Pre-operative electrodiagnostic studies and intraoperative neurophysiologic monitoring: power and pitfalls

Nerve conduction studies

Nerve conduction studies (NCS) are performed using peripheral nerve stimulation to record an evoked sensory or motor potential. This is achieved by applying a brief electrical pulse to the skin overlying the nerve where it becomes superficial. A motor, sensory or mixed potential is obtained by recording electrodes on the skin over a muscle, cutaneous sensory, or mixed nerve^[2,3,8]. The result is a recorded compound muscle (CMAP) or sensory nerve action potential (SNAP), the former representing a summated potential derived from muscle fibers and the latter a summated response from nerve fibers [Figure 1]. The amplitude of the response reflects the electrical activity of all the individual muscle or nerve fiber action potentials; a reduced amplitude is, therefore, indicative of axonal loss of nerve or muscle fibers. In the case of the CMAP, the low amplitude can also result from a drop out of signal between the stimulation site and recorded muscle, or conduction block, due to segmental demyelination^[8]. Conduction block is essential in the electrodiagnosis of nerve entrapments not only as a diagnostic hallmark but also because of the implication for better outcomes^[9,10]. The CMAP amplitude can also be affected early on by local or distal sprouting from adjacent and preserved nerve axons to the recorded muscle. This can account for amplitude improvement in the first year after injury^[11]. Recognition of the effect of distal sprouting on CMAP is essential because it does not reflect an effect of axon regrowth from the proximal site of injury. Therefore, CMAP amplitude in the months after an injury may increase without meaning that axons are being regenerated from the injury site, whereas CMAP amplitude increases in a chronic injury that must occur because of regrowth from the original injury. This timing is essential in distinguishing true nerve recovery from nerve compensation for injury^[11,12].

Figure 1. Nerve conduction responses. A: Stimulation of motor axons recording over the muscle with detecting a compound muscle action potential (CMAP). Amplitude reflects the number of muscle fibers contributing to the response; distal latency reflects the time (msec) it takes for a stimulus to pass along a nerve to produce the CMAP. B: Stimulation of a sensory nerve recording a sensory nerve action potential (SNAP). Amplitude reflects the number of nerve axons contributing to the response. Distal latency is equivalent to conduction velocity.

The conduction speed of an impulse (velocity, m/s) between two points on a nerve, on the other hand, is generally a reflection of myelin integrity. This is affected early in the setting of focal compression, even before evidence of axonal loss. Myelin loss due to compression can slow the response early on and block the impulse’s conduction to individual fibers. In this case (conduction block), the proximal CMAP is reduced, but the distal CMAP is not, and there is clinical weakness without distal loss of amplitude (axonal loss) or other indicators of axonal loss^[9,10]. This is because weakness is due to a drop out of signal in otherwise intact nerve fiber axons. Conduction block is an essential source of weakness that carries much better outcomes than axonal injury; therefore, its detection on NCS can be very informative for surgical planning^[4]. Nerve conduction studies, therefore, allow for identifying the fiber type involved (motor, sensory or mixed), the pathophysiology of the injury (axonal or demyelinating), localization, and outcomes. The nerves selected for the study will depend on the clinical context derived from the initial clinical question.

Needle electromyography

Needle electromyography (EMG) is subsequently performed on selected muscles to localize a lesion further and determine its age^[13]. This is achieved by detecting aberrant electrical activity from muscle fibers and morphological changes in the structure and recruitment of the motor unit action potential (MUAP) during voluntary contraction. In this portion of electrodiagnostic testing, a needle electrode is placed into a muscle of interest. Movement of the electrode through the muscle at rest results in a burst of muscle fiber potentials that generally last no longer than 300 ms after needle movement. In denervated muscle, single muscle fibers spontaneously twitch at rest due to denervation hypersensitivity; this is detected as fibrillation potentials and positive sharp waves that reflect the activity of denervated muscle fibers due to nerve fiber axonal loss^{[2,4,8,13,14]}.

When the muscle is contracted, the morphology of individual MUAPs and the recruitment of those potentials are studied [Figure 2]. Recruitment reflects the gradual addition of more MUAPs as increasing force is generated in a muscle. Reduced recruitment occurs when an insufficient number of MUAPs are recruited for a given muscle contraction; this occurs in the setting of a neurogenic lesion or with sufficient demyelination to cause conduction block (drop out of the nerve action potential due to demyelination). Changes in the duration, number of phases (polyphasia), or amplitude of a MUAP reflect chronic changes resulting from reinnervation. These changes can reflect the reinnervation of muscle fibers as collateral nerve fibers sprout from adjacent intact axons. The conduction speed of sprouting nerve fibers is slower concerning mature branches due to their size and incomplete or absent myelination; therefore, the potential from the muscle fibers innervated by new sprouts is slightly delayed increasing in phases, turns, and the duration of the MUAP [Figure 2]. This reflects the asynchronous firing of potentials from a reinnervated muscle^{[2,4,8,13-15]}. Therefore, polyphasia can occur in the setting of early reinnervation with poorly myelinated axonal sprouts that eventually become more established, allowing for the polyphasia to resolve (see “Time course of MUAP changes”).

Figure 2. Motor unit action potential. Amplitude reflects the number of muscle fibers contributing to a potential elicited from a single nerve fiber. Duration reflects dispersion in neuromuscular activation of branches to muscle fibers. Phases reflect baseline crossings and relate to the increased complexity of motor units due to reinnervation. Amplitude is 100 µV per division. The sweep speed is 5 msec per division.

Time course of MUAP changes

A predictable pattern of MUAP changes is seen after nerve injury resulting in axonal loss. Within the first week to 2 weeks, MUAP morphology is normal, but recruitment is reduced, reflecting drop out of MUAPs innervated by injured axons. By 4 weeks after injury, distal sprouting to denervated muscle fibers occurs from adjacent axons; this results in polyphasic MUAPs because of unstable neuromuscular transmission and asynchronous conduction along immature, poorly myelinated/unmyelinated nerve fiber sprouts branching from intact axons, as described above. These are referred to as “nascent units”. Because reinnervation is still early and nascent units have reinnervated a few muscle fibers, motor units can appear small. As more distant muscle fibers are reinnervated over the subsequent several months, the MUAP size and duration grow while remaining polyphasic. After ~6 months, the degree of polyphasia will diminish as nascent units mature and develop more stable neuromuscular transmission and conduction. However, MUAPs will continue to appear long in duration and large amplitude, reflecting the more significant number of muscle fibers taken over by the reinnervating motor unit^[2,4,13-15].

Free-running spontaneous electromyography

Continuous electromyography is monitored from muscles innervated by nerves or nerve roots at risk for injury during surgery. Usually, the recorded EMG signals will consist of a broad mixture of motor units. Surgical manipulation, such as stretching or compression of peripheral nerves, branches, or nerve roots, can result in high-frequency bursts or trains of neurotonic discharges in the corresponding muscle(s). Such changes in muscle activity provide valuable instantaneous feedback to the monitoring team and the surgeons regarding the surgical manipulation and actions and their proximity to neural structures. In this manner, EMG monitoring is beneficial for preventing severe or irreversible nerve injury from inadvertent trauma.

Irritation related to the surgical procedure activates motor units in various patterns that several factors, such as the nerve condition, the degree of surgical manipulation, and the neuromuscular integrity, can influence. Ischemia from compression of the local blood supply can be another cause of such changes in neural activity. Simultaneous activation of multiple motor units can result in EMG “bursts”, which are brief periods of polyphasic EMG activity [Figure 3]. EMG bursts can often inform the surgeon of proximity to a peripheral nerve or nerve root. Continuous compression or traction of nerves or nerve roots can result in the repetitive firing of one or more motor units for seconds to minutes producing an EMG “train” [Figure 4]. The higher amplitude or frequency trains are often due to significant nerve fiber recruitment from the excessive local force, which could be a harbinger of nerve injury if sustained. In general, the degree of motor nerve irritation correlates roughly with the intensity of EMG activity, i.e., the duration, frequency, and complexity of neurotonic activity are often associated with its severity. Despite the utility of neurotonic discharges, their presence does not necessarily indicate nerve damage, and their absence does not exclude this possibility. Mechanical stimulation can be sufficient to induce EMG discharges but not permanent neural injury. In addition, damaged nerves are less likely than healthy nerves to produce discharges^[18-20]. However, persistent neurotonic EMG activity, despite cessation of the surgical factor(s), can indicate the degree of local nerve injury. The EMG may become quiescent after the transection of a nerve with or without an initial large burst or train of neurotonic activity. In addition, manipulation and stimulation of the distal nerve stump may still activate the muscle, leading the surgical team to erroneously conclude that the motor nerve remains intact^[18-20].

Figure 3. Spontaneous EMG recording demonstrating two large bursts of neurotonic activity in the orbicularis oris muscle during facial nerve retraction.Amplitude is 20 µV per division. The sweep speed is 400 msec per division.

Figure 4. Spontaneous EMG recording demonstrating trains of neurotonic activity from several motor units in the orbicularis oculi and multiple motor units in the orbicularis oris muscles during facial nerve retraction. Note the superimposed electrocardiogram artifact in the trapezius muscle recording. Amplitude is 20 µV per division. The sweep speed is 400 msec per division.

It can be determined whether or not the change in EMG activity is related to a surgical action by temporal and anatomic correlation. If there is no relation of the activated muscles to the timing and location, e.g., spinal nerve root or peripheral nerve, of the surgical manipulation, it is unlikely that there is a causative relationship. For example, increased EMG activity in the abductor pollicis brevis muscle (C8, T1) cannot be explained by surgical manipulation near the ipsilateral C5 spinal nerve root. Other potential causes of neurotonic discharges include temperature changes related to local cool saline irrigation, heat from electrocautery, or local osmotic changes. Determining mechanical versus non-mechanical causes is essential since it is less common for the latter to result in neurotonic activity or neural damage. However, mechanical irritation can pose a significant risk of injury, particularly with repetitive trauma.

Electrodes detect interference or ambient electrical noise from various sources within the operating room, such as bed warmers, pumps, electrocautery devices, drills, or microscopes. It is essential to distinguish such artifacts from EMG activity [Figure 5]. In addition, various neuromuscular conditions and treatments can interfere with EMG recording, e.g., muscular dystrophy, myasthenia gravis, chronic neurogenic conditions, and botulinum toxin. EMG activity can also change due to several non-surgical factors, including the level of sedation and pharmacologic neuromuscular blockade. Motor unit potentials may saturate the recording due to irritation of multiple axons or voluntary muscle activation as the patient begins to awaken from the sedated state [Figure 6]. This EMG activity is usually bilateral and often involves the axial muscles. In addition, it is an anesthetic requirement that no paralytic agents are administered when monitoring intraoperative EMG. The absence of neuromuscular blockade can be verified with train-of-four (TOF) testing. Four successive supramaximal electrical stimuli at 2 Hz are applied to a peripheral nerve (e.g., ulnar nerve), and the resultant CMAPs are recorded from a corresponding muscle (e.g., abductor digiti minimi). Four robust CMAPs of equal amplitude should be induced for the EMG recording to be of sufficient sensitivity to detect intraoperative changes reliably. For short-acting agents, the neuromuscular blockade effects can dissipate during surgical exposure. However, if TOF testing indicates that neuromuscular blockade is still present, this can be resolved by administering reversal agents such as sugammadex^[18,20-23].

Figure 5. EMG recording demonstrating EMG activity with a superimposed artifact in the masseter, orbicularis oculi, and orbicularis oris muscles during the surgical procedure. Amplitude is 20 µV per division. The sweep speed is 400 msec per division.

Figure 6. EMG recording demonstrating a train of increased spontaneous EMG activity in the orbicularis oris and trapezius (control) muscles during wearing off of pharmacologic sedation. Amplitude is 20 µV per division (orbicularis oculi and oris) and 50 µV per division (trapezius). The sweep speed is 400 msec per division.

Stimulation-triggered electromyography

Polyphasic CMAPs in innervated muscles can be evoked with motor nerve stimulation, either directly or through intervening tissue, e.g., tumor or bone [Figure 7]. To localize a nerve or nerve root intraoperatively, a sterile handheld monopolar or bipolar stimulating probe is used within the operative field by the surgeon. Typically, the surgeon will use the monopolar stimulator during dissection for stimulation through tissue, such as bone or tumor, to determine if a neural structure, i.e., nerve root, plexus, or peripheral nerve, is nearby. The bipolar stimulator is used when the neural structure is in clear view. An absent CMAP may indicate the absence of motor nerve function, technical problems (poor nerve contact, current shunting, subthreshold stimulation intensity, erroneous recording settings), or pharmacologic neuromuscular blockade. Figure 8 illustrates how the surgeon can sweep the stimulator across the surgical field to locate the peripheral nerve precisely; the stimulation location corresponding to the largest amplitude CMAP would be closest to this neural structure. In addition to providing information regarding localization, the integrity of the nerve or nerve root can be quantified by determining the stimulation threshold and how consistently this is maintained during the procedure. Another way to quantify nerve function is to apply supramaximal stimulation to the motor nerve: the size of the CMAP generated is correlated with the number of functioning axons between the stimulation and recording locations. Stimulation can be performed intermittently during surgery to monitor the integrity of the nerve. Preservation of the response and consistency of the threshold with proximal stimulation provides objective evidence that no significant neural injury has occurred. If a decrement of the CMAP amplitude or increase in the stimulation threshold occurs, the surgeon should consider altering the surgical approach. A significant reduction or complete disappearance of the response often correlates with the severity of the postoperative neural deficit in the short-term and potentially long-term^[18,19].

Figure 7. Bottom four tracings - Triggered EMG recordings demonstrating large amplitude, polyphasic compound muscle action potentials (CMAPs) in the orbicularis oculi and oris muscles during facial nerve stimulation within the surgical field. The top four tracings show the corresponding free-running EMGs with superimposed stimulation artifacts. Please note the different time bases for the recordings, i.e., 400 ms/division for the four top tracings and 1.5 ms/division for the bottom four tracings.Amplitude is 20 µV per division.

Figure 8. Triggered EMG recordings demonstrating large amplitude, polyphasic compound muscle action potentials (CMAPs) in the orbicularis oculi muscle during facial nerve branch stimulation, sweeping across the surgical field. The largest amplitude response corresponds to the location closest to the facial nerve branch. Amplitude is 20 µV per division.

Bipolar probe utilization results in localized current stimulation that minimizes undesired current spread to adjacent nerves. This can be advantageous when multiple nerves are nearby, such as in the brachial or lumbosacral plexus. However, current shunting between the two electrodes can occur within conductive fluids, e.g., blood in the surgical field, resulting in a false negative motor nerve response. This can be avoided by ensuring that the surgical field is sufficiently dry to avoid this aberrant local conduction. For monopolar stimulation, a reference anode electrode is placed remotely at least several centimeters away from the stimulation site, which results in a broader stimulation field. Monopolar stimulation is not prone to current shunting and can help localize deep neural structures during dissection^[18,21-23].

Intraoperative nerve conduction studies

Peripheral and cranial nerves and nerve roots can be electrically stimulated intraoperatively, recording the compound nerve action potential (NAP) at another location along the course of the sensory or motor nerve. Plastic surgeons often use intraoperative stimulation to identify fascicles before nerve transfer. Handheld bipolar electrodes are utilized within the surgical field for stimulation and recording. Such recordings can assess conduction across the nerve’s scarred, injured, or compressed regions. Motor or sensory nerve conduction changes can occur during the procedure as indicated by alterations in the latency, amplitude, or conduction velocity of the NAP [ Figure 1]^[18,20,22]. An absent NAP may be due to the lack of motor or sensory nerve function or could be related to technical problems (poor nerve contact, current shunting, subthreshold stimulation intensity, erroneous recording settings). As with the CMAP, nerve function can be quantified during surgery using supramaximal sensory or motor nerve stimulation, where the amplitude of the NAP is proportional to the number of functioning axons between the stimulating and recording electrodes^[19].

Errors in electrodiagnosis of radiculopathies

EMG results that do not correlate well with clinical features or inconsistencies in electromyographic data underscore the need to understand potential sources for error in electrodiagnostic studies. Errors can result in underdiagnosis (type 2 error or false negative), overdiagnosis (type 1 error or false positive), or misdiagnosis [Table 2]. One of the more important limitations is the nerve lesion’s age. Studies performed during the hyperacute or acute period after injury (0-3 weeks) are likely to be normal. Reduced recruitment of MUAPs can be seen if there is sufficient root demyelination to cause conduction block, but this is not sensitive; conduction block requires significant demyelination, and the dual innervation of muscles by more than one nerve root further dilutes the impact. Rarely, very brief trains of positive sharp waves can be seen in the acute period reflecting motor axon loss. Increased abnormal spontaneous activity, fibrillation potentials, and positive sharp waves appear in the acute to the subacute period (> 3 weeks post-injury), and this is the phase in which studies can be most helpful. If fibrillation potentials and positive sharp waves are seen without associated changes in motor unit action potential morphology, the lesion represents active injury and is termed acute, active radiculopathy. The presence of changes in motor unit architecture with increased duration, phases, and amplitude establishes the radiculopathy as chronic; if fibrillation potentials and positive sharp waves are present, it becomes a chronic, active radiculopathy.

Another source of underdiagnosis occurs if only sensory fibers are involved in the root injury. The nerve root’s ventral (motor) branch is adjacent to the protruding intervertebral disk and, therefore, at greater risk than the dorsal (sensory) branch in the setting of a disk herniation. However, if a root injury affects only the sensory branch, sparing motor axons, needle EMG will be normal even in the setting of radicular pain and sensation change. This is similar to the situation if the root injury is purely demyelinating due to nerve root compression. If there is insufficient conduction block to cause reduced recruitment, the electrodiagnostic studies will also be normal. If there is only intermittent nerve root compression, as can be seen with spinal stenosis where the symptoms are purely positional, fixed EMG changes may not occur^[8].

Most errors in electrodiagnostic testing for radiculopathy affect the sensitivity rather than the specificity of testing [Table 2]. However, specificity can be affected if the diagnosis relies solely on MUAP morphology changes (49%-50% specificity) rather than fibrillation potentials and positive sharp waves (71%-89% specificity). Similarly, if too few or inappropriate muscles are sampled during the needle EMG, a false negative study may result. Therefore, the diagnosis should be based on a minimum of two muscles within the same myotome with different peripheral nerve innervation. Needle EMG of paraspinal muscles revealing active denervation potentials confirms the lesion is at or proximal to the nerve root level^[24,25]. A caveat to the interpretation of paraspinal muscle abnormalities is overlapping innervation, making it difficult to localize a root lesion based on paraspinal denervation alone^[8]. Localization in the extremity can also be limited by sampling error, myotomal overlap or variability in the innervation of muscles. A range like L4-L5 might be the most precise localization electrical studies can offer. Finally, fibrillation potentials and positive sharp waves are nonspecific and may be seen in various conditions, including peripheral neuropathy. This can limit electrodiagnostic specificity in the setting of co-morbid conditions.

Pedicle screw testing

Intraoperative pedicle screw stimulation is a highly sensitive and specific method for detecting pedicle wall breach into the spinal canal during screw placement. A malpositioned pedicle screw poses a significant risk to the nerve root. Although the secure placement of pedicle screws can be determined by fluoroscopy, the local neuroanatomy cannot be visualized, so triggered EMG provides an essential complementary tool for this purpose. The holes and screws correctly positioned within the pedicle wall are separated from the adjacent nerve roots by a high-impedance cortical bony layer. When a pedicle screw is placed correctly, the surrounding bone impedes electrical conduction to adjacent nerve roots. Therefore, a higher electrical current will be needed to stimulate the local nerve root. During testing, a monopolar electrical stimulator is used to apply a series of increasing currents to the pedicle screws. Electrodes placed into muscles innervated by the respective spinal nerve root levels will record CMAPs synchronized with the stimulation. If a pedicle wall breach occurs, the stimulation threshold is significantly reduced. A stimulation threshold greater than 15-20 mA reliably indicates adequate screw position. A stimulation threshold of less than 10 mA is often associated with significant pedicle cortical perforation by the screw. When less bone separates the screw from the nerve root, the stimulation threshold will be reduced due to lower electrical impedance. This should prompt the surgeon to remove or redirect the screw at that site, particularly if there is concurrent neurotonic EMG activity in a corresponding muscle. Stimulation thresholds between these two current levels usually indicate adequate screw position, although surgical exploration to determine the integrity of the pedicle wall should be considered. False negatives can occur due to various factors, such as the use of muscle relaxants or pre-existing nerve damage, which can result in a higher stimulation threshold or no CMAP response at all. In addition, the region around the pedicle screws should be sufficiently dried to avoid current shunting, which can also result in an erroneously high stimulation threshold. In such cases, if possible, one should first determine the stimulation threshold to evoke a CMAP by directly stimulating the exposed chronically compressed spinal nerve root and then performing pedicle screw testing^{[18,20,27,28]}.

Errors in the electrodiagnosis of plexopathy

An advantage of the brachial plexus study over radiculopathy is the availability of sensory responses as part of the diagnostic arsenal. This not only helps to distinguish a lesion as localized distal to the dorsal root ganglion (DRG) but also increases the number of responses that can be obtained to detect an abnormality. Similar considerations regarding limitations in plexus studies apply, as have been discussed for radiculopathy studies. As in radiculopathy studies, there exist similar timing considerations, with poor sensitivity during the hyperacute period (< 3 weeks). The injury must either result in axonal loss or conduction block to be detected. An adequate sample of muscles representing trunks, cords, and nerves must be tested to be confident of localization. Brachial plexus lesions can often be patchy to complicate these challenges further, leading to incomplete lesions of specific anatomic branches. There can also be variation in the spinal levels supplying the plexus, leading to an incomplete denervation picture or denervation in unexpected muscles. As discussed above, differentiation of plexus injury from radicular or polyradicular involvement depends on the sensory NCS, the pattern of denervation on needle EMG, and, essentially, denervation in paraspinal muscles. However, errors can occur in these metrics: (i) sensory nerve action potentials can be affected if the DRG is intraspinal or very proximal; (ii) needle study can be insensitive in the absence of active denervation; (iii) fascicular involvement can mislead localization or result in incomplete data; and (iv) paraspinal muscle study may be contraindicated due to anticoagulation or limited by incomplete relaxation. Therefore, plexus studies are often more extensive and difficult to localize.

Errors in the electrodiagnosis of mononeuropathies

Several potential pitfalls exist in the routine electrodiagnostic evaluation of errors in the electrodiagnosis of mononeuropathies (UNE). In particular, the position of the upper extremity when performing the studies can have a significant effect on measurements used to calculate conduction velocities^[34]. The elbow should be kept in a moderate flexion position to provide the greatest correlation between the surface skin measurement and true nerve length^[35]. The distance between below and above elbow nerve conduction testing sites should be 10 cm, as shorter segments may amplify errors and longer segments may dilute abnormal findings^[17]. The below-elbow stimulation site should be 3 cm distal to the ulnar groove to include the cubital tunnel across the elbow; however, it should not be too distal to possibly include a Martin-Gruber anastomosis (MGA) in the forearm^[8]. MGA must be excluded in the setting of conduction block across the forearm segment since there are rare reports of a proximal MGA mimicking UNE^[36]. Finally, subluxation of the ulnar nerve with elbow flexion is a potential cause of a false negative study due to erroneous overestimation of nerve length across the elbow.