IOM During Pedicular Screw Placement

The use of pedicle screws for spinal stabilization has become commonplace during various spinal surgical procedures. However, the placement of these screws is largely done blindly, and even in the hands of experienced surgeons, the incidence of misplaced pedicle screws resulting in neurological impairment has been reported to be quite high, despite the use of surgical inspection and imaging techniques. Although new imaging techniques have been developed that may help to reduce the incidence of misplaced hardware, the equipment needed to implement their usage is generally costly, and the techniques themselves are still not developed to the point where they are free from error. As a result, surgeons and clinical neurophysiologists have used various electrophysiological monitoring techniques for assessing nerve root function and pedicle screw placements. To be widely and effectively used, these techniques must meet certain criteria; the strengths and limitations of each of the techniques will be discussed in terms of these criteria. The combined use of spontaneous and triggered myogenic activity for intraoperative monitoring purposes satisfies all the criteria that a monitoring technique should meet. This technique is cost-effective and improves surgical outcomes.

Introduction

Through the ages, various treatments for spinal deformity have evolved. In
1962, Harrington ushered in the revolutionary use of metallic, internal fixation devices for spinal deformity when he reported on the use of a distraction rod construct for the treatment of scoliosis. In 1982, Luque demonstrated how spinal deformity could be corrected by the use of segmental fixation and the application of transverse forces. In the thoracolumbar region of the spine, spinal instrumentation constructs consisting of hooks and rods have now become the standard of care for the surgical management of degenerative spinal disease and traumatic insults. In the lumbosacral region, it has become very popular to use pedicle screws rather than hooks to hold the rods in place for the purpose of segmental transpedicular fixation.
Although pedicle screws can be placed in the thoracic and lumbosacral spine, they are generally placed in the most caudal segments of the spine:
L2–S1. The cross-sectional area of the pedicles is smaller in the thoracic segments of the spine. Because of this, and because it is common for spinal instabilities to occur in the lumbosacral region, it is unusual for pedicle screws to be placed in the thoracic region. From a monitoring perspective, it is important to remember that the spinal cord ends in the conus medullaris at about the T12–L1 level of the spine. Therefore, the placement of pedicle screws below these levels potentially places nerve root rather than spinal cord function at risk.
Proper placement of pedicle screws requires that a surgeon be extremely knowledgeable about the anatomical characteristics of the thoracic, lumbar, and sacral vertebrae. Despite the use of anatomical landmarks and fluoroscopy, the placement of pedicle screws is largely done blindly. Ideally, they should be placed so that they pass through the pedicle with about 1 mm to spare on both the medial and lateral walls and without any breach of the pedicle walls. In addition, they should be placed well into the vertebral bodies without any breach of the vertebral body walls. Nerve roots tend to position themselves near the medial and inferior aspects of the pedicles as they exit the spinal canal through the spinal foramen. Therefore, screws that are placed so that they protrude or are exposed from the medial or inferior pedicle walls can cause nerve root irritation or injury. In preparation for the placement of pedicle screws, markers are generally placed into the pedicles in order to visualize, via radiographs, the trajectories that the pedicle screws will take. These trajectories may place nerve roots at risk for injury, because both lateral and anterior–posterior radiographs are subject to reading errors. As a result, undesirable medial placements of both markers and screws may not be identified. Such readings are followed by removal of the markers and tapping of the holes made by the markers. These holes can then be palpated to detect holes in the pedicle walls. The pedicle screws are then placed. This placement can result in fractures of the pedicle, breakthroughs of the pedicle walls, and/or extrusion of pedicle fragments. Even in the hands of experienced surgeons, the current literature reports pedicle cortical perforation rates that have ranged from 5.4 to 40%. Such events may go undetected unless the pedicle walls are visualized. However, most surgeons are reluctant to routinely visualize screw placements unless this action is warranted, since doing so would require that the surgeon do multiple laminotomies.
This is time-consuming, and these additional procedures by themselves could also affect postoperative outcome. However, despite the use of surgical inspection and imaging techniques, misplaced screws have still been frequently associated with neurological functional impairment. The incidence has ranged from 1% to more than 11%.
New imaging technologies aimed at reducing these incidences continue to evolve. Thus far, they have been somewhat cumbersome, costly, and time consuming, and the end result is that they are still not free of error. Therefore, existing electrophysiological techniques have been used, as well as others that have evolved for monitoring neurological function during pedicle screw placement and for assessing these placements. They include mixed nerve somatosensory evoked potentials (SEPs), dermatomal somatosensory evoked potentials (DSEPs), and techniques that rely upon both spontaneous and triggered myogenic activity. In addition, the measurement of electrical tissue impedance has been suggested as another means for assessing placements.
Other techniques, which include transcranial and spinal stimulation, can also be used to test nerve root function during pedicle screw placement; however, although feasible, these techniques are rarely used for this purpose.

Techniques For Assessing Nerve Root Function And Pedicle Screw Placement

All nerve roots consist of both sensory and motor fibers. The monitoring techniques that are used to assess nerve root function during surgery produce either sensory or motor responses. Sensory responses that are mediated by a single specific nerve root can be elicited by stimulating a specific body surface area known as a dermatome. Motor activity that is mediated by a single specific motor nerve root can elicit myogenic responses from a group of muscles known as a myotome. Therefore, the responses that are acquired to assess the sensory and motor function of a single nerve root are known as dermatomal and myotomal responses, respectively.
There are certain criteria that monitoring techniques should meet if they are to be widely and effectively used to assess pedicle screw placements and preserve nerve root function. First of all, implementation of the techniques should be practical; in other words, they should not require special equipment or expertise. Otherwise, these factors will be a deterrent to their use. For economic and practical reasons, the techniques should utilize standard equipment that may already be used for monitoring purposes, they should be easy to perform, and the anesthetic requirements should not be unusual. Second, the techniques must be effective. They should provide an instantaneous indication of nerve root irritation in order to prevent injury or further damage of a nerve root that is already irritated. They should also be able to detect the presence of a misplaced screw that is not causing nerve root irritation but that may have the potential to do so. Finally, they should produce accurate results that make a difference in patient outcomes and that are cost-effective. A discussion of each of the following techniques will address these requirements.

Sensory Pathway Assessment Techniques

Somatosensory Evoked Potentials (SEPs)

Since the late 1970s, mixed nerve somatosensory evoked potentials (SEPs) have been used to monitor spinal cord function during spinal instrumentation procedures in order to minimize the probability of postoperative neurological deficits. SEPs are elicited by stimulating a peripheral nerve at a distal site: typically the median or ulnar nerves at the wrist for acquiring SEPs from the upper extremities, and the posterior tibial nerve at the ankle or the peroneal nerve at the fibular head for acquiring lower-extremity SEPs. The ascending sensory volley that contributes to the SEP enters the spinal cord through dorsal nerve roots at several segmental levels and may ascend the spinal cord via multiple pathways. The general consensus is that the dorsal or posterior column spinal pathways primarily mediate the SEPs. Other pathways, such as the dorsal spinocerebellar tracts and the anterolateral columns, may contribute to the early SEP responses that are used for monitoring purposes.
Despite the fact that SEPs are primarily mediated by the dorsal columns and therefore are a means of directly assessing only sensory and not motor
pathway function, they have proven to be extremely useful as a clinical tool for detecting changes in spinal cord motor function, particularly when these changes result from mechanical insults.
It is important to realize that mixed nerves receive sensory and motor fibers from multiple nerve roots. Therefore, when mixed nerves are stimulated, the electrophysiological responses that result from the stimulation (known as SEPs) are mediated by more than one nerve root prior to being mediated by the spinal cord. It is not unreasonable to expect that SEP changes should occur when the function of one of the contributing nerve roots becomes abnormal. However, the usefulness of SEPs for assessing spinal root function in patients diagnosed as having cervical spondylosis and lumbar root lesions has been limited. In addition, when used as a neurophysiological monitoring tool during pedicle screw placements, they appear to be totally insensitive to changes in nerve root function, largely because several nerve roots typically contribute to the composition of a peripheral nerve.
For example, the posterior tibial nerve receives contributions from the L4, L5, S1, S2, and S3 nerve roots. As a result, a monoradicular functional abnormality may not be apparent when mixed-nerve evoked potentials are used to evaluate a patient because abnormal nerve root function may be masked by the normal activity mediated via unaffected spinal nerve roots. Therefore, mixed nerve SEPs may be insensitive to irritation or injury to a single nerve root. For this reason, they should not be used to monitor spinal nerve root function, since other techniques are much better suited for this purpose. On the other hand, since SEPs were developed and continue to be used as a technique for assessing spinal cord function, they might be useful during pedicle screw placements if spinal cord rather than nerve root function is placed at risk. However, since pedicle screws are usually placed at levels caudal to the conus medullaris (i.e., L2–S1), screw placement would not place spinal cord function at risk, and, indeed, there have been no published reports of the loss of spinal cord function during such procedures. Therefore, unless there is reason to believe that spinal cord function is at risk, there appears to be no basis for the use of SEPs as a monitoring tool during pedicle screw procedures.

Dermatomal Somatosensory Evoked Potentials (DSEPs)

A dermatome is defined as a body surface area that receives its cutaneous sensory innervation from a single spinal nerve root. It has been demonstrated that DSEPs arise from stimulation of receptors in the skin rather than from subcutaneous digital nerves. As a result, they are normally elicited using some form of surface electrodes and are probably mediated via the same pathways as mixed-nerve SEPs.
The first reported use of dermatomal or segmental SEPs (as they were initially named) was by European investigators. Since these first studies, they have been used to assess children with myelomeningocele, to evaluate patients with spinal cord injuries, as a monitoring tool during spinal
surgery to determine the adequacy of spinal nerve root decompression and to detect nerve root functional impairment during pedicle screw placement. DSEPs are acquired using the same stimulation and recording techniques and equipment that are used to acquire mixed-nerve SEPs. Unlike SEPs, however, the only DSEP responses that are clinically useful are recorded from the scalp, because it is normally very difficult to record either peripheral or subcortical DSEP responses. This may be a function of the relative number of afferent nerve fibers that mediate and contribute to DSEPs as compared to mixed nerve SEP responses. As a result, like mixed-nerve cortical SEP responses, DSEP responses can be very susceptible to the anesthetic drugs that are used during surgery. In addition, it has been shown that the latency and amplitude of dermatomal responses are a function of the stimulation intensity. Therefore, for monitoring purposes, stimulation intensity should remain constant throughout a surgical procedure so that one does not attribute response changes to surgical events.
DSEP responses, at least ideally, are considered to be nerve root–specific. However, this may not always be the case. Dermatomes tend to overlap, and their spatial distributions vary from person to person. Besides their susceptibility to typical anesthetic drugs, this is another minor shortcoming of this technique. One of the two major limitations associated with the use of DSEPs is that, because of their small amplitude, they can only be acquired using an averaging technique; hence, their acquisition, like that of SEPs, may require a few minutes. During this period of time, functional changes can occur that may go undetected until another average is acquired. At that point in time, nerve root damage may have occurred and the associated functional changes may be irreversible.
For pedicle screw placement, the second major limitation of the DSEP technique is that changes in dermatomal responses will only occur if a pedicle screw actually makes contact with a nerve root. Misplaced screws that do not make contact with a nerve root may still represent a potential source of nerve root irritation or damage and will go undetected with the dermatomal technique.

Anesthetic Management For Sensory Techniques

Although the monitoring of SEPs has clearly been beneficial during many surgical procedures, the anesthesia used to facilitate these procedures produces effects that alter the evoked potentials. These effects are well documented. They are most prominent on the cortically generated responses and less so on the subcortical and peripheral responses. They are generally dose related, and their effects on cortical SEPs tend to parallel their effects on EEG.
Most of the commonly used anesthetic drugs produce dose-related SEP changes that include amplitude decreases and latency increases. The relative degree of change differs between anesthetic agents. The drug dosage that causes a 50% decrease of cortical SEP amplitude correlates with the lipid solubility of the agent and therefore its anesthetic potency. Therefore, when anesthetic techniques are being considered, the effect of each anesthetic agent on specific monitoring modalities must be considered. Probably the most commonly used anesthetics are the halogenated inhalational agents (desflurane, enflurane, halothane, isoflurane, sevoflurane). All these agents produce a dose-related increase in latency and reduction in amplitude of the cortically recorded SEP responses. Several studies have demonstrated that halogenated agents differ in their potency of effect on cortical SEPs.
Isoflurane has been reported to be the most potent, and enflurane and halothane the least potent. At steady state, the potency of sevoflurane and desflurane appears to be similar to that of isoflurane. The effects are less on the subcortical SEP responses recorded over cervical spine and are minimal on spinal responses recorded epidurally or on peripherally recorded responses. If it is essential to monitor cortical SEPs, the use of halogenated inhalational agents may need to be restricted or eliminated entirely. However, if the recording of subcortical responses is adequate for monitoring purposes, halogenated agents may be acceptable anesthetic choices.
Nitrous oxide produces decreases in cortical SEP amplitude and increases in
cortical SEP latencies when used alone or in conjunction with halogenated inhalational agents or opioid anesthetics. When compared to other inhalational anesthetic agents at equipotent anesthetic concentrations, nitrous oxide produces the most profound cortical SEP changes. Like halogenated agents, the effects of nitrous oxide on subcortical and peripheral sensory responses are minimal. However, nitrous oxide has been reported to have a synergistic effect on cortical SEPs when used in conjunction with other inhalational agents.
Although the use of DSEPs represents an improvement over the use of SEPs since one is able to detect single nerve root functional changes, both SEP and DSEP cortical responses are sensitive to standard anesthetic management that includes the use of halogenated gases and nitrous oxide; the DSEPs are relatively more sensitive to these anesthetic agents than are mixed-nerve SEPS because they are smaller in amplitude to begin with. This minor limitation can be minimized by using anesthetic agents that are administered intravenously (total intravenous anesthesia–TIVA), but these anesthetic agents tend to be more costly than the anesthetic gases.
A number of factors determine the choice of anesthetic agents when monitoring is to be performed. These include (1) how anesthetic agents may interact with a patient’s pathophysiology, (2) surgical requirements (i.e., performance of a stagnara wake-up test, awake during a carotid endarterectomy procedure), and (3) the specific monitoring modalities to be used.
In general, anesthetic agents produce an alteration in the evoked responses consistent with their clinical effects on the CNS. Several important generalizations can be made regarding the effects of anesthetic agents on SEPs. First, most tend to decrease neural conduction and synaptic transmission. As a result, they tend to decrease the amplitude and increase the latency of SEPs. Second, the effects of anesthetics on SEPs appear to be most prominent in regions where synaptic transmission is prominent. Therefore, their effects are most pronounced on cortically generated peaks and least effective on brainstem, spinal cord, and peripheral responses. Third, anesthetic effects appear to be dose related, although many agents have a disproportionate effect at low dosages in the range where major clinical anesthetic effects are occurring. Fourth, just as
patients react differently to the same dose of an anesthetic drug, so also their SEPs are affected differently. Finally, during periods when neurological function is acutely at risk, it is important to maintain a steady state of anesthesia.
Taking into consideration all these factors, an anesthetic regimen can usually be chosen that will permit effective monitoring.

Motor Pathway Assessment Techniques

Myotomes are the motor complement to dermatomes, and myotomal distributions are also quite variable between individuals. Whereas a myotome is a group of muscles that receive their motor innervation from a specific spinal nerve root, most muscles receive efferent innervation from several spinal nerve root levels. The amount and type of innervation to specific muscle groups will vary from person to person.
Myotomal activity can be spontaneously elicited by mechanical stimulation or triggered by electrical stimulation. Typically, the myotomal activity from several muscle groups is monitored at any given time, and the activity is recorded using either surface or subdermal needle electrodes placed over or into the various muscle groups. The selection of the muscle groups to monitor is made on the basis of which spinal nerve roots are at risk for irritation or injury.
Muscles typically receive their innervation from several spinal levels, although one spinal level generally predominates in terms of the amount of innervation it provides to any given muscle group. Activity can be recorded from muscles innervated by the cervical, thoracic, lumbar, and sacral spinal nerve roots. In addition to paraspinal muscles, the muscles commonly used for these recordings and their innervation appear in Table1.

The responses that are elicited when nerve roots are mechanically or electrically stimulated are summed responses from many muscle fibers known as compound muscle action potentials (CMAPs). They can be recorded using pairs of surface or needle electrodes that are placed over or into the belly of a muscle. Recordings should be made continuously throughout a surgical procedure.
Assuming that excessive amounts of muscle relaxants have not been administered to a patient during surgery and that muscles are adequately unrelaxed, spontaneous activity will be elicited when mechanical activation results in nerve root irritation or injury. This spontaneous activity, suggestive of nerve root irritation, can be recorded when train-of-four testing of target muscles (i.e., muscle groups innervated by the nerve roots at risk) produces only one CMAP. The activity will typically be elicited from one or more muscle groups, depending on the activated nerve root, the muscle groups being monitored, and the placement of the recording electrodes on these muscle groups.
The EMG activity from each electrode pair is recorded using differential amplification and is filtered using a wide bandpass filter (30 Hz–3 kHz). When spontaneous myogenic activity is recorded to detect mechanical nerve root irritation, the data acquisition system should be set to operate in the free run mode. In this mode, the sweep time is typically 1 s and any elicited activity can easily be visualized and evaluated. Typically, several channels of myogenic activity should be monitored simultaneously, depending on the number of channels available, but six or more should be monitored.
When interpreting spontaneous activity, there are several factors to take into consideration. First of all, normal nerve roots and irritated or regenerating nerves in continuity react differently to mechanical forces. When mechanical forces are statically or rapidly applied to normal nerve roots, they induce no nerve root activity or trains of impulses of short duration. When the same forces are applied to irritated or regenerating nerves, they induce long periods of repetitive impulses. Minimal acute compression of normal dorsal root ganglion also induces prolonged repetitive firing of nerve roots. When interpreting intraoperative motor nerve root activity, it is important to understand the pathophysiological mechanisms of nerve root injury and to understand the response of normal and pathological nerve to not only different types of mechanical force but also to electrical stimulation.
Normally, the recordings of spontaneous activity will demonstrate the lack of activity. However, when preexisting nerve root irritation has been present, the recordings will often consist of low-amplitude periodic firing patterns. Mechanically elicited activity consists of either short bursts of activity that can last a fraction of a second or long trains of activity that can last up to several minutes. The short aperiodic bursts of activity are common. Attention should be paid to these, but they are normally not cause for alarm and are rarely indicative of a neural insult. The long trains are more serious, may be indicative of neural injury, and are causes for alarm. The short bursts are associated with direct nerve trauma such as tugging and displacement, irrigation, electrocautery, metal-to-metal contact, or application of soaked pledgets. Train activity is commonly related to sustained traction and compression. The more sustained the activity, the greater the likelihood of nerve root damage. When train activity occurs, the surgeon must be notified so that corrective measures can immediately be taken.

Triggered Myogenic Activity

As mentioned earlier, triggered myogenic activity can be elicited in several ways: through direct nerve root stimulation, indirect nerve root stimulation
by means of stimulation of spinal instrumentation, direct spinal cord stimulation (myogenic motor evoked potentials), and transcranial motor evoked potentials elicited by either electrical or magnetic transcranial stimulation.

Transcranial and Spinal Stimulation

Both spinal cord and transcranial stimulation are typically used to elicit myogenic responses from lower-extremity muscle groups in order to assess spinal cord motor function. They can also be used to assess individual nerve root function during pedicle screw placements. However, they typically are not used in this fashion, for several reasons. First, these techniques are more complex than other techniques that are currently available; they require special equipment, electrode placement skills, anesthetic management, and/or consent for their implementation. Second, functional status can only be determined when stimulation occurs, and not continuously. Third, although myogenic responses can generally be elicited from some designated target muscle groups such as the tibialis anterior muscles, the threshold stimulation intensities needed to elicit myogenic responses from other designated target muscles vary from muscle to muscle. These thresholds can vary during a procedure, and these changes may be unrelated to surgical causes. Therefore, their reliability in determining when a surgical event has caused a functional change is in question. Finally, the techniques can only detect when functional changes have already occurred; they are not able to detect potential causes of functional changes. As a result, investigators have turned to other techniques for monitoring nerve root function during pedicle screw placements. These techniques include direct nerve root stimulation and indirect nerve root stimulation by means of stimulation of spinal instrumentation.

Direct Nerve Root Stimulation

Direct nerve root stimulation is sometimes used to determine the stimulation thresholds of nerve roots placed at risk during screw placement. Ideally, this technique should be used in conjunction with indirect nerve root stimulation when stimulation thresholds are in question, either as a result of chronic nerve root compression (particularly when a radiculopathy is present) or when disease processes are present, such as diabetes, that may effect nerve root function. It is generally assumed that when nerve roots are indirectly stimulated via the spinal instrumentation, the nerve roots that are being excited are healthy and function normally. The constant current stimulation threshold for eliciting responses from normal nerve roots range from 0.2 to 5.7 mA, with an average stimulation intensity of 2.2 mA. However, investigators have reported that chronically compressed nerve roots have elevated stimulation thresholds, and  stimulation thresholds greater than 20 mA may be necessary to elicit myogenic responses from such nerve roots. These findings indicate that if test parameters that have been developed from testing pedicle screw placements in patients with normal nerve root function are used to test screw placements involving chronically compressed nerve roots, false-negative findings can result.
As indicated earlier, elevated stimulation thresholds may also occur in patients with metabolic disorders such as diabetes. However, no reports have appeared in the literature to support this supposition. Experience with such patients is limited, but some of the diabetic patients tested have had only moderately elevated stimulation thresholds. Rather than having thresholds of 2 mA or less, the diabetic patients have been found to have thresholds of 4–5 mA—still in the normal range reported by others.
One way to avoid false-negative findings is to directly stimulate each nerve
root at risk to ensure that it is functioning normally before testing the placement of each pedicle screw. Although this may be a reasonable step if decompressions are being done, routine laminotomies to explore each nerve root are time-consuming and can be associated with undue risk. Therefore, most surgeons may prefer not to use direct stimulation of nerve roots in conjunction with indirect nerve root stimulation techniques. However, in patients exhibiting signs of nerve root malfunction as a result of either compression or disease processes, it is strongly suggested that direct nerve root testing be performed to establish stimulation thresholds when indirect nerve root stimulation techniques are being used.

Indirect Nerve Root Stimulation Techniques

Indirect nerve root stimulation is performed by electrically stimulating bone or hardware in order to elicit nerve root responses. Some surgeons favor such testing during every aspect of pedicle screw placement. They test the probe used to make the initial hole into the pedicle for marker placement, the markers, the taps used to make the holes for the pedicle screws, the pedicle screw holes, and the pedicle screws themselves. Other surgeons may prefer to test only the screw placements. The assessment criteria are similar in all cases.
The published stimulation parameters that have been used have varied. These studies have used either constant current or constant voltage stimulation to assess placements. Although similar, these two forms of stimulation are not equivalent. The flow of electrons, also known as current
flow, is what actually causes a nerve or nerve root to depolarize. Voltage is only the driving force that causes the electrons to flow through the resistance or impedance of biological tissue. When testing pedicle screw placements, tissue impedance includes that of pedicle and vertebral body bone in addition to the impedance of muscle, vascular tissue, and blood. Although the latter impedances probably remain relatively constant between individuals, bone density and therefore bone impedance is known to vary between individuals as a result of osteoporosis and other factors. Therefore, it takes more or less voltage to cause the same current to flow in various individuals. Therefore, it would be expected that the results of using constant voltage for testing pedicle screw placements would be more variable than constant current stimulation. Constant current stimulation appears to be superior to constant voltage stimulation for assessing pedicle screw placements. Various stimulation parameters and techniques have been used to electrically assess pedicle screw placements. A probe of some type such as a nasopharyngeal electrode functions as the cathode and is placed within the pedicle screw holes and/or on hardware, and a needle electrode is typically placed in muscle near the surgical site. It is used as the anode and provides are turn path for the stimulation current. Rates of pulsatile stimulation have ranged from 1 to 5 Hz with pulse durations of 50–300 ms. Typically, when testing, the intensity of the stimulation is gradually increased from 0 mA until a current threshold is reached at which a reliable and repeatable EMG response is elicited from at least one of the monitored muscle groups or a predetermined maximum stimulus intensity is reached. For safety reasons, we generally use 50 mA as a maximum stimulation intensity. If EMG responses are elicited at a stimulus intensity that is lower than a predetermined “warning threshold,” i.e., the stimulus intensity that is used to warn of a possible breach of the pedicle wall, the surgeon is advised to examine the hole or hardware placement. In such instances, radiographs may falsely suggest adequate screw placements. These “warning thresholds” have varied between groups of investigators.
Some have used stimulus intensities of 10 mA or higher, whereas others have used intensities as low as 6 mA.

Anesthetic Management For Motor Techniques

In order to provide an optimal surgical field, the anesthesiologist must render a patient unconscious and free from pain and must also control muscle tone. The degree of muscle relaxation is the only anesthetic factor of concern when myogenic activity is used for monitoring purposes. One way to suppress muscle tone is to suppress it at its origin, within the cerebral cortex, with deep anesthesia. Although nitrous oxide does not produce muscle relaxation, the administration of the halogenated agents such as halothane, enflurane, and isoflurane does have a dose-related effect. However, because of cardiovascular depression, these agents cannot be used by themselves to produce the amount of relaxation necessary for abdominal surgery. A second means of diminishing muscle tone is to block the signals from the brain to the muscles as they traverse the spinal canal by using either spinal or epidural anesthesia. A third means of doing so is to use neuromuscular blocking agents that interfere with the transmission of signals from motor nerves to muscle fibers. In order to avoid major arterial hypotension, neuromuscular blockade is achieved by the use of a neuromuscular blocking agent in conjunction with a volatile halogenated agent. In this way, the anesthetic is used to produce only unconsciousness and analgesia and can be administered at low safe concentrations.
When monitoring nerve root function using the spontaneous or triggered myogenic activity from specific muscle groups, it is imperative that these muscle groups be sensitive to changes in nerve root function resulting from
traction or compression of the roots. The level of muscle relaxation significantly affects myogenic responses. The greater the degree of muscle relaxation for the muscle groups of interest, the less likely they will be to respond to changes in nerve root function. As a result, it would be ideal if no muscle relaxation were used to interfere with elicited activity. To be absolutely sure that relaxation levels play no part in determining response thresholds, some neurophysiologists insist on patients being totally unrelaxed when testing. However, in many clinical settings, this level of relaxation may be difficult if not impossible to achieve, particularly if surgeons feel that it compromises their ability to adequately perform surgery.
An effective means of assessing the degree of muscle relaxation is to use a
train-of-four technique, which consists of electrically stimulating a peripheral nerve four times and recording the four CMAPs (T1, T2, T3, and T4) that result from target muscle groups. For the hands, the ulnar nerve could be stimulated at the wrist with CMAPs recorded from the adductor digiti minimi muscle. For the legs, the peroneal nerve could be stimulated at the fibular head and CMAPs could be recorded from the tibialis anterior muscle. Typically, 2 Hz, 0.2 ms pulses of supramaximal stimulation intensity are used to elicit the CMAPs. The T4 CMAP disappears with a 75% blockade, the T3 with 80%, T2 with 90%, and T1 with 100%.

Factors That Can Contribute To False-negative Findings

When pedicle screws have breached pedicle walls, these events should be detectable when using electrical stimulation for test purposes. When this form of testing fails to detect these events, false-negative findings in the form of nerve root irritation or damage can result. Several factors, both technical and physiological, can contribute to such findings. The following is a discussion of thesefactors.

Degree of Muscle Relaxation
Probably the most important factor when testing during the placement of pedicle screws is the degree of muscle relaxation, because it can significantly influence the stimulation thresholds at which responses are elicited. Therefore, an accurate assessment of muscle relaxation is essential. Train-of-four testing is a common way for anesthesiologists to make these assessments. Although this testing technique appears to be a reasonable way to make these assessments, testing should be done by the person providing the monitoring rather than by the anesthesiologist, for several reasons. First, the anesthesiologist typically does a train-of-four assessment using a small portable battery-driven device. These devices may not always work properly and should not be relied upon.
Second, the anesthesiologist’s assessment of train-of-four test results is a subjective one based on visible twitches from muscle groups that the anesthesiologist has access to—either hand or facial muscles. It is unlikely that the responses from these muscle groups will be the same as those from the leg muscles from which responses are elicited when pedicle screw testing is performed, because these muscle groups react differently to the relaxant levels. Finally, it is appropriate that the person providing the monitoring be responsible for guaranteeing that the test results are as accurate as possible by doing his or her own train-of-four testing of leg musculature.
The issue of what are adequate relaxant levels for accurately assessing threshold stimulation intensities remains controversial. Clearly, no spontaneous or triggered myogenic activity will be present with 100% blockade. On the other hand, it has been reported that it is not necessary to have the absence of any blockade to effectively monitor spontaneous and triggered myogenic activity. Spontaneous activity can be elicited with one twitch present during train-of-four testing, or up to 90% neuromuscular blockade.
However, when using indirect stimulation during pedicle screw placement, the assessment criteria make a determination of relaxant levels much more critical. It cannot be overstated how important it is to have the patient adequately unrelaxed when testing. If a patient is too relaxed when testing is performed, the stimulation thresholds for eliciting responses will be artificially elevated and may lead to false-negative findings. One example of this, when a patient had only one large twitch during train-of four testing and pedicle screw testing was performed. The stimulation threshold was found to be 50 mA. Further screw testing was delayed until a small fourth twitch became evident during train-of-four testing. Stimulation of the same screw then resulted in a threshold of 12 mA! Clearly, the relaxant level associated with the presence of only one twitch during train-of-four testing may be adequate for eliciting spontaneous activity, but it is not adequate for testing during screw placement. It has been reported that the minimal criterion for making such assessments is the presence of a fourth twitch from a target muscle. It has also been proposed that a postinduction–preinduction CMAP amplitude ratio from a hand muscle that is greater than 0.8 is a better measure for determining adequacy of relaxation. Our own service is currently using an amplitude ratio of the fourth to the first twitch to determine adequacy of relaxation criteria for assessing screw placements. Based on surgical findings when visualizing screw placements after experiencing stimulation thresholds below “warning thresholds,” we now feel that the value of this ratio must be at least 0.1. The direct electrical stimulation of nerve roots is one means of determining
if the level of muscle relaxation is adequate for using myotomal responses to assess nerve root function. If myogenic responses cannot be elicited using a constant current stimulation level of 2–4 mA, it is likely that the muscle relaxant level is too high to effectively monitor nerve root function using myogenic techniques. Considering the importance of proper relaxation, further studies need to be done to correlate the results of direct nerve root stimulation with noninvasive twitch monitoring.

Current Shunting

It was pointed out earlier that excitation of a nerve root only occurs when a portion of the current being applied in a pedicle hole or to pedicle hardware is adequate to excite and depolarize the nerve root. The stimulation current that is used to test pedicle screw placements can exit the screw through several different pathways and will seek the pathways of least resistance as the current returns to the anodal electrode. When a pedicle screw breaches a pedicle wall, it provides a pathway for current to exit. The larger the breach, the lower the resistance to current flow, and the greater the amount of current that will flow through the breach. If a nerve root is located close to the breach, excitation of the nerve root will occur. However, if fluid is allowed to accumulate at the surgical site so that the fluid makes contact with the stud of a pedicle screw, that fluid will provide another low-resistance pathway for current to flow away from
the screw. Less current will exit through the pedicle wall, and the amount of current needed to flow into the pedicle screw and cause depolarization of the nerve root will increase. As a result, stimulation intensities needed to elicit myogenic responses generally increased between 12 and 20 mA. It is interesting that despite the low resistance of the fluid, current is not completely but only partially shunted away from the pedicle screw.
Therefore, if shunting is present, it appears that it can mask the presence of a breached pedicle and result in false-negative findings when stimulation intensities are less than 30 mA. However, at stimulation intensities greater than this, such an occurrence seems unlikely.

Physiologic Factors

The physiologic factors that can contribute to false-negative findings largely pertain to the health status of the stimulated nerve roots. The threshold criteria for all of the stimulation techniques that are used for testing purposes are based on the assumption that the nerve roots that are being excited by the pedicle screw stimulation are healthy and function normally. These nerve roots, when directly stimulated, have excitation thresholds of about 2 mA. However, it has been reported that chronically compressed nerve roots have elevated stimulation thresholds. For some of these nerve roots, thresholds may even exceed 20 mA. Therefore, the use of the “warning threshold” for normal nerve roots when testing chronically compressed nerve roots may also contribute to false-negative findings.
In two cases of chronic nerve root compression in which nerve root thresholds were determined as a result of direct nerve root stimulation and were found to be elevated. In the first case, the patient was diagnosed with spinal stenosis, and a right L5 nerve root that had been chronically compressed had a stimulation threshold of 5.5 mA. In the second case, the patient presented with a left-sided drop foot of about 1 month duration that occurred immediately after a previous back surgery. The placement of the left L5 screw was visually examined during surgery and was found to be in contact with the left L5 nerve root. Direct stimulation of the left L5 nerve root just outside the foramen resulted in a stimulation threshold of 13.7 mA.
Stimulation thresholds may also be elevated when testing patients with metabolic disorders such as diabetes. Those tested have either exhibited normal thresholds or thresholds that have been only slightly elevated (thresholds of 4–6 mA).

IMPEDANCE TESTING

An alternative approach to pedicle screw stimulation is to use the electrical
impedance of biological tissues as a means of assessing pedicle screw placements. Using a porcine model, Myers et al. reported on a method for assessing pedicle wall thickness using impedance techniques. They were able to determine that the impedance of intact vertebral bone was about 400 Ω (400 ± 156 Ω) when a probe was first inserted into a pedicle and decreased as the depth of insertion increased. For an intact pedicle, the vertebral impedance decreased to 100 ± 22 Ω at maximum probe penetration. The accuracy of the technique was determined using postmortem anatomical confirmation of the pedicle probe placement and regression analysis of the impedance data. Based on this model, it was determined that impedance values below 58 Ω were associated with a 100% likelihood of a breach in the pedicle wall. These data were gathered from probing pedicle holes and measuring the impedance of the walls.
Although very promising, the authors recognized that the technique’s utility
needed to be demonstrated for implanted pedicle screws. Thus far, this utility has not been demonstrated. When testing implanted pedicle screws, the factors that contribute to the measured impedance become much more complex than simply measuring the tissue impedance at various points on the pedicle wall. Other investigators have compared the impedance measurements taken from pedicle screws to the results obtained via electrical stimulation. The impedance readings were very variable, with no correlation to electrical stimulation data or findings from visual observation. At the present time, impedance measurements do not appear useful for assessing pedicle screw placements. Further refinements in the technique may be necessary to make this technique useful.

Richard J. Toleikis Experience.

For many years, mixed-nerve SEPs have traditionally been used to monitor spinal cord function during spinal instrumentation procedures in order to minimize the probability of postoperative neurological deficits. In the past 10–15 years, intrapedicular fixation of the thoracolumbar and lumbosacral spine by means of pedicle screw instrumentation has become increasingly popular. With increased use of pedicle screw instrumentation came varying degrees of neurological impairment. However, when the implementation of pedicle screws was increasing, the only forms of neurophysiological intraoperative monitoring that were available to avoid postoperative neurological deficits were mixed-nerve SEPs. One of the patients that was monitored in this fashion using SEPs elicited by posterior tibial nerve stimulation had complaints of paresthesia and numbness of the right great toe immediately after surgery. The scope and intensity of these symptoms increased, and the patient was clinically found to also have increasing weakness of the dorsiflexors and extensor hallucis longus muscle on the right side. A computed tomography (CT) scan was subsequently obtained that provided evidence of pedicular screw irritation of the nerve root. Subsequent surgery revealed that the right L5 pedicular screw was located medial to the pedicle and juxtaposed to the right L5 nerve root. Review of the intraoperative monitoring tracings revealed no significant changes in the monitored responses throughout the surgical procedure. In addition, tracings acquired during postoperative testing were comparable to those obtained intraoperatively. SEPs are typically mediated by several spinal nerve roots. It was surmised that the reason the monitored SEPs did not demonstrate any changes was that the functional compromise of the single L5 nerve root was masked by the normal volleys mediated by unaffected nerve roots. On the basis of this one case, it was determined that mixed-nerve SEPs might be an inappropriate tool for monitoring procedures in which nerve root rather than spinal cord function is at risk. The use of SEPs for monitoring purposes was abandoned, and other methodologies were sought for monitoring during pedicle screw procedures. As a result, DSEPs, which previous studies had already indicated were an effective means of assessing single nerve root function, were used to monitor pedicle screw procedures.
Subsequently, the results of our experience using DSEPs as a monitoring tool during surgical intrapedicular fixation procedures were published. They indicated that the loss of DSEP responses appeared to be a sensitive indicator of mechanical root compression, whereas DSEP responses were rarely found to change to any significant degree during root decompression
or even several days post surgery. However, because of the major shortcomings associated with DSEPs (i.e., the time-consuming need for averaging to acquire responses and the insensitivity to potential sources of nerve root irritation or damage), their use was later abandoned in favor of monitoring spontaneous myogenic activity in conjunction with indirect nerve root stimulation responses. Monitoring using a combination of both these techniques appears to have adequately addressed all of the DSEP shortcomings.
Although the monitoring of spontaneous myogenic activity is used to safeguard nerve roots during pedicle screw placement, the actual probability of nerve root irritation or injury during these placements, although finite, appears to be very small. Having kept data from over 1000 surgical procedures that have involved the placement of over 5000 pedicle screws, the author have never observed any sustained (longer than 2 s) spontaneous activity that was associated with the tapping of screw holes or the placement of markers or screws. Sustained spontaneous activity has been observed in less than 4% of the patients that have been monitored, and this activity has always been associated with mechanical nerve root irritation during traction or decompression.
Based on the data that were acquired from 662 patients using a “warning threshold” of 10 mA (using a stimulus duration of 0.2 ms), correlation between responses elicited at or below warning threshold, the surgical findings, and the actions surgeons took based on visual inspection of the screw placements. When EMG responses were elicited at or below 5 mA, screws were almost always removed and might be redirected. If responses were elicited at intensities at or greater than 8 mA, screws were generally left in place. Between 5 and 8 mA, screws were equally likely to be removed or left in place. Therefore, despite the fact that stimulation thresholds less than 5 mA generally resulted in screw removal, screw removal also occurred at stimulation intensities up to and including the “warning threshold” of 10 mA. These findings support the findings of others, which indicate a close correlation between the intensity of screw stimulation needed to elicit myogenic responses and the risk for neurological injury associated with the screw placements.
When electrical stimulation is used to assess hardware placement or the integrity of a pedicle hole, the stimulation current can take many pathways as it returns to the anodal needle electrode, but it will follow those pathways that provide the least resistance. When hardware or pedicle hole stimulation is associated with low-threshold stimulation intensities, the results suggest that a path of least resistance is located near a nerve root, but one cannot tell whether the pathway is through a cracked pedicle, a thin wall of osteoporotic bone, or an exposed pedicle screw. Responses elicited at stimulation intensities below the “warning threshold” could result from current flow through any of these pathways. However, there does appear to be a relationship between threshold stimulation intensities and the exposure of a pedicle screw; i.e., cracked pedicles or a minimally exposed screw tends to be associated with stimulation thresholds greater than 7 mA, whereas exposed screws near a nerve root tend to have thresholds less than 5 mA. However, not all low-threshold readings are associated with screw placements that represent threats of potential neurological injuries. Testing of pedicle screws in two patients resulted in each having a screw with a stimulation threshold less than 5 mA. After visual inspection, neither was removed because their placement did not appear to represent a threat of potential neurological injury. Neither patient experienced any postoperative pedicle screw–related neurological deficits.
Based on the author’s experience and the results of other investigators, screw placements that are associated with stimulation thresholds greater than 10 mA are unlikely to represent a risk to neurological function if normal healthy nerve roots are involved and testing conditions are adequate. However, several factors, both technical and physiologic, can contribute to false-negative findings when stimulation thresholds exceed “warning thresholds.” These include excessive muscle relaxation, current shunting as a result of excessive fluid in the surgical site, and chronic nerve root compression.
Since 1995, author service has monitored well over 1000 patients during surgery in which over 5000 pedicle screws were placed. For each patient, a
physical therapist and the attending physician routinely performed postoperative assessments. These practitioners were asked to inform the monitoring staff of any imaging or surgical evidence of misplaced hardware or any functional deficits that could be attributed to hardware placement. Based on their information, only one patient has experienced a postoperative neurological deficit directly attributable to a misplaced pedicle screw. In that patient, pedicle screws were placed from L3 to L5. The patient had stimulation thresholds that exceeded the “warning threshold” of 10 mA in all cases except for the left and right L3 screws, which had thresholds of 6.7 and 5.1 mA, respectively. The surgeon elected to leave these screws in place. Immediately after the operation, the patient experienced symptoms of low back pain and right leg pain. The patient was brought back to surgery, and both L3 pedicle screws were removed and replaced with sublaminar hooks at L3 and pedicle screws at S1 using Texas Scottish Rite Hospital (TSRH) instrumentation and a crosslink. Following this procedure, the patient’s leg pain completely resolved. This is considered a good example of a true-positive finding. The author have also had what he considered to be one false-negative finding, although the patient did not experience any postoperative neurological symptoms as a result of the screw placement. After operation, the patient reported unilateral back and leg pain. Postoperative CT scans were
not routinely performed for patients, but in this case one was ordered and revealed a screw that was positioned medial to the pedicle in the spinal canal on the asymptomatic side. Because of the location of the screw, it was removed less than 1 week after it had been placed and before it caused any postoperative nerve root irritation. Unfortunately, in this one case, the screw was removed without any repeated testing to confirm earlier monitoring findings. A retrospective review of these findings indicated that the stimulation thresholds for the four placed screws were 40–50 mA. Routine train-of-four testing of the leg musculature performed by the monitoring staff just before and after screw placement indicated that the level of paralysis was adequate for accurate assessments because four full twitches could be elicited from the tibialis anterior muscle. Although this patient did not experience any new postoperative deficits as a result of screw placement, the electrical stimulation technique should have detected the misplaced screw. Thus this result is considered a false-negative finding. None of the factors that have been discussed earlier can explain these results. Therefore, it is possible that other factors that are not obvious to the author may also contribute to false-negative findings.

Conclusions

The incidence of neurological complications associated with the placement of pedicle screws has been reported as 2–10%. Based on these estimates of incidence, we would then expect that for every 1000 patients in which pedicle screws were placed, between 20 and 100 should have exhibited some new postoperative neurological deficits directly attributable to screw placement.
By R.J. Toleikis outcome data, which indicate that only 1 patient in his population of over 1000 patients has thus far exhibited such symptoms (actually, the result of a screw with a low test threshold that was left in place), clearly suggest that the use of pedicle screw stimulation to monitor screw placements has played an important role in minimizing the incidence of such deficits. The technique appears to be very reliable for detecting breaches of the pedicle wall, even those that may pose no threat of causing neurological irritation or injury. It provides an easy, quick, and accurate means to assess pedicle screw placements and to safeguard neurological function.
In the real world, it is also essential that monitoring be cost-effective. That is, the overall costs of monitoring should not exceed the costs associated with patient care if monitoring is not provided. In most institutions at this time, the cost of monitoring for a typical instrumented fusion involving pedicle screw placement with spontaneous and triggered myogenic techniques is generally $1000 or less. Therefore, the cost associated with monitoring 1000 procedures would be $1 million. However, as indicated earlier, the minimal expected incidence of postoperative neurological deficits resulting from the placement of pedicle screws is 2%, and it would involve at least 20 patients. Therefore, if the average medical costs to correct a patient’s postoperative outcome and to rehabilitate that patient are more than $50,000, then the monitoring is cost-effective.
It is very unlikely that $50,000 would cover all of the resultant medical costs. This discussion of cost-effectiveness does not even take into account the medical and legal costs associated with each of these occurrences. Clearly, monitoring during pedicle screw placement is cost-effective.
All the techniques that can be used to monitor during pedicle screw placements have some limitations. The combined use of spontaneous and triggered myogenic activity is the only technique that meets all the necessary criteria if it is to be widely and effectively used to assess pedicle screw placements and to preserve nerve root function.