Abstract: Spinal cord disease is the most common cause of neurological problems in dogs. Clinical examination gives a broad localisation of the spinal cord dysfunction and a fairly precise evaluation of its severity; imaging techniques help to refine the lesion localisation. However, clinical examination reaches its limits under several circumstances because it is subjective and relies on cooperation by the animal. Imaging may be useless if the lesions are microscopic, multifocal, diffuse or purely functional.

Somatosensory evoked potentials recorded at the spine and scalp level may help in the precise localisation of a spinal cord problem, and in the objective evaluation of its severity. This paper reviews the technical and physiological background needed for this type of investigation and progress in its clinical use.

Spinal cord dysfunction is the most common problem in veterinary neurology. A fairly precise evaluation of the spinal cord status can be gained through clinical evaluation. Indeed, nerve fibers with clearly different mean axon diameters carry out several cl senhe graded disappearance of proprioception, motor command and dull pain parallels the increasing severity of cord lesions. Dull (deep, burning) pain is believed to arise from activation of unmyelinated (C) fibers by noxious stimuli (Price 1977).¹ A prognosis can be formulated based on this clinical grading and on the large body of experience gained with previous cases. The clinical grading system has limited precision because it is discontinuous and subjective. Moreover, it depends on observation of the animal, whose temperament can modify the reactions.

In addition to determining the severity of the lesion, clinical evaluation is also aimed at determining the location of the problem. Imaging plays a major role in localizing causative lesions but is dependent on presence of a macroscopic lesion. However, spinal cord dysfunction can originate in a purely functional problem, which can not be localised by imaging, e.g. spinal shock. Also, it can arise from microscopic lesions such as inflammatory lesions or infarcts. Conversely, there can be radiographic lesions that do not explain the functional impairment and therefore are irrelevant. Finally, at the present time, imaging has only a marginal role in assessing the severity of most cord dysfunction.

There are some specific factors which limit the clinical examination:

1.  Borderline cases can be difficult to classify, especially those with slight or intermittent neurological impairment and those, at the opposite end of the spectrum, which that have reached the prognostically relevant border between presence or absence of dull pain perception.

2.  Some doubt can appear when patients suffer from diseases of more than one body system: for example, cauda equina compression and concurrent hip osteoarthrosis.

3.  Decisions must sometimes be reached with uncooperative, stoic patients or even in comatose or anaesthetised patients.

4.  Patients may suffer from two concurrent neurological diseases. Old dogs with concurrent degenerative myelopathy and disk protrusion are sometimes encountered; animals with clinical signs indicating one lesion sometimes have two radiographic lesions. In such patients, a functional test could be helpful in determining whether a radiographically identified lesion is indeed the cause of the signs, or to determine which of two lesions is most likely to be the cause.

Procedures which complement the clinical examination and imaging techniques could be of value in dealing with these limitations. Such procedures should be reasonably easy and fast to perform and should provide clear answers. Some evoked potential recordings might well help this process.

An evoked potential is any electrical activity generated by excitable cells in response to a distant stimulation. With regard to spinal cord functions, two methods have been evaluated for possible clinical use in veterinary medicine: somatosensory evoked potentials (SSEP) ^2-15 and motor evoked potentials (MEP).^16-21 Somatosensory evoked potentials are recorded over the spine and skull in response to the stimulation of a peripheral sensory or mixed nerve. Motor evoked potentials are recorded in limb muscles in response to magnetic stimulation of the head or spine.

The purpose of this paper is to familiarise the reader with SSEP recordings. It is necessary first to describe certain technical and physiological factors about artificial nervous tissue stimulation and about recording from excitable tissues. After that, the clinical relevance of these methods in various spinal cord problems can be reviewed.

Peripheral Nerve Stimulation For SSEP Recordings

The same kind of stimulators are used in eliciting somatosensory potentials as are used for peripheral nervous system evaluation, either motor nerve conduction, repetitive stimulations or sensory electroneuronography. These stimulators deliver rectangular pulses between two electrodes.

Stimulation electrodes setting

Usually, needle electrodes placed in line along the nerve, i.e., in a "bipolar" setting, will be more efficient because the current is more likely to flow through the nerve fibres than if an electrode is at a distance from the nerve ("monopolar" setting). The axon membrane will be depolarised under the most negative electrode (cathode) because it brings the extra- and intra-axonal potentials closer to each other. The axonal membrane can be hyperpolarised under the most positive electrode resulting in "anodal" block of conduction; consequently, the cathode should be placed closer to the recording electrodes than the anode.

Stimulus intensity: nominal versus effective

The stimulus delivered by the stimulator is under an electronic feedback control to overcome the short-circuiting effect of the tissues interposed between the electrodes and the nerve aimed at. For example, a constant voltage unit needs current peaks to build up the sharp rise and fall of the square wave. Constant current units, with feedback control on the voltage are available also.

Regardless of the type of stimulator, the effective stimulus reaching the nerve diminishes as the distance between the electrodes and the nerve increases. Consequently, the stimulus intensity actually delivered to the nerve itself is always less than the nominal stimulus (i.e. the stimulus setting of the equipment). It is beneficial to define the stimulus intensity by a physiological effect such as the intensity needed to observe a threshold response, or the intensity that gives just a maximal muscle potential when a mixed nerve is used. Because of the constant relation between the nominal and effective voltage, using multiples or fractions of this reference voltage allows accurate determination of the effects of varying stimulus intensity.

Stimulus intensity: recruitment of nerve fibers

Increasing the stimulus intensity recruits more nerve fibres. The larger the diameter of a nerve fibre, the more sensitive it is to the stimulus. Indeed, large diameter fibres have a lower longitudinal resistance and more current flows through them. As a general rule, large afferent fibres synapse on large higher-order fibres.

A study of the SSEP evoked by tibial nerve stimulation and recorded at T13/L1 demonstrated that the measured characteristics of the potential (latency and amplitude of the first peaks) rapidly reached a stable value when the stimulus intensity was increased, although the waveform changed further. ²² Consequently, spine-recorded SSEP seems relatively insensitive to small changes in the stimulus intensity above this point.

Stimulus duration

Stimulus duration is a compromise between the time needed to overcome the capacitive properties of nerve fibres and surrounding tissues, a minimum of 50 µs, and the necessity of eliciting a synchronous volley in the nerve; 100 µs is the stimulus duration most widely used. The capacitance of a nerve fibre is also related to its diameter. For a given current, the potential change is slower in a large fibre. It is possible to preferentially recruit large fibres through the use of low intensity, long duration stimuli.

Stimulus frequency

The stimulus frequency should not be increased above 5 Hz since nerve impulses cause an inhibition of the afferents (the primary afferent depolarisation, PAD) that can last tens of milliseconds. Consequently, increasing stimulus frequency can lower the amplitude of the responses. The amplitude of the scalp-recorded SSEP may be modified by stimulus rates above 2 Hz²³ and late components of the spine recorded SSEP by stimulus rates above 4 Hz.³

Candidate nerves for stimulation

Various nerves have been stimulated, including sensory, muscle and mixed nerves. Limb nerves (tibial, peroneal, cubital and radial nerves) are most often used, but pudendal and coccygeal nerves have also been used. Proximal stimulation recruits more fibres, and gives larger potentials. But proximal stimulation induces muscle activity reflexively, and directly also with mixed nerves.⁴ Stimulation just above the carpus or tarsus compromises between activating a sufficient number of fibres and avoiding excess muscle activity. Neuromuscular blocking may be helpful when muscle artefact problems can not be solved.

The Neural Generators

Adequate stimulation of a peripheral nerve depolarizes numbers of axons simultaneously. The depolarized area is propagated centrally (and also peripherally) and can be recorded as a compound action potential from electrodes placed along the nerve. The peripheral nerve activity also evokes compound action potentials in spinal cord tracts. Peripheral nerve stimulation also leads to synaptic activity in the spinal cord, first in spinal cord segments receiving axons from the stimulated nerve. The synaptic activity leads to propagation of compound action potentials ascending spinal cord tracts, which also are recordable. The ascending activity in turn leads to synaptic activity in nuclei in the medulla and is relayed to higher centres, including cerebral cortex where synapses also are activated. Synaptic activity at all levels also is recordable and reflects arrival of the evoked activity at the synapses and the health of the cells bearing the synapses. It is important to emphasize these basic differences: compound action potentials are moving, travelling along the activated axons, whereas synaptic potentials are fixed, immobile events. They are referred to as field potentials, (or sometimes as "sink-source potentials").

Somatosensory Neural Pathways

First order sensory fibres branch profusely in the grey matter of the spinal cord segment they reach and the branches may travel cranially and caudally through several segments before terminating. In addition, many fibres from skin and joint receptors send a branch in the dorsal column up to the medulla without synapsing at the spinal cord segments. Terminals from the group Ia muscle afferents ascend in the dorsal column several cord segments before relaying in the Clarke's column.^{24, 25} A copy of this latter information travels in second order fibres in the dorsolateral fasciculus.²⁶ Pain sensation also is carried in part in the dorso-lateral fasciculus.²⁷

The information from the hind limb and caudal half of the body is conveyed to synapses in the medullary (nn. gracilis and z) and lateral cervical nuclei. From there the pathway decussates in the medulla and continues rostrally in the medial lemniscus to reach the thalamus, mostly the ventrocaudolateral nucleus.^28-30 A large fraction of the terminals from the Clark's column (dorsal spino-cerebellar tract) reaches the cerebellum.²⁶

For the information from the forelimb and cranial part of the body, the role of Clark's column and the cervical, gracilis and z nuclei is played by the cuneate and lateral cuneate nuclei.³¹ From the thalamus, the information is relayed to the primary somato-sensory cortex, which is centred on the coronal sulcus in carnivores.³²

The dorsolateral fasciculus and dorsal columns generate the earliest deflections of the spine-recorded SSEP. However, the ventral cord, comprising the ill defined spino-thalamic tract of non-primate mammals, the spino-tectal and spino-reticular pathways, as well as the ventral spino-cerebellar tract, may contribute to the later deflections of the spine-recorded SEP.^33,34 The dorsal column and dorsolateral fasciculus are also of primary importance for the scalp-recorded SSEP. However, the ventral cord might contribute as well.³⁵

Compound Action Potentials

The action potential travels along a nerve fibre. The localised potential change on the nerve fibre induces a potential change in the surrounding medium. A vector can be associated with the depolarisation, and the potential at a distant point of the surrounding medium can be calculated.³⁶ A vector with the opposite polarity is associated with the repolarisation and closely follows the first vector. The combination of these two electrical events (depolarisation followed by repolarisation) gives rise to the characteristic, triphasic positive/negative/positive wave as the action potential approaches, passes by, and travels away from the recording point (Figure 1). The compound action potential is the sum of the more or less synchronised contributions from fibres in different tracts. The compound action potential in peripheral nerves of the dog has been modelled.³⁷ An important characteristic of the compound action potential is that its latency increases when the recording point is further from the stimulus site. Its amplitude and configuration (waveform) are much more constant although dispersion of components is observed with more distant recording locations because of differences in conduction velocities among axons of different diameters.

Importantly, the configuration of the recorded potential can change in a predictable way because of physiologic or pathologic causes. If the action potential approaches but does not reach the level of the recording point, e.g., when blocked by a lesion, only a positive deflection will be recorded. If the action potential passes by but is blocked a little past the recording site, a positive/ negative wave will be recorded³⁸ (Figure 2). These two aspects of the "evoked injury potential" (EIP) can be used for clinical purposes.¹⁵

Click on any image to see a larger view

Figure 1. Evoked action potential model. Single fibre model. In this and all Figures in this paper, an upward deflection indicates positivity of the exploring electrode with respect to the reference electrode.
	A. Action potential approaching the recording point.
	B. Action potential after passing the recording point.
	C. Sum of the two contributions. Note triphasic configuration.

Figure 2. Models of evoked injury potentials (EIP) and EIP recorded from a clinical case.
	A. Conduction velocity: 100 m/s; conduction block at 500 mm from the stimulation point; recording caudad to the conduction block; contributions from the depolarisation and repolarisation, and sum. From Poncelet et al, J Am Vet Res, 1998, 59, 300-6., with permission
	B. Same conditions, recording point craniad to the conduction block. From Poncelet et al, J Am VetRes, 1998, 59, 300-6., with permission
	C. Recording from a dog with a suspected spinal cord thromboembolism in the thoraco-lumbar area: stimulation of tibial nerve, typical appearances of the evoked injury potential caudad, at the level of, and craniad to a conduction block.

Close ("Near") Field Potentials

In the grey matter, action potentials travel only short distances. The amplitude of a field potential will diminish as the recording point is moved away from the source. However, it is important to note that the latency of field potentials does not change with distance of the recording point from the source of the potentials. Predicting the way synapses, dendrites, cell bodies and axons can influence a distant recording electrode is much more difficult. However, based on the same principle as that of cable potentials, we can expect that information traveling toward the recording electrode before being cancelled by the spread through the axon terminals will give a monophasic positive potential. Conversely, information traveling away from the recording electrode will give a negative potential (Figure 2).

Excitatory synapses induce a current flow in relay neurones. These currents make the potential of a nearby electrode more negative. Physiologists using depth microelectrodes speak of current sinks for this observation. Currents also leak out of the neurones up to some distance creating an extracellular area which is relatively more positive, leading to the concept of current sources (inhibitory synapses also create a current source). Sinks are recorded near the soma and dendrites of interneurones and sources somewhere along their axon. A recording electrode closer to the sink has a negative potential whereas a recording electrode closer to the source has a positive potential. "Sink-source potential" is used as a synonym of close field potential.

If the axons in the grey matter are oriented in the same direction, their individual contributions summate and a potential can be recorded from a distance. They are said to form an open field. If they transmit the information in various directions, the individual contributions may cancel and will not be recorded from a distance; they are said to form a closed field.

Far Field Potentials

Far field potentials can be recorded under certain circumstances. They share with the close field potential the characteristic of having a fixed latency no matter the location of the recording electrode, but they do not originate in nuclei. They result from the fact that the medium surrounding the potential sources is neither homogenous nor infinite, as hypothesised for modeling compound action potentials. Six causes for far field potentials have been identified and modelled.³⁹ Changes in the conductivity of the medium, in the volume conductor size or in the direction of the nerve fibres are the most often recognised causes. Far field potentials are used for clinical purposes in human patients; they make it possible to monitor the functioning of several points along the somatosensory pathways with a fixed electrode setting.^{40, 41} They can be recorded regularly in dogs although no specific study has been published in this species.

Effect of the distance between the recording point and the generator.

The amplitude of the recorded potential decreases rapidly as the distance between the recording electrode and the generator increases. The recording electrode should be inserted as close as possible to the generator without endangering the nervous tissue, at the interarcuate ligament or lamina level for example. Actually, it is the difference between the potentials of the two electrodes which is measured. In a monopolar setting, one electrode (the recording electrode) is seated as close as possible to the generator whereas the second one (the reference electrode) is seated as far away as possible. However, the benefit of moving this second electrode away diminishes with the distance. A large interelectrode distance makes the recording of artifacts more likely. However, such a setting is needed when looking for far-field potentials.

In a bipolar setting, both electrodes are seated near the generator. This setting is not expected to give maximal amplitudes, and the interpretation of the observed deflections is more complicated (Figure 3). Efforts should be made to keep the distance between the generator and the recording electrode constant, not only to make the amplitudes comparable but also because it modifies the measured latency of compound action potentials. ⁴²

Click on the image to see a larger view

Figure 3. Effect on the potential (V) of an electrode (E) as a function of its distance from the generator. The actual measured potential difference (ΔV) depends on the reference electrode (R) location.

Recording From Nervous Tissue

Differential amplifiers.

Differential amplifiers are universally used in clinical neurophysiology. A differential amplifier looks like two amplifiers built together as mirror images; the output of these first stages are then mixed. There are two inputs that are completely equivalent, except that they influence the single output in opposite directions. Hence, the output is proportional to the potential difference between the two inputs.

Filtering

It is detrimental to amplify potentials detected by the electrodes and wires that do not originate from the generator under study; they are just "noise" (electronic or physiologic) that obscures the desired signal. The ability of the amplifier to reject these "common-mode" potentials is not infinite. Filtering is helpful in reducing "noise. In addition, a high frequency filter is needed before the next step, the analogue to digital conversion, which would otherwise generate artifacts (a phenomenon called "aliasing"). However, excessive filtering will lower, distort and suppress some components of the signal and can make the measurements pointless. Consequently, it is wise to keep the filter window as open as possible. We use a window from 20 Hz to 4 kHz in spine recorded SSEP and 10 Hz to 2 kHz for the scalp-recorded SSEP. The high frequency limit is needlessly high in this latter case: the waveforms have mostly a low frequencies content and 300 Hz is often used in other laboratories.

Analog to digital (A-D) conversion and signal averaging

The signals under study are in the µV range or even lower. Electronic and physiologic noise is of the same order of magnitude. "Signal averaging" is used to detect the signal amidst the noise. The recording is sampled at high frequency, every 30 µs for example, and the value is stored. The responses to stimulation, and the signal processing, are "time locked" to the stimulation. Stimulation is repeated and the values of each sample point of the successive trials are added. Each sample value is made of two potentials: one is the value at that moment of the signal under consideration and the second is a contribution from the noise. The former has the same sign and value on each trial (because it is time locked) while the sign and the value of the second is a random event. With successive sums, the signal amplitude grows and the noise tends towards zero. However, this trend, which may be expressed as a "signal to noise ratio"(S/N), is not linear and the improvement of the S/N fades out progressively.

It is possible to control in real time the improvement of the S/N. While the values of each sample point are added after each trial in one memory area, they can be successively added and subtracted in another memory area. Changing the signs on each successive trial cancel the contribution of the time locked signal while the random contributions still tends towards zero. This calculation of the residual noise by the "plus/minus" method adds an objective criterion when looking for signals with very low amplitude.⁴³ Indeed, the S/N equals one if no signal is present. The examiner can place a limit, for example a S/N lower than 1.2, to decide that no response exists. The statistical arguments behind such a value are debated.⁴⁴

Recording electrodes: which is which?

A recording (exploring, active, different) electrode is seated near the assumed generator and a reference (inactive, indifferent) electrode is more distant from this generator. If the examiner wants positivity of the potential of the recording electrode to be displayed as a downward deflection on the screen and printer, he must connect the recording electrode to the inverting input of the amplifier. This setting is used in EMG and most work with SSEP, but is not routinely used for some sensory evoked potentials e.g., auditory evoked potentials. This is just a matter of convention. (Positivity is displayed upwards in all the tracings of the present paper.)

When the recording electrodes are close to one another, as in several works about the scalp-recorded SSEP in the dog^5-7 designating one electrode "recording" and the other "reference" is, in fact, somewhat arbitrary. This explains in part the discrepancies among authors about the polarity of the scalp-recorded SSEP waves.

4- Spine-recorded SSEP in intact animals

Proceeding from caudal to rostral, four contributions to the SSEP recorded along the spine can be recognised. At the L7-S1 level, the root component is recorded; in the caudal lumbar area, the cord dorsum potential is prominent, and at more rostral levels, the ascending evoked potential (AEP) can be followed; in addition, the medullary component can be recorded at the level of the cisterna magna ^3,4,12,46 (Figure 4).

Spine recorded SSEP amplitude can be increased by simultaneous bilateral stimulation.¹² The amplitude of the early peaks reaches a maximum rapidly when the stimulus intensity is raised, although later peaks can be added thereafter.²² The earliest events are not sensitive to the anesthesia depth.⁴

Click on the image to see a larger view

Figure 4. Typical appearance of the spine-recorded somatosensory evoked potential in response to tibial nerve stimulation. Lowermost trace: L6-L7: root component; Middle trace: L2-L3: interneuronal component (Ni: negative waves, P1: positive wave); Uppermost trace: T12-T13: ascending evoked potential.

The root component

The root component is not very different from the potential recorded near a nerve trunk. It is a compound action potential that originates in the cauda equina nerve roots, and is usually made of three successive positive-negative deflections. It can be detected above the two to three most caudal intervertebral spaces as it merges with the interneuronal component. The latencies of its peaks increase incrementally in more rostral recordings.

The cord dorsum potential

The cord dorsum potential is best recorded over the lumbosacral enlargement. It is made of a relatively short compound action potential, also called the triphasic potential, that precedes the interneuronal component and merges with it. This potential originates in the ascending collaterals of incoming axones. ⁴⁵

The amplitude of the interneuronal component is several times larger than that of the root component and of the ascending evoked potential. The latency of its peaks does not vary, but the amplitude diminishes when the recording electrode is moved away. This close field potential is made up of a long duration negative peak followed by a blunted positive wave of even longer duration. The negative peak (action potentials traveling away from the recording electrode) originates in several places in the spinal cord grey matter depending on the kind of afferents (low or high threshold cutaneous, muscular) stimulated.⁴⁶ Increasing the strength of the stimulus evokes later negative waves that indent the long positive wave. The duration of this peak has been attributed to extensive relaying or to repetitive firing⁴⁵. The late positive wave following the large, negative interneuronal component originates from primary afferent depolarisation (PAD) (action potentials traveling toward the recording electrode). This is a presynaptic inhibitory mechanism that partially or completely blocks synaptic transmission at the axon terminal in the dorsal horn grey matter.⁴⁷ This highly generalised process filters out some inputs (while postsynaptic inhibition acts on all inputs). It is the first of several central inhibitory mechanisms that allow the organism to cope with the excess of information that reaches it.

The ascending evoked potential

The AEP has the characteristics of a compound action potential. It is of small amplitude and the components segregate (disperse) as the recordings are attempted more cranially. It can be difficult to record in the cranial thoracic area and in the cervical area owing to the difficulty of finding a proper recording electrode position. This is especially true in large dogs.¹³ The generator-electrode distance can be large in these areas and recording can be demanding.

Conduction velocity values of the ascending evoked potential have been published.^7-9,14 Mean values varied from 66 m/s to 127 m/s. Several sources of variation in the measurement of the AEP conduction velocity have been identified. The depth of the recording electrode has an influence on the measured latencies as already mentioned. If the recording electrode locations used for the conduction velocity measurement span the lumbosacral enlargement, synaptic delays are incorporated in the cranially measured latency, hence diminishing the calculated velocity. ¹² The fast, synaptically-delayed information traveling in the dorsolateral funiculi overtakes the slower, uninterrupted information traveling in the dorsal columns. This introduces a change in the maximum conduction velocity that takes place at a variable location along the thoracolumbar area, depending on the size of the dog. ⁴ Finally, the diameter of ascending fibres tapers with distance; consequently, the conduction velocity, diminishes in the cervical area.⁴⁸

Stimulating a forelimb nerve gives rise to an interneuronal component at the spinal cord cervical enlargement level, and more cranially, to an ascending evoked potential, which can be obscured by the medullary component.²

The medullary component

At the level of the cisterna magna, component arising from the nuclei in the medulla can be recorded.^2,11 It is made of a negative peak followed by a positive wave. This close field potential may originate in the cervical and medullary nuclei. Contribution from the cuneate nucleus has been demonstrated in cats. ⁴⁹ The information plunges from the dorsally seated relay nuclei to the more ventral medial lemniscus. The waveform is supposed to be caused by mechanisms similar to those generating the interneuronal component: relaying in interneurons and PAD.⁴⁷

Scalp Recorded SSEP in Intact Dogs

These potentials can be recorded using scalp electrodes for exploring and reference electrodes. Also, far field potentials can be recorded with an extracephalic reference electrode while using the same cephalic locations for exploring electrodes. The earliest close field potentials (P18 and N 30) are maximum between the nasion and the bregma. Their amplitude decreases rapidly if the recording electrode is moved away (Figure 5). They keep the same appearance no matter the position of the reference electrode (tip of the nose or base of the neck).¹³ From what is known about the somatotopic projections on the primary somatosensory cortex, recording off the midline, contralateral to the stimulated nerve, might give larger P18-N30 potentials. However, no systematic study on the effect of the recording position location in dogs has been published. Longer latency positive/negative potentials can be recorded in animals that are only sedated or anaesthetised with narcotics.^5,6 The latter potentials are suppressed by barbiturates and are quite sensitive to halogenated anaesthetics.⁵⁰ Even the waveform of the first deflections is modified by the kind of sedation or anaesthesia.⁶ In addition to variable use of sedatives and anaesthetics, the recording electrode locations, often of the bipolar type, varied among existing studies, making comparisons difficult.

The inter-individual reproducibility of the amplitude and even of the latency of the first waves is not as good as with the spine recorded SSEP. Latency is of course related to the body size, but the amplitude is influenced too. Reference values for latency and amplitude as a function of body size are available.¹³

Click on the image to see a larger view

Figure 5. Typical appearance of the scalp-recorded somatosensory evoked potential in a barbiturate-anesthetized dog. Origins of the arrows correspond to locations of the recording electrodes; reference at base of the neck. P18, N30: near field potentials, unaffected by the location of the reference electrode; P13 far field potential recorded with only an extracephalic reference electrode.

SSEP in Dogs with Spinal Cord Compression

There is only a limited number of observations about the SSEP changes in dogs with naturally acquired spinal cord disease.^4,9,14,51 This contrasts with the abundant literature on SSEP changes after experimentally inflicted spinal cord lesions in laboratory species and with the studies in human.

SSEP cranial to the cord lesion and AEP conduction velocity through the damaged area

Scalp recorded SSEP are more resistant than spine recorded SSEP when the severity of experimental cord lesion increases.⁵² The observation of scalp recorded SSEP or their reappearance anticipated the functional recovery in a series of dogs with naturally acquired spinal cord compression,⁵¹ a result consistent with the observation of experimental compression of the cervical cord in dogs.⁵³ In another study involving dogs with naturally acquired spinal cord compression, ¹⁴ conduction through the damaged area could not be detected with the spine recorded SSEP in 40% of dogs with grade 3 clinical disability (paraplegic, with some voluntary motor activity present) and in 70 % of the grade 4 dogs (paraplegic with some pain perception present). The scalp-recorded SSEP were more resistant with a potential recordable in all grade 3 dogs and in 60% of the grade 4 dogs. Consequently, a sizeable proportion of cases should be considered as false negative with these tests when compared with the clinical evaluation. No scalp-recorded SSEP was recordable in dogs without pain perception (grade 5). While failure to record SSEP from the head does not necessarily carry the same desperate functional prognosis as the absence of dull pain perception, the recording of any activity cranial to the lesion carries a good prognosis. ⁴

An interesting feature of spine-recorded SSEP was the finding that an evoked injury potential (EIP) could be recorded in most dogs that had focal lesions.

Dispersion of the AEP recorded cranial to an old lesion¹⁴ or a slowly growing mass (Figure 6) is obvious. These features can be of value when one old, clinically silent and one acute, contemporary lesion coexist in the thoraco-lumbar area: most probably the contemporary one is close to an EIP; also the duration of the AEP will be increased cranial to the old one (Figure 7, Figure 8).

A significant diminution of the AEP conduction velocity across the damaged cord area is found in ataxic, ambulatory patients.¹⁴ Experimental work has demonstrated that sustained compression typically causes latency changes craniad to a sustained compression.⁵⁴ The ratio of the conduction velocity across the damaged area to the duration of the SSEP recorded from more a more cranial site has been used to evaluate the severity of spinal cord lesions in dogs.⁹ However, the cranially recorded SSEP was actually an EIP in severely affected dogs in this study. The confusion made this attempt useless.

In contrast to focal lesions, in diffuse neurological problems such as degenerative radiculomyelopathy, the AEP may be expected to progressively lose amplitude and become more dispersed as the recording site is moved more cranially. However, the difficulty in obtaining interpretable recordings in the thoracic area of large dogs can make this difficult to demonstrate convincingly.

Click on any image to see a larger view

Figure 6. Dispersion of ascending evoked potential components and increased duration of the AEP craniad to a slowly growing spinal cord tumor located on the right side at T13-L1. (Radiograph below)
	Upper traces: Stimulating left, then right, tibial nerve. There is a delayed scalp-recorded evoked potential with right stimulation as compared with left. Middle traces: Bilateral tibial nerve stimulation (simultaneous): There is an AEP superimposed on an EIP. The T12-T13 recording was repeated and the traces superimposed to show the high degree of replicability of the spine-recorded SSEP. Bottom traces: Stimulation of the left tibial nerve, then the right. At this level of the spine, superimposition of the AEP on the EIP is more obvious when the left tibial nerve is stimulated than when the right tibial nerve is stimulated and an asymmetry of the responses is evident.
	Myelogram of the same dog. Radiograph from Poncelet et al., J Am Anim Hosp Assoc, 1994, 30, 213-6,

Figure 7. Functional significance of radiographic findings. Pekinese, male, 8 years, pain followed by hind limb ataxia, subtle forelimb involvement evident in the hopping reactions, 2 months duration.
	A. Thoracolumbar radiograph with an old disc herniation at the L2-L3 intervertebral space
	B. Cervical spine myelogram with a C4-C5 disk extrusion
	C. Somatosensory evoked potentials in the thoracolumbar area are essentially normal; lesion in this area is not likely to contribute to the current signs.

Figure 8. Functional significance of radiographic findings. Danish Dog, female, 9 years of age, with a two months history of hind limb ataxia followed by paraplegia for one week, with upper motor neuron signs.
	Myelogram revealed two small disk protrusions at the L1-L2 and T13-L1 intervertebral spaces.
	At these same spinal levels, two evoked injury potentials with different onset latencies can be seen; both locations are probably clinically significant. All traces show excellent replicability.

Evoked Injury Potentials

The occurrence of the EIP craniad to a cord lesion was well described from experimental work in cats.⁵⁵ It has also been used as an intraoperative localising tool for acute cervical spinal cord injury⁵⁶ and during cervical spinal cord tumour removal⁵⁷ in man.

Unlike SSEP recorded from the head, the EIP has a tendency to be larger in more severe cases. A study of the EIP involving 25 dogs with spontaneous spinal cord compression was performed.¹⁵ The aim of the study was to look for a maximal conduction block location. Indeed, the EIP waveform changes were found to depend on the location of the recording electrode relative to the conduction block location (Figure 2). Recording sites were placed every 5 to 10 mm to follow the changes in the EIP configuration. Maximal conduction block was assumed to be midway between the two electrodes recording the change of the EIP from biphasic to monophasic. This point, measured on a radiograph with the recording electrodes in place, was always caudal to the actual radiographic location of the spinal cord compression. The distance between the maximal conduction block and the spinal cord compression was larger in the most affected dogs, probably reflecting the caudal spread of the impairment caused by the secondary lesion phenomenon. This distance measurement may allow a further grading among the dogs without pain perception, something that is completely out of reach with the clinical examination. In this study, no statistical difference could be found between the dogs with clinical grades 4 and 5, but the sample size for the grade 4 was rather small. The distance, conduction block to mechanical compression, is the most promising objective prognostic criterion in severe spinal cord compressions to date.

Other clinical uses of the SSEP

SSEP has been used as a diagnostic part of brachial plexus problems evaluation,⁵⁸ for functional monitoring of the sciatic nerve during hip surgery⁵⁹ and in the evaluation of dogs with cauda equina syndrome.⁶⁰

There is a potential use of scalp-recorded SSEP in brain problems that has not been worked out. The medullary component could be used in such cases as an internal control and to exclude a concurrent spinal cord involvement.

References

1.  Price DD, Dubner R. Neurons that subserve the sensory-discriminative aspects of pain. Pain 1977; 3: 307-38.

2.  Parker AJ. Evoked cisterna cerebellomedularis potentials in the clinically normal dog. Am J Vet Res 1978; 39: 1811-5.

3.  Holliday TA, Weldon NE, Ealand BJ. Percutaneous recording of evoked spinal cord potentials of dogs. Am J Vet Res 1979; 40: 326-33.

4.  Holliday TA. Electrodiagnostic examination: Somatosensory Evoked Potentials and Electromyography. Vet Clin North Am 1992; 22: 833-57.

5.  Kornegay JN, Marshall AE, Purinton PT,et al. Somatosensory-evoked potentials in clinically normal dogs. Am J Vet Res 1981; 42: 70-3.

6.  Purinton PT, Oliver JE, Kornegay JN, et al. Cortical averaged potentials produced by pudendal nerve stimulation in dogs. Am J Vet Res 1983; 44: 446-8.

7.  Sims MH, Selcer RR. Somatosensory-evoked and spinal cord-evoked potentials in response to pudendal and tibial nerve stimulation in cats. Am J Vet Res 1989; 50: 542-5.

8.  Redding RW. Spinal evoked potentials in the dog and cat. Proc Am Col Vet Int Med, 1985, 161.

9.  Shores A, Redding RW, Knecht CD. Spinal-evoked potentials in dogs with acute compressive thoracolumbar spinal cord disease. Am J Vet Res 1987; 48: 1525-30.

10. Steiss JE, Wrigth JC. Maturation of spinal-evoked potentials to tibial and ulnar nerve stimulation in clinically normal dogs. Am J Vet Res 1990; 51: 1427.

11. Oliver JE, Purinton PT, Brown J. Somatosensory evoked potentials from stimulation of thoracic limb nerves of the dog. Prog Vet Neurol 1990; 1: 433-43.

12. Poncelet L, Delauche A, Vinals C,et al. Effect of bilateral tibial nerve stimulation on the spinal evoked potential in dogs. Am J Vet Res 1992; 53: 1305-8.

13. Poncelet L, Michaux Ch, Balligand M. Effect of body size on tibial somatosensory evoked potentials in dogs. Am J Vet Res 1993; 54: 178-182.

14. Poncelet L, Michaux Ch, Balligand M. Somatosensory potentials in dogs suffering naturally-acquired thoracolumbar spinal cord disease. Am J Vet Res 1993; 54: 1935-40.

15. Poncelet L, Michaux Ch, Balligand M. A study on the evoked-injury potential using dogs with naturally acquired thoraco-lumbar spinal cord disease and computer modeling. Am J Vet Res 1998; 59: 300-6.

16. Heckman R, Hess CW, Hogg HP et al. Transkranielle Magnetstimulation des motorische Kortex und perkutan Magnetstimulation peripher-nervöser Strukturen beim Hund. Schweis Arch Tierheilk 1989; 131:341-50.

17. Sylvester AM, Cockshutt JR, Parent JM et al. Magnetic motor evoked potentials for assessing spinal cord integrity in dogs with intervertebral disc disease. Vet Surg 1993; 22: 5-10.

18. Van Ham LML, Vanderstraeten GGW, Mattheeuws DRG et al. Transcranial magnetic motor evoked potentials in sedated dogs. Prog Vet Neurol 1994; 5: 147-54.

19. Van Ham LML, Mattheeuws DRG, Vanderstraeten GGW. Transcranial magnetic motor evoked potentials in anesthetized dogs. Prog Vet Neurol 1995; 6: 5-12.

20. Van Ham LML, Mattheeuws DRG, Vanderstraeten GGW. Sufentanil and nitrous oxide anaesthesia for the recording of transcranial magnetic motor evoked potentials in dogs. Vet Rec 1996; 138: 642-45.

21. Van Ham LML, Nijs J, Vanderstraeten GGW et al. Comparison of two techniques of narcotic-induced anaesthesia for use during recording of magnetic motor evoked potentials in dogs. Am J Vet Res 1996; 57: 142-6.

22. Cozzi F, Poncelet L, Michaux Ch, et al. Effect of stimulus intensity on the tibial somatosensory evoked potential in dogs. Am J Vet Res 1998; 59: 217-20.

23. Wiederholt WC. Recovery function of short latency components of surface and depth recorded somatosensory evoked potentials in the cat. Electroencephalography Clin Neurophysiol 1978; 49: 259-65.

24. Horch KW, Burgess PR, Whitehorn D. Ascending collaterals of cutaneous neurons in the fasciculus gracilis of the cat. Brain Res 1976; 117: 1-17.

25. Brown AG. Ascending and long spinal pathways: dorsal columns, spinocervical tract and spinothalamic tract. In Handbook of Sensory Physiology. Vol. II, Somatosensory System, A Iggo, ed., Springer, New York, 1973, pp 315-338.

26. Johansson H, Silfvenius H. Axon-collateral activation by dorsal spino-cerebellar tract fibers of group I relay cells of nucleus Z in the cat medulla oblongata. J Physiol 1977; 265: 341-69.

27. Kennard MA. The course of ascending fibers in the spinal cord of the cat essential to the recognition of painful stimuli. J comp Neurol 1954; 100: 511-24.

28. Brown AG, House CR, Rose PK, et al. The morphology of spinocervical tract neurons in the cat. J Physiol 1976; 260: 719-38.

29. Ha H, Liu CN. Organization of the spino-cervico-thalamic system. J Comp Neurol 1966; 127: 445-70.

30. Hagg S, Ha H. Cervicothalamic tract in the dog. J Comp Neurol 1970; 139: 357-74.

31. Rosén I, Sjölund B. Organization of group I activated cells in the main and external cuneate nuclei of the cat: identification muscle receptors. Exp Brain Res 1973; 16: 221-37.

32. Kappers CUA, Huber GC, Crosby EC. The development of the cortex in mammals. In the comparative anatomy of the nervous system of vertebrates, including man. Vol. III. Hafner Publishing Company, New York, 1967, pp 1517-1674.

33. Weldon NE, Holliday TA. Pathways of evoked spinal cord potentials after forelimb stimulation. J Am Vet Med Assoc 1979; 174: 954-8.

34. Ealand Snyder BG, Holliday TA. Pathways of ascending evoked potentials of dogs. Electroencephalography and Clin Neurophysiol 1984; 58: 140-54.

35. Ducati A, Schieppati M. Spinal pathways mediating somatosensory evoked potentials from cutaneous and muscle nerves in the cat. Acta Neurochir 1980; 52: 99-104.

36. Brown BH. Theoretical and experimental waveform analysis of human compound nerve action potentials using surface electrodes. Med & biol Eng 1968; 6: 375-86.

37. Niederhauser UB., Holliday TA, Hyde DM.,et al. Correlation of sensory electroneurographic recordings and myelinated fiber diameters of the superficial peroneal nerve of dogs. Am J Vet Res 1990; 51: 1587-95.

38. Tani T, Ushida T, Yamamoto, H. A computer simulation of conduction block: loss of phase cancellation as the cause of killed-end potential and increased amplitude. Fifth international symposium on spinal cord monitoring, June 2-5, 1992, London. p13.

39. Dumitru D, King JC. Far-field potential production by quadrupole generators in cylindrical volume conductors. Electroencephalography Clin Neurophysiol 1993; 88: 421-31.

40. Lueders H, Andrish J, Gurd A, et al. Origin of far field subcortical potentials evoked by the stimulation of the posterior tibial nerve. Electroenceph Clin Neurophysiol 1981; 52: 336-44.

41. Yamada T, Mashida M, Kimura J. Far-field somatosensory evoked potentials after stimulation of the tibial nerve. Neurology 1982; 32: 1151-8.

42. Sarnovsky RJ, Cracco RQ, Vogel HB etal. Spinal evoked response in the cat. J Neurosurg 1975; 43: 329-36.

43. Wong PKH, Bickford RG. Brain stem auditory evoked potentials: the use of noise estimate. Electroenceph Clin Neurophysiol 1980; 50: 25-34.

44. Don M, Elberling C, Waring M. Objective detection of averaged auditory brain stem responses. Scand Audiol 1984; 13: 219-28.

45. Yates BJ, Thompson FJ., Mickle JP. Origin and properties of spinal cord field potentials. Neurosurg 1982; 11: 439-50.

46. Beall JE, Applebaum AE, Foreman RD et al. Spinal cord potentials evoked by cutaneous afferents in the monkey. J Neurophysiol 1977; 40: 199-211.

47. Schmidt RF. Presynaptic inhibition in the vertebrate central nervous system. Ergebn Physiol 1971; 63: 20-101.

48. Maturation of spinal-evoked potentials to tibial and ulnar nerve stimulation in clinically normal dogs. Am J Vet Res 1990; 51: 1427-32.

49. Andersen P, Eccles JC, Schmidt RF et al. Slow potential waves produced in the cuneate nucleus by cutaneous volleys and by cortical stimulation. J Neurophysiol 1964; 27: 78-91.

50. Allison T, Hume AL. A comparative analysis of short latency somatosensory evoked potentials in man, monkey, cat and rat. Exp neurol 1981; 72: 592-6.

51. Bright RM, Breazile JE, Bojrab MJ. Prognostic application of cortical evoked responses in dogs with spinal cord injury. J Vet Surg 1977; 6: 55-9.

52. Schramm J, Shigeno T, Brock M. Clinical signs and evoked response alterations associated with chronic experimental cord compression. J Neurosurg 1983; 58: 734-41.

53. Kojima Y, Yamamoto T, Ogino H et al. Evoked spinal potentials as a monitor of spinal cord viability. Spine 1979; 4: 471-7.

54. LeCouteur RA, Holliday TA. Electrophysiological studies of experimental spinal cord injury in dogs. Proc Am Col Vet Int Med 1984: 170-7.

55. Schramm J, Kraus R, Shigeno T et al. Experimental investigation on the spinal cord evoked injury potential. J Neurosurg 1983; 59: 485-92.

56. Katayama Y, Tsubokawa T, Yamamoto T et al. Preoperative determination of the level of spinal cord lesions from the killed end potential. Surg Neurol 1988; 29: 91-4.

57. Morioka T, Kiyotaka F, Mitani M et al. Intraoperative localization of a cervicomedullary glioma from the killed end potential: illustrative case. Neurosurg 1990; 26: 1038-41.

58. Bailey CS. Patterns of cutaneous anaesthesia associated with brachial plexus avulsion in the dog. J Am Vet Med Assoc 1984; 185: 889-99.

59. Thompson SE, Moore MP, Lincoln JD et al. Intraoperative monitoring of ischiatic nerve function with somatosensory evoked potentials. Vet Surg 1990; 19: 276-82.

60. Kornberg M, Bichsel P, Lang J. Electromyographie und spinal evozierte Potentiale beim Cauda equina Syndrom des Hundes. Sweiz Arch Tierheilk 1989; 131: 287-98.