Peripheral Neuropathies: Proceedings of the International Symposium on Peripheral Neuropathies Held in Milan, Italy, on June 26–28, 1978

Pozza

INTRODUCTORY LECTURE

Outline

Chapter 1: SENSORY ACTION POTENTIALS AND BIOPSY OF THE SURAL NERVE IN NEUROPATHY

SENSORY ACTION POTENTIALS AND BIOPSY OF THE SURAL NERVE IN NEUROPATHY

FRITZ BUCHTHAL and FRIEDRICH BEHSE,     Institute of Neurophysiology, Laboratory of Clinical Neurophysiology and Research Laboratories of the Rigshospital, section 4112, Copenhagen, Denmark 2100

ABSTRACT

In 167 consecutive patients with various types of neuropathy the amplitude of the sensory potential and the maximum conduction velocity along the sural nerve were compared with conduction in other sensory nerves and related to nerve biopsy. Electrophysiological findings in the sural nerve were similar to those in the superficial peroneal and the median nerve, though the distal segment of the median nerve was normal in 20% of the patients when it was abnormal in the sural nerve. Quantification of histological findings was a more sensitive method than the electrophysiological study in that two-thirds of 33 patients with normal electrophysiology in the sural nerve showed slight loss of fibres or signs of remyelination in teased fibres. Of the 95 nerves from which teased fibres were obtained, maximum conduction velocity was abnormal in 43. In 18 nerves slowing in conduction was due to axonal degeneration: the velocity was as to be expected from the diameter of the largest fibres in the biopsy (proportionate slowing). In nine nerves slowing was severe and more marked than to be expected from loss of the largest fibres (disproportionate slowing); these nerves showed paranodal or segmental demyelination in more than 30% of the fibres. In 16 nerves from patients with neuropathy of different ætiology, among these three from patients with diabetic neuropathy, neither loss of fibres nor demyelination could explain the moderate slowing. When an incidence of teased fibres with demyelination over a length of 10 mm of the nerve is extrapolated to 140 mm of the nerve, 50% of the fibres would have no myelin defects (either de. or remyelination). The cause of the slowing is possibly a functional defect in the excitable membrane of the nerve.Finally the limitations are discussed and minimal requirements are suggested for the electrophysiological diagnosis of a peripheral neuropathy with discrete clinical symptoms and signs.

INTRODUCTION

Experimental and clinical science advances as much by the development and application of new techniques as by new theories. In the field of peripheral neuropathy the progress in techniques lies essentially in the quantification of morphological and electrophysiological findings. The progress in theory was inspired by the concept that neuropathies can be divided into two types: One with extensive demyelination and marked reduction in conduction velocity primarily affecting the Schwann cell. Another larger group is characterized by axonal degeneration and little or no demyelination. Slowing in conduction is less than in nerves with extensive demyelination. This concept, originally based on findings in experimental neuropathy, has greatly stimulated work in the field. It was, however, soon recognized that a neuropathy with extensive demyelination is nearly always associated with marked loss of fibres¹,²,³,⁴. In our material of 85 nerves from which teased fibres were obtained, segmental demyelination without fibre loss was present in two nerves. Moreover, signs of paranodal demyelination were found in axonal neuropathy, possibly indicating involvement of the Schwann cell secondary to the involvement of the axon⁵,⁶. To suggest demyelination as the underlying pathology for slowing in conduction encounters difficulties when slowing is of the order of 30 to 40%. This degree of slowing occurs in axonal neuropathy in the absence of segmental or paranodal demyelination.

Based on electrophysiological and biopsy findings in 167 patients with neuropathy (Table 1) we shall adress two problems, the one concerns the relation between morphometric findings in the sural nerve, conduction velocity and amplitude of the evoked potentials. The other problem concerns the early recognition of neuropathy. This problem has attracted interest because of growing awareness of possible toxic agents in the environment.

Table 1

⁷167 PATIENTS WITH POLYNEUROPATHY

*Peroneal muscular atrophy of the neuronal type plus involvement of the central nervous system.

METHODS

The method of stimulation and recording has been described⁸,⁹,¹⁰. Fig. 1 shows the sites of recording in the median and sural nerves and the site of the biopsy of the sural nerve and Fig. 2 shows the position and dimensions of the near-nerve electrode at an optimal distance from the sural nerve. Electronic averaging of 500 to 2000 responses was used to estimate amplitudes of less than 3 μV (lower limit 0.02 μV) and to record the conduction velocity of the fastest and of the slow components.

Fig. 1 ) needle electrodes to record from different segments of the median and sural nerves. S, stimulating cathode (surface or needle). S1, cathode to stimulate the sural nerve. Sura denotes the site of recording 12 to 14 cm proximal to the lateral malleolus.

The shaded area shows the 3-5 cm long segment of the sural nerve taken in toto as biopsy¹¹.

Fig. 2 Recording electrode (right) at an optimal distance from the sural nerve (left) to ascertain that the action currents from the fibres of all fascicles are about equally represented when they reach the leading-off surface of the electrode. Above, left, cross-section through the sural nerve at a 10 times higher magnification (courtesy of Professor Annelise Rosenfalck).

Biopsy: 3 to 5 cm of the sural nerve were removed in toto and prepared for light and electron microscopy and for preparation of 50 to 70 teased fibres as described¹²,¹³,¹⁴,¹⁵. In each biopsy the transverse endoneurial area was measured. Moreover, the total number and size distribution of myelinated fibres and of groups of three or more regenerating fibres (clusters) were determined within an area of 0.4 to 0.6 mm², sampled from all fascicles.

In addition, we counted sites with clumps of myelin and determined the incidence of degenerated fibres and of bands of Büngner in electron micrographs. In teased fibres the incidence of fibres, of segments with segmental and paranodal demyelination, of remyelination and of regenerated fibres was determined.

RESULTS

The most frequent histological abnormality that changes nerve conduction and amplitude of the sensory potentials is loss of myelinated fibres. The amplitude of the potential recorded via needle electrodes from the sural nerve increases with the number of large myelinated fibres¹¹,¹⁴,¹⁸,⁷. Since loss of large fibres was equally prominent in nerves with axonal degeneration as in those with demyelination¹⁵, the question arises whether the type of pathology can be predicted from the degree of slowing in conduction. We have tried to answer this question by determining whether and when the slowing in conduction in a given sural nerve from a patient with polyneuropathy could be predicted from the largest fibres found in the biopsy of the same nerve. The prerequisite for the prediction of the maximum conduction velocity is the fact that conduction velocity varies proportionally with the diameter of the nerve fibre¹⁶. The fastest component of the sensory potentials was related to the myelinated fibres of largest diameter in the histogram of diameter of the same nerve as shown in Fig. 3. In the human sural nerve the maximum conduction velocity was 4.3 (S.D. 0.3) times the fibre diameter. Findings in nerves with purely axonal degeneration showed that the conversion factor was the same for fibres of more than 7 εm in diameter¹¹,¹⁷.

Fig. 3 Components of the sensory potential and distribution of diameters of myelinated fibres in a normal sural nerve. Above and middle: The nerve was stimulated maximally at the lateral malleolus and the potential was recorded 15 cm (sura) and 50 cm (fossa poplitea) proximally to it. The dashed lines connect components conducted at the same velocity and point to the corresponding fibre diameter in the histogram below (6600 fibres). Note that the diameter of the fibres is plotted from right to left¹¹,¹⁷.

To calculate conduction velocity from the diameter of the largest fibres, we have assumed that at least 10 fibres must be present to give a response distinguishable from noise when 500-2000 responses are averaged. This assumption is based on the relation between amplitude of the sensory potential and number of large myelinated fibres in the sural nerve¹¹,¹⁴,¹⁷,¹⁸.

When slowing in maximum conduction velocity deviates by less than 20% from that to be expected from the fibres of largest diameter, it can be explained by axonal loss and demyelination does not slow the maximum conduction velocity. The scatter of 20% is derived from findings in controls¹⁴,¹⁸.

From counts of myelinated fibres and from quantitation of abnormalities in teased fibres axonal degeneration was shown to be the dominating pathology in 20 sural nerves from patients with alcoholic neuropathy¹⁹,¹⁴. The incidence and extent of paranodal and segmental demyelination (0.3%) and remyelination (20%, compared with 5-20% in controls) was unimpressive¹⁴. This was the case even when fibre loss was insignificant and when the neuropathy had lasted for only a few weeks. When present at all, myelin damage was distributed over multiple sites of a given fibre in one third of the fibres, interpreted to be secondary to beginning axonal degeneration⁵⁶. Axonal degeneration was also the main pathology in 19 biopsies from patients with the neuronal type of peroneal muscular atrophy¹⁵, in Friedreich’s ataxia²⁰, in three of four patients with paraneoplastic neuropathy, in patients with rheumatoid neuropathy, polyarteritis nodosa, hepatic neuropathy, acute intermittent porphyria⁷ and in lead neuropathy²¹.

In alcoholic neuropathy, in the neuronal type of peroneal muscular atrophy, in paraneoplastic neuropathy, in lead neuropathy and in lead-exposed men the recorded conduction velocity was equal to the velocity expected from the fibres of largest diameter, i.e. the slowing in conduction could be explained by axonal degeneration (Fig. 4). A decrease in conduction velocity from the normal 53 m/s to 30 m/s (to 55% of normal) could be due solely to loss of the largest fibres. Not unless the recorded sensory conduction velocity was disproportionally slower than expected or - when no biopsy was available - not unless the velocity was slowed to less than 60% of normal is it justified to assume causes other than axonal degeneration for the diminution in conduction velocity.

Fig. 4 Maximum conduction velocity determined from the sensory potential of the sural nerve (ordinate) as a function of the velocity expected from the diameter of the largest myelinated fibres (abscissa) in polyneuropathies of different ætiology. The full line indicates equality between the recorded and the expected velocity. Loss of large fibres can explain the slowing in conduction when the recorded velocity is within 20% of that expected from fibre diameter (dashed line lower limit of normal range).

a: Peroneal muscular atrophy of the neuronal type and of type neuronal plus (Table 1).

b: Peroneal muscular atrophy of the hypertrophic type.

c: Nerves from patients with discrete (uncertain) clinical symptoms and signs in neuropathy of unknown ætiology²²,⁷.

A reduction in amplitude and conduction velocity is also seen in regenerating fibres. This is illustrated by the recovery of the sensory potential after section and suture of the nerve²³. The sensory potential of regenerating nerve is characterized by 30 to 50 components of low amplitude. Five months after suture of the median nerve at wrist, when the first response could be distinguished, the cumulative amplitude, obtained by adding the amplitude of component potentials, was 1 to 3 μV, the fastest component was conducted at 10 to 25 m/s, the slowest at 2 to 3 m/s. 40 months after suture, when tactile sensibility had become normal, the potential was still split-up in 20 to 30 components, the cumulative amplitude was normal as was the maximum conduction velocity. The velocity of the slowest components was still markedly diminished (Fig. 5). Fig. 6 shows the time course of recovery of the cumulative amplitude of the sensory potential recorded just proximal to the site of the end-to-end suture at wrist. The left plot shows the much faster recovery after a transient compression of the ulnar nerve at the elbow, that presumably has caused segmental demyelination²³.

Fig. 5 Sensory potential evoked by maximal stimuli to the proximal phalanx of digit III and recorded at wrist, proximal to the site of suture of the median nerve. The recording was obtained 40 months after suture, when the cumulative amplitude and tactile sensibility were normal and the maximum sensory conduction velocity 80% of normal. The upper scale below the traces indicates the latency, the lower scale the conduction velocity. The figures to the left denote the number of responses that were averaged. The patient was 20 years old at the time of the suture²³.

Fig. 6 Recovery of the cumulative amplitude of the sensory potentials recorded just proximal to the suture of the median nerve at wrist (right) compared with the recovery of the sensory potential recorded proximal to the elbow (left) after transient compression of the ulnar nerve at the elbow. The patients were 20 and 33 years old. A: Normal tactile sensibility. B: The patient could distinguish light touch (0.3 g) from pin-prick but was unable to localize the stimulus. C: The patient perceived touch as an uncharacteristic stimulus and was unable to distinguish touch from pin-prick and to localize the stimulus.

The curves were drawn by eye²³.

In the biopsy regeneration is indicated by an incidence of three or more closely packed small myelinated fibres (clusters) and in teased fibres by the occurrence of fibres in which all internodal segments are shorter than in normal fibres of the same diameter. The degree of regeneration differs in different types of neuropathy. Thus, in the neuronal type of peroneal muscular atrophy the number of the largest fibres was diminished whereas the number of small fibres was as in controls, because their loss was compensated by regeneration¹⁵. In the hypertrophic type of peroneal muscular atrophy¹⁵ and in alcoholic¹⁴ and diabetic²⁴ neuropathy the number of small fibres was diminished because of lack of regeneration. When the histology suggests purely axonal degeneration components may occur conducted at extremely slow rates side by side with components conducted at a near normal or moderately slowed rate. They must be attributed to regeneration. This is illustrated in Fig. 7, which shows the sensory potentials of the sural nerve in a patient with discrete signs and symptoms of neuropathy. The patient was a 45-year-old woman who complained of weakness in the legs for the past 14 months. The family history was negative and the glucose tolerance test and vitamin levels of B12 and B6 in blood were normal. The force was graded as being normal, there was no wasting, tendon jerks in the legs were weak to absent, and tactile, vibrational and postural sensibility were normal. The maximum conduction velocity and the amplitude of the sensory potential were normal in the sural nerve, the only abnormality being components conducted at 3 to 9 m/s that do not occur in normal nerve. The sensory conduction velocity along the distal segment of the posterior tibial nerve was slowed to 78% of normal and the amplitude of the sensory potential recorded at the medial malleolus was diminished to 8% of normal. The biopy of the sural nerve showed a diminished number of large fibres (1400, normal range 1650-3300) and an increased incidence of clusters of regenerating fibres (63, normal range 0-38). Teased fibres were not obtained.

Fig. 7 Slow components conducted at 2.7-9 m/s (arrows) as the only electrophysiological abnormality in the sural nerve of a patient with discrete clinical signs and symptoms of neuropathy of unknown ætiology. The potential was evoked by maximal stimuli at the lateral malleolus and recorded at midcalf. The upper trace shows the potential, recorded by a single sweep, conducted at a maximum speed of 52 m/s. The lower traces show slow components after averaging of 1000 and 2000 responses (36°C).

When slowing of the fastest components is to more than 60% of normal, the sensory potential does not give a clue as to whether the late components are originally fast components or are due to conduction along immature remyelinating fibres. There is experimental evidence that demyelination is associated with slowing in conduction and may lead to block of nerve fibres. The muscle weakness in the absence of wasting in idiopathic poly-radiculo-neuropathy is due to block in conduction of demyelinated fibres¹,²,²⁵,²⁶. The rapid recovery in muscle force is compatible with remyelination rather than with regeneration after axonal loss.

Teased fibres are the only way to quantitate abnormalities in the myelin sheath²⁸,²⁹,³⁰. In evaluating myelin defects one must consider that some abnormalities occur in controls¹⁴. Thus, in the seven controls we have analyzed (15 to 54 years old), rows of myelinated segments occurred in up to 20% and solitary intercalated segments in up to 10% of the teased fibres. Segmental and paranodal demyelination were not encountered (Fig. 8). Segmental demyelination was found in 45 nerves of 167 patients with neuropathy. It usually occurred in less than 10% of teased fibres except in the hypertrophic type of peroneal muscular atrophy¹⁵, in the neuropathy after gastrectomy¹⁴ and in one patient with neuropathy of unknown ætiology⁷. In these conditions segmental demyelination was present in 30 to 100% of the fibres. Paranodal demyelination occurred in a quarter of the nerves, the incidence rarely exceeding 15%. We have obtained teased fibres from 9 nerves of patients with diabetic neuropathy²⁴. Segmental and paranodal demyelination were rare: Segmental demyelination occurred in 8 of 141 teased fibres from 2 nerves and paranodal demyelination in 17 of 198 teased fibres from 3 nerves. Both were present when the neuropathy was of short duration. Since axonal degeneration was present in all nerves, also in those from patients with the shortest duration of the neuropathy, demyelination and axonal degeneration seem to proceed side by side rather than demyelination being secondary to axonal degeneration.

Fig. 8 )²⁴.

Segmental remyelination was increased above the level of controls in 33 nerves but only in 5 did more than half of the fibres show this abnormality. The most frequent abnormality, seen in half the nerves, is paranodal remyelination, i.e. solitary intercalated segments. They occurred in half the nerves, twice the incidence in controls.

In this connection the question arises whether findings obtained over 10 mm of 50 teased nerve fibres are representative for the 120 to 150 mm of the 100 to 3500 fibres over which conduction velocity was determined⁷. Teased fibres were obtained from 3-5 different fascicles. This probably compensates for the differences in myelin abnormalities in different fascicles.

The incidence of fibres with myelin abnormalities did not increase proportionately with the length over which the fibres were teased. Thus, in 16 nerves of patients with slowing that could not be explained by loss of the largest fibres, the incidence of normal fibres, of fibres with myelin abnormalities and of myelin defects per fibre (including remyelination) obtained over 5 mm of teased fibres was compared with that obtained over 10 mm of teased fibres⁵. When the length of nerve teased was 10 mm, the incidence of abnormalities was 30% higher instead of doubled. By extrapolation Mr. Steenstrup found (unpublished) that the incidence of fibres with myelin damage increases from e.g. 20% in 10-mm-long teased fibres to 50% over a 130-mm-long portion of the nerve, i.e. with this assumption half of the fibres do not have segments with myelin abnormalities⁷.

It is unlikely that the diminished conduction velocity in demyelinating neuropathy is due to selective loss or block of the fastest fibres. This would result in a shortening of the duration of the sensory potential. In fact, the slowed potentials were prolonged. Nor can the decreased conduction velocity be due to selective slight slowing of conduction in the fastest fibres (9 to 14 μm in diameter) because this would result in an increased amplitude. In fact, the amplitude was decreased. Therefore either all components of the potential are conducted at a diminished velocity, or the fastest fibres are slowed such that they contribute or give rise to components throughout the range of velocities represented in the prolonged potential. Slowing in conduction per se causes an increased temporal dispersion and thereby a diminished amplitude. Reconstruction of the compound sensory potential from its components showed that an amplitude of less than 40% of normal could not be accounted for solely by temporal dispersion⁸,¹⁰.

The relation of histological abnormalities to slowing in conduction was evaluated in the 43 nerves with slowing from which teased fibres were obtained (Fig. 9). In 18 nerves slowing was moderate and was due to axonal loss; demyelination occurred in less than 3% of teased fibres. Conversely, in 9 nerves, axonal loss could not explain the marked slowing; demyelination was seen in many teased fibres and could account for the marked slowing.

Fig. 9 Maximum sensory conduction velocity as a function of the incidence of segmental and paranodal demyelination in teased fibres of 43 sural nerves with slowed conduction from the lateral malleolus to midcalf: •18 nerves in which slowing in conduction was as to be expected from the diameter of the largest myelinated fibres. Δ 9 nerves with a high incidence of demyelination and disproportionate slowing (p. 16 nerves with disproportionate slowing and an incidence of demyelination of 19% or less⁷,²².

There remain 16 nerves (37%), among these three from diabetic neuropathy, in which slowing could be explained neither by fibre loss nor by the incidence of demyelination present in at most 20% of the fibres. Remyelination occurred as frequently in nerves with as in those without slowing. With remyelination included in the incidence of myelin defects, the model on which the extrapolation and the statistical calculations were based shows that so many fibres do not have myelin defects over the length along which conduction is measured that morphologically evident defects cannot explain the slowing in conduction. There are conditions in which slowing occurs in the absence of structural changes: i) In the hypoxic segment of the nerve conduction decreased to 40% of the pre-hypoxic level³¹,³², whereas it regained normal velocity proximal to the hypoxic segment, indicating that there was true slowing before conduction was blocked (Fig. 10)³². Presumably hypoxia prolonged the rate of rise of the nodal action current. ii) In experimental diabetes³⁴,³⁵ and in galactosæmia³⁶ slowing in conduction has been found in the absence of fibre loss or other structural or ultrastructural changes. Some kind of a functional defect in the nerve membrane may be more common than hitherto assumed. When superimposed on the slowing due to loss of fast fibres the functional defect may explain why in nerves with proportionate slowing the recorded velocity tended to be slightly lower than that calculated from diameter. Depending on the severity of the defect, nerves in alcoholic, diabetic or postinfectious neuropathy may exhibit proportionate or disproportionate slowing.

Fig. 10 The normal maximal sensory conduction velocity (Vs) proximal to the region of hypoxia (X) for the 15 min of hypoxia (abscissa) during which a sensory potential could be discriminated outside the region of hypoxia, illustrates that the same fibres along which conduction was slowed in the hypoxic portion of the nerve conducted at a normal rate proximal to it. Thus, slowing in the hypoxic segment is not due to block of the largest fibres³³.

MINIMUM ELECTROPHYSIOLOGICAL REQUIREMENTS FOR THE DIAGNOSIS OF POLYNEUROPATHY

In most patients the clinical examination can establish whether or not a patient has polyneuropathy. When the clinical investigation shows discrete abnormalities and leaves doubt the electrophysiological study and nerve biopsy or both may help to substantiate a diagnosis. It is then important to appreciate the limitations and the diagnostic significance of these studies. Even though the temperature is kept constant, there is a considerable scatter in the conduction velocity and particularly in the amplitude of the evoked responses in normal subjects matched for age. With respect to motor conduction velocity we have to consider that one large motor nerve fibre results in the measurement of a normal maximum conduction velocity even if all other fibres conduct abnormally slowly. In general, abnormalities in sensory or motor conduction in a single nerve are insufficient evidence to diagnose a systemic neuropathy. A low amplitude of the sensory potential of the ulnar nerve at wrist or mild slowing in motor or sensory conduction velocity along the ulnar nerve in the forearm may be secondary to local damage of the nerve at the elbow. Similarly, slowing along the distal sensory or motor branch of the deep peroneal nerve is a poor indicator of a systemic neuropathy since this nerve is often damaged by pressure.

The most prominent abnormality in men with increased levels of lead in the blood was a prolonged latency from the ankle to the extensor digitorum brevis muscle²¹. This turned out to be due to pressure by the metal-lined safety shoes rather than to a subclinical lead neuropathy. For the same reason, electromyographic signs of partial denervation in the extensor digitorum brevis muscle (diminished number of motor units or fibrillation potentials or both) are unreliable signs of systemic disease.

The nerve biopsy can give information as to the nature and the severity of pathological changes; only rarely can it contribute to elucidate the ætiology (amyloid neuropathy, metachromatic leucodystrophy, hereditary neuropathy with liability to pressure palsies etc.). Onion-bulb formations, once thought to be specific of hypertrophic neuropathy, have been observed in neuropathies of different ætiology. An increase in the perivascular space of endoneurial vessels thought to characterize diabetic neuropathy has been found in other neuropathies as well²⁴. Unless histological abnormalities are advanced, it is necessary to quantitate biopsy findings in cross-sections and among teased fibres. We have determined the total number of myelinated nerve fibres from transverse sections of the nerve in toto rather than their number per mm². In two-thirds of the sural nerves taken from 100 consecutive neuropathies the endoneurial area was increased above that in 10 controls. An increased endoneurial area may simulate (19% of the nerves) or erroneously accentuate (21% of the nerves) loss of nerve fibres¹³.

In summary: If diabetes, uræmia and alcoholism can be excluded and the patient shows discrete neurological symptoms and signs we investigate (i) the sural nerve, from the lateral malleolus to midcalf, (ii) motor and sensory conduction along the peroneal nerve from ankle to capitulum fibulae (on the contralateral leg) and (iii) motor and sensory conduction along the distal segments of the median and posterior tibial nerves.

These studies we supplement by electromyography of a muscle with moderate weakness in the lower and the upper extremities (e.g. m.tibialis anterior and m.abductor pollicis brevis).

Unless we find abnormalities in conduction (velocity or amplitude) in at least two nerves and electromyographic abnormalities in at least one muscle, we consider the electrophysiological study inconclusive. One or the other of these patients may non the less have a neuropathy which only can be demonstrated by quantified biopsy findings.

ACKNOWLEDGEMENTS

We are indebted to the Departments of Neurology of the Rigshospital and of the Municipal and County Hospitals, Copenhagen, for referring patients under their care. The sural biopsies were skillfully performed by the staff of the Department of Neurosurgery, Rigshospital, Copenhagen. We thank Mr. S. Stenstrup of the Institute of Physics II for calculating the incidence of nerve fibres with myelin abnormalities with increasing fibre length.

The work was supported by grants from the Michaelsen Foundation, Copenhagen and the Muscular Dystrophy Associations of America, New York.

REFERENCES

1. Gilliatt, R.W. Proc. Roy. Soc. Med.. 1966; 59:989–993.

2. Gilliatt, R.W.Desmedt, J.E., eds. in New Developments in Electromyography and Clinical Neurophysiology; 2. S. Karger, Basel, 1973:2–18.

3. Dyck, P.J., Gutrecht, J.A., Bastron, J.A., Karnes, W.E., Dale, A.J.D. Mayo Clin. Proc.. 1968; 43:81–123.

4. Thomas, P.K. Proc. Roy. Soc. Med.. 1971; 64:295–298.

5. Dyck, P.J., Johnson, W.J., Lambert, E.H., O’Brien, P.C. Mayo Clin. Proc.. 1971; 46:400–431.

6. Hopkins, A. J. Neurol. Neurosurg. Psychiat.. 1970; 33:805–816.

7. Behse, F., Buchthal, F. Brain. 1978; 101:473–493.

8. Buchthal, F., Rosenfalck, A. Brain Res. (Amsterdam). 1966;