Sensory Nerve Action Potential

NCS of the sensory nerves generates a recording referred to as a sensory nerve action potential (SNAP). In this study, the sensory nerve is stimulated with sufficient electrical current that all the large-diameter sensory axons are simultaneously depolarized (Fig. 232-2). This stimulation is referred to as supramaximal stimulation because a higher electrical current than the minimum required for stimulation of all the axons is used to ensure that all the axons are depolarized. Action potentials of the depolarized axons immediately travel away from the site of stimulation at various velocities, depending on a number of factors. For instance, conduction velocity increases with larger axon diameter, increased myelination, and higher nerve temperature. The action potentials travel along the axons and are recorded by the recording electrodes over the nerve (Fig. 232-2). Each individual action potential generates a triphasic recording (see Fig. 232-1). A typical sensory nerve contains up to several hundred sensory axons, and an equal number of action potentials is recorded from the nerve. The SNAP is the sum of the individual action potentials recorded from each sensory axon. Under normal conditions, the action potentials of large-diameter sensory axons travel at similar velocities and thus pass under the recording electrode nearly simultaneously. The sum of these action potentials results in the SNAP (Fig. 232-3). The action potentials of small-diameter myelinated and unmyelinated axons travel at slower and more variable velocities. Thus, these action potentials pass under the recording electrode at variable times and do not summate sufficiently to generate enough amplitude to be a visible waveform. The amplitude of the SNAP is calculated from the baseline, or first positive peak, to the negative peak (while keeping in mind that negative is upward) and is a reflection of the number of normal large-diameter myelinated sensory axons. Conduction velocity is calculated by dividing the distance between the sites of stimulation and recording by the time that the first action potentials reach the recording electrodes, which is represented as the beginning of the upward slope from the baseline or first positive peak. Under normal circumstances, a longer distance between the stimulation and recording sites results in less synchronized action potentials from the variety of slow- and fast-conducting axons. This phenomenon is referred to as temporal dispersion and results in a decreased SNAP amplitude and increased duration when recording over large distances (Fig. 232-3).

FIGURE 232-2 Typical setup of a sensory nerve action potential.

FIGURE 232-3 Sensory nerve action potential of a normal median nerve when recording at the index finger and stimulating at the wrist or elbow.

In the presence of nerve injury or disease, there is often a change in SNAP conduction velocity or amplitude. Changes in the SNAP depend on the site and mechanism of the lesion, but all such lesions cause numbness, paresthesias, and other sensory symptoms. Any lesion at or distal to the dorsal root ganglion that causes wallerian degeneration results in fewer sensory axons, fewer action potentials recorded by the electrodes, and a decreased SNAP amplitude (Fig. 232-4A). Conduction velocity is normal or nearly normal because the remaining axons are myelinated and function normally. With marked axonal loss, conduction velocity may be slightly decreased because of the loss of faster conducting axons. Conduction velocity can be decreased to 80% of the lower limit of normal for mild axonal loss and to 70% of normal when SNAP amplitude is less than 50% of the lower limit of normal.

FIGURE 232-4 Numbness from different mechanisms and sites of lesions and the corresponding sensory nerve action potential (SNAP). A, Axonal degeneration from a lesion distal to the dorsal root ganglion, such as a severed nerve or dying-back axonopathy; such lesions result in a low-amplitude SNAP with normal or nearly normal conduction velocity. B, Distal demyelination between the site of nerve stimulation and recording, which results in conduction slowing and possibly a low-amplitude SNAP, depending on the degree of temporal dispersion or conduction block. C, Proximal demyelination or radiculopathy leaving the distal axons and myelin intact and thereby resulting in a normal SNAP.

Demyelination distally between the sites of nerve stimulation and recording leads to slowing of the individual action potentials and thus more markedly reduced conduction velocity or increased latency (time from stimulation to the initial waveform) of the SNAP (Fig. 232-4B). Typical clinical manifestations include carpal tunnel syndrome with demyelination of the median sensory nerve at the carpal tunnel or a demyelinating polyneuropathy such as Guillain-Barré syndrome. Because acquired demyelinating lesions frequently result in varying degrees of slowing of the individual axons, there is often increased temporal dispersion. If the demyelination is severe enough, the action potential may be unable to continue propagating down the axon across the site of demyelination to the recording electrode, thereby resulting in conduction block. Conduction block, which is almost always caused by demyelination, can result in a low-amplitude or absent SNAP. In most cases of conduction block there is concomitant conduction slowing because some axons are demyelinated to the point of conduction block but others are only partially demyelinated, which causes conduction slowing. Thus, a low-amplitude SNAP with conduction slowing suggests a demyelinating lesion, whereas a low-amplitude SNAP without conduction slowing suggests a lesion causing axonal degeneration without primary demyelination.

A demyelinating lesion proximal to the point of nerve stimulation and recording leaves the distal nerve intact. A lesion causing axonal loss proximal to the dorsal root ganglion, such as at the root, results in wallerian degeneration of the proximal axon. However, the sensory axons distal to the dorsal root ganglion maintain their continuity to the cell body and thus remain normal. The SNAP remains normal under these circumstances (Fig. 232-4C), even in the presence of anesthetic sensations. A normal SNAP in a patient with numbness suggests that the lesion is proximal to the dorsal root ganglion (root or CNS) or is a proximal demyelinating lesion.

Compound Muscle Action Potential

NCS of the motor nerves generates a recording referred to as the compound muscle action potential (CMAP). Unlike the situation with a sensory NCS, the CMAP is recorded from the muscle and not the motor nerve. Similar to a sensory NCS, the motor nerve is stimulated with supramaximal stimulation such that all the motor axons are simultaneously depolarized. Action potentials from the depolarized axons immediately travel away from the site of stimulation at nearly identical velocities. The action potentials travel along the axons to the neuromuscular junction. Each motor axon innervates up to several hundred muscle fibers. Because the typical motor nerve contains up to a few hundred motor axons, the amplitude of the CMAP—a summation of action potentials from muscle fibers—is 100 to 1000 times the magnitude of the SNAP. When compared with sensory nerves, the motor axons of motor nerves are much more similar in diameter and degree of myelination, thereby resulting in more similar individual axonal conduction velocities and very little temporal dispersion (Fig. 232-5). Moreover, the duration of motor unit action potentials (MUAPs) is long enough that the degree of overlap is affected relatively little by temporal dispersion.

FIGURE 232-5 Compound muscle action potential (CMAP) of a normal median nerve when recording at the abductor pollicis brevis (APB) and stimulating at the wrist and elbow. The distal motor latency is the time from stimulation at the wrist (the most distal point of stimulation) to onset of the CMAP at the APB; this time is called t₁, and in this case it is 3.8 msec. The time from stimulation at the elbow to onset of the CMAP at the APB is t₁ + t₂; in this case it is 7.7 msec. Subtracting the two times (7.7 msec − 3.8 msec = 10.9 msec) is equivalent to t₂, which is the time for the action potential to travel from the elbow to the wrist. The distance between stimulation points at the elbow and wrist can be measured (d = 220 mm), and the conduction velocity can be calculated by dividing the distance by time (d/t₂ = 220 mm/3.9 msec = 59 m/sec).

The conduction velocity of the distal motor nerve segment cannot be calculated for the CMAP as it can for the SNAP because the time from nerve stimulation to recording of muscle fiber action potentials includes the release of acetylcholine. Acetylcholine must diffuse across the neuromuscular junction and bind to the acetylcholine receptor before the muscle fiber action potential can be recorded. This process takes about 0.5 to 1 msec. Instead of calculating conduction velocity from a point of stimulation to the muscle, the time from stimulation at a distal point of the nerve to the onset of the CMAP—referred to as the distal motor latency—is compared with that in normal controls to determine whether distal conduction slowing is present. The conduction velocity of the proximal portion of the motor nerve can be calculated as described in Figure 232-5. The three most important aspects of the CMAP are the amplitude, conduction velocity, and distal motor latency.

In the presence of injury or disease of the motor nerve or muscle, there is often a change in CMAP conduction velocity, amplitude, or distal motor latency. Changes in CMAP depend on the site and mechanism of the lesion, but all such lesions cause weakness and possibly muscle atrophy in the case of denervation. The more common scenarios are presented in Figure 232-6. For the sake of illustration, consider a motor nerve consisting of two motor axons, each of which innervates two muscle fibers. Under conditions of normal PNS functioning, supramaximal stimulation of the motor nerve distally and proximally leads to action potentials in both motor axons and all four muscle fibers. The CMAP recordings consist of the sum of the muscle action potentials from all four muscle fibers (Fig. 232-6A).

FIGURE 232-6 Various compound muscle action potential (CMAP) amplitude and velocity results, depending on the site of the lesion. A, Normal strength or a central nervous system cause of weakness. B, Axonal degeneration without reinnervation results in only one of the two motor axons and two of the four muscle fibers being able to generate action potentials. The CMAP amplitude is half of normal when stimulating either distally (Stim₁) or proximally (Stim₂). C, Demyelination and conduction block between sites of stimulation result in no motor axonal loss. CMAP amplitude is normal when stimulating distal (Stim₁) to the site of the lesion because the motor axons are normal there and all four muscle fibers are able to generate action potentials. When stimulating proximally, action potentials are generated in both motor axons. However, one of the action potentials is blocked at the site of demyelination and cannot proceed to the muscle fibers. Thus, only one of the motor axons can transmit the action potential to two muscle fibers, and the CMAP amplitude is half of normal. The decline in CMAP amplitude from a point of distal stimulation to one of proximal stimulation is characteristic of conduction block.

When there is motor axonal degeneration from distal dying-back axonopathy, wallerian degeneration from a proximal lesion, or loss of motoneurons, only one of the two motor axons is present and can transmit an action potential to its two innervated muscle fibers (Fig. 232-6B). The other two muscle fibers cannot be stimulated, which leads to decreased CMAP amplitudes from both the distal and proximal stimulation sites.

In this model, conduction block between the distal and proximal sites of stimulation results in a characteristic CMAP abnormality (Fig. 232-6C). A typical demyelinating lesion, such as from entrapment neuropathy, may result in conduction block of one motor axon and conduction slowing of the other. Distal to the site of demyelination, the motor axons and myelin are normal. Thus, stimulation of the motor nerve distally results in activation of both motor axons and all four muscle fibers, which yields a normal CMAP amplitude and distal motor latency. Stimulation of the motor nerve proximal to the site of demyelination leads to action potentials in both motor axons. However, because one of the action potentials is blocked at the site of demyelination, it is unable to proceed with activation of the corresponding muscle fibers. The other action potential is slowed but can still activate two of the muscle fibers. Hence, the CMAP from the proximal point of stimulation is the sum of two muscle fiber action potentials; its amplitude and area are smaller than normal, and it is delayed because of conduction slowing. Therefore a scenario in which CMAP amplitude and area at the proximal point of stimulation are decreased in comparison to the distal site is indicative of a conduction block between the two points of stimulation. If there is a decrease in CMAP amplitude but area is maintained, temporal dispersion is suggested. Unlike sensory nerves, there is normally very little temporal dispersion of motor nerves, and its presence indicates an area of demyelination.

Demyelination between the point of distal stimulation and the muscle can lead to distal conduction slowing and result in prolonged distal motor latency. However, the proximal conduction velocity and CMAP amplitudes are normal because all the muscle fibers are activated and the nerve is normal proximally.

Late Responses

SNAP and CMAP studies are best at evaluating distal nerves. Stimulation and recording from nerves at proximal sites, such as the root or plexus (e.g., Erb’s point), are often unreliable. When stimulating proximally, it is often difficult to ensure supramaximal stimulation or to limit stimulation to one nerve. Alternatively, evaluation of late responses can provide useful data regarding the proximal portions of nerves. The two most commonly evaluated late responses are the F wave and H-reflex (Fig. 232-7).

FIGURE 232-7 F-wave and H-reflex pathways.

F waves result from the late response of a motor unit, defined as a single motoneuron and all the muscle fibers that it innervates. The F-wave response is generated by stimulation of motor axons. Action potentials normally travel distally and proximally from the site of stimulation. The distally traveling action potential results in the CMAP, or M wave. The proximally traveling action potential reaches the motoneuron cell bodies in the anterior horn of the spinal cord. For unknown reasons, one or a few of the motoneurons often immediately generate an action potential that travels back from the cell body to the muscle, as though the action potential had bounced from the cell body back down the axon. The action potential from one or a few motoneurons activates the innervated muscle fibers, thus generating an F wave consisting of the sum of the action potentials from the motor unit. The most reliable value of the F-wave response is the latency, or the time that it takes the action potential to travel from the site of stimulation proximally to the motoneuron cell body and then distally from the cell body to the muscle. Comparisons to normal controls or to the contralateral side are used to help identify conduction slowing anywhere along the motor axon. F-wave latencies are most sensitive for disorders causing generalized or multifocal demyelination, such as Guillain-Barré syndrome. They are less sensitive for focal demyelinating disorders, such as a radiculopathy, because any focal conduction slowing is diluted by the normal conduction velocity over most of the F-wave pathway.

H-reflexes are equivalent to an electrophysiologic ankle reflex. In adults, the H-reflex can be reliably elicited only in the soleus muscle. The tibial nerve is stimulated at the popliteal fossa. An action potential travels proximally along the sensory pathway of the tibial nerve, similar to the pathway of an action potential generated by stretching of the ankle tendon during testing of the ankle reflex. The action potential enters the spinal cord along the sensory axons, which synapse with the anterior horn cells. These sensory action potentials result in the release of neurotransmitter from the sensory end terminals and activation of the anterior horn cells and motor axons to the soleus muscle, similar to the sensory-to-motor monosynaptic ankle reflex pathway. The CMAP amplitude and latency generated by the soleus are recorded and compared with those of normal controls or the contralateral side. A delay in latency or a diminished amplitude suggests a lesion anywhere along the H-reflex pathway.

Repetitive Stimulation

Repetitive stimulation is used to diagnose abnormalities of the neuromuscular junction, such as myasthenia gravis, botulism, Lambert-Eaton myasthenic syndrome, and congenital myasthenia. The technique involves rapid, repetitive CMAP recordings from 2 to 50 Hz. Defects in neuromuscular transmission, especially at the postsynaptic site, lead to successive decrements in CMAP amplitude with stimulation at low frequencies of 2 to 3 Hz. An increase in CMAP amplitude at high frequencies such as 50 Hz is indicative of a presynaptic neuromuscular transmission defect, such as Lambert-Eaton syndrome. The principles underlying these abnormalities can be found in other textbooks.^1,2

Needle Electromyography

EMG examination consists of inserting a recording needle into a muscle to measure the action potentials from muscle fibers. This information is used to determine the functioning of motoneurons and muscle. There are three aspects of an EMG examination: (1) assessment of spontaneous muscle fiber action potential activity; (2) measurement of MUAP duration, amplitude, and phases; and (3) recruitment and interference pattern of MUAPs.

Spontaneous Activity

Under normal conditions, when a patient is relaxed, muscle fibers are electrically silent, with no significant spontaneous muscle fiber action potentials. Different types of abnormal spontaneous muscle fiber action potentials can be seen and indicate specific abnormalities of the PNS (Table 232-1). The most common and significant spontaneous activities consist of fibrillation potentials and positive sharp waves (Fig. 232-8A and B). These are spontaneous action potentials from individual muscle fibers in response to either acute denervation or acute muscle fiber injury. Muscle disorders associated with these discharges include muscle fiber necrosis from muscle trauma, muscular dystrophies or inflammatory myopathies, and other muscle diseases, such as acid maltase deficiency or hyperkalemic periodic paralysis. Because fibrillation potentials and positive sharp waves are caused by the same group of PNS abnormalities, they usually occur together. Denervation of muscle results in fibrillation potentials and positive sharp waves within 2 to 3 weeks. These findings persist until the muscle fiber is reinnervated, usually within 3 to 4 months in mild injuries, or until the denervated muscle fiber undergoes complete atrophy after up to a few years of persistent denervation without reinnervation.

TABLE 232-1 Abnormal Spontaneous Muscle Fiber Action Potentials and Associated Peripheral Nervous System Abnormalities

MUSCLE FIBER ACTION POTENTIAL	ABNORMALITY
Fibrillation potential	Acute denervation, acute muscle fiber necrosis
Positive sharp wave	Acute denervation, acute muscle fiber necrosis
Complex repetitive discharge	Chronic denervation, chronic muscle fiber necrosis
Fasciculation potential	Normal finding, motoneuron disease, radiculopathy, neuropathy
Myotonic discharge	Myotonic dystrophy, myotonia congenita, paramyotonia
Myokymic discharge	Radiation plexopathy or myelopathy, multiple sclerosis, brainstem glioma with facial myokymia
Cramp discharge	Normal finding, motoneuron disease, radiculopathy, neuropathy
Neuromyotonic discharge	Isaac’s disease, neuropathy

FIGURE 232-8 Fibrillation potentials (A), positive sharp waves (B), and motor unit action potentials of normal (C), reinnervated (D), and myopathic (E) motor units. The sensitivities are 50 µV per division for A and B and 200 µV per division for C to E.

Complex repetitive discharges are generated from muscle fibers that have been denervated for more than 2 months or from injured muscle fibers, usually associated with muscle fiber necrosis. The neurological disorders that cause complex repetitive discharges are similar to those associated with fibrillation potentials and positive sharp waves, except that complex repetitive discharges occur under chronic conditions. The other spontaneous abnormalities listed in Table 232-1 are seldom encountered in patients with neurosurgical conditions and are not discussed.

Carpal Tunnel Syndrome

NCSs are very sensitive for the diagnosis of median neuropathy at the wrist, such as carpal tunnel syndrome. Together with needle EMG, they are useful for making a diagnosis, differentiating among various possible causes of hand numbness (e.g., cervical radiculopathy, brachial plexopathy, distal entrapment), and delineating the extent of nerve demyelination or axon loss.

Most studies have shown that large-diameter, highly myelinated sensory fibers are affected before motor fibers in entrapment neuropathies. Hence, sensory conduction studies are more commonly affected than motor studies. There are a number of approaches for measuring sensory conduction of the median nerve across the wrist, most of which involve comparing the sensory latency with “normal” (more appropriately called “reference”) values or with another nearby nerve that does not traverse the carpal tunnel (e.g., radial or ulnar nerve). The former approach—simply measuring median sensory latency—is less satisfactory because of all the nonpathologic factors that can prolong sensory latency, including cool limb temperature, increasing age, and greater height. Comparing median latency with that of another nearby nerve avoids these factors because both nerves will be affected equally.

Three conduction studies are most commonly performed to evaluate median sensory latency across the wrist.^3,4

Commonly, one or more of these studies are performed to evaluate patients referred for possible carpal tunnel syndrome. One should be aware, however, that the more studies one performs, the greater the chance of false-positive results.^5,6 Recent studies have shown that performing all three studies and simply adding together the differences in latency is more sensitive, specific, and reliable than performing a single study or performing multiple studies and considering them independently.^7,8 When the three latency comparisons are added, a sum (referred to as the combined sensory index) of 1 msec or greater is considered abnormal and is suggestive of carpal tunnel syndrome.

Motor conduction studies are less commonly abnormal than sensory studies in patients with carpal tunnel syndrome. When abnormalities are present, they probably represent more severe electrophysiologic abnormalities than sensory slowing alone. Median motor latencies also vary with age, temperature, and height, so comparison with the ulnar nerve is usually helpful. A median-ulnar motor latency difference greater than 1.5 msec is probably abnormal.

Needle EMG of the thenar muscles is sometimes useful for detecting motor axon loss in the thenar muscles. Denervation is usually seen in more severe cases of entrapment or in traumatic median neuropathy at the wrist. Depending on the clinical findings, needle EMG might not be needed (e.g., when NCSs show only mild sensory slowing), may be limited to the thenar muscles, or might be used to examine a number of muscles in the limb to detect possible radiculopathy or plexopathy.

It should be noted that some improvement in latencies is usually expected after surgical release of the median nerve at the wrist. However, in many cases, latencies do not return to normal despite a good postsurgical clinical outcome.³ Thus, in patients with persistent postoperative symptoms, it is important to compare the results with preoperative conduction studies. If preoperative results are not available, postoperative testing separated by several months should be performed to see whether the latencies are getting better or worse over time.

Ulnar Neuropathy at the Elbow

Motor NCSs are often the most useful technique for localizing the site of ulnar neuropathy at the elbow and determining the pathophysiology of the lesion. Recording from the abductor digiti minimi is the most common method. Some authors, however, have found that recording from the first dorsal interosseous muscle, the most distal muscle supplied by the ulnar nerve, is more sensitive.^9–11 A two-channel technique may be used to record from both muscles simultaneously so that extra stimulation is not required.

Stimulation is usually performed at the wrist, below the elbow, above the elbow, and sometimes at the axilla. Study of the across-elbow segment requires much care in technique and interpretation. The position of the elbow greatly influences the measured conduction velocity. When the elbow is extended, the ulnar nerve may become redundant in the ulnar groove, and surface measurements may not reflect the true distance of the underlying nerve. Flexing the elbow stretches the nerve to its full length, and measurement of the distance over the ulnar groove more closely reflects the distance along the nerve. Because there is room for considerable error in measurement of across-elbow conduction velocity as a result of distance measurements and elbow position, many electromyographers allow up to an 11- to 15-m/sec difference between the across-elbow and forearm segments before calling the finding “abnormal.”¹² Recent studies, however, have demonstrated that it is better to compare the velocity with established reference values than to compare it with forearm velocity because the latter also slows in ulnar neuropathy.¹³

Slowed conduction velocity is not the only finding that should be considered diagnostic of ulnar neuropathy at the elbow. Such patients may also have a drop in amplitude in the across-elbow segment. A reduction in amplitude of more than 10% in the across-elbow segment is probably abnormal.¹²

It is often found that studying very short segments yields higher sensitivity for focal lesions. With short-segment studies, the area of demyelination occupies a higher percentage of the distance studied than with studies of longer segments, in which normal nerve dilutes the measurement. Inching studies (or perhaps more appropriately called “centimetering” studies) can be performed by stimulating the nerve at 2-cm increments across the elbow.^14,15 With this technique, a conduction delay of more than 0.7 msec across 2-cm segments is probably abnormal.¹⁴ More impressive are focal changes in amplitude or waveform morphology across a segment.

Most of the aforementioned abnormalities require the presence of demyelination for localization. However, in many traumatic ulnar neuropathies in which there is only axon loss without demyelination, localization of ulnar neuropathy is far more difficult. In such cases, there is diffuse, mild slowing of conduction velocity without focal slowing or conduction block; there are no focal nerve conduction changes across the lesion. Therefore, despite one’s best technique, localization cannot be precisely determined in a significant number of patients with traumatic or vasculitic lesions of the ulnar nerve (in which only axon loss is present).

Sensory NCSs are often of less localizing value than motor studies. Nevertheless, sensory responses are frequently helpful for measuring the degree of sensory axon loss. A drop in amplitude of the ulnar SNAP is probably one of the more sensitive indicators of ulnar neuropathy at the elbow.¹⁶

Needle EMG of the ulnar-innervated muscle is critical, both to determine whether any axon loss has occurred and to help localize lesions that may be purely axonal in nature. Thus, even if NCS results are entirely normal, when ulnar neuropathy is clinically suspected, needle EMG should still be performed. The most helpful hand muscles to assess are the abductor digiti minimi and the first dorsal interosseus, two muscles commonly involved in ulnar neuropathy at the elbow.⁹ Study of the flexor carpi ulnaris and the ulnar half of the flexor digitorum profundus is marginally helpful. Although the branch to these muscles usually comes off distal to most entrapment sites at the elbow, the fascicles supplying these muscles are in a relatively protected position within the nerve, so these muscles are frequently spared.

Needle EMG of non–ulnar-innervated muscles is often useful to rule out other lesions that may mimic ulnar neuropathy. Examination of the thenar muscles or the extensor indicis proprius offers the opportunity to compare C8-T1 muscles not innervated by the ulnar nerve, which can be useful to rule out lower cervical radiculopathies, as well as lower brachial plexopathies.

Radiculopathies

In most cases, radiculopathies are a result of nerve root compression proximal to the dorsal root ganglion. Mild cases may consist of only demyelination or irritation of the nerve root, whereas more severe cases demonstrate motor and sensory axon loss.

The practitioner should keep in mind the relative sensitivity and specificity of various imaging and electrodiagnostic testing. Although magnetic resonance imaging (MRI) provides a very sensitive method for assessing nerve roots in the back and neck, it is a highly nonspecific technique. Many asymptomatic people have disk bulges and disk protrusions, and their frequency increases with age. In one study, 61% of asymptomatic 40- to 49-year-olds had a disk bulge on MRI and 33% had a disk protrusion; the incidence is considerably higher in older individuals.¹⁷ Other studies have shown a specificity of only about 50%.^18,19 Hence, there is about an equal chance that a disk abnormality will or will not correspond with clinical symptoms at that site. Electrophysiologic studies are somewhat less sensitive than MRI in detecting mild root compression, but their specificity is considerably higher—probably greater than 85% to 90%. It is often useful to combine the highly sensitive but nonspecific imaging modalities with the more specific electrophysiologic testing when evaluating someone with possible radiculopathy.

Needle EMG is probably the best electrophysiologic test for detecting radiculopathy.²⁰ After the onset of radiculopathy, evidence of denervation can be seen in proximal muscles such as the paraspinal muscles in as little as 10 to 14 days. More distal muscles in the limb become abnormal later, with up to 3 to 4 weeks needed to show evidence of denervation. For the diagnosis of radiculopathy, at least two muscles in the same myotome, but supplied by different peripheral nerves, should show evidence of denervation (fibrillations, positive sharp waves). In chronic root lesions, evidence of reinnervation (long-duration, polyphasic, large-amplitude MUAPs) may be seen in a myotomal distribution; however, this is a softer finding than evidence of recent denervation.

It is helpful to demonstrate paraspinal muscle involvement, as well as limb muscle abnormalities, although a significant proportion of patients with radiculopathies do not have abnormalities in the paraspinal muscles. Needle EMG of the paraspinal muscles has some specific limitations. False-positive findings can be seen after laminectomy and recent myelography, as well as in patients with some metabolic diseases (e.g., diabetes). Hence, abnormalities limited to the paraspinal muscles may be suggestive of some level of nerve root irritation but should not be considered diagnostic.

Sensory NCS results should usually be normal in patients with radiculopathies because compression occurs proximal to the dorsal root ganglion and the distal sensory axons remain in continuity with their cell bodies. Nevertheless, sensory conduction studies are often helpful to rule out a more distal lesion, such as plexopathy or entrapment neuropathy, both of which should affect sensory conduction studies in the appropriate distribution.

Motor NCS results are frequently normal unless severe axon loss has occurred. When severe motor axon loss is present and sufficient time has passed for axonal degeneration, the motor nerve response falls in amplitude, roughly in proportion to the degree of axon loss. For example, if half the motor axons in the L5 root were lost recently, the motor response from the extensor digitorum brevis (predominantly innervated by the L5 root) with stimulation of the peroneal nerve would be about half that of the other side.

Late responses can sometimes be helpful in assessing patients with possible radiculopathies. The F wave is usually normal or only mildly affected. The H wave, in contrast, is probably more sensitive than needle EMG for detecting S1 root lesions because it can detect demyelination whereas needle EMG detects primarily motor axon loss.

Somatosensory evoked potentials (SEPs) should theoretically be better at detecting root abnormalities affecting sensory fibers. However, data indicate that SEPs are not as good as needle EMG in detecting isolated radiculopathy, probably because of the overlap of dermatomes, such that multiple roots are stimulated simultaneously during SEPs; normal roots can produce a normal result. In contrast, SEPs are probably better than EMG at detecting spinal stenosis in which more than one root is involved.²¹

Timing

The time course of electrodiagnostic changes after the onset of a traumatic nerve lesion should always be considered when interpreting the electrophysiologic findings. Neurapraxia (conduction block), demyelination, and severe axon loss produce electrophysiologic changes immediately if one can stimulate proximal to the lesion to detect conduction block. More proximal lesions, in which one cannot easily stimulate proximal to the lesion, do not immediately produce changes on NCSs or EMG. Moreover, distinction between neurapraxia and axonotmesis or neurotmesis cannot be made until time for wallerian degeneration to occur has passed. Optimal timing of electrodiagnostic studies varies according to the clinical circumstances. When it is important to define a lesion early, initial studies 7 to 10 days after injury may be useful for localizing the lesion and distinguishing conduction block from axonotmesis. When clinical circumstances permit waiting, studies performed 3 to 4 weeks after injury provide much more diagnostic information because fibrillations will be apparent on needle EMG. Finally, when a nerve lesion is surgically confirmed and EMG is used primarily to document recovery, initial studies a few months after injury may be most useful.

Nerve Conduction Studies

In purely neurapraxic lesions, the motor response changes immediately after injury, assuming that one can stimulate both above and below the site of the lesion. When recording from distal muscles and stimulating distal to the site of the lesion, the response should always be normal because no axonal loss and no wallerian degeneration have occurred. Moving the stimulation proximal to the lesion produces a small or absent motor response because conduction in some or all fibers is blocked. In addition to conduction block, partial lesions often demonstrate concomitant slowing across the lesion. This slowing may be due to either loss of faster conducting fibers or demyelination of surviving fibers.

Electrodiagnostically, complete axonotmesis and complete neurotmesis look the same; the difference between these lesions lies in the integrity of the supporting structures, which have no electrophysiologic function. Thus, these lesions can be grouped together as axonotmesis for purposes of this discussion. Immediately after axonotmesis and for a few days thereafter, motor conduction studies look the same as those seen in a neurapraxic lesion. Nerve segments distal to the lesion remain excitable and demonstrate normal conduction, whereas proximal stimulation results in an absent or small response from distal muscles. Early on, this picture looks the same as conduction block and can be confused with neurapraxia. Hence, neurapraxia and axonotmesis cannot be distinguished until sufficient time for the occurrence of wallerian degeneration in all motor fibers has passed, typically about 9 days after injury.²²

After this time, the amplitude of the motor response elicited with distal stimulation falls. This decrease in amplitude starts at about day 3 and is complete by about day 9.²² Thus, in complete axonotmesis, by day 9 the picture is very different from that of neurapraxia. Responses are absent both above and below the lesion. Lesions with partial axon loss produce small-amplitude motor responses, with the amplitude being roughly proportional to the number of surviving axons.

Lesions that have a mixture of axon loss and conduction block pose a unique challenge. They can usually be sorted out by carefully examining CMAP amplitudes elicited from stimulation both above and below the lesion and by comparing the amplitude after distal stimulation with that obtained from the other side. The proportion of axon loss is best estimated by comparing the CMAP amplitude from distal stimulation with that obtained on the contralateral side. Of the remaining axons, the proportion with conduction block is best estimated by comparing amplitudes, or areas, obtained with stimulation distal and proximal to the lesion.

Needle Electromyography

Needle EMG examination of purely neurapraxic lesions shows changes in recruitment but usually no abnormalities in spontaneous activity (i.e., no fibrillations or positive sharp waves).

After a lesion causing axon loss (axonotmesis or neurotmesis), needle EMG demonstrates fibrillation potentials and positive sharp waves a number of days after injury. The time between injury and the onset of fibrillation potentials depends in part on the length of the distal nerve stump. When the distal stump is short, it takes only 10 to 14 days for fibrillations to develop. With a longer distal stump (e.g., ulnar-innervated hand muscles in a brachial plexopathy), 21 to 30 days is required for the full development of fibrillation potentials and positive sharp waves.²³

Fibrillation and positive sharp wave density are usually graded on a scale of 1 to 4. This is an ordinal scale, which means that as numbers increase, the findings are worse. However, it is not an interval or ratio scale; that is, 4+ is not twice as bad as 2+ or four times as bad as 1+. Moreover, 4+ fibrillation potentials do not indicate complete axon loss and in fact may represent only a minority of axons lost.^9,24 Evaluation of recruitment and particularly of distally elicited CMAP amplitude is necessary before one can decide whether complete axon loss has occurred.

When there are surviving axons after an incomplete axonal injury, the remaining MUAPs are initially normal in morphology but demonstrate reduced or discrete recruitment. Axonal sprouting is manifested by changes in the morphology of existing motor units. Amplitude increases, duration becomes prolonged, and the percentage of polyphasic MUAPs increases as motor unit territory increases.^25,26

In complete lesions, the only possible mechanism of recovery is axonal regrowth. The earliest needle EMG finding in this case is the presence of small, polyphasic, often unstable motor unit potentials previously referred to as “nascent potentials.” (This term is now discouraged because it implies a cause; it is preferable to simply describe the size, duration, and phasic nature of the MUAP.) Observation of these potentials is dependent on establishing axon regeneration, as well as new neuromuscular junctions, and such observation represents the earliest evidence of reinnervation, usually preceding the onset of clinically evident voluntary movement.²⁶ These potentials represent the earliest definitive evidence of axonal reinnervation in complete lesions.

Localization

Localization of peripheral nerve injuries is usually straightforward, but can be complicated by a variety of pitfalls. Localization is usually performed by two methods: detecting focal slowing or conduction block on NCSs or assessing the pattern of denervation on needle EMG.

Localizing peripheral nerve lesions by NCSs usually requires that there be focal slowing or conduction block as one stimulates above and below the lesion. To see such a change, focal demyelination must be present, or the lesion must be so acute that degeneration of the distal stump has not yet occurred. Thus, lesions with partial or complete neurapraxia (because of demyelination) can be well localized with motor NCSs, as can very acute axonal injuries.

In pure axonotmetic or neurotmetic lesions, it is difficult if not impossible to localize the lesion with NCSs. In such cases, there is mild and diffuse slowing in the entire nerve as a result of loss of the fastest fibers, or there is no response at all. Conduction across the lesion site is no slower than that across other segments. In addition, if enough time for the development of wallerian degeneration has elapsed (at least 9 days for motor fibers and 11 days for sensory fibers), there will be no change in amplitude as one traverses the site of the lesion. Thus, lesions causing purely axon loss are not well localized along a nerve by NCSs.

Another indirect inference that can be made based on sensory NCSs is localization of the lesion at a preganglionic versus postganglionic site. Lesions proximal to the dorsal root ganglion—that is, at the preganglionic level (proximal root, cauda equina, spinal cord)—tend to have normal SNAP amplitudes, even if there is reduced or absent sensation.^27,28 This is a particularly bad prognostic sign when seen in the setting of possible root avulsion. Conversely, lesions occurring distal to the dorsal root ganglion have small or absent sensory responses (when recorded in the appropriate distribution).

The other major electrodiagnostic method for determining the site of nerve injury is needle EMG. Conceptually, if one knows the branching order to various muscles under study, one can determine that the nerve injury is between the branches to the most distal normal muscle and the most proximal abnormal muscle. There are, however, a number of potential limitations with this approach. First, the branching and innervation for muscles are not necessarily consistent from one person to another. Sunderland demonstrated a great deal of variability in the branching order to muscles in the limbs, in the number of branches going to each muscle, and in which nerve or nerves supply each muscle.²⁹ Thus, the typical branching scheme may not apply to the patient being studied, and the site of the lesion can be misconstrued.

Additionally, the existence of partial lesions can lead to misdiagnosis of more distal sites. In partial ulnar nerve lesions at the elbow, for example, the forearm ulnar-innervated muscles are often spared.²⁴ This is thought to be due at least partially to sparing of the fascicles in the nerve that are preparing to branch to the flexor digitorum profundus and flexor carpi ulnaris (i.e., they are in a relatively protected position). This finding could lead one to inadvertently localize the lesion distal to the distal part of the forearm or wrist. Intraneural topography needs to be considered when making a diagnosis based on branching.³⁰

Localization of brachial plexus lesions deserves special consideration. In such cases it is important to differentiate injury to the root (e.g., avulsion) from plexus injuries and multiple peripheral nerve injuries. Differentiation between root and plexus lesions is accomplished primarily by examination of the paraspinal muscles and sensory amplitudes. Both of these methods are subject to the limitations mentioned earlier for needle EMG and sensory conduction studies. Distinguishing between plexus and peripheral nerve lesions is sometimes more complex. Intimate knowledge of brachial plexus anatomy is required to distinguish between a peripheral nerve distribution of abnormalities and a plexus distribution. Sampling of muscles from the cord and trunk levels of the plexus (e.g., latissimus dorsi, pectoralis major, infraspinatus) is often helpful. Even with this knowledge, however, multiple peripheral nerve lesions (e.g., axillary, radial) can be erroneously ascribed to a single plexus insult (e.g., posterior cord).