Effects of temperature on the excitability properties ... - Semantic Scholar

Brain (2001), 124, 816–825

Effects of temperature on the excitability properties of human motor axons Matthew C. Kiernan,1 Katia Cikurel2 and Hugh Bostock1 1Sobell

Department of Neurophysiology, Institute of Neurology and 2Department of Neurology, Royal London Hospital, London, UK

Correspondence to: Professor Hugh Bostock, Sobell Department of Neurophysiology, Institute of Neurology, Queen Square, London WC1N 3BG, UK E-mail: [email protected]

Summary The effects of temperature on parameters of motor nerve excitability were investigated in 10 healthy human subjects. The median nerve was stimulated at the wrist and compound muscle action potentials were recorded from the abductor pollicis brevis. Multiple excitability measures were recorded: stimulus–response curves, the strength– duration time constant (τSD), threshold electrotonus, a current–threshold relationship and the recovery of excitability following supramaximal activation. Recordings were made at wrist temperatures of 35, 32 and 29°C by immersing the arm proximal to the wrist in a waterbath. Cooling increased the relative refractory period by 7.8% per degree C (P < 0.0001), slowed the accommodation to depolarizing currents by 4.0% per degree C (P < 0.0001) and increased τSD by 2.6% per degree C (P < 0.01), but most other excitability parameters were

not affected significantly. The effects of temperature on threshold electrotonus were investigated further in separate studies on two subjects over the range 28– 36°C and found to be complex. Whereas the rate of accommodation to depolarizing current was closely related to instantaneous temperature, the threshold increase induced by hyperpolarizing current was most sensitive to changes in temperature, probably because warming the nerve causes a transient hyperpolarization by accelerating the electrogenic sodium pump. Consequently, it may be preferable to make allowances for differences in skin temperature when testing patients for abnormal excitability parameters, rather than to change the temperature to a standard value. For most excitability parameters, however, temperature control is not as important as it is for conduction velocity measurements.

Keywords: excitability; motor axons; temperature Abbreviations: CMAP ⫽ compound muscle action potential; ETD ⫽ equivalent temperature deviation; IH ⫽ inward rectification; I/V ⫽ current/threshold relationship; TEd ⫽ depolarizing threshold electrotonus; TEh ⫽ hyperpolarizing threshold electrotonus; τSD ⫽ strength–duration time constant.

Introduction The effects of temperature on standard measures of nerve conduction are well established. The conduction velocity increases by ~5% per degree C as the temperature of the nerve increases from 29 to 38°C (Johnson and Olsen, 1960; De Jesus et al., 1973; Lowitzsch et al., 1977), whereas the amplitude of nerve and muscle action potentials decreases (Buchthal and Rosenfalck, 1966; Bolton et al., 1981; Lang and Puusa, 1981). It is therefore customary to ensure that skin temperature is within defined limits before clinical nerve conduction measurements. In contrast, the effects of temperature on nerve excitability measures, which are being introduced into the clinic to provide complementary information about nerve dysfunction (Bostock et al., 1998; Kiernan et al., 2000), have been much less studied. Burke and colleagues studied the effects of temperature on several © Oxford University Press 2001

excitability measures of human cutaneous afferents (Burke et al., 1999), although the parameters selected did not include the accommodative processes accessible to measurement using threshold electrotonus. Comparable information on any excitability parameter is not available for human motor axons. Kiernan and colleagues noted that, of 24 excitability parameters measured in 29 subjects, only relative refractory period was significantly related to temperature (Kiernan et al., 2000). However, the range of temperatures in that study was limited to the spontaneous variations between normal subjects. Since patients, especially those with wasting or autonomic disorders, may exhibit much wider variations in temperature, it was important to establish how excitability parameters for individual subjects depend on temperature. We have therefore repeated the sequence of excitability

Temperature and nerve excitability tests described previously (Kiernan et al., 2000) at temperatures from 29 to 35°C, to encompass the range typically encountered in routine electrodiagnostic clinics. Nerve excitability parameters are strongly dependent on membrane potential (Kiernan and Bostock, 2000), but no clear relationship of membrane potential to temperature was evident from these tests. In two subjects we therefore made repeated measurements of threshold electrotonus, which is particularly sensitive to membrane potential, as temperature was raised and lowered over a slightly wider range. The results provide evidence that membrane potential is more sensitive to changes in temperature than to steady-state temperature and suggest that the strategy of warming a limb before commencing a recording may be inappropriate for nerve excitability studies.

Methods Studies were performed on 10 healthy volunteers of both sexes (aged 24–54 years). All subjects gave informed consent and the study was approved by the Joint Research Ethics Committee of the National Hospital for Neurology and Neurosurgery and the Institute of Neurology, London. Two series of recordings were made. The first used a recently described protocol designed to measure a number of different nerve excitability parameters rapidly (Kiernan et al., 2000), whereas in the second series repeated measurements solely of threshold electrotonus were made at short intervals. The excitability of median nerve motor axons was tested while compound muscle action potentials (CMAPs) were being recorded from thenar muscles, using surface electrodes over the abductor pollicis brevis, with the active electrode at the motor point and the reference on the proximal phalanx. The EMG signals were amplified (gain 1000, bandwidth 1.6 Hz to 2 kHz) and digitized by computer (486 PC) with data acquisition (DT2812; Data Translation, Marlboro, Mass., USA), using a sampling rate of 10 kHz. Stimulus waveforms generated by the computer were converted to current with a purpose-built isolated linear bipolar constant current stimulator (maximum output ⫾50 mA). The stimulus currents were applied via nonpolarizable electrodes (Red Dot; 3M Health Care, Borken, Germany) with the active electrode over the nerve at the wrist and the reference electrode ~10 cm proximal over muscle. Stimulation and recording were controlled by new software, written in BASIC [QTRAC version 4.3 (Institute of Neurology, London) with multiple excitability protocol TRONDXM]. Test current pulses of 0.2 or 1 ms were applied regularly at 0.8 s intervals and were combined with suprathreshold conditioning stimuli or subthreshold polarizing currents as required. The amplitude of the CMAP was measured from baseline to negative peak. For all tracking studies, the target CMAP was set to 40% of the peak response. For recording multiple excitability properties, stimulus– response curves were first recorded separately for test stimuli

817

of durations 0.2 and 1 ms. The stimuli were increased in 6% steps, with two responses averaged for each step, until three averages were considered maximal. The stimulus–response data were used for several purposes. First, the 1 ms peak response was used to set the target submaximal response (40% of peak) for threshold tracking. Secondly, the slope of the 1 ms stimulus–response curve was used in conjunction with the tracking error (deviation from the target) to optimize the subsequent threshold tracking. Finally, when the data were analysed, the ratio between the 0.2 ms and 1 ms stimuli required to evoke the same responses were used to estimate the strength–duration time constants and rheobases of axons of different threshold. The current–threshold relationship was tested with 1 ms pulses at the end of subthreshold polarizing currents lasting 200 ms. The polarizing current was altered in a ramp fashion from ⫹50% (depolarizing) to –100% (hyperpolarizing) of the control threshold in 10% steps. As with the conventional threshold electrotonus protocol, stimuli with conditioning currents were alternated with test stimuli alone and each stimulus combination was repeated until three valid threshold estimates had been obtained. Prolonged subthreshold currents were used to alter the potential difference across the internodal axonal membrane, a process referred to as electrotonus. The changes in threshold associated with electrotonus normally have a similar time course to the changes in membrane potential and are known as threshold electrotonus (Bostock and Baker, 1988). In the present protocol, test stimuli of 1 ms duration were used to produce the target CMAP (40% of maximal) and threshold tracking was used to record the changes in threshold induced by subthreshold polarizing currents, 100 ms in duration, set to be ⫹40% (depolarizing) and –40% (hyperpolarizing) of the control threshold current. The three stimulus combinations were tested in turn: test stimulus alone (to measure the control threshold current); test stimulus plus depolarizing conditioning current; and test stimulus plus hyperpolarizing conditioning current. Threshold was tested at 26 time points (maximum separation 10 ms) before, during and after the 100 ms conditioning currents. Each stimulus combination was repeated until three valid threshold estimates had been recorded, as judged by the response being within 15% of the target response, or alternate responses being either side of the target. The final part of the protocol recorded the recovery of excitability after a supramaximal conditioning stimulus. These changes were recorded at 18 conditioning–test intervals, decreasing from 200 to 2 ms in approximately geometric progression. Three stimulus combinations were tested in turn: (i) unconditioned test stimulus (duration 1 ms) tracking the control threshold; (ii) supramaximal conditioning stimulus (duration 1 ms) alone; and (iii) conditioning plus test stimuli. The response to (ii) was subtracted on-line from the response to (iii) before the test CMAP was measured, so that the conditioning maximal CMAP did not contaminate the measured response when the conditioning–test interval

818

M. C. Kiernan et al.

was short. Each stimulus combination was repeated until four valid threshold estimates had been obtained.

Temperature control In all studies, skin temperature was monitored close to the stimulation site. Temperature was varied by immersing the arm in a water-bath. The region of the arm immersed went from above the elbow to a point just proximal the wrist such that the hand and wrist were kept outside of the waterbath, to limit the temperature gradient between skin and nerve and to limit temperature changes in the neuromuscular junction and muscle. However, the changes in latency and the recorded CMAP waveform showed that the distal nerve and muscle were also affected by the temperature changes. In the first series of recordings, complete sequences of multiple excitability measurements were made at three temperatures, close to 29, 32 and 35°C, in each of 10 subjects. An average of 35 min was allowed between recordings for the temperature to stabilize at the next value, and the ordering of the three temperatures was randomized between subjects. Both cooling and heating were deliberately slow, so that skin temperature would reflect neural temperature more accurately. Despite these precautions, skin temperature was inevitably lower than core temperature and therefore provides only an approximate index of relative temperature rather than absolute nerve temperature. In the second series, repeated recordings of threshold electrotonus only were made in two subjects as the temperature was raised and lowered over the range 28–36°C.

Data analysis Values for the following variables were abstracted from each recording for numerical analysis. The variables correspond to those shown in Table 1. Latency (ms) was measured from 1 ms stimulus onset to peak amplitude, averaged over the whole recording period. Peak CMAP (mV) was the average of the three largest responses on each stimulus–response curve. Threshold (mA) was the 1 ms stimulus required to elicit a 50% maximal CMAP. SDTC (ms) was the strength– duration time constant (τSD) estimated, as described above, for a 40% maximal CMAP and rheobase (mA) was the corresponding rheobase. The dimensionless variable stimulus–response slope was estimated from the normalized stimulus–response relationship (Fig. 1B) as the stimulus evoking a 75% maximal response minus that evoking a 25% maximal response, divided by that evoking a 50% maximal response. The resting I/V slope was the dimensionless slope of the current–threshold relationship shown in Fig. 1C, calculated from the polarizing currents between –10% and ⫹10% of resting threshold; the minimal I/V slope was calculated by fitting a straight line to each three adjacent points in turn; and the hyperpolarizing I/V slope was derived from the most hyperpolarized three points. RRP (ms) was

the relative refractory period, calculated from the recovery cycle data in Fig. 1F as the first intercept on the x-axis; superexcitability (%) was measured as the minimum mean of three adjacent points, and the late subexcitability (%) as the maximum mean of three adjacent points after 10 ms. The remaining 10 excitability parameters were derived from the threshold electrotonus data in Fig. 1E. TEd (peak) was the peak percentage threshold reduction to the depolarizing current, averaged over 20 ms (referred to as S1 by Kuwabara et al., 2000). TEd (10–20 ms), TEd (40–60 ms) and TEd (90–100 ms) were the mean threshold reductions between the specified latencies after the onset of depolarizing current, and TEh (10–20 ms) and TEh (90–100 ms) were the corresponding threshold changes after the onset of hyperpolarizing current. TEd (undershoot) was the peak threshold increase after the end of the depolarizing current and TEh (overshoot) the peak threshold reduction after the end of the hyperpolarizing current, each averaged over 20 ms. S2 accommodation (Kuwabara et al., 2000) was the difference between the peak threshold reduction on depolarization and the plateau value, i.e. TEd (peak) – TEd (90–100 ms), and the accommodation half-time was the time from the start of the depolarizing current until the threshold reduction returned to half way between the peak and plateau levels. The temperature sensitivity of each excitability parameter was expressed as the percentage change per degree C. This was obtained for each subject from linear regression of the parameter on temperature. Temperature sensitivity was calculated from a and b in the estimating equation yT ⫽ a ⫹ b(T – 32) as b/a ⫻ 100, where yT is the estimated value of the parameter at skin temperature T°C. Student’s t-test was then used to estimate the probability of the mean temperature sensitivity differing from zero by the amount found. These statistics indicate the extent to which each parameter varies with temperature and the statistical significance of that variation, but do not provide a comparison with the normal variation between subjects. For this purpose we also computed an ‘equivalent temperature deviation’ (ETD) for each parameter by dividing the coefficient of variation of y32 (expressed as a percentage) by the temperature sensitivity defined above. This measure, which has the dimensions of temperature, is the temperature range estimated to cause variation in the parameter similar to that which occurs by chance between normal subjects at 32°C. It therefore provides an indication of how closely temperature has to be matched between two groups for an apparently significant difference in the excitability parameter to arise because of the temperature difference alone. For example, when comparing two groups of 10 subjects using the t-test, the parameter means would appear to differ at the 5% level if their mean temperatures differed by more than 0.94 ETD, even though there was no real difference. If two groups of 100 subjects are compared, the critical temperature difference falls to 0.28 ETD. [In general, if groups of N1 and N2 subjects are compared, the difference in the mean will be significant at the 5% level if it exceeds t0.05 ⫻ α(1/N1 ⫹ 1/N2) ⫻ ETD.]

Temperature and nerve excitability

819

Table 1 Sensitivity of various excitability parameters to temperature in the range 29–35°C (A) Mean value (n ⫽ 10)

Latency (ms) Peak CMAP (mV) Threshold (mA) Rheobase (mA) SDTC (ms) Stimulus⫺response slope Resting I/V slope Minimum I/V slope Hyperpolarizing IV slope RRP (mS) Super-excitability (%) Subexcitability (%) TEd (peak) TEd (10–20ms) (%) TEd (40–60ms) (%) TEd (90–100ms) (%) TEh (10–20ms) (%) TEh (90–100ms) (%) TEd (undershoot) TEh (overshoot) S2 accommodation Accommodation half-time (ms)

~29°C

~32°C

~35°C

(B) Coefficient of variation at 32°C (%)

7.49 9.43 4.22 2.77 0.49 5.01 0.62 0.22 0.31 4.78 ⫺22.10 12.37 69.10 66.72 55.12 39.32 ⫺73.44 ⫺114.81 ⫺23.24 19.87 29.78 51.65

6.96 7.93 4.16 2.76 0.46 4.82 0.61 0.22 0.38 3.47 ⫺25.79 13.74 68.49 67.43 51.99 39.61 ⫺73.86 ⫺120.07 ⫺21.29 17.21 28.88 46.86

6.33 8.03 4.19 2.87 0.42 4.75 0.57 0.22 0.37 2.96 ⫺25.07 14.01 67.07 67.24 48.88 40.67 ⫺73.81 ⫺125.32 ⫺19.27 14.91 26.40 40.66

7 28 23 28 22 23 13 17 18 22 27 44 8 8 10 11 7 16 20 24 14 15

(C) Temperature sensitivity (% change/°C) (mean ⫾ SE)

(D) Equivalent temperature deviation (B/C)°C

⫺2.9 ⫺2.9 ⫺0.04 0.8 ⫺2.6 ⫺0.9 ⫺1.5 0.4 1.3 ⫺7.8 3.3 1.7 ⫺0.5 0.2 ⫺2.1 0.6 0.04 1.4 ⫺3.2 ⫺5.1 ⫺1.8 ⫺4.0

⫺2.6 ⫺9.4 ⬎100 34.9 ⫺8.3 25.2 ⫺9.0 40.0 13.8 ⫺2.8 ⫺8.5 25.8 ⫺17.4 51.5 ⫺4.7 17.6 ⬎100 ⫺11.6 ⫺6.1 ⫺4.6 ⫺7.8 ⫺3.8

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

0.3**** 0.7** 1.5 1.5 0.7** 0.8 0.6* 0.9 0.9 1.1**** 1.8 2.0 0.3 0.3 0.3*** 0.4 0.3 0.5 0.9** 1.5** 1.2 0.4****

(A) Mean values of excitability parameters recorded at temperatures of 29, 32 and 35°C. (B) Coefficient of variation (SD/mean⫻100) determined for a temperature of 32°C. (C) Mean percentage change in parameter per °C rise in temperature, averaged across 10 subjects, ⫾ standard error of the mean. The probability of this occurring by chance alone is represented as follows: *P ⬍ 0.05; **P ⬍ 0.01; ***P ⬍ 0.001; ****P ⬍ 0.0001. (D) The equivalent temperature deviation was obtained by dividing the coefficient of variation of the parameter across subjects by the temperature sensitivity within subjects (i.e. column B/column C); it is a measure of how tightly the temperature of two groups of recordings must be controlled to avoid spurious significant differences (see Methods).

The equivalent temperature deviation figures listed in Table 1 thus indicate the importance of the temperature sensitivities from a practical point of view and how tightly skin temperature has to be controlled before the differences can be ignored.

Results The full sequence of excitability measurements, as described in Methods, was recorded in 10 subjects at 29, 32 and 35°C [actual temperatures (mean ⫾ SD): 28.99 ⫾ 0.27, 31.97 ⫾ 0.12, 34.84 ⫾ 0.25, respectively]. The recordings were averaged across subjects and mean waveforms for the three temperatures are compared in Fig. 1, with the same format of six plots as used by Kiernan and colleagues (Kiernan et al., 2000). Numerical data on the CMAP latencies and amplitudes and 19 excitability parameters derived from the recordings are summarized in Table 1. Variation in six of these parameters with temperature and across subjects is also presented graphically in Fig. 2.

Changes in stimulus–response behaviour, strength–duration time constant and rheobase Absolute stimulus response curves, recorded for test stimuli of durations 0.2 and 1 ms, are plotted in Fig. 1A on double

logarithmic coordinates, while normalized stimulus–response relationships are plotted on linear axes in Fig. 1B. There was a small drop in CMAP amplitude at higher temperatures (see below), but the changes in temperature had no significant effect on the slope of the normalized stimulus–response relationship or on the absolute thresholds (Table 1). The insensitivity of the stimulus–response slope to temperatures enhances the utility of this measure as a screening test for disease states, particularly those that affect axons in a nonuniform manner, such as the demyelinating neuropathies (Meulstee et al., 1997). Strength–duration time constants (τSD) were calculated using the stimulus–response curves recorded for 0.2 and 1 ms test durations (Fig. 1D). There was a significant reduction in τSD with increasing temperature (Table 1 and Fig. 2E). The temperature sensitivity of –2.6 ⫾ 0.7% per degree C is not very different from the 18% increase found on cooling from 32 to ~22°C by Burke and colleagues for human cutaneous afferents (Burke et al., 1999). Data from previous studies have shown that there is considerable inter- and intrasubject variability in measurements of τSD (Mogyoros et al., 1996; Kiernan et al., 2000). Consequently, although the variation of τSD with temperature within subjects is clear, the variation between subjects is such that a temperature discrepancy of 8°C between groups of 10 subjects would be required before

820


Fig. 1 Excitability data at three temperatures, averaged over 10 subjects. (A) Absolute stimulus– response relationships for pulse widths of 1 and 0.2 ms. (B) Normalized stimulus–response relationship (1 ms). (C) Current–threshold relationship. (D) Distribution of strength–duration time constants. (E) Threshold electrotonus. (F) Recovery cycle. Thick lines, 29°C; dashed lines, 32°C; thin lines, 35°C.

it could cause spurious evidence of a difference in τSD (see Equivalent temperature deviation in Table 1 and Methods). The temperature dependence of τSD can therefore usually be ignored. As found for cutaneous afferents (Burke et al., 1999), there was a small increase in rheobase, but this was not statistically significant.

Change in current–threshold relationship and threshold electrotonus with temperature The current–threshold relationship reflects the rectifying properties of the axon, both nodal and internodal (Fig. 1C).

The increase in slope in the top right quadrant indicates outward rectification due to activation of voltage-dependent potassium channels on depolarization, while the smaller increase in slope towards the bottom left indicates inward rectification due to activation of the hyperpolarizationactivated current IH (Kiernan et al., 2000). In Fig. 1C the curves for the temperatures from 29 to 35°C almost superimpose, indicating little change in the rectifying properties. Despite this, statistical analysis indicated a small decrease in resting slope with increasing temperature (Table 1). The resting slope is particularly sensitive to membrane potential (Kiernan and Bostock, 2000), but


821

Fig. 2 Effects of temperature on selected parameters. Circles represent the mean data for the 10 subjects, error bars indicate 95% confidence limits for the mean. Dashed lines indicate 95% confidence limits for an individual measurement. (A) Relative refractory period. (B) Latency. (C) Accommodation half-time. (D) Peak threshold increase after depolarizing current. (E) Strength–duration time constant. (F) Peak threshold decrease after hyperpolarizing current.

other potential-dependent excitability parameters [e.g. TEd (90–100 ms), TEh (90–100 ms), superexcitability] were not significantly temperature-sensitive, so any membrane hyperpolarization on warming must have been very slight. The changes in excitability occurring during and after long-duration (100 ms) subthreshold depolarizing and hyperpolarizing currents are collectively known as threshold electrotonus (Bostock and Baker, 1988). The fast changes in threshold that occur at delays of 0 and 100 ms are due to the rapid changes in potential occurring at the nodes of Ranvier at the onset and offset of the polarizing currents. The slower changes in excitability are caused by slower potential changes occurring passively on the internodal

membrane and by ion channels with slow kinetics, especially slow potassium channels at the nodes (Bostock, 1995). The effects of changing temperature on threshold electrotonus were modest (Fig. 1E). The most significant effects were a slight acceleration of the accommodative ‘sag’ to depolarizing currents on warming [indicated by TEd (40– 60 ms) and the accommodation half-time] and a reduction in the amplitudes of the off-responses [TEd (undershoot) and TEh (overshoot)] (Table 1 and Fig. 2C–F). These effects are all understandable in terms of the acceleration of the gating of slow potassium channels (see Discussion). Although there appears in Fig. 1E to be an effect of temperature on hyperpolarizing electrotonus at 90–100 ms, this was not

822


Fig. 3 Effects of temperature and temperature changes on threshold electrotonus in two subjects. For each subject, recordings are compared (i) at temperature extremes (solid line, cool; dotted line, warm) when temperature was changing slowly, showing differences in the rate of accommodation to depolarizing current and in the undershoot and overshoot after current offset; and (ii) at similar temperatures, while cooling (solid line) and warming (dotted line), showing different responses to hyperpolarization. The letters A and B represent two subjects. Approximate skin temperatures were (solid line/dotted line) (i) 24.4/33.5°C and (ii) 29.1/29.2°C for Subject A, and (i) 26.9/35.5°C and (ii) 28.6/28.2°C for Subject B.

statistically significant (Table 1). However, by making repeated recordings of threshold electrotonus on the same nerve while temperature was altered over a wider range, we found that temperature does have consistent, but complex, effects on hyperpolarizing threshold electrotonus. In separate studies on two subjects using a cooling– warming cycle, we found that the effects of temperature on the depolarizing responses could be almost entirely separated from the effects on the hyperpolarizing responses. The depolarizing responses were closely related to temperature itself, while the hyperpolarizing responses were much more sensitive to the direction of temperature change. Figure 3A(i) compares threshold electrotonus in one subject at 24.4 and 33.5°C. The hyperpolarizing response curves superimpose, but accommodation to depolarizing currents occurred appreciably faster at the higher temperature. A similar difference was found for the second subject in Fig. 3B(i) for the temperatures of 26.9 and 35.5°C. In this experiment,

thresholds were tracked for a further 50 ms after the ends of the polarizing currents to show that recovery was also appreciably delayed at the lower temperature, presumably because this also depended on temperature-dependent channel gating. Panels A(ii) and B(ii) in Fig. 3 illustrate different pairs of recordings from the same two experiments, in which the skin temperature was approximately the same for both members of the pair but the temperature was changing in different directions. For each subject, the depolarizing responses were similar, but the increase in threshold on hyperpolarization was greater when the temperature was increasing than when it was decreasing. Figure 4A shows the plots of the time courses of the changes in temperature, CMAP latency and threshold electrotonus for the experiment shown in Fig. 3A(i) and A(ii). There is a good correspondence between the time course of skin temperature and both latency to the peak of the CMAP and depolarizing electrotonus response 40–60 ms after the start of the polarizing current.


823

Fig. 4 Time relationship of temperature and electrotonus changes. Panels A and B illustrate the same experiments as Fig. 3A and B, respectively. (i) Skin temperature measured close to stimulating cathode. (ii) Latency to half peak of compound action potential. (iii) Depolarizing electrotonus averaged between 40 and 60 ms to detect change in rate of accommodation, showing close parallel to latency and temperature. (iv) Hyperpolarizing electrotonus averaged between 90 and 100 ms, showing influence of rate of change of temperature.

In contrast, the hyperpolarizing response, measured at 90– 100 ms, changes most at the time the direction of temperature change is reversed. The hyperpolarizing response was more nearly proportional to the rate of change of temperature than to temperature itself. A similar relationship between the time courses of changes in temperature and threshold electrotonus was found for the second subject (Fig. 4B).

Change in the recovery cycle due to temperature The effects of temperature on the recovery cycle were more conspicuous than on any of the other excitability measures (Fig. 1F). The relative refractory period, i.e. the period after an action potential during which threshold is elevated, was the most temperature-sensitive parameter measured (Table 1). This is understandable, because refractoriness is due primarily to sodium channel inactivation, and recovery from inactivation, like other channel gating functions, is strongly temperature-dependent, with a Q10 close to 3. The relative refractory period and the measure of latency were the two most temperature-sensitive parameters when compared with the intersubject variability (Table 1D). The equivalent temperature deviation of 2.8°C means that

the standard deviation between subjects was equivalent to the effect of this temperature change on a single subject, or that skin temperature must be controlled within similar limits when small groups of subjects are compared (see Methods). After the refractory period, axons enter a superexcitable phase produced by the depolarizing afterpotential (Bergmans, 1970; Barrett and Barrett, 1982), followed by a late subexcitable period due to activation of nodal voltagedependent slow potassium channels (Baker et al., 1987; Kiernan et al., 1996). The times to peak superexcitability and subexcitability were also clearly delayed by cooling, but these delays were not measured. The small increases in amplitude of superexcitability and subexcitability with temperature were not statistically significant (Table 1).

Change in latency and amplitude due to temperature Unlike the various measures of axonal excitability mentioned to date—measures that reflect the excitability of the nerve at the point of stimulation—both the latency and the amplitude of the CMAP reflect the changing temperature of the nerve

824


throughout its course from the point of stimulation to the recording site. By implication, measures of these parameters include contributions from temperature effects on both the neuromuscular junction and the muscle fibres that contribute to the generation of the target CMAP. Consequently, while the present study can provide qualitative information about latency and amplitude, it is not possible to determine quantitative information about the contribution of individual components. As temperature decreases, sodium channel gating is slowed, leading to slowing of the inward sodium current responsible for depolarization of the axonal membrane (Schwarz and Eikhof, 1987). In routine nerve conduction studies, this is manifested by prolongation in latency of the compound potential and this will occur at each node of Ranvier along the nerve segment measured (Burke et al., 1999). In the present study, latency increased by 2.9% per degree C fall in temperature (Table 1 and Fig. 2A). The amplitude of the compound potential is also inversely related to temperature (Ludin and Beyeler, 1977; Bolton et al., 1981). As temperature decreases, inactivation of sodium channels slows and repolarization becomes delayed. There is therefore an increase in both the amplitude and the duration of the individual action potentials that contribute to the compound potential, in both the nerve segment and the contributing muscle fibres. As the duration of the individual potentials increases, the effects of temporal dispersion are reduced, with less phase cancellation, resulting in larger compound potentials (Table 1) (Kiernan et al., 1995). However, temporal dispersion is much less important for CMAPs than for sensory nerve action potentials.

Discussion The present study compares the effects of temperature on different excitability properties of human motor axons. The results are important for the interpretation of excitability measurements from patients with peripheral nerve disorders and for deciding the best strategy to adopt when testing patients with different limb temperatures. The principle finding is that most excitability parameters are not very sensitive to temperature over the range encountered clinically, the conspicuous exception being the relative refractory period. This result is in agreement with a study of the effects of temperature on the excitability properties of cutaneous afferents (Burke et al., 1999). The high sensitivity of refractoriness to temperature means that, to obtain meaningful comparisons of refractoriness between a single patient or a group of patients and controls, it is necessary either that all measurements be made at the same temperature or that temperature differences be taken into account. The former option is often regarded as preferable in principle, even though a specified temperature can only ever be approximated and the extra time required to stabilize temperature at a new level adds to the expense and inconvenience of the test. However, this study has shown

that changing the temperature of a nerve to a standard temperature can introduce previously unexpected errors. If the sensitivity of hyperpolarizing electrotonus [TEh (90– 100 ms)] to temperature changes reflects transient changes in membrane potential, then other potential-sensitive excitability parameters may also be sensitive to temperature changes, though this remains to be confirmed. We suggest, therefore, that for clinical studies of nerve excitability patients should be kept in a constant-temperature environment before testing and that skin temperature should be measured close to the site of stimulation during the test, but not altered if it is abnormal. Temperature corrections are best applied if temperature was stable at the time of the recordings. The primary effects of temperature on nerve function occur by altering the kinetics of channel gating. The acceleration of sodium channel activation with warming increases conduction velocity, while the acceleration of sodium channel inactivation shortens the relative refractory period. A new finding in this study was that the acceleration of potassium channel gating with warming also produces detectable effects: accommodation to depolarizing currents through the activation of slow potassium channels is accelerated, thereby reducing the accommodation half-time and TEd (40–60 ms). On release of the depolarizing current, deactivation of potassium channels is also accelerated, reducing the amplitude and duration of the afterhyperpolarization or undershoot. Similarly, on release of the hyperpolarizing current, reactivation of slow potassium channels is accelerated, reducing the amplitude and duration of the afterdepolarization or overshoot. It is to be presumed that activation and deactivation of the hyperpolarizationactivated current, IH, were also accelerated, but the effects of this were not evident with 100 ms current pulses. Other excitability parameters were little affected, presumably because they depend primarily on the degree of activation of ion channels at the resting potential and neither membrane potential nor the sensitivity to membrane potential of ion channels varies appreciably with temperature as long as it is held constant. The remaining new finding to be explained is the sensitivity of hyperpolarizing electrotonus to changes in temperature. We hypothesize that, because the rate of electrogenic pumping by Na⫹/K⫹-ATPase is temperaturedependent, the first effect of a rise in temperature would be an increase in pumping, leading to membrane hyperpolarization. In the long run, however, the rate of pumping must match the rate of sodium entry. So that, unless sodium influx is strongly temperature-dependent, the rate of pumping and membrane potential should return close to their starting values. Binder and Raymond reported transient changes in the threshold of frog sciatic nerve fibres to changes in temperature (Binder and Raymond, 1975), which would fit with this hypothesis. Excitability studies are now being used to provide functional information about neurological disease, changes in measures of excitability being interpreted in terms of altered ion channel function. The present study advances

Temperature and nerve excitability this possibility by exploring the variations in excitability properties with temperature. Different excitability parameters respond very differently to changes in temperature and these effects can be related to the behaviour of the ion channels involved. As in routine nerve conduction studies, measures of the refractory period must be conducted under stable temperature conditions and are prone to inaccurate interpretation if such conditions are not strictly adhered to. By contrast, the relative insensitivity of the majority of excitability measures to temperature serves only to enhance their utility in the clinical setting as an electrodiagnostic tool.

Acknowledgements M.C.K. was supported by a C. J. Martin/R. G. Menzies Fellowship from The National Health and Medical Research Council of Australia. References Baker M, Bostock H, Grafe P, Martius P. Function and distribution of three types of rectifying channel in rat spinal root myelinated axons. J Physiol (Lond) 1987; 383: 45–67. Barrett EF, Barrett JN. Intracellular recording from vertebrate myelinated axons: mechanism of the depolarizing afterpotential. J Physiol (Lond) 1982; 323: 117–44. Bergmans J. The physiology of single human nerve fibres. Louvain (Belgium): Vander; 1970. Binder MJ, Raymond SA. Some effects of temperature changes on the threshold of sciatic nerve fibers. Quarterly Progress Report No. 116. Research Laboratory of Electronics, Cambridge, Mass., USA: The MIT Press; 1975. Bolton CF, Sawa GM, Carter K. The effects of temperature on human compound action potentials. J Neurol Neurosurg Psychiatry 1981; 44: 407–13. Bostock H. Mechanisms of accommodation and adaptation in myelinated axons. In: Waxman SG, Kocsis JD, Stys PK, editors. Axon. New York: Oxford University Press; 1995. p. 311–27. Bostock H, Baker M. Evidence for two types of potassium channel in human motor axons in vivo. Brain Res 1988; 462: 354–8. Bostock H, Cikurel K, Burke D. Threshold tracking techniques in the study of human peripheral nerve. [Review]. Muscle Nerve 1998; 21: 137–58. Buchthal F, Rosenfalck A. Evoked action potentials and conduction velocity in human sensory nerves. Brain Res 1966; 3: 1–122.

825

Burke D, Mogyoros I, Vagg R, Kiernan MC. Temperature dependence of excitability indices of human cutaneous afferents. Muscle Nerve 1999; 22: 51–60. De Jesus PV, Hausmanowa-Petrusewicz I, Barchi RL. The effect of cold on nerve conduction of human slow and fast nerve fibers. Neurology 1973; 23: 1182–9. Johnson EW, Olsen KJ. Clinical value of motor nerve conduction velocity determination. JAMA 1960; 172: 2030–5. Kiernan MC, Bostock H. Effects of membrane polarization and ischaemia on the excitability properties of human motor axons. Brain 2000; 123: 2542–51. Kiernan MC, Mogyoros I Miller TA, Burke D. The significance of temporal dispersion in the measurement of compound nerve action potentials [abstract]. Muscle Nerve 1995; 18: 1062. Kiernan MC, Mogyoros I, Burke D. Differences in the recovery of excitability in sensory and motor axons of human median nerve. Brain 1996; 119: 1099–105. Kiernan MC, Burke D, Andersen KV, Bostock H. Multiple measures of axonal excitability: a new approach in clinical testing. Muscle Nerve 2000; 23: 399–409. Kuwabara S, Cappelen-Smith C, Lin CS, Mogyoros I, Bostock H, Burke D. Excitability properties of median and peroneal motor axons. Muscle Nerve 2000; 23: 1365–73. Lang AH, Puusa A. Dual influence of temperature on compound nerve action potential. J Neurol Sci 1981; 51: 81–8. Lowitzsch K, Hopf HC, Galland J. Changes of sensory conduction velocity and refractory periods with decreasing tissue temperature in man. J Neurol 1977; 216: 181–8. Ludin HP, Beyeler F. Temperature dependence of normal sensory nerve action potentials. J Neurol 1977; 216: 173–80. Meulstee J, Darbas A, van Doorn PA, van Briemen L, van der Meche´ FG. Decreased electrical excitability of peripheral nerves in demyelinating polyneuropathies. J Neurol Neurosurg Psychiatry 1997; 62: 398–400. Mogyoros I, Kiernan MC, Burke D. Strength–duration properties of human peripheral nerve. Brain 1996; 119: 439–47. Schwarz JR, Eikhof G. Na currents and action potentials in rat myelinated nerve fibres at 20 and 37 degrees C. Pflu¨ gers Arch 1987; 409: 569–77.

Received October 10, 2000. Revised December 1, 2000. Accepted December 4, 2000