental conduction [10]combined with an increased probability of blocking three consecutive nodes at local anesthetic concentrations near the threshold for conduction block: Tasaki [17]demonstrated that action potentials in large myelinated amphibian axons could breach a two-node block but never three blocked nodes. In the shorter spinal roots, large diameter axons will have fewer nodes exposed to local anesthetic and, therefore, may be more resistant to conduction block than their smaller diameter counterparts, wherein more nodes would be exposed to anesthetic. In the current study, there was no significant effect of exposure length on the incidence of conduction block in either myelinated or unmyelinated dorsal root axons. Raymond et al. [10]defined "critical length" as the length of axon exposed to a specific concentration of lidocaine through which 50% of the action potentials fail to propagate. In six frog sciatic nerve myelinated axons where serial measurements of critical length were made, they demonstrated that at exposure lengths greater than 10 mm, anesthetic requirements changed minimally by comparison with anesthetic requirements at critical lengths less than 10 mm. It is possible that our shortest fibers (11–15 mm exposed length) were too long to manifest any differential block effects.

Alternatively, differential block among sensory axons may not depend on differential conduction block but on differential effects on perception. [18]Using microneurographic techniques, MacKenzie et al. demonstrated that the perception of local skin cooling in human volunteers could be prevented with concentrations of lidocaine insufficient to block low-frequency conduction in the A-sigma fibers known to mediate this sensation. Thus, perception of a specific sensation may depend primarily on the transmission of meaningful patterns of pulse-coded data. Local anesthetics, by transiently affecting postimpulse membrane excitability and through use-dependent effects on sodium channels [19]may significantly disrupt firing patterns in sensory nerves, thereby preventing perception without blocking conduction.

Finally, the clinical appearance of differential nerve block may be the result of local anesthetic effects at sites proximal to the dorsal root axon. Local anesthetics are known to have many effects in addition to their well studied action at sodium channels (see review by Butterworth and Strichartz [20]). These effects include presynaptic inhibition of calcium channels and neurotransmitter release and effects on membrane associated enzymes (e.g., sodium-potassium ATPase-coupled ion pumps) and second messenger systems (e.g., adenylate cyclase). We recently demonstrated that subanesthetic concentrations of lidocaine (3.6–36 micro Meter) were able to selectively inhibit a C-fiber mediated nociceptive spinal cord potential. [21]This antinociceptive effect of lidocaine was not mediated by opioid receptors or by fractional blockade of voltage activated sodium channels. This selective effect of lidocaine was likely mediated at sites within the spinal cord, independent of axonal sodium channels.

In summary, we presented the first measurements of local anesthetic sensitivity in individual myelinated and unmyelinated dorsal root axons. Furthermore, we demonstrated that, compared to unmyelinated axons, myelinated dorsal root axons are significantly more sensitive to the steady-state sodium channel blocking effects of lidocaine. However, within the myelinated and unmyelinated axon groups, no size-related differences in lidocaine sensitivity were detected. Comparing dorsal root axons with vagus axons revealed that dorsal root axons are intrinsically more sensitive to lidocaine than their peripheral nerve counterparts. We propose that differential nerve block may depend on multiple factors indirectly related to axon diameter, including myelination, specific sensory function, level of resting activity (phasic block), and the extra-axonal effects of local anesthetics.