The development of sophisticated imaging techniques has had a great impact on the role of somatosensory evoked potentials (SEPs) in the clinical setting. Their role in the evaluation of most neurologic diseases is limited; nevertheless, as described in this chapter, SEPs are valuable as a diagnostic and prognostic test in several clinical situations. Their role in the operating room and intensive care unit has expanded, and interest remains high in SEPs as a research tool for unraveling fundamental aspects of central sensory physiology.
Stimulation of a mixed nerve is used most commonly to evoke the SEP. This type of stimulus activates predominantly—if not entirely—the large-diameter, fast-conducting group Ia muscle and group II cutaneous afferent fibers. The number of axons that are activated synchronously is large, so the response is relatively large and easy to record. The SEP amplitude is almost maximal when the peripheral nerve action potential is only 50 percent of its maximum. This translates into a stimulus intensity of about twice sensory threshold, which is well tolerated by patients. Selective intrafascicular nerve stimulation has provided evidence for a direct muscle afferent fiber (Ia) projection to the human somatosensory cortex. However, when a mixed nerve is stimulated, both group Ia muscle afferents and cutaneous group II afferents contribute to the resulting SEP, and this method detects selective impairment of sensory modalities poorly. Results can be normal even when selective sensory deficits are considerable. Discrimination can be improved by using a variety of mechanical or thermal stimuli. These techniques selectively activate specific sensory end-organs. Unfortunately, because of the rather small number of axons stimulated and the wide range of their conduction velocities, the SEPs elicited are usually small in amplitude and desynchronized. Many hundreds of responses need to be averaged, and even then the onset latency may be difficult to identify. These issues have limited the clinical utility of SEPs elicited by mechanical and thermal stimulation.
Selective ablation of the dorsal columns in animals greatly attenuates or abolishes the SEP, thus indicating that within the spinal cord the SEP is mediated predominantly via the dorsal columns. Therefore, diseases of the dorsal columns, with associated impairment of vibration and position sensation, invariably are associated with abnormalities of the SEP. Conversely, spinal cord lesions that do not interrupt the dorsal columns are often associated with a relatively normal SEP after mixed nerve stimulation. Tourniquet-induced ischemia in humans has been shown to abolish short-latency before long-latency SEP components, which suggests that the short-latency components are mediated by different centrally conducting afferent tracts.
Although, in general, the SEP is best recorded over the somatosensory cortex, topographic mapping indicates that several of its components are distributed widely over the scalp, and some are recorded maximally outside the somatosensory cortex. The SEP monitors more than just the somatosensory pathways. For example, abnormalities occur commonly in certain primary diseases of the motor system (e.g., amyotrophic lateral sclerosis) and should not cause concern.
Mixed Nerve Stimulation
Electrical stimulation of a mixed nerve initiates a relatively synchronous volley that elicits a sizeable SEP. It therefore has become the standard for clinical use. The stimulus intensity required to elicit an SEP of maximum amplitude should be one that induces a mixed nerve action potential just exceeding 50 percent of its maximum amplitude. This will elicit a slight twitch of the muscle, indicating that motor as well as sensory axons are activated. Stimulation of Ia afferents is not uncomfortable, even at repetition rates of 5 Hz. Their threshold for activation is lower than that of motor axons (as indicated by the intensity required to induce an H reflex). If the stimulus intensity is too high, occlusion of the rapidly converging afferent impulses may occur, depending on the limb that is stimulated. A short-duration (200 to 300 μsec) stimulus is popular, but stimuli of longer duration (1,000 μsec) and appropriately lower intensity are more likely to activate preferentially the group Ia and II afferents. A repetition rate of 3 Hz is convenient and without discomfort, and faster rates are usually tolerable. Repetition rates of up to 15 Hz do not alter the SEP. Sometimes electrocardiographic artifacts are troublesome, a point particularly relevant when trying to obtain noncephalic, referential recordings of SEPs elicited by stimulation of nerves in the lower limbs. This problem usually can be solved by using the electrocardiogram to trigger the stimulus.
Cutaneous Nerve Stimulation
Use of cutaneous nerve stimulation has particular application to peripheral nerve diseases and should be considered in several circumstances:
To assess the integrity of specific cutaneous nerves that are not studied readily by conventional sensory nerve conduction techniques (e.g., the lateral femoral cutaneous, saphenous, and antebrachial cutaneous nerves). Sensory nerve action potentials can be difficult to record from these nerves, whereas the corresponding SEP is usually of sufficient size to measure with ease.
To measure peripheral sensory conduction when not otherwise possible because the sensory nerve action potential is small or absent. In hereditary sensory motor neuropathy types I, II, and III, sensory nerve action potentials are often absent; but the SEP, although dispersed, can be elicited from stimulation of two or more sites along a cutaneous nerve, and thereby allows one to deduce a conduction velocity.
To evaluate isolated root function. When compared with mixed nerve stimulation, the segmental specificity of cutaneous stimulation is increased, even though most cutaneous nerves are derived from at least two sensory segments.
To assess dubious patchy numbness for medicolegal reasons by stimulating homologous areas of affected and normal skin supplied through cutaneous terminals.
Dermatomal stimulation is even more segmentally specific than is cutaneous nerve stimulation, which invariably activates two or more dermatomes. However, the ascending volley is sometimes very desynchronized, making the SEP difficult to interpret. The technique has been used most often to assess function of the lumbosacral roots. For L5, the medial side of the first metatarsophalangeal joint or the dorsal surface of the foot between the first and second toes is stimulated. For S1, the lateral side of the fifth metatarsophalangeal joint is stimulated. Care must be taken to avoid stimulus spread to neighboring dermatomes, underlying muscle (which induces activity of Ia afferents), and digital cutaneous nerves. Such stimulus specificity can be achieved if the stimulus is kept at 2.5 times the sensory threshold, which gives about 80 percent of the maximum amplitude. Normative data for the L5 and S1 dermatomes are well established.
Because of the high specificity of radiant heat for nociceptor activation, laser-evoked brain potentials (LEPs) are a suitable tool for testing small-fiber and spinothalamic tract function. The CO 2 laser pulse is absorbed completely within the epidermis and activates a very limited number of superficial afferents, mostly A-delta fibers, belonging predominantly to the pain system.
A pathway from cortical layer 6 to the thalamus directly excites relay cells and indirectly inhibits them via the thalamic reticular nucleus. Laser-scanning photostimulation, which specifically activates cell bodies or dendrites, has been used to understand the circuit organization of this cortical feedback stimulating the primary somatosensory cortex. LEP components have been recorded by intracranial electrodes and these mirrored the ones picked up from the Cz lead, thus suggesting that they probably are generated by the opposite pole of the same cortical sources producing the scalp responses. In the intracranial traces, there was no evidence of earlier potentials possibly generated within the thalamus or of subcortical far-field responses. This means that the nociceptive signal amplification occurring within the cerebral cortex is necessary to produce identifiable LEP components.
Based on studies in normal subjects, as well as in patients with sensory loss, the ascending signals following CO 2 laser stimulation are conducted through A-delta fibers and the spinothalamic tract. The peripheral and spinal cord components of the LEP are not detectable, whereas a cortical potential can be recorded with a wide distribution on the scalp and with a maximal amplitude at the vertex. Using a modified technique, conduction velocity has been determined in C-fibers (2 m/sec) as well as in A-delta fibers (10 m/sec) of the spinothalamic tract.
LEPs have been demonstrated to complement quantitative sensory testing in assessing the relationship between skin nerve denervation and regeneration. They have also been used to document posterior root impairment objectively in patients with acute monosegmental radiculopathy. Such studies open the perspective of electrophysiologically differentiating the presence or absence of posterior root pathology in patients with similar clinical symptoms but possibly different prognoses.
LEPs have been evaluated during sleep. In the awake subject, nonpainful stimulation evokes early- and middle-latency components (N20, P30, and N60), and painful stimulation additionally evokes later pain-specific components (N130 and P240) at the Cz electrode. During sleep, N20 and P30 do not change in amplitude, N60 shows a slight but significant amplitude reduction, and N130 and P240 decrease markedly or disappear entirely.
Contact Heat Stimulation
A specially designed, commercially available, contact heat stimulator can be used to deliver recurrent heat stimuli at a specified temperature to the skin while the cerebral responses, mediated primarily by A-delta fibers, are recorded over the scalp. The thermode has a circular contact area of 572.5 mm with stimulators in two layers working together; two thermocouples embedded within its outer coating measure the skin temperature. The thermode allows for very rapid heating and cooling of the skin. SEPs elicited by contact heat stimulators have been used to assess the function of A-delta fibers in patients with neuropathy. The response latencies of such SEPs are longer than those elicited by laser stimulation, perhaps because of the longer duration of the stimuli and the time required to conduct the heat through the skin to stimulate nociceptors. The approach is easier to use in the clinic than laser stimulation, does not require eye protection, and has a lesser risk of inducing burns or erythema.
Recording and filtering of SEPs
Surface or needle electrodes can be used for recording SEPs. Scalp needle electrodes are inserted easily into the scalp but are not popular because of their higher impedance, the discomfort caused during insertion, and concern about infection. Now that disposable needles are widely available, they are preferred when needle electrodes are to be used. Recording montages are either “cephalic bipolar,” in which both electrodes are placed on the head; or “referential,” with the reference electrode placed at a noncephalic site. A cephalic bipolar montage has the advantage of being relatively free from noise and is preferred for routine clinical use; however, small-amplitude, far-field potentials, which reflect activity within subcortical structures, are largely canceled. The Cz′–Fz derivation generally is used for recording the cortical component of tibial SEPs, but considerable intersubject and intrasubject variation exists in the cortical distribution of tibial SEP components. In particular, P37 in the Cz′–Fz recording may be very small, even in normal subjects, and a Cz′–Cc derivation as the single cortical bipolar recording arrangement may be better suited for routine clinical examinations. Far-field potentials can be recorded only with noncephalic references (e.g., with the reference electrode placed on the opposite mastoid, shoulder, arm, hand, or knee; linked mastoids or ear lobes may also be used).
Multichannel recording using both cephalic and noncephalic montages is advantageous; however, the number of channels (recording derivations) used should be dictated by the specific reason for performing the SEP study. For example, in field distribution and brain-mapping studies, 16 or more channels may be important, whereas one channel is sufficient when the SEP is being used to assess peripheral nerve disease.
The near-field potentials recorded with a bipolar cephalic derivation are characteristically of negative polarity and of relatively large amplitude. The amplitude of near-field potentials falls rapidly as the electrode is moved away from the generator source. Some small-amplitude far-field potentials may also be recorded with bipolar cephalic derivations, but this is too variable for clinical purposes, and referential recording is required to identify far-field potentials with confidence. These potentials are characteristically of small amplitude, and are recorded with equal ease and amplitude over a wide area of the scalp. They are usually positive and monophasic at the active electrode, reflecting a moving front approaching the recording electrode, but under some circumstances they may be biphasic and of either polarity.
Different recording montages are used in different laboratories, depending on the purpose of the study, and recommendations by various professional organizations also differ in their details. For general clinical purposes, a common but minimal montage for recording median or other upper-limb SEPs is between the ipsilateral and contralateral Erb’s point (channel 1); the fifth cervical spine and contralateral Erb’s point (channel 2); midway between the ipsilateral C3/4 and P3/4 (CP placement; also known as C3′ or C4′, located 2 cm behind the C3 or C4 placements) and the contralateral Erb’s point (channel 3); and contralateral and ipsilateral CP placements (channel 4). For the tibial SEP, recordings may be made between the distal and proximal regions of the popliteal fossa (channel 1); the L1 and L3 spine (channel 2); Cz or CPz (Cz′) and Fz or Fpz (channel 3); and the ipsilateral CP placement and Fpz (channel 4).
Small-amplitude components of the SEP are composed of both high and low frequencies, and filtering can be problematic. Too wide a bandpass results in a “noisy” SEP, whereas a restrictive bandpass attenuates either the high- or low-frequency components, depending on the settings chosen. There is no “correct” filter setting; the choice is best related to the particular task at hand. For general purposes, a relatively broad bandpass (10 to 2,500 Hz) is suitable, but even the same bandpass has differing effects, depending on the amplifiers used. It is reasonable to experiment with different filter settings for a given recording device. Restrictive filtering of between 150 and 300 to 3,000 Hz enhances high-frequency, small-amplitude, near- and far-field components, but does so at the expense of the low-frequency components.
Restrictive analog filtering induces a phase shift of components, which invariably causes distortion of these components and may create artifactual components. Digital filtering, available on some equipment, is designed for zero phase shift and does not distort components. Digital filtering does not create new peaks, but it greatly enhances those normally present ( Fig. 26-1 ). This characteristic is of particular importance for recording SEPs evoked by leg stimulation because it makes the small-amplitude components more easily identifiable ( Fig. 26-2 ). Human median nerve SEPs exhibit a brief oscillatory burst of low amplitude (less than 500 nV) and high frequency (about 600 Hz), which can be isolated by digital filtering. These oscillations are superimposed on the primary cortical response N20, but the N20 component and the high-frequency oscillations are dissociated functionally and represent independent SEP components. The N20 response is generated mainly by excitatory postsynaptic potentials in pyramidal cells of Brodman area 3b, whereas the high-frequency oscillations seem to be generated intracortically by postsynaptic inhibitory interneurons.
Measurement of the SEP
Several characteristics of the SEP can be measured, including onset latency, peak latency, interpeak latency, amplitude (including the presence or absence of components), morphology, and dispersion of the SEP. Side-to-side comparisons may be useful. Latency is the easiest SEP feature to measure and standardize, but it gives rather limited information. Other characteristics (e.g., morphology or dispersion) are more variable and difficult to interpret.
Latency varies with limb length, whereas interpeak transit (conduction) times are reliable parameters that are independent of limb length and are usually independent of peripheral nerve disease. Central afferent pathways do not mature at the same rate as peripheral pathways, and adult values for conduction velocity are not attained until 7 or 8 years of age. Aging is associated with prolongation of SEP latencies. This prolongation is not simply a reflection of slowed peripheral conduction because central conduction times are also slowed significantly. Latency or interpeak conduction times are considered abnormal when they are more than 3 standard deviations above the normal mean. They are measured easily but may be normal in the face of obvious clinical impairment. It is important to be aware that SEP components may come to be so attenuated that they cannot be recognized, particularly in the case of SEPs evoked by stimulation of leg nerves. Care is required to distinguish between a delayed SEP and an absent component; when a component is absent, the subsequent component may be mistaken for the missing potential and the latency of the SEP erroneously considered to be delayed.
There is an asymmetry of the median-derived SEP in relation to handedness. This occurs only at the cortical level. However, both functional and morphologic cortical asymmetry of somatosensory representation appears to vary independently of motor and language functions. Other factors that modify the SEP include acute and regular exercise, which shortens SEP latency and decreases its amplitude, and attention. There are differences between elderly and young subjects in the brain mechanisms underlying selective attention. The median SEP amplitude increases during a selective attention condition in young subjects but not in the elderly, who are unable to divert their attention from the electrical stimuli being delivered to the median nerve. These observations agree with recent hypotheses that suggest a decrease of the inhibitory control of attention mechanisms during aging.
Absence of components normally present is regarded as abnormal. The absolute amplitude of SEP components is quite variable, but an interside difference exceeding 50 percent is regarded as abnormal by many clinical neurophysiologists. A difference of this extent may be indicative either of central conduction block or of considerable neuronal/axonal loss. However, complex facilitatory and occlusive synaptic interactions can result in “central gain” within the nervous system and may prevent amplitude reduction in the SEP and thereby mask axonal or neuronal loss. Amplitude reduction of the SEPs also occurs during voluntary movement, a phenomenon commonly known as “gating.” This presumably prevents irrelevant afferent inputs during movement from reaching consciousness. The gating effect takes place at the cortical level; subcortical SEP components remain unchanged during movement. Gating of SEPs is selective to muscles utilized for specific tasks and may persist after repetitive movements, leading to an altered pattern and level of cortical processing. It has been proposed that this results in plastic changes in the cortex that may be relevant to the development of overuse syndromes. SEP amplitude increases in the elderly, and “giant” potentials typify hereditary myoclonic epilepsy.
The morphology (shape) and dispersion of the SEP are difficult features to quantitate, but both may be abnormal before or in the absence of latency prolongation or amplitude reduction. Computer-assisted methods for quantifying both dispersion and morphology of the SEP have been described, but the methodology is not suitable for routine clinical use.
Neural generators of the SEP
The different SEP components are designated by their polarity and latency. Polarity at the active electrode is indicated as either positive (P) or negative (N). Debate continues regarding the origin of some SEP components, and no agreement has been reached regarding the numerical latency value ascribed to each component. The origin of specific, early-latency SEP components elicited by median and tibial nerve stimulation and recorded over the scalp is shown in Table 26-1 , which summarizes the most widely held views at this time.
|P9||Just distal to the brachial plexus||Sometimes bilobed P9a and P9b (depending on arm position)|
|P11||Posterior root entry (presynaptic)|
|P13/P14||Medial lemniscus (postsynaptic)||P13 may reflect synaptic activity of posterior horn interneurons|
|N18||Between the upper pons and midbrain||N18 remains intact after thalamic lesions|
|N20||Area 3b in the posterior bank of the rolandic fissure||Diagnostically the most useful peak to measure|
|P22||Motor area 4||Increases with age|
|N30||Supplementary motor area||Decreases with age|
|P18||Sacral plexus||Analogous to median P9|
|P31||Gracile nucleus||Analogous to median P14|
|N34||Brainstem||Analogous to median N18|
|P37/N37||Primary somatosensory cortex||Analogous to median N20|
|Multiple cortical generators are involved|
It is accepted generally that the different components of the SEP predominantly reflect sequential activation of neural generators excited by the ascending volley. This concept is appealing because it provides a rational base for interpreting an absent or abnormal component in relation to a specific anatomic lesion. However, factors other than neural generators (i.e., synapses in relay nuclei) are important in the origin of some SEP components, particularly the small far-field potentials recorded with noncephalic references. In particular, stationary far-field peaks (e.g., P9, P11, P13, and P14 evoked by median nerve stimulation) reflect propagated volleys of action potentials traveling in axons and can be recorded as the traveling volley approaches but before it actually reaches the active recording electrode. As shown in Figure 26-3 , this property results from physical changes in the surrounding volume conductor, including the resistance or impedance of the volume conductor, sites of axonal branching, and anatomic orientation of the traveling impulse.
Neurosurgical interventions allow for intracerebral near-field recordings of SEPs from the deep lemniscal and thalamocortical system. Such studies of subcortically recorded SEPs have helped to establish concepts on the generators of the components that correspond to the scalp responses peaking in the first 15 msec after median nerve stimulation. These reflect successive activation of the lemniscal system at the spinal entry zone of the peripheral nerve, the posterior column, the cuneate nucleus, and the medial lemniscus.
Of the far-field potentials in the median and tibial SEPs, it seems likely that relay nuclei are responsible only for the generation of P13 and P31, respectively ; the other potentials probably reflect electrophysical change in the surrounding volume conductor. The near-field components N20 and N37 are generated neurally but probably reflect multiple and even independent thalamocortical projections.
The clinical usefulness of middle-latency SEPs (P40, N60) is still controversial because these components show considerable interindividual variability even in healthy subjects. Nevertheless, middle-latency SEPs following stimulation of the median nerve are considered a sensitive measure of cortical function, and several studies have reported their reliability for objective assessment and quantification of cerebral dysfunction and prognostic evaluation in patients with a variety of diseases. Amplitude and latency of middle-latency SEPs increase gradually with age in normal subjects.
Long-latency components are distributed bilaterally, with the greatest amplitude being at the vertex. They relate to nonspecific thalamocortical projection systems involved in habituation, adaptation, and arousal. These components are of large amplitude (between 10 and 40 μV), so as few as 50 averaged epochs are required to obtain a measurable response. Unfortunately, their recovery time takes several seconds, which necessitates a stimulus repetition rate that is lower than once every 5 seconds to avoid significant attenuation of the response. Recovery function of given SEP components is not determined simply by the number of synapses interposed between the stimulus site and the generator source of the response in the central nervous system (CNS). There appears to be a structural or functional process of low-cut filtering in the primary sensory cortex.
Spinal SEPs can be recorded over the cervical and thoracolumbar regions. They are considerably smaller in amplitude than SEPs recorded over the scalp, which limits their clinical usefulness. However, the difference in latency between the scalp and cervical or lumbar SEPs is a measure of central sensory conduction, which remains a major clinical application of spinal SEPs.When a combined vertical and horizontal array of electrodes is placed over the neck, and a noncephalic reference is used, the cervical SEP elicited by median nerve stimulation consists of three distinct components ( Fig. 26-4 ): (1) the proximal plexus volley (PPV); (2) the posterior (dorsal) column volley (DCV); and (3) a cervical component (CERV N13/P13) that possibly is generated at the level of the posterior gray matter of the cervical cord. These occur with onset latencies of approximately 10, 12, and 13 msec, respectively. With a cephalic reference, these components become fused, although some or all of the subcomponents may be appreciated as small deflections superimposed on the main negativity. The latency difference between the peak of the main negativity of the spine-recorded response, usually measuring between 12 and 14 msec, and the N20 cortical SEP component gives a “central conduction time” between the lower part of the brainstem and the primary sensory cortex that measures about 5.5 msec.
Thoracolumbar spinal SEPs are even smaller than cervical spinal SEPs and can be difficult to elicit in obese subjects. Nevertheless, by using a knee or iliac crest reference, an SEP elicited by tibial nerve stimulation at the ankle can be recorded over the thoracolumbar spine and, like the cervical SEP, has several subcomponents. An initial traveling wave, N18, can be recorded over the lower lumbar spine and is akin to the cervical PPV evoked by median nerve stimulation; it indicates that the volley has passed through the sacral plexus. It can be seen in scalp-recorded far-field potentials as a small positivity designated P18. A second component, N22, is recorded with maximal amplitude over the lower thoracic spine (T10 to L1). It is a stationary peak and its latency remains constant even when the recording electrodes are moved several segments rostrally or caudally. It is probably akin to the median-evoked DCV and represents postsynaptic activity in the lumbar gray matter generated in response to input from axon collaterals. The third component is a negative traveling wave, N24, with a latency that shortens from the caudal to the rostral thoracic cord; it is most evident over the lower thoracic cord, however, and reflects a propagated volley in the posterior columns that is equivalent to the DCV potential recorded over the neck with median nerve stimulation.
SEPs in disorders of the peripheral nervous system
The main clinical reason to record SEPs is to identify and localize a lesion involving the somatosensory pathways. The presence of an SEP abnormality, however, may provide no further information than could have been obtained by clinical examination, and normal findings do not exclude the possibility of organic disease as a cause of the symptoms. Furthermore, the presence of an SEP abnormality does not indicate the nature of the disease process. Nevertheless, the recording of SEPs may provide information of diagnostic relevance, may help characterize disease processes, and may determine the extent of pathologic involvement.
Disorders of the Peripheral Nerves
In general, SEPs have a definite role in the investigation of the peripheral nervous system when conventional techniques (see Chapter 13 ) cannot be used because the site of pathology is so proximal or because the nature of the lesion is such that responses cannot otherwise be recorded.
SEPs have been used to determine sensory conduction velocity in distal segments of a nerve and can provide results comparable to those obtained by peripheral nerve conduction studies if the stimulus is applied in turn to two or more sites along the course of the nerve. In several early studies, for example, scalp-recorded SEPs were found to be absent or delayed in patients with a variety of polyneuropathies and mononeuropathies. Such a sophisticated approach is unnecessary if similar information can be obtained by routine conduction studies, but it may be an important means of identifying extensive or localized lesions of peripheral nerves and of determining sensory conduction velocity when sensory nerve action potentials cannot be recorded peripherally by conventional techniques. In such circumstances, afferent conduction velocity, as determined by the SEPs, is slowed, usually in proportion to the degree of slowing in motor conduction velocity, but occasionally is within the normal range. The absence of peripheral responses presumably results from dispersal of impulse traffic by the underlying pathology, and the preservation of cortically generated SEPs probably reflects central amplification by a reorganization and synchronization of afferent volleys at different synaptic levels. Central amplification of a preserved but attenuated response derived from a few normally conducting axons is probably also responsible for the apparently normal afferent conduction velocity found occasionally. Accordingly, the very factor that permits afferent conduction velocity to be determined by the SEP technique when it otherwise cannot be measured sometimes may result in misleadingly normal values.
In patients in whom a diagnosis of a demyelinating sensorimotor polyneuropathy cannot be confirmed by conventional motor and sensory nerve conduction studies—either because they are normal or because distal sensory studies are normal but motor studies are abnormal—SEP studies may reveal proximal involvement of sensory nerves, thereby suggesting the correct diagnosis. SEP studies may also be helpful in showing proximal sensory nerve dysfunction in patients with a pure sensory ataxia. Such proximal involvement typically is not found in diabetic polyneuropathy and other length-dependent neuropathies.
In Guillain–Barré syndrome, conduction is normal in distal segments of the nerves in up to 20 percent of patients at an early stage of the disorder. Conduction slowing in the proximal segments of peripheral nerves has been demonstrated with SEPs (e.g., by an increase in the interpeak interval between the responses recorded over Erb’s point and the cervical spine or by the finding of a small dispersed cervical response) in a number of patients with Guillain–Barré syndrome. In general, however, the median-derived SEPs are less sensitive than F-waves for detecting proximal pathology in patients with this syndrome. Accordingly, in patients with suspected Guillain–Barré syndrome, SEP studies probably are best reserved for occasions when peripheral nerve conduction and F-wave studies are normal.
The SEP findings have been used with those of other electrodiagnostic tests to support the belief that a continuity exists between Bickerstaff brainstem encephalitis and Miller Fisher syndrome. In one instance, for example, an absent median-derived N20 (despite normal peripheral function), together with a diffusely slowed electroencephalogram (EEG) and absent R2 of the blink reflex, suggested central involvement, whereas facial conduction studies and F-wave and H-reflex studies suggested peripheral pathology.
Skin denervation often is associated with reduced thermal sensitivity, and this has led to studies of contact heat evoked potentials in patients with neuropathy. A specially designed, commercially available contact heat stimulator delivers recurrent heat stimuli at a specified temperature to the skin while the cerebral responses, mediated by A-delta or C fibers, are recorded over the scalp. In patients with pathologic evidence of skin denervation, the amplitude of the SEPs is reduced and correlates with the degree of denervation. Latencies are prolonged. The technique thus provides a means of evaluating small-fiber function in patients with peripheral neuropathy. It also shows promise for distinguishing neuropathy from other chronic pain states, and sometimes shows abnormalities in symptomatic patients without significant loss of intraepidermal fibers or abnormality of quantitative sensory testing.
Another approach to assessing small-fiber function is with pain-related SEPs elicited by painful electrical stimuli. A concentric planar electrode has been devised that delivers stimuli limited to the superficial layers of the dermis and that therefore depolarizes mainly superficial nociceptive A-delta fibers. In a study of HIV-associated sensory neuropathy, all 19 had abnormal pain-related SEPs, with the latency and amplitude of the responses correlating with the intraepidermal nerve-fiber density, whereas standard nerve conduction studies were abnormal in only 8 instances.
Focal nerve lesions may be detected and evaluated by SEP studies. Thus, common entrapment syndromes (e.g., median nerve compression in the carpal tunnel ) have been diagnosed by this means, although such an approach is usually unnecessary. However, the technique may be used to monitor recovery in patients after complete section of a peripheral nerve because a response may be recorded from the scalp at a time when a potential cannot be recorded from the nerve itself. SEPs have been used to evaluate focal lesions involving the proximal segments of sensory nerves, which may not be accessible for conventional nerve conduction studies. Such an application complements the use of F waves to evaluate conduction through proximal motor fibers. The approach may be useful, for example, in documenting abnormalities in patients with such disorders as meralgia paresthetica.
Finally, SEPs may be useful in the evaluation of patients with distal axonopathies (e.g., those related to toxin exposure) inasmuch as these conditions may involve the terminal portions of central as well as peripheral axons. Indeed, in patients with cisplatin neuropathy, CNS involvement—as evidenced by slowing of central conduction of the tibial-derived SEP—occurred with doses of cisplatin that caused peripheral nerve changes, suggesting that loss of large fibers in the posterior columns was secondary to degeneration of posterior root ganglion cells. In the authors’ experience, however, SEPs are always normal in patients with symptoms possibly related to occupational toxic exposure when peripheral nerve conduction studies are normal.
In patients with injuries to the brachial plexus, the location of the lesion influences the outcome; root avulsion carries a poor prognosis, whereas a postganglionic lesion may require surgical treatment and may be followed by recovery. Conventional nerve conduction studies can help in determining the probable site of injury. Patients with preganglionic lesions have preserved sensory nerve action potentials despite clinical sensory loss. However, the recording of SEPs over the scalp and spine following the stimulation of arm nerves has been held to enhance the accuracy of electrophysiologic assessment because any attenuation of N13 relates to the total proportion of damaged fibers, whereas attenuation of N9 reflects the proportion damaged postganglionically. Unfortunately, such a theoretical approach has not been of practical use in a clinical context and has now been abandoned.
A preganglionic lesion may be partially or completely obscured electrophysiologically by coexisting postganglionic damage to the same fibers; this is a major limitation of electrophysiologic techniques for evaluating plexus lesions. In occasional patients, however, the findings are of help in localizing the lesion, as when peripheral sensory nerve action potentials are attenuated (but present) but SEPs are absent over the spine and scalp, a situation suggesting that multiple root avulsions have occurred in addition to more peripheral lesions. Nevertheless, in most instances, the information derived from SEP studies concerning the site and extent of plexus injury can be obtained more quickly and conveniently by needle electromyography (EMG).
The recording of SEPs may be useful in certain other contexts. The operative appearance of the brachial plexus may be misleading because nerve fascicles may appear intact even though anatomic disruption has occurred. In this regard, intraoperative stimulation of specific roots while responses are recorded from the contralateral portion of the scalp may help in establishing their functional continuity with the cord. More useful in this context is the epidural recording of evoked spinal cord potentials. Intraoperative SEP recording can be especially helpful when ruptured nerves are to be treated by grafting. The presence of a cortical response to stimulation of the proximal stump makes a second, more rostral lesion unlikely, whereas the lack of a response usually is associated with a poor outcome.
Thoracic Outlet Syndrome
Characteristic electrophysiologic changes may occur in patients with cervical ribs or bands that are causing neurogenic thoracic outlet syndrome. Typical findings include a small or absent ulnar sensory nerve action potential; chronic partial denervation in the small muscles of the hand, especially in the muscles of the thenar eminence; and prolongation of the F-wave latencies from these muscles. In many patients with suspected thoracic outlet syndrome, however, clinical examination is normal, and needle EMG and nerve conduction studies reveal no abnormalities. The designation non-neurogenic syndrome is used in such circumstances.
In patients with suspected thoracic outlet syndrome, the ulnar-derived SEP is abnormal in many—but not necessarily all—patients with objective neurologic signs. It typically has a poorly formed potential at Erb’s point, with or without abnormalities of the responses over the spine or scalp, such as an attenuated N13 response or a prolonged N9–N13 interpeak interval. Occasionally, needle EMG is abnormal but the SEP is normal, or vice versa. Thus, the recording of SEPs is of limited diagnostic importance in patients with the neurogenic variety of thoracic outlet syndrome.
In patients with the non-neurogenic syndrome, the SEPs elicited by median, ulnar, or superficial radial nerve stimulation are usually normal in the authors’ experience and that of others. Reports of abnormalities in this syndrome are usually difficult to interpret because of the nature of the claimed abnormalities or because the clinical and electrophysiologic findings were not provided in sufficient detail to allow for independent evaluation.
Cervical Spondylotic Radiculopathy or Myelopathy
Short-latency SEPs have been used to evaluate patients with cervical spondylosis, but the findings do not distinguish spondylosis from other cervical lesions. Most of the published studies concern SEPs elicited by stimulation of the median or ulnar nerve trunks. Patients with pain and paresthesias but without neurologic signs often have normal responses, whereas those with a more severe spondylotic radiculopathy causing objective neurologic signs may have abnormal SEPs, with delayed or lost components, regardless of whether a myelopathy is also present. Some authors emphasize that amplitude rather than latency of the responses reflects nerve lesions caused by spondylotic deformities. In general, however, the nature of the surface-recorded SEP abnormality does not help in indicating either the severity of the neurologic disorder or the long-term prognosis, and in some patients with a spondylotic myelopathy, the SEP findings may be entirely normal. Thus, the SEPs elicited by stimulation of nerve trunks in the arms are no better than careful clinical examination in determining the severity and prognosis of cervical spondylosis. They are of no help in the selection of patients for surgery, and such a decision should be made on clinical grounds.
It is possible that axons arising in the lumbosacral region are more likely to be affected by cervical compression than are fibers from the arms. Whether the recording of sural-derived SEPs in patients without clinical evidence of myelopathy will help in determining which patients with cervical spondylosis are at risk for the development of a cord deficit clinically is not known.
It is hard to account for reports that fibular- or tibial-derived SEPs are abnormal in a high percentage of patients with lumbosacral radiculopathy because stimulation of a nerve derived from several segments would not be expected to lead to an SEP abnormality in patients with an isolated root lesion. Fibers traversing the unaffected roots would presumably generate a normal response, and this accords with the authors’ experience.
Cutaneous nerves can be stimulated at selected sites so that segmental specificity is better than with stimulation of mixed nerve trunks. For example, the musculocutaneous nerve can be stimulated in the forearm for testing C5, median nerve fibers in the thumb to test C6, median nerve fibers along the adjoining surfaces of the second and third fingers for C7, and the ulnar nerve in the little finger to test C8. For evaluating lumbosacral roots, the saphenous nerve can be stimulated at the knee and ankle for L3 and L4, respectively; the superficial fibular (peroneal) nerve at the ankle to test L5; and the sural nerve at the ankle for S1. In our experience, however, only about 50 to 60 percent of patients with lumbosacral radiculopathies confirmed by imaging studies have abnormal scalp-recorded SEPs derived from such stimulation. Abnormalities usually involve parameters that vary considerably in normal subjects or that are hard to quantify (e.g., amplitude reduction and abnormal morphology), thereby limiting the clinical relevance of the technique; prolongation in latency is found only occasionally. Although some have claimed a higher yield for the technique in detecting radicular involvement, the sensitivity and specificity are insufficient to make it clinically useful.
SEPs have been recorded in response to dermatomal stimulation in the L5 or S1 territory. The typical response recorded at the vertex (referenced to the midfrontal region or the contralateral C3′/C4′ position) consists of a positive–negative-positive complex, the initial positivity having a latency of about 50 msec ( Fig. 26-5 ). Early reports claimed a high yield of abnormalities in patients with surgically verified L5 or S1 root entrapment from a herniated disc, but it is unclear how criteria of abnormality were selected. In our experience, dermatomal SEPs enable an isolated radicular lesion to be localized correctly in only about 25 percent of patients. A latency abnormality is relatively uncommon, the usual finding being loss or marked attenuation of the response (see Fig. 26-5 ). In general, needle EMG is the single most useful electrophysiologic approach for evaluating root lesions, but dermatomal SEPs sometimes provided complementary information by revealing unsuspected root dysfunction. Dumitru and Dreyfuss similarly found the utility of dermatomal SEPs was limited in patients with suspected unilateral L5 or S1 root lesions because they do not have both high specificity and high sensitivity.