Transcranial magnetic stimulation identifies upper motor neuron involvement in motor neuron disease

Transcranial magnetic stimulation.

We used Magstim 200 magnetic stimulators (Magstim; Whitland, Wales, UK). We used a 9-cm mean diameter circular coil centered over the vertex of the scalp for all studies. Viewed from above, current direction in the coil was counterclockwise for stimulation of the left hemisphere and clockwise for stimulation of the right hemisphere. Twenty-seven patients were tested with a low-power (peak 1.5 T) magnetic coil between 1989 and 1992, and 94 patients were tested with a high-power (peak 2.0 T) coil thereafter.

Subjects were seated comfortably in a chair with Ag/AgCl electroencephalographic electrodes over the biceps, triceps, abductor pollicis brevis (APB), and abductor digiti minimi (ADM) muscles in belly-tendon derivation. On average, we used three of these four target muscles per limb, per patient. Surface EMG signals were recorded using a bandpass of 10 to 10,000 Hz, inspected on-line, and stored on EMG hard drives (Mystro [Teca, Pleasantville, NY] and Viking IIe [Nicolet, Madison, WI]) for analysis. We determined resting motor evoked potential (MEP) threshold in 5% increments of maximum stimulator output as the minimum stimulus intensity that evoked at least three discernible MEPs in six consecutive stimulations using a display gain of 100 μV/cm. Threshold was recorded as 100% if no MEP was elicited with 100% stimulus intensity. After threshold was recorded, we elicited MEPs during modest tonic isometric contraction (10 to 20% maximal effort) using TMS 25% of maximum stimulator output above threshold (within the limits of stimulator output). We expressed the baseline-to-peak amplitude of ADM MEPs as a percentage of the baseline-to-peak amplitude of the compound muscle action potential (CMAP) obtained with supramaximal electrical stimulation of the ulnar nerve. We used MEP latencies and cervical magnetic root stimulation to calculate CMCT.18 We used MEP and F-wave latencies to calculate the CMCT to APB and ADM in a small proportion of patients who were intolerant to cervical root stimulation. After eliciting MEPs, we then looked for dissociation between MEP threshold and the cortical stimulation silent period (CSSP) by reducing stimulus intensity in 5% increments of stimulator output until TMS no longer altered the appearance of the averaged rectified ADM EMG, as described previously.12 We defined dissociation between excitatory and inhibitory effects of TMS (hereafter termed failure of MEP facilitation) as EMG inhibition without a preceding MEP at two or more stimulus intensities.

Longitudinal transcranial magnetic stimulation.

A subset of 12 patients were studied with TMS longitudinally, at intervals ranging from 2 to 18 (mean, 11 ± 6) months.

Limits of normality.

We used findings in normal volunteers to define limits of normality for MEP variables. We used a conservative criterion of three SDs relative to normal means for each MEP variable. We tested subgroups of volunteers on at least two occasions to determine test-retest reliability for MEP threshold and CSSP duration.

In 60 normal volunteers studied with the low-power magnetic coil, the MEP threshold was 47.7 ± 7.5% of stimulator output and was not significantly related to age (r = 0.22, p < 0.10). Because we measured threshold in 5% increments, we considered low-power coil MEP thresholds of no more than 25% or no less than 70% of stimulator output to be abnormal. In our experience, the MEP threshold with the high-power coil averages 5% of stimulator output lower than the MEP threshold measured with the low-power coil. In 24 normal volunteers studied with the high-power magnetic coil, the MEP threshold was 41.2 ± 7.3% of stimulator output. Measured in 5% increments, we considered high-power coil MEP thresholds of no more than 20% or no less than 65% of stimulator output to be abnormal.

In ADM, three SDs below the mean established a lower normal limit for the MEP/CMAP ratio of 11%. Upper limits for CMCT were 9.5 milliseconds (biceps), 9.0 milliseconds (triceps), and 10.0 milliseconds (APB and ADM). The MEP threshold test-retest reliability was r = 0.95 (Pearson product-moment correlation; n = 16; p < 0.0001). The test-retest MEP threshold difference was 2.5 ± 2.6% (mean ± SD) of the stimulator output. Thus, in longitudinal patient studies, an MEP threshold change of at least 10% of the stimulator output was defined as significant. CSSP duration test-retest reliability was r = 0.98 (Pearson product-moment correlation; n = 5; p < 0.005). Test-retest CSSP duration difference was 9 ± 6 (mean ± SD) milliseconds. Thus, in longitudinal patient studies, a change in the CSSP duration greater than 30 milliseconds was considered significant.

Statistical methods.

We compared the clinical characteristics and TMS results among patients with ALS, ALS-PUMNS, PBP, and PMA using the chi-square test, unpaired t-test, or Mann-Whitney U test, depending on the analysis conducted.

In patients, we averaged the MEP threshold and CMAP amplitude for the right and left ADM. We then used Pearson product-moment correlation coefficients to test the relationship between each subjects’s averaged ADM MEP threshold and CMAP amplitude. We also tested the relationship of MEP threshold and CMAP amplitude to symptom duration.

We considered p values less than 0.05 significant.

Pathologic correlations.

In four patients, neuropathologic examinations were performed on formalin-fixed and processed paraffin-embedded sections, 6 to 8 μm in thickness, from multiple brain areas including primary motor cortex, brainstem (midbrain, pons, medulla), and spinal cord. The sections were stained with luxol fast blue, counterstained with hematoxylin and eosin, and examined without reference to clinical examinations. Sections were evaluated for neuronal loss, gliosis, presence of inflammatory cells, particularly macrophages, and myelin loss in the corticospinal pathway.

Clinical findings.

Hand function in our patients measured using the ordinal scale defined in the Methods section was 1.1 ± 0.8 (mean ± SD; range, 0 to 3). Table 1 shows the clinical characteristics of our patients according to clinical classification. Several findings emerged from analysis of these data. Patients with PBP were older than patients with clinically definite ALS. Patients with PMA were less likely to be women and had longer symptom duration than patients with clinically definite ALS.

Abnormalities of transcranial magnetic stimulation.

Table 2 shows TMS results in patients with MND. Patients with PMA were less likely to have abnormal TMS than patients with clinically definite ALS. However, the frequency of abnormal TMS was not significantly different among patients with ALS, ALS-PUMNS, and PBP. Table 2 categorizes abnormal TMS results into abnormal excitation (MEP threshold and MEP/CMAP ratio), failure of MEP facilitation, and abnormal CMCT. The abnormality observed most frequently was reduction in excitation, manifested predominantly by increased MEP threshold (figure 1). Thirty-two of the 70 patients in this category had absent MEPs. Only 8 of the 70 patients with abnormal excitation had decreased ADM MEP/CMAP ratios independent of absent MEPs; 3 of these 8 patients also had abnormal MEP thresholds. Most (41/70; 58%) of the patients with abnormal excitation also showed failure of MEP facilitation. However, failure of MEP facilitation also occurred in eight patients in whom the MEP threshold was normal (figure 2). Abnormal CMCT was relatively infrequent in our patients. Furthermore, fully half of the patients with abnormal CMCT also had abnormal excitation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Univariate scattergram showing the distribution of abductor pollicis brevis motor evoked potential (MEP) threshold to transcranial magnetic stimulation in 84 control subjects (open circles); 99 patients with amyotrophic lateral sclerosis (ALS), amyotrophic lateral sclerosis with probable upper motor neuron signs (ALS-PUMNS), or progressive bulbar palsy (PBP) (closed circles); and 22 patients with progressive muscular atrophy (PMA) (closed squares).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Averaged (n = 5) rectified surface electromyography recorded over the abductor pollicis brevis muscle during moderate (10 to 20% maximum) voluntary contraction in a normal volunteer and a patient with ALS and probable upper motor neuron signs (ALS-PUMNS), comparing excitatory and inhibitory effects of transcranial magnetic stimulation (TMS) given at time 0. Successive traces show the effects of decreasing intensities of TMS expressed as percentage of maximum stimulator output above the threshold (T) for eliciting a motor evoked potential (MEP) at rest. Resting thresholds were comparable in both subjects. However, MEPs elicited during voluntary contraction (arrows) were discernible only at the highest intensity of TMS in the patient with ALS-PUMNS. At lower intensities, MEPs in this patient failed to facilitate relative to background contraction. Note that cortical stimulation silent periods induced by TMS are comparable in both subjects.

The MEP threshold was inversely related to CMAP amplitude (r = 0.25; p < 0.01). However, although statistically significant, the strength of the correlation indicates that CMAP amplitude accounts for only (r² = 0.06) 6% of the variance in MEP threshold. This suggests that abnormal MEP threshold was largely independent of CMAP amplitude as an index of lower motor neuron loss. Furthermore, 66 of 84 (78%) patients in whom TMS was abnormal showed abnormalities in at least one target muscle with normal CMAP amplitude (≥5 mV). This figure includes all six patients with PMA in whom TMS detected subclinical UMN dysfunction.

Clinical–neurophysiologic correlations.

Neither ADM MEP threshold nor CMAP amplitude was significantly related to symptom duration (r = 0.07, p < 0.44, MEP threshold; r = 0.06, p < 0.50, CMAP amplitude).

Thirty-two patients (21 ALS, 7 ALS-PUMNS, 2 PBP, and 2 PMA) had absent MEPs at maximum stimulator output. Nine of these 32 patients (6 ALS, 2 ALS-PUMNS, and 1 PBP) also had no CSSP—that is, TMS at maximum stimulator output did not alter the EMG. Comparing these two patient groups (absent MEPs with and without a discernible CSSP) revealed no significant differences in hand function scores (p < 0.62, Mann-Whitney U test) or presence of bulbar signs (p < 0.15, chi-square test). However, patients with a CSSP had a shorter symptom duration than patients with no discernible CSSP (19 ± 11 versus 33 ± 15 months; p < 0.009).

Longitudinal transcranial magnetic stimulation.

Twelve patients were studied with TMS longitudinally. Two of these 12 had PMA and initially had normal studies. Longitudinal studies of these two patients showed no significant changes at 12 and 18 months. However, excluding these two patients with PMA, longitudinal studies showed significant interval changes in 9 of 10 patients, including 2 patients in whom TMS was initially normal. The changes observed most frequently between studies included increased MEP threshold (six patients) and decreased CSSP duration (six patients; figure 3). Neither decreased MEP threshold nor increased CSSP duration occurred in any patient.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Averaged (n = 5) rectified surface electromyography recorded over the abductor pollicis brevis (A) and abductor digiti minimi (B) muscles during moderate (10 to 20% maximum) voluntary contraction showing progressive loss of the cortical stimulation silent period (CSSP) in two patients with ALS and probable upper motor neuron signs. In each trace, transcranial magnetic stimulation (TMS) at maximum stimulator output was given at time 0. The patient in (A) showed an elevated threshold for eliciting a motor evoked potential (MEP) at rest. TMS at 100% of stimulator output initially elicited a small MEP (arrow) followed by the CSSP. In the second study, TMS elicited neither response. Note slight differences in display gain. In the patient in (B), TMS initially elicited only a CSSP, without a definite MEP. In the second study, note the decrease in the duration of the CSSP.

Pathologic correlations.

Four patients in this investigation were studied postmortem. Neuropathologic evaluations of these patients confirmed the presence or absence of UMN involvement identified with TMS.

Failure of motor evoked potential facilitation.

In some patients, failure of MEP facilitation was the sole abnormality detected with TMS. In these patients, the threshold for eliciting an MEP at rest was normal. However, when TMS was administered during voluntary muscle contraction, the MEP became obscured in the background contraction, leaving only a CSSP. This failure of MEPs to facilitate during voluntary contraction has been noted incidentally in previous studies.7,10,11 However, considering that voluntary contraction of the target muscle causes dramatic facilitation of MEPs elicited in normal people,26 the absence of such facilitation in clinical studies has received surprisingly little attention. Uozumi et al.11 reported an increased ratio of background EMG activity to MEPs elicited in patients with ALS and implied that failure of MEP facilitation resulted from increased motor unit size. We believe that this possibility is unlikely because we never observed failure of MEP facilitation in patients with PMA, despite obvious electrophysiologic evidence of chronic denervation and reinnervation. We cannot exclude the possibility that patients with ALS recruited a higher proportion of available motor units than did normal volunteers. However, MEPs are readily recorded during even maximum voluntary contraction in patients with ALS.11 Furthermore, we and others10 have observed failure of MEP facilitation in patients with MS, in whom lower motor neuron function is presumably normal. Therefore, we suggest that failure of MEPs to facilitate during voluntary contraction is a sign of UMN impairment. Mills27 reached a similar conclusion in a subgroup of patients with ALS in whom TMS with single motor unit analysis revealed excitation thresholds that were normal at rest but were increased during voluntary motor unit activation.

Progression of cortical dysfunction in ALS.

The abnormality observed most frequently in this study of patients with MND was an increase in MEP threshold. However, several investigators have actually observed reduced excitation thresholds in patients with ALS,10,27,28 particularly in the early stages of the disease. Reduction in MEP threshold in these patients has been interpreted as reflecting spinal10 or cortical motor neuron hyperexcitability or defective intracortical inhibition.28-33 It has been suggested that MEP thresholds are initially reduced in ALS, increasing as the disease progresses.13,28 We failed to observe patients with significant reductions in MEP thresholds. This discrepancy may have occurred for several reasons. First, mean symptom duration in our patients was 25 months. In contrast, Mills and Nithi28 have emphasized the brief symptom duration of patients with reduced MEP thresholds. Second, we used a relatively conservative definition of abnormality (mean ± 3 SDs) in analyzing MEP variables. We were also unable to replicate any meaningful correlation between symptom duration and MEP threshold.13 However, we suggest that questions regarding neurophysiologic changes associated with disease progression may best be answered through longitudinal assessment of individual patients.

This investigation is the first to use longitudinal TMS studies to document changes in cortical motor excitatory and inhibitory function during progression of ALS. We found that the cortical elements mediating MEPs are affected relatively early in the course of the disease, manifested largely by progressive increases in threshold.12 This is consistent with previous observations of increased MEP thresholds in patients with ALS with normal M waves13 and of progressive inexcitability of central motor pathways in the course of ALS.34 Similarly, Uozumi et al.11 illustrated progressive loss of MEP amplitude with longitudinal studies in a single patient with ALS. However, previous investigators did not examine changes in the inhibitory effects of TMS with disease progression. Serial studies of our patients indicated that the neural substrates mediating the inhibitory effects of TMS are affected relatively later in the course of the disease, manifested by loss of the CSSP in patients in whom initial studies showed increased MEP thresholds.