Clinical manifestations

Presentation and course

Postoperative paraparesis, or other serious neurologic deficits, had been a feared complication stemming from spinal surgery, especially consequent to corrective intervention for scoliosis (142). The advent of intraoperative monitoring has reduced the risk of serious neurologic deficits (114). Somatosensory-evoked potential monitoring is now a standard of care for monitoring the dorsal column sensory pathways. This approach can detect, early on, potential impending damage to neural structures; thus, it is possible that injury may be averted. Ischemia and mechanical injury are the most likely mechanisms. Damage may arise from direct blunt trauma, excessive compression, distraction, stretching, or vascular insufficiency via embolus or thrombus formation. It should be noted that normal somatosensory potentials or motor-evoked potentials at the end of surgery do not guarantee the absence of delayed paraplegia. Thus, postoperative monitoring, especially after vascular procedures, is occasionally indicated.

Sensory-evoked potentials involve the electrical stimulation of peripheral nerves creating action potentials that propagate cephalad from the periphery, subsequently over time, to the then central nervous system. The amplitude of the sensory-evoked potentials is small when compared to relative EEG activity (93). Computer averaging allows the differentiation of the somatosensory-evoked potential signals from the background electroencephalographic activity (which is random in nature, rather than event-related; ie, “noise” rather than true signal).

Monitoring provides services beyond simply the warning of the possibility of ensuing complications. It offers advance insight toward prompt intervention (94). A surgeon can feel reassured about the integrity of the spinal cord and can, therefore, extend the procedure to a greater degree. Patients and families can be relieved knowing that certain feared complications are screened for during surgery. Further, some patients may receive technically challenging procedures that would have been avoided in the absence of such feedback about the status of the nervous system. As Muthukumar states: “Considering the enormous costs of health care and the human suffering related to the development of postoperative paraplegia/quadriplegia, there is enough evidence to prove that the cost of performing IONM does not exceed that of providing health care to the injured patients” (89). Ibrahim and colleagues at Neurosurgery, Penn State, take a different position in expressing: “The aim of neuromonitoring during an operation is to provide the surgeon with a real-time analysis of spinal cord function at a time when there is still a possibility to correct any possibility of morbidity. Changes in intraoperative neuromonitoring measurements can be due to changes in arterial pressure, cardiopulmonary function, and spinal cord function. Potentials can also be influenced by anesthetic regimen, perfusion pressure, hypothermia and hyperthermia. Intraoperative neuromonitoring has been utilized in many contexts, including spine surgery, arterio-venous malformations, thyroid and parathyroid surgery, pediatric deformity correction surgery, epilepsy surgery, subarachnoid hemorrhage repair and others. Although it has been used in the numerous contexts shown above, an obvious benefit of IONM providing optimal functional outcomes in patients has NOT been demonstrated. Both low sensitivity and low specificity can have detrimental effects on the surgery and adversely affect patient outcomes” (52). The debate burns on with reference to thyroid/parathyroid surgery as well. Lombardi and associates state: “IONM should NOT be considered the standard care in preventing recurrent laryngeal nerve palsy” (72) whereas Sun defends its use: “IONM has become an effective adjunct for the golden standard of naked-eye protection. It is simple, effective and practicable” (129). Jahangiri and colleagues promote multimodality intraoperative neurophysiological monitoring during shoulder surgeries in that it grants the identification – in real time – of signal changes consistent with likely impending nerve injury (54).

Clinical vignette

G.W. was a neurodevelopmentally challenged teenager who suffered from global developmental delay, a mixed seizure disorder, and progressive thoracolumbar scoliosis with associated issues of pain and positioning. Her treating orthopedic physician elected to surgically intervene with planned anterior and posterior correction of the otherwise progressive, clinically significant spinal curvature or kyphoscoliosis. The extensive distraction procedure, with instrumentation, was undertaken with assistance from the neurophysiology monitoring team. Intraoperative monitoring included both upper extremity and lower extremity somatosensory-evoked potentials. Transcranial motor-evoked potentials were also performed as well as two channels of EEG monitoring. The patient had been medicated with valproate acid for her outstanding epilepsy. Her antiepileptic drug levels were therapeutic, and she had no history of bruisability or thrombocytopenia. Intraoperatively, the surgical team was notified of a sudden loss in the TcMEPs followed by a significant increase in latency of the cortical waves of the lower extremity SSEP’s as well as a greater than 50% to 60% reduction in amplitude. This was noted at a time when the deformity was being corrected and there was some significant bleeding. Coagulation studies were sent to the lab and found to be unremarkable. The surgical team decided to be less aggressive with the corrective procedure and reduced the amount by which the deformity was corrected because of the change in the evoked potentials. Postoperatively, G.W. had a spinal straightening of 15 degrees (preoperative curve was 60 degrees). She was able to be more aptly positioned in her wheelchair but, unfortunately, had developed new-onset bowel and bladder incontinence, presumably from an intraoperative myelopathic insult. Some practitioners speculated that the disodium valproate, known to potentially induce a thrombocytopathy without demonstrable thrombocytopenia, played a role--combined with the mechanical stress of surgical distraction--in the initiation of an ischemic event, as detected by the sensitive intraoperative monitoring of the dorsal column pathways.

Biological basis

Etiology and pathogenesis

Intraoperative neuromonitoring (IONM) aims to reduce the possibility of spinal cord injury during procedures indicated toward overcoming symptoms related to spinal cord deformity. The deformity may be congenital, acquired, traumatic, or neoplastic; the procedure could include decompression, correction, instrumentation, or fusion, all of which are hazardous but less so in the clinical or operative setting of intraoperative neuromonitoring. Intraoperative neuromonitoring can consist of SSEP, TcMEP, and/or EMG--a multimodality combination is preferred (28). To broaden, intraoperative neuromonitoring is also utilized in thyroid, carotid, and aortic surgical cases, as well as central operations.

Somatosensory-evoked potentials (SSEPs). Somatosensory-evoked potentials are obtained by repetitive electrical stimulations of peripheral nerves. The electrical signals generated are detected at the level of the peripheral nerves (over the brachial plexus or popliteal fossa), spinal cord (lumbar or high cervical), and cortex (scalp). The signals generated by hundreds of stimulations are averaged to make the time-locked components more evident and to produce cancellation of the random electrical activity that may otherwise obscure the low amplitude signal of the potentials. This method relies chiefly on the stimulation of the large myelinated somatosensory fibers, which transmit the impulses orthodromically to the spinal cord via the posterior column medial lemniscus system, although some have ascribed a role of the dorsal spinocerebellar pathways as well (112).

Methodology of intraoperative somatosensory-evoked potentials.

Intraoperative somatosensory-evoked potential monitoring should follow the American Clinical Neurophysiology Society guidelines for intraoperative monitoring (02) and those of the American Society of Neurophysiologic Monitoring (136). The American Scoliosis Society’s position statement of 2009 in favor of intraoperative neuromonitoring should be heeded as well. The electrode sites are demarcated using the International 10-20 system of electrode placement. The preferred recording sites are C3, C4, CZ, FPZ, FZ, A1, and A2. In most instances, utilization of gold cup electrodes in gauze soaked with collodion is favored although subdermal needle electrodes may be used. Additional gold cup electrodes are placed over the cervical spine at the level of C2 and C3 and over the shoulder to serve as grounds. The use of properly sized electrodes appropriately proportioned to the child's head size is essential toward obtaining optimally recorded data. Electrode impedances should be maintained between 2000 and 5000 ohms.

The recommended stimulus is a monophasic pulse of 10 to 25 mA and 100 µs of duration although this may be increased if clinically indicated (44). The stimulus pulse is applied to the median nerve at the volar wrist or at the posterior medial malleolar region of the respective ankle. Spinal cord intraoperative somatosensory-evoked potential monitoring often involves the posterior tibial nerve; however, when surgery is caudad to the eighth cervical spinal cord level, median nerve intraoperative somatosensory-evoked potential monitoring is used above that level (17; 69). Rates of stimulation are typically near 5 Hz for median nerve and 2 Hz for the posterior tibial nerve; however, exact divisors of 60 Hz are not used so that 60 Hz noise in the signal will cancel out after averaging (44; 125). Each limb should have separate evoked potential testing performed. Generally, 350 to 500 repetitive stimulations are averaged; however, up to 2000 repetitions may be necessary in order to elicit optimal waveform morphology and, thus, reliable and reproducible data (17). An analysis time of 100 msec is sufficient for the lower extremity SSEPs and for the upper extremity SSEPs 50 msec is appropriate (17; 44). The filter settings are often 30 Hz (low pass) and 3000 Hz (high pass), but many groups use high frequency filters as low as 250 Hz in order to reduce noise levels. Most institutions employ the convention of negative potentials producing upgoing deflections, but the opposite convention is used in certain parts of the United States and Europe. In regard to the source generators of key signals, “N” denotes the negative deflection and “P” denotes, conversely, the positive deflection.

When using the posterior tibial nerve as the stimulation site, a channel derived from a popliteal fossa recording electrode is referenced to one that is placed 4 cm above this topical recording landmark. That particular pair of electrodes is used to verify the integrity of the signal generated distally and to rule out problems that could interfere with the nerve conduction, such as clinically significant changes in temperature. Other important recording montages include the upper cervical cord (CS lead) referenced to CZ and the ipsilateral ear as well as a scalp lead at CZ referenced to FZ. The upper cervical cord recording may require an inserted needle electrode due to the low amplitude of the signal stemming from posterior tibial nerve stimulation.

Advocated recommendations include the use of at least three channels for median nerve intraoperative somatosensory-evoked potential monitoring (44). The first channel records over the ipsilateral Erb's point at the level of the brachial plexus. The more proximal channels include mid-cervical cord referenced to CZ and, subsequently, in a cephalad array, contralateral central cortex to contralateral Erb's point or to FZ.

Median nerve somatosensory-evoked potential waveforms and generators monitored during surgery. Erb’s point potential (P9), is the name given to the waveform generated by the passage of the stimulus volley through the distal brachial plexus. The montage used to record this potential is ipsilateral to contralateral Erb’s point. The Erb’s point potential may manifest itself as a double negativity, especially in children.

Cervical potential (N13). This potential is maximal at the low- to mid-cervical region. It is stationary because its latency does not change when one moves the recording electrode proximally or distally. It shows negativity posteriorly and shows positivity anteriorly (ie, an anterior-to-posterior phase reversal in the cervical area). The N13 represents the postsynaptic activity in the dorsal gray matter of the cervical cord or dorsal column as the sensory volleys reach the cervicomedullary junction. N13 is recognized as the most prominent of several negative potentials, but it may be confused with the N11, which is generated in the root entry zone. Unlike N13, the N11 does not show anterior-to-posterior phase reversal in the cervical area.

N20 is a contralateral parietal near field potential. It is the initial corticothalamic response to the sensory volley. The use of the contralateral central cortex to reference and the contralateral central cortex allows the separation of N18 (a subcortical brainstem potential) from N20. It is generated in the post-central gyrus with an activation of the 3b area (posterior bank of the central sulcus) with subsequent spread of the depolarization to area 1 and area 2 (post-central sulcus), area 3a (bottom of the central sulcus), and area 4 (anterior bank of the central sulcus) (75). There is also thought to be a thalamic component to the N20. Around 1.5 to 2 ms after N20, a positive deflection is seen over the centroparietal region (P22) that is also thought to have a cortical source generator, but the exact location remains controversial.

Posterior tibial nerve somatosensory-evoked potential waveforms and generators monitored during surgery. This includes the popliteal fossa potential that has a triphasic morphology. This peak is also known as the knee potential, and its negative peak is alluded to as N7 or N8 (81). This waveform is generated locally as the stimulus traverses the nerve at the region of the popliteal fossa.

N22 is a negative potential maximally recorded between T10 and L1, using the contralateral iliac crest as reference. It reflects the postsynaptic activity at the level of anatomically normal lumbar enlargement in the spinal cord. This potential is important when monitoring surgical procedures near the pelvic area.

P31 is a component of the posterior tibial somatosensory-evoked potentials that reflects the activity on the medial lemniscal system. It is characterized by a small positive deflection, and it is a far field potential.

P37 (also known as P38, P39, or P40) is a potential recorded over CZ-PZ or the ipsilateral central region due to the positivity generated as a dipole in layer 4 of the contralateral leg, cytoarchitecturally, area of the primary sensory cortex. P37 is a near-field potential restricted to the central and parasagittal region. FZ is inactive after 34 ms; therefore, it can be utilized selectively as a reference for this particular component.

After P37, two other main waveforms are recordable in a caudo-cephalad direction: (1) N45 and (2) P60. P60 is also referred to as P2 (148); it is maximal at the vertex and may persist when P37 is absent (81). The P37, N45, and P60 combination produces the typical W-shaped morphology of the evoke potential.

Cervical potentials. One of the main differences between the potentials generated by the median nerve and by the homologous posterior tibial nerve is the presence of decided cervical potentials. Median nerve somatosensory-evoked potentials generate clearly identifiable cervical potentials (N13). On the other hand, by the time the stimulus generated at the posterior tibial nerve reaches the cervical spinal cord, the temporal dispersion will cause a significant amplitude reduction of the cervical signal. The use of needle electrodes implanted at the interspinal ligament can aid in the accurate detection of these otherwise small waveforms. When needle electrodes are used, the cervical potentials are seen as a major negativity followed by a positive wave (74; 75; 17). Throughout the surgery (from wound opening to wound closure), the waveforms are recorded at approximately 3- to 5-minute intervals, but more frequent (and nearly continuous) recordings should be routinely performed during the more intensified surgical periods, such as the critical times of either instrumentation or of distraction, both being mechanically intrusive and, thus, potentially necrogenic, given the heightened risk of ischemic insult.

Criteria for abnormality (aka, what constitutes warning criteria or an ALERT). It is important to obtain baseline data, including pre-incision tracings and early (surgical site preparation) intraoperative data. All subsequent electrographic findings, as collected, are then compared to the baseline. In essence, the subject serves as their own ‘internal’ control. That is, the usual subject variables, such as age, limb length, and body height, play a minor role in the operative setting and are eclipsed by the patient’s own individual baseline values and deviation from that index criterion. Stimulus intensity as well as rate and duration of the square wave need to be regarded (101). A multifactorial analysis of the effects of physiologic parameters on intraoperative somatosensory-evoked potentials yielded a correlation between variations in latency and amplitude in the order of influence as temperature, then paCO2, then heart rate, followed by diastolic blood pressure, and then systolic blood pressure (23). The patient's underlying condition may prevent the recording of intraoperative somatosensory-evoked potentials in about 5% of the cases (46). The most frequent conditions leading to the problem of poor data acquisition are neural tube defects and severe spastic quadriparesis with atrophy of the lower extremities. Other disorders associated with unrecordable intraoperative somatosensory-evoked potentials include sequelae of spinal cord trauma, advanced spondylosis with myelopathy, scoliosis with myelopathy, peripheral neuropathy, and spinal cord tumor (46). In one study, spinal cord monitoring using somatosensory-evoked potentials was reliably achieved in 31 out of 34 patients with cerebral palsy undergoing corrective surgery for progressive scoliosis (30). Proposed pathogenetic mechanisms of signal change during monitoring include temporal dispersion of the afferent volley and conduction block in damaged axons stemming from ischemia and anoxemia with resultant abnormalities in the electrophysiological parameters of amplitude of the response potential or prolongation of onset time of the evoked signal. Further, hypoperfusion can injure the white matter of the central nervous system with loss of waveforms arising from calcium influx into the intracellular space. Hypothermia after exposure of the spine but before instrumentation may render an increase in false negative outcomes (117).

Intraoperative neuromonitoring, multimodality, provides data that prompts reevaluation in approximately 10% of patients with pediatric spinal deformity (106).

Some variability in latencies and amplitudes between different stages of scoliosis surgery is normal. Principally, the SSEP (when the spine is exposed) may be used as the reference baseline to determine whether future potentials are subnormal at the subsequent stages of surgery. Of note, the amplitude of the SSEP decreased in most patients when the spine was exposed, although there was no injury per se to the spinal cord. This must be heeded in order to avoid declaring undue false positives.

The criteria commonly used in significant wave form change (in relation to the baseline tracing) are either an increase in latency equal to or greater than 10% of the preoperative baseline (95; 44), or a decrease in amplitude of more than 50% (95; 92; 148; 44). Of late, there is an advocacy movement toward revising the threshold for alarm to, more adequately, a 60% amplitude attenuation compared to baseline (after skin incision or spine exposure) rather than 50%, based on data (49). The literature suggests that amplitude may be the more sensitive indicator of neurologic deficit, but the criteria for change vary with the type of procedure and the type of expected injury, so careful consideration must be made regarding what criteria will be used in each case.

Nuwer's experience was that patients with persistent amplitude reduction of greater than 50% maintain a 25% chance toward a new neurologic deficit in the postoperative period (92). The same author mentions that amplitude reductions of less than 50% are not as concerning but do warrant careful neurologic follow up; a similar conclusion was reached by another study including 81 patients (148).

Regarding indications of intraoperative neurophysiological monitoring, the neurosurgical team at the University of Wisconsin examined the diagnostic and therapeutic utility of intraoperative neurophysiological monitoring in the surgical treatment of cervical degenerative disease (109). They deemed evoked potential monitoring as a sensitive tool during anterior spinal surgery for cervical spondylotic myelopathy. Further, they concluded that intraoperative signal worsening does not tightly correlate with clinical worsening and that its recognition does not necessarily prevent neurologic insult. That is, intraoperative neurophysiological monitoring does not seem to forecast outcome with reliability. This must be considered in the appropriate context. Anterior cervical discectomy and fusion is associated with a low risk of neurologic injury, so it will be difficult for any tool to improve outcomes (133; 123). When there is a higher risk of neurologic injury, any tool becomes much more useful.

The value or power of intraoperative neurophysiological monitoring has been evaluated in the setting of surgical remediation of tethered cord syndrome in which the risk of injury to nerves embedded in the tether is significant. Of note, neurologists at the Aga Khan University investigatively utilized – beyond the routine tibial SSEPs – clitoral and dorsal penile SSEPs during monitoring (59). They deduced that few data exist to support the merits of intraoperative neurophysiological monitoring in tethered cord syndrome. This has been challenged by data from the University of Virginia (107). That center contends that electrophysiological monitoring provides, for untethering, an efficient, effective, and reliable method for intraoperative guidance with the goal of reducing iatrogenic injury. They also purport that monitoring, in the form of both a threshold-based interpretation system and continuous EMG, can localize the “autonomous placode” in secondary tethered spinal cord syndrome. For example, clinically, if a newborn has an operation in the perinatal period for myelomeningocele, there is the possibility of subsequent onset secondary tethered spinal cord syndrome. The symptoms, of remote or latent onset, can typically include progressive bowel or bladder dysfunction with associated lumbago and lower extremity paresthesia and spasticity. The monitoring can detect the tethering placode, a combination of scar and neural structures. The surgeon can, more aggressively, section out the placode and, thus, grant the patient more postoperative relief. In Pouratian and colleagues’ series, the patients benefitted greatly in terms of both intraoperative utility and postoperative outcome (107).

Again, as a guidepost for the valid interpretation of EP data, when alluding to cervical and cortical signals, suspect potential impending pathologic alteration if: (1) there is an increase in latency equal to greater than 10% of the baseline value or (2) there is evidence of a decrease in amplitude greater than 50%. These are certainly, at the least, warning criteria (91; 94).

Transient and relatively less significant changes in the amplitude (30% to 50% of baseline) or latency (less than 2 ms) of intraoperative somatosensory-evoked potentials may be seen during surgery due to hypotension/rendering of ischemia, hemodilution, hypothermia/anemia, hypo- or hyper-thermia, hemodilution, hypothermia, or irrigation with cold fluid or due to certain anesthetics; return to the baseline values is seen when these alterations are corrected (46; 44). These alterations often occur gradually over a period of 30 to 60 minutes. On the other hand, clinically significant changes, such as intraoperative somatosensory-evoked potential amplitude reductions of greater than 50% of baseline or increases in latency of 10% (2 ms or greater), tend to be acute and not associated with any change in temperature, blood pressure, or the amount or nature of anesthetic administered (44). The Scottish National Spine Deformity Centre espouses an algorithm of action in response to intraoperative monitoring events; in the patient consent forms, they state the possible need to abandon surgery depending on the intraoperative monitoring findings and complete it in a staged manner at a later date (137). One should realize that increasing concentrations of halothane can quickly produce a decrease in the amplitude of the cortical potentials, which is directly proportional to the end tidal concentration of that gas (148). Patients younger than 10 years of age are particularly susceptible to the effects of high concentration--or boluses--of general anesthetics, producing attenuation of the cortical potentials (46). Cortical potentials in children are also more likely to be attenuated by a combination of anesthetics, such as isoflurane and nitrous oxide, especially in high doses (46). Improved and more assured scalp recordings are obtained through the avoidance of combination anesthetics or by keeping the concentration of nitrous oxide at less than 50% and isoflurane at less than 0.6% when these agents are used together (46). When one uses the latter range of anesthetic doses and adds narcotics (ie, fentanyl) during the induction, intraoperative somatosensory-evoked potentials remain unaltered (46; 44). Other anesthetics that have been shown not to affect the intraoperative somatosensory-evoked potential recordings are propofol, etomidate (44), and ketamine (04). A bolus of lidocaine may produce a transient, but significant change in the cortical response (19). The greater instability of the cortical responses in children is thought to be due to the lack of symmetry and synchrony in the myelination process (36). Due to the unreliability of the cortical response in children, recording of the cervical potential has been recommended (46). Cervical potentials are more resistant to the effects of general anesthetics and can be used to monitor the integrity of the spinal cord above the surgical level when the cortical potentials are absent (57; 39). If the posterior cervical recording electrode is within the actual operative field, an anterior neck recording electrode can be placed, which is usually referenced to CZ or FZ (45).

When interpreting intraoperative somatosensory-evoked potential abnormalities, in real-time, one has to be attentive to certain factors that may generate recording artifacts. These include electrical stimulator failure, electrode problems (high impedance, detachment from the skin), and electrical or magnetic interference from other equipment in the typically electrically “hostile” operating room setting. Additionally, certain physiological factors, such as obesity, diabetes with comorbid neuropathy, peripheral vascular disease, seizure disorders, and closed head injuries have been associated with less than optimal responses (154).

When cervical or cortical potential abnormalities (with an amplitude drop greater than 50% or latency increase greater than 2 ms from baseline) are seen during surgery, the patient and the recording apparatus should be checked for sources of artifact; the anesthesiologist should be questioned about the subject's vital signs (especially blood pressure and body temperature) and relevant changes in the pharmaco-anesthetic regimen. Remediable causes, such as hypotension should be dealt with appropriately. Intraoperative somatosensory-evoked potentials should be monitored frequently to realize, optimistically, a return to baseline. The entire review process should be accomplished within 5 to 10 minutes, and, if no other probable reason is present to explain the intraoperative SSEP alterations, the surgical team should be notified promptly. It is also sound practice to have someone from the surgical team tell the neurophysiology monitoring team of the impending onset of high-risk manipulations, such as spinal distraction, instrumentation, and sublaminar wire placement (144; 46). When cortical or cervical intraoperative somatosensory-evoked potential abnormalities return to baseline within 15 minutes of the acute changes, postoperative neurologic sequelae are unlikely to be realized. The opposite is true if the waveform abnormalities persist for periods of time greater than 15 minutes (44).

In circles of physicians performing cervical spine surgery, controversy and debate surround the issue of whether intraoperative somatosensory-evoked potentials are clinically useful in uncomplicated, non-upper cervical spine procedures. One group, launching a retrospective review of a large number of cases of anterior cervical discectomy with fusion (ACDF), concluded that the surgical procedure itself was safe and that intraoperative somatosensory-evoked potentials had no utility and was, thus, withdrawn from use (133). Certain groups are exploring S100B, a serum marker for glial injury, as an adjunct with evoked potentials, to predict long-term neurologic alteration postoperatively (131).

Some surgeons favor somatosensory-evoked potential monitoring during lumbar root decompression. Yue and Martinez found that during L4-5 root decompression, superficial peroneal nerve SSEP (SPN-SSEP) is more reliable than post tibial nerve SSEPs (PTN-SSEP) (150).

A survey was orchestrated from the neurosurgery department at the University of Saskatchewan (104). Canadian neurosurgeons and orthopedic surgeons were asked revealing questions, through the Canadian Spine Society. Respondents stated that monitoring was performed to reduce the risk of an adverse operative event rather than because of liability concerns, and they collectively favored monitoring in cases of reduction of major deformity (scoliosis), symptomatic and asymptomatic spinal cord compression, spinal cord tumors, and instrumentation. Availability was an issue, as was the lack of neurophysiology specialists within neurology.

Motor-evoked potentials (MEPs). Over time, direct electrical stimulation and, to a lesser extent, magnetic transcranial stimulation of the cortex or spinal cord have become a more standard method of evaluation of the integrity of the motor pathways during spinal surgery, and, thus, we have the term “transcranial motor-evoked potential” or “TcMEP” (68; 03). Customarily, two main reasons have ushered in the need for intraoperative motor monitoring. The first stems from the fact that postoperative motor deficits are often profoundly clinically disabling. The second reason centers on the selective vulnerability of the anterior spinal cord to hypoperfusion when compared, in contradistinction, to the posterior columns (87). The ventral spinal cord seems to have fewer anastomotic vascular communications and, further, contains gray matter, predisposing it to hypotensive damage during surgery (87). From a practical standpoint, the MEP changes tend to precede the SSEP changes, making them an earlier warning sign for the neurophysiology team to deal with; this allows greater time for the neuro-orthopedic surgeon and anesthesiologist to rectify aspects of the operation and, thus, mitigate irreversible injury (87).

Transcranial electrical stimulation of the motor cortex via motor-evoked potentials (TcMEPs) is a desirable method, but its interpretation has several caveats. This method requires careful assessment of the degree of neuromuscular blockade during surgery and of the type of anesthetics used (03). In general, inhalational anesthetics and neuromuscular blockade have been shown to limit the ability of the TcMEP monitoring to detect significant changes. Hypothermia, as well, can negatively affect intraoperative neurophysiological monitoring. Opioids have little influence on TcMEPs. Further, a stable concentration of inhalational or intravenous anesthetics optimizes TcMEP monitoring (140). TcMEPs have been successfully used to monitor intramedullary spinal surgery in children (61). The combination of electrical TcMEPs and SSEPs in spinal surgery appears to be safe and accurate to predict neurologic deficits in children (127).

A relatively high voltage is utilized to generate the transcranial motor response, often a few hundred volts (88). This level of stimulation can be painful in an awake patient. The charge density utilized for TcMEPs is 10% to 15% of that required to induce seizures in humans (78). Reportedly, the incidence of seizures during MEP monitoring for cranial procedures was overall 1.8%. The incidence rose to 5.4% when direct cortical stimulation was additionally applied (138).

Voltage-based stimulation is not commonly undertaken in other settings due to safety issues. Current intensity-based parameters are used for direct cortical stimulation (see “Functional brain mapping” section below). The stimulating electrodes are located at C3-C4 (of the International 10-20 system of electrode placement) for elicitation of upper extremity responses and C1-C2 when looking for activity over innervated lower extremity musculature. The anode (positive electrode) is the one stimulating as cathodal (negative electrode) stimulation requires relatively higher currents, so the anode should be placed at C1 or C3 for left lower or upper extremity responses (88). When lower extremity response is not seen with C1-C2 placement, an alternative electrode positioning using CZ as the anode and with the cathode located on the midline 6 cm anterior to CZ (between CZ and FPZ) can be effective. These techniques allow a more vertically oriented vector, which enables a better excitation of the descending axons’ cortical spinal tract neurons (88). The closer the stimulating electrodes are, the more horizontal the stimulus is and more robust the I waves are.

TcMEPs use somewhat short pulse durations ranging from 50 to 100 microseconds for recovery of the D waves between the also short interpulse intervals of 2 milliseconds (88). Using trains of 4 to 9 stimuli with 2 to 4 milliseconds interpulse intervals tends to overcome anesthetic inhibition of anterior horn cells (87).

Stimulation of the motor cortex produces two patterns of negative waveforms D and I waves with direct recording from the dorsal spinal cord using epidural electrodes. The D wave can be identified as a single negativity, and the I waves are characterized by up to 4 negative peaks named N2 to N5 (88). D waves are caused by direct stimulation of the corticospinal tract. I waves are caused by stimulation of neurons of deeper layers of the primary motor cortex, which ultimately will cause indirect and “transsynaptic” (cortical-cortical connections) activation of the corticospinal tract.

D waves are less influenced by anesthetic or neuromuscular blocking agents, especially when they are averaged (87).

For practical reasons, often only peripheral recordings are used through needle electrodes inserted on the muscles of interest (the muscle-MEP). Because the focus is on the corticospinal tract, the preference is to monitor the distal muscles of the extremities with EMG needles. The hand muscles used are usually the abductor pollicis brevis (APB) and abductor digiti minimi (ADM). On the lower extremities, corticospinal tract converges over the abductor hallucis brevis and tibialis anterior, which are often used during TcMEPs (88).

The muscle responses vary more than the direct spinal cord waveforms (D and I waves) because of the synapse at the anterior horn cell. Three criteria for interpreting changes in the Tc-MEP are used. The first is the threshold method (14) in which the typical criterion for abnormality is the need to increase the stimulating voltage more than 100 V from baseline to obtain responses. Another criterion is either absence or a significant reduction in the amplitude of the Tc-MEP (65; 62). The third criterion is a change in the complexity of the Tc-MEP waveform (108). The criterion for the change in the D-wave is generally a 50% decline in amplitude (61).

The interpretation of the TcMEPs should take into account all the factors that may alter the signal, such as anesthesia and neuromuscular blocking agents. Barbiturates, benzodiazepines, and propofol are the intravenous anesthetics that are more likely to decrease the amplitude and increase the latency of the TcMEPs (55). Inhalatory anesthesia, either halogenated or nitrous oxide, may also decrease the amplitude of the TcMEPs in a concentration-dependent manner (55).

The neuroscience group at Singapore General Hospital discerned that cross-scalp stimulation was often an augmentative approach, especially in infratentorial and spinal cord surgical cases, because the summation of the ipsilateral and contralateral stimulation signals would be easier to realize (71).

A practical way of troubleshooting a poor signal from myogenic TcMEPs is to consider the main causes (87):

Calancie and colleagues found that transcranial electrical stimulation of the motor cortex was more efficient in detection of postoperative motor deficits but missed some of the sensory deficits; the opposite was true for intraoperative somatosensory-evoked potentials (15).

Electrical stimulation of pedicle screws. The goal of neurophysiological testing during screw pedicle placement is to attempt to find the tip (135). The closer the screw tip is to the nerve root, the higher the likelihood of neurologic sequelae. Cadaver studies have demonstrated a 20% chance of screw misplacement, such as misdirection and piercing of the bone wall (48). Radiological verification of the screw placement with plain x-rays is not more precise that neurophysiological testing. The “gold standard” is the computed tomography in real time that is not a practical alternative in the operating room.

The threshold to a response is recorded with progressively higher current intensity, which is delivered at the outer end of the screw. The spontaneous muscle (EMG) activity is also recorded during the screw instrumentation.

The surgeon and the neurophysiology team should be aware of the factors that alter the bone and soft tissue impedance during the surgery. A “wet” surgical field tends to lower the response threshold (88). Some authorities have used a voltage-constant system to circumvent this problem.

After drilling the hole in the pedicle, a needle is inserted just half-way (48). This method allows for greater proximity to the nearby nerve root. Inserting the tip of the needle all the way into the vertebral body would make it more removed from the site; this is where many of the bone wall perforations occur. When using the needle electrode, an EMG response threshold of 4 mA or lower has been associated with bone perforation, provided that the adjacent nerve is normal (48).

Subsequently the screw itself is electrically stimulated. When stimulating the screw, an EMG response threshold of 6 mA or lower has been associated with bone perforation, provided that the adjacent nerve is normal (48).

The effect of pedicle screw instrumentation on the functional outcome has been studied primarily in lumbar surgery (135). The lack of electrical stimulation of the pedicle screws is associated with higher risk of neurologic sequelae after spinal surgery (135; 88).

The data related to cervical spine surgery are more limited, but even the most optimistic evaluations show pedicle perforation with 4% to 10% of the screws placed and a smaller incidence (approximately 1%) of radiculopathy as a consequence of the procedure (01). More screws inducing bone wall perforation are seen with surgery at the C4 and C7 levels (01). Thus, there is some room for improvement, and further studies are necessary to establish the role of pedicle testing in cervical surgery.

Transcranial magnetic motor-evoked potentials (TcMMEPs). Transcranial magnetic stimulation produces a motor response by inducing an electrical current on the nerve. Magnetic stimulation causes less discomfort than its electrical counterpart, so it can be done while the patient is awake.

TcMMEPs recorded from the spinal cord will elicit the D and I waves similar to but different from the responses to electrical stimulation (88). As with intraoperative somatosensory-evoked potentials, transient changes in the magnetically-evoked response were not associated with postoperative neurologic deficits. The preliminary experience using cortical-spinal magnetic stimulation in children during spinal surgery is encouraging. In a study with a mixed adult-pediatric population, the use of motor-evoked potentials seemed to improve the sensitivity and sensibility of the intraoperative somatosensory-evoked potentials to detect postoperative neurologic deficits (102). Among the 500 cases in the latter study, no case of false negative results (normal monitoring recording with postoperative neurologic deficit) was seen when intraoperative somatosensory-evoked potentials were used in combination with motor-evoked potentials, and the specificity of normal data predicting normal findings in a neurologic examination was 100%. The Department of Neurosurgery at Baylor College of Medicine set an index whereby a MEP signal decrease of 50% or greater was predictive of a postoperative neurologic deficit involving the motor- or corticospinal pathways (32). Krieg and colleagues found that continuous MEP monitoring provides reliable data during resection of metastases in motor-eloquent brain regions. However, they established the warning criteria of an amplitude decline greater than 80% of the baseline rather than the conventional 50% as quoted by most others (63).

EMG. Electromyographic monitoring in the form of free-running EMG is also commonly used in a number of surgeries, including spinal surgery for scoliosis and anterior discectomy with fusion. Cranial nerve distribution muscles are monitored during some intracranial operations, such as surgery at the base of the skull (acoustic neuroma, meningioma). Whether the cranial nerves are monitored depends on the surgery. The muscles innervated by cranial nerves III to XII are the ones monitored most often during surgery. Additionally, the optic nerve (cranial nerve II) and the visual system can be monitored by the use of visual-evoked potentials (VEPs, see below).

During motor mapping, the response can be further substantiated by using EMG, which also allows for lower stimulus intensity with a visible response (see Motor and cognitive mapping through direct cortical electrical stimulation below).

Electromyographic monitoring also has been useful during selective dorsal rhizotomy for the treatment of spasticity. Stimulation of the dorsal rootlets with a 50-Hz frequency is used when looking for exaggerated responses, such as lower stimulus threshold, sustained activity after discharges, or the spread of the response to other muscles (44).

Several texts provide data regarding the innervation of muscles used (24; 82), some citing their practical application for intraoperative monitoring (87).

Table 1. EMG Monitoring of Muscles According to Spinal Levels

Visual-evoked potentials (VEPs). The optic nerve (cranial nerve II) and the visual system can be monitored intraoperatively by the use of visual-evoked potentials. Nonetheless, intraoperative visual-evoked potentials are somewhat challenging due to poor reproducibility of flash visual-evoked potentials in this setting (88). Visual-evoked potentials performed with high-intensity light flashes can improve the reliability of this test during surgery. Visual-evoked potentials with direct recording from the optic nerve also may be easier to obtain during surgery than with scalp leads.

When used during surgery, visual-evoked potentials are done through light-emitting diodes that are attached to contact lenses (88). Using green or high-intensity lights rather than a red-colored source will decrease the intraoperative dark adaptation.

Alternative monitoring methods for spine surgery. Direct electrical stimulation of the spinal cord during surgery has also been performed. The stimulation may be either epidural or transosseous (115). The transosseous method employs a needle near the spinous process of a vertebra above the level of surgical intervention when concurrently monitoring below that level (115). False positives may be seen with this technique due to shunting of electrical current from the stimulating needles through the metal rod to the ground electrode, thus, preventing adequate stimulation to the spinal cord. A modification of the transosseous technique with epidural stimulation has been suggested to improve reliability of the procedure (115). Pereon and colleagues used direct electrical stimulation of the spinal cord via needle electrodes in the rostral part of the surgical field and recorded from both sciatic nerves (09; 105; 73). The latter study found no false-negative cases of intraoperative spinal cord damage among the 112 consecutive patients being monitored during surgery for spinal deformity. Neurogenic motor-evoked potentials appear to be faster than intraoperative somatosensory-evoked potentials in the detection of spinal insults during surgery (105).

From a frontier biomedical engineering standpoint, the neuroengineering laboratory at the Georgia Institute of Technology devised a superior stimulator that not only is exacting in the specific tract and white matter funiculi stimulated but also is stretchable and conformational to the spinal cord, so it is less apt to induce mechanical damage to the soft tissue. The SMEA, or stretchable microelectrode array, derived the same range of evoked cap conduction velocities and stimulus resolution as rigid tungsten microelectrodes but afforded the circumferential contact with less mechanical incompatibility to the cord itself (86).

Dermatomal somatosensory-evoked potentials could be considered ideal for monitoring the avoidance of radicular injury that is subadequately detected by intraoperative somatosensory-evoked potentials. However, dermatomal responses do not have good reproducibility. This technique has not been explored sufficiently in children (44). Anesthesia can readily compromise dermatomal responses.

During carotid endarterectomy (CAE) some surgeons advocate triple monitoring; namely, combined somatosensory-evoked potential, motor-evoked potential, and electroencephalography monitoring, citing evidence of predicable synergy (05). In the setting of CAE, the diagnostic accuracy of MULTImodality intraoperative neuromonitoring was higher than an approach using solely single modality intraoperative neuromonitoring such that neuroprotective therapies to prevent periprocedural strokes could be based on the changes in SSEP and EEG, according to the University of Pittsburgh study (134). Additionally, during carotid endarterectomy, some vascular surgeons rely on vagus nerve neuromonitoring as well in order to avoid potential vocal fold paralysis postoperatively.

Intraoperative brainstem auditory-evoked potential monitoring. Intraoperative brainstem auditory-evoked potential monitoring has been shown to be useful for preservation of hearing and vestibular nerve function during the resection of acoustic neuroma/intracanalicular schwannoma and other posterior fossa surgeries (42). The experience with this procedure in children is limited. The types of operations in which the brainstem auditory-evoked potentials are used include not only acoustic neuroma resections, but also: extirpation or revision of posterior fossa and petroclival skull-base tumors, arteriovenous malformations, and aneurysms. Also, microvascular decompression, and decompressive procedures in patients with Chiari malformations (suboccipital craniectomy).

As in diagnostic recordings done in the laboratory, the patients receive auditory stimulation delivered by a series of clicks at intensities of 60 to 70 dB hearing level (17). Earphones, transducers, or even direct middle ear inserts deliver the sounds (80). The signals from many stimuli are averaged due to the low amplitude of each individual auditory-evoked response, which is often less than 0.1 µv. The recording can be done on the scalp in the nonoperative brainstem auditory-evoked potentials or directly from the acoustic nerve with special cotton wick electrodes (69). The scalp electrodes are placed on both ears or mastoids, and vertex, and ground. A contralateral (to the side of stimulus) ear-mastoid to vertex montage can help differentiate the wave form IV and V peaks that may be fused in the ipsilateral channel. The signal phase (rarefaction, condensation, or mixed) should be chosen to maximize brainstem auditory-evoked potential wave forms. The American Clinical Neurophysiology Society Committee on Guidelines for Intraoperative Monitoring of Sensory-Evoked Potentials (02) suggests using click frequencies between 5 and 50 per second. Higher frequencies allow for faster results, which are important in the operating room setting. Nonetheless, signal frequency equal to or higher than 30 clicks per second may produce degradation of the wave forms (20). Using click frequencies that are not multiples of 60 (ie, 11.3 clicks per second) may prevent time-locked summation effects from 60-Hz artifacts generated by multiple electrical devices used in the operating room (69). Overall, the goal should be the identification of the most important components of the brainstem auditory-evoked potentials, namely wave I and wave V, in the shortest time possible (17).

The effects of surgery on the brainstem auditory-evoked potentials are interpreted noting the generators of the wave forms to help localize where the problem is taking place.

Table 2. Brainstem Auditory-Evoked Potential Wave Generators

Similar to intraoperative somatosensory-evoked potentials, intraoperative brainstem auditory-evoked potentials can be influenced by temperature. When the temperature drops below 35°C, the latency of all the wave forms will increase wave V (the most sensitive wave to the thermal effect will disappear with temperatures below 28°C) (44).

Age also has a strong influence in the brainstem auditory-evoked potentials, with latencies getting shorter and the wave forms becoming better formed with maturation. Wave form I, wave form III, and wave form V achieve mature (adult) latencies sequentially.

Table 3. Brainstem Auditory-Evoked Potential Waveform Maturation

The brainstem auditory-evoked potentials are resistant to the effects of medications including benzodiazepines, barbiturates, narcotics, and nitrous oxide. Inhalation general anesthetics (ie, isoflurane, halothane, and enflurane) produce only a mild latency delay and mild decrease in amplitude. Wave V is the most sensitive to the effect from these agents.

Intraoperative brainstem auditory-evoked potential abnormalities. Absolute criteria for intraoperative brainstem auditory-evoked potential abnormalities are not available (17). Reversible disappearance of all brainstem auditory-evoked potential wave forms is compatible with complete recovery, and persistent loss of the brainstem auditory-evoked potential pattern is usually associated with lingering neurologic deficit (17; 56). Typically, changes in wave V are most easily monitorable, and prolongation in latency of more than 10% are considered early changes. Changes up to 20% or 1 msec are considered more significant. Brainstem auditory-evoked potential changes seen during surgery can be divided into 3 types (44):

Other strategies used to monitor posterior fossa surgery. Somatosensory-evoked potentials and EMG of the muscles within the cranial nerve distribution usually have been monitored in co-association with brainstem auditory-evoked potentials. Somatosensory-evoked potential parameters used during posterior fossa, regarding infratentorial lesions, surgery are similar to the ones used for spinal cord monitoring (44). The EMG parameters used for posterior fossa surgery monitoring include a vertical display range of 200 to 500 µv, sweeps of 10 msec per screen, low frequency filter of 30 Hz, and high frequency at 3 KHz to 10 kHz.

The cranial nerve distribution muscles commonly sampled for intraoperative monitoring EMG are: inferior rectus (cranial nerve, or CN, III), superior oblique (cranial nerve IV), masseter (cranial nerve V), lateral rectus (cranial nerve VI), orbicularis oris (cranial nerve VII), stylopharyngeal (cranial nerve IX), cricothyroid or vocalis (cranial nerve X), trapezius (cranial nerve XI), and tongue (cranial nerve XII) (44; 87).

Table 4. Recordings for Cranial Nerve Monitoring, Including EMG, Visual-Evoked Potentials, and Direct Nerve/Nucleus Compound Action Potential

Intraoperative-EMG monitoring evaluation includes survey for the presence of spontaneous electrical activity from muscles that may indicate injury or depolarization of the innervating nerve. Intraoperative monitoring of BAEP is of established benefit in the prevention of hearing loss or cochlear nerve impairment during microvascular decompression of the eighth cranial nerve for primary hemifacial spasm (119).

Direct recording from the exposed VIIIth cranial nerve or nearby cochlear nucleus through monitoring of compound action potentials can be also helpful (88). Compound action potentials (CAPs) monitoring had shown that the cochlear component of cranial nerve VIII is sensitive to stretching and heat injuries, which can be a problem due to the use of bipolar electrocoagulation near the auditory portion of cochlear nucleus VIII.

During surgery for the resection of acoustic neuromas, in addition to monitoring the VIIIth nerve function with intraoperative BAEP, free running EMG of cranial nerve VII innervated muscles such as orbicularis oris, orbicularis oculus, mentalis and frontalis should be used. Because often the surgeon needs to inflict damage on the acoustic nerve to resect the tumor, monitoring the facial nerve distribution muscles may be the most important task to perform during this type of procedure.

Laryngeal nerve monitoring. Intraoperative neuromonitoring (IONM) has gained widespread acceptance among head and neck surgeons as an adjunct, during thyroid surgery, to direct visual (recurrent laryngeal) nerve identification (“gold standard”). Laryngeal nerve monitoring is becoming essentially routine practice for both select and standard thyroidectomy and parathyroidectomy (120). Dralle and colleagues summarize the “Recommendations of the Surgical Working Group for Endocrinology” regarding intraoperative neuromonitoring in thyroid and parathyroid surgery (29), and Barczynski and colleagues review the standards guideline statement from the “International Neural Monitoring Study Group” in reference to monitoring of the external branch of the superior laryngeal nerve during thyroid and parathyroid surgery (10). Some authors stress that neuromonitoring does not indeed reduce the risk of postoperative laryngeal nerve palsy, at least in complicated cases such as redo thyroid surgery or thyroid cancer surgery (06). The general consensus, however, is that intraoperative neuromonitoring has a favorable effect in terms of decreasing the prevalence and severity of upper aerodigestive symptoms typically involving altered swallowing and change of phonation or voice pattern (118). Dionigi and colleagues espouse that continuous, as compared to intermittent, monitoring of the recurrent laryngeal nerve in thyroid surgery is cutting-edge technology and needs further assessment in an evidence-based fashion (26).

Intraoperative neuromonitoring or laryngeal nerve monitoring is then a risk minimization tool. It can be used effectively to verify the functional integrity of the recurrent laryngeal nerve (27). Recurrent laryngeal nerve palsy, a potentially catastrophic complication of thyroid surgery, can be averted with laryngeal nerve monitoring; the incidence without monitoring is up to 3% for permanent palsy and 5% to 8% for transient palsy. For persons who professionally depend on their voice this would be especially taxing. Bilateral recurrent laryngeal nerve trauma can lend toward the eventual need for tracheostomy, stemming from vocal cord paralysis. Intraoperative neuromonitoring can also be beneficial in identification of the anatomy and physiology of the nonrecurrent laryngeal nerve. The movement is toward C-IONM (continuous), over I-IONM (intermittent), to potentially enable the surgeon to react before irreversible damage to the recurrent laryngeal nerve arises.

Regarding animal studies, Wu and colleagues investigated with a porcine model. Recurrent laryngeal nerve traction resulted in eventual loss of signal, similar to the change rendered from clamping or transection or electrothermal injury. But if the traction stress was relieved before the loss of signal, there was typically recovery of the EMG tracings recorded. Repeated traction resulted in lost amplitudes, lending support to the notion that undue retraction of the tissue is deleterious (145).

Electrocorticography (ECoG). Electrocorticography is commonly used during epilepsy surgery. The goal of the electrocorticography is to help determine the areas that need to be resected during epilepsy surgery (99; 97; 98). Electrocorticography can give prognostic information during surgery by indicating the areas of residual discharges after resection of a brain tumor or epileptogenic focus. Nonetheless, this is controversial; some believe that when no interictal discharges are seen after resection, the patients are more likely to be seizure-free than those with persisting discharges (99; 97). In one study of electrocorticography in intractable frontal lobe epilepsy, a higher percentage of Engel’s classification Class I outcome was associated with pre-excision spikes recorded from two gyri or less (p < 0.05) and post-excision spikes absent or limited to the resection border with a p < 0.01 (143). Complete lesion excision correlated with Class I outcome with a p < 0.001 (143). The same study found that only 2% of the cases had Class I outcome when spikes were seen distant from the lesion border and no good outcome if more than 24 spikes/minute were seen; combining completeness of lesion excision with electrocorticography risk factors was highly correlated with class I/II outcome.

In a series of patients with temporal lobe epilepsy, electrocorticography was found to help by predicting poor outcome (83). Baseline electrocorticography with less than one spike per 4 minutes was associated with a poorer prognosis. Conversely, pre-resection electrocorticography showing more than 18 spikes per minute was typically associated with a good outcome.

Spikes with a major positive component are commonly derived from both depth electrodes (78%) and from subdural monitoring strips (72%) (83). Berger and colleagues also found that using electrocorticography to define areas of epileptogenic cortex in and around brain tumors increases the likelihood of satisfactory postoperative seizure control (12).

Using electrocorticography to define the seizure focus during low-grade glioma removal may be more effective in children (11).

Intraoperative ECoG is now being utilized during MRI-guided Laser-Interstitial Thermal Therapy (LITT) for intractable epilepsy, which is an exciting frontier (76).

Asano and colleagues found that intraoperative electrocorticography in children with intractable neocortical epilepsy is reliable only when spike frequency is greater than 10 spikes/minute (07). A spike frequency of less than one spike/minute is largely unreliable for localization of seizure foci.

Salanova and colleagues found that in patients undergoing surgery for the treatment of medically refractory occipital lobe epilepsy, electrocorticography helped improve the outcome (111). Residual spikes on the post-resection electrocorticography were associated with worse outcome.

The adequacy of resection of temporal-lobe mass lesions such as gangliogliomas, cavernous angiomas, and dysembryoplastic neuroepithelial tumors can be aided by intraoperative electrocorticography. Resection of spike foci after lesionectomy improved the 3-year seizure-free outcome (128).

One study used intraoperative hyperventilation or overbreathing to increase the amount of epileptiform discharges in children with relative success after removal of those foci, which were not located in eloquent cortex (132). Occasionally, infusions of the proconvulsant barbiturate methohexital are also used.

In electrocorticography of cortical dysplasia, certain patterns appear to predict the outcome more precisely after resections for intractable seizures. In 1995, Palmini and colleagues found that when areas in which the ECoG showed ictal-like or continuous or quasi-continuous patterns were left unresected, the outcome for seizure-freedom was worse.

Table 5. ECoG Patterns Associated with Cortical Dysplasia

Although intraoperative electrocorticography is not often used by many epilepsy centers and has been found non-useful in cases of surgical management of symptomatic mesial temporal sclerosis (116). Hippocampal electrocorticography (HECoG) may permit a temporal lobectomy to be performed in a tailored fashion. Guided by hippocampal electrocorticography in a temporal lobectomy, a surgeon can minimize the amount of hippocampus removed to minimize postoperative memory decline while maximizing seizure-free outcome (84).

Several combinations of anesthetics have been administered to facilitate intraoperative electrocorticography with and without functional brain mapping. The alpha 2-adrenergic receptor agonist dexmedetomidine has been useful (124), and dexmedetomidine tends to produce a natural sleep pattern on the EEG and tends to reduce the need for propofol as well as inhalational and opiated anesthetics (124).

Functional brain mapping. Intraoperative somatosensory-evoked potentials are a valuable tool for functional brain mapping during various types of resective surgery. The SSEPs are often the first step in cortical localization of the eloquent cortex. SSEPs are recorded during intracranial surgery with a 6- or 8-contact strip of subdural electrodes. Median nerve SSEPs show a phase reversal over the central sulcus near the area of cortical representation of the hand.

The negativity located over the sensory cortex is often alluded to as N1 (75) and corresponds to the N20 on scalp recordings. N1 tends to be relatively small but has a more gradual spatial fall off when compared to P2. Thus, N1 is seen several centimeters around its point of maximal negativity. A positivity located over the hand somatomotor cortex is noted 1 to 2 milliseconds later and corresponds to the P22 of the scalp recordings. The component has been named P2, with a major positivity peaking around 2 to 3 msec after N1 and which was maximal range over the post-central gyrus but may extend to the precentral gyrus. The absolute latency of P2 was calculated to be 22.3+/-1.6 msec (25). P2 has a fast fall-off, disappearing 1 to 2 cm posterior to the central sulcus (75). The post-rolandic potential (N1) can be differentiated because it is negative and peaks earlier and has amplitude that is twice that of the pre-rolandic primary cortical potential (25). Waveform morphology is best when subdural strip electrodes are perpendicular to the central fissure.

The position of the phase reversal of the cortical potentials of the median SSEP (recorded in a reference montage) tends to be across the rolandic fissure as verified by motor response to direct electrical stimulation (25). Nonetheless, the phase reversal is sometimes anterior to the rolandic fissure (25). In these recordings, it is important to make sure that the recording strip is oriented perpendicular to the sulcus being studied.

Motor and cognitive mapping through direct cortical electrical stimulation. During neurosurgical procedures, brain mapping can be also done using direct electrical cortical stimulation to provide an improvement in otherwise intractable epilepsy, facilitating more aggressive and complete removal of the epileptogenic tissue, a brain tumor, or both.

When performing electrical stimulation, a couple of electrical principles are helpful to remember: (1) the stimulus intensity decreases (attenuates) with the square of the distance from the stimulating electrode; and (2) the strength of a stimulus is thought to be dependent on the charge density. Most cortical stimulation is done with systems that vary the current intensity.

The charge (in Coulombs) during cortical stimulation also can be calculated by measuring the “area under the curve” of one of the phases of the pulse. So, for the usual rectangular pulse, the charge equals the current intensity multiplied by the pulse duration.

Charge C= I x D (I = current intensity; D = pulse duration).

Ohms law relates the current intensity (I) to the resistance (R) and voltage (V).

Ohms law: I = V/R

The current intensity (I) is measured in Amperes (Amp) OR milliamperes (mAmp). The resistance is measured in Ohms (Ω), and the voltage is measured in Volts (V) or millivolts (mV).

The energy (E) required to move an electrical current through the tissue is expressed (in Joules) and is related to the square current intensity, pulse duration, and resistance.

E = I2 x D x R.

The charge density (CD) is calculated dividing the charge (C) by the area of the stimulating electrode (ASE) assuming the transmission media is uniform.

Charge density CD = C/ASE

The surface of a bipolar wand electrode (4 cm2 surface) is higher when compared with a subdural grid lead (12 to 13 cm2 surface). The charge density of a hemispheric ball electrode is 159 to 796 microcoulombs/cm2 per phase for peak currents of 13.6 to 15 mA (66). Subdural grid leads will deliver charge densities of 54 to 57 microcoulombs/cm2 per phase for peak currents of 13.6 to 15 mA.

While in the operating room, the surgeon often uses a bipolar stimulator with a range of current intensity from 1 to 15 mA and with a pulse duration of 200 to 1000 microseconds, using a biphasic pulse set at a 50 to 60 Hz frequency. Young children and toddlers may require longer pulse durations of 500 to 1000 microseconds and somewhat higher currents of 10 to 15 mA. The stimulus delivery is generally three seconds long. One should be careful with longer durations of four to five seconds and amperage higher than 15 mA as they may cause activation of neighboring tissues. Some stimulators such as the Ojemann Cortical Stimulator are set to deliver an amperage that is half of the total current because the dial takes into consideration the current from the baseline to the first peak and not from peak to trough.

Cortical Stimulation Parameters:

This type of mapping minimizes the risk of neurologic morbidity, whether it be speech and language deficit or motor or sensory loss. Adults and older children (usually 12 years and older) can undergo “awake craniotomies” in which they are under anesthesia for opening and closing of the skull and dura but awake during the resection of the tumor or epileptic focus.

Our approach entails the use of two modalities of motor-mapping. During awake craniotomy, if our patients are completely cooperative, we perform motor and cognitive testing (such as picture naming) during most of the surgery after cortical exposure. In some cases, we also perform ongoing motor stimulation as witnessed by EMG recordings, which allows for the use of a weaker cortical stimulus and fewer patient movements, thus, enhancing the safety of the procedure. Furthermore, we can also stimulate the white matter before resections looking for motor and cognitive dysfunction. This is especially important during temporal and paracentral resections in which the fibers connecting eloquent cortex are vulnerable to operative insult.

In cases of tumor surgery, brain mapping allows more aggressive resections. The decrease of the tumor burden or metastatic disease affords a survival advantage and lesser likelihood of subsequent risky operative intervention. Because, in the presence of tumor and cerebral edema, the CNS anatomy can be distorted, it becomes critical to rely on intraoperative-evoked potentials for more accurate identification of sensorimotor cortex. One particular surgical procedure seems to warrant MEP over SSEP. With aneurysm repair is the risk of perforating arteries at the corticospinal tract region, within the corona radiate of the internal capsule. This vulnerable area is best monitored with motor- rather than sensory-evoked potentials, given the neuroanatomy and neurocerebrovasculature (41).

The multipulse technique used to obtain the Tc-MEP has many advantages over the Penfield technique discussed above. In particular, it can map motor pathways during general anesthesia. It also has a lower likelihood of generating seizures. However, the longer stimulation times used with the Penfield technique are useful in studying cognitive functions.

Prevention

The goal of intraoperative somatosensory-evoked potential monitoring is prevention of neurologic deficits during and after surgery.

Management

Outcomes

The limitations of intraoperative evoked potential monitoring must be considered when used during neurosurgical orthopedic, vascular, or other resective or ablative procedures. One factor that must be taken into account when monitoring intraoperative somatosensory-evoked potentials is that only the posterior column somatosensory pathways are assessed. Intraoperative somatosensory-evoked potentials may be misleading at times; there may be false positives (18), and postoperative neurologic deficits may be seen despite unchanged findings during monitoring (67). Overall, intraoperative somatosensory-evoked potentials tend to be more sensitive for the spinal insults that involve multiple regions. More focal intraoperative spinal lesions, either compressive or vascular in nature, are less likely to be detected by intraoperative somatosensory-evoked potentials (44). To optimize care, certain neurosurgery groups have designed, developed, and implemented a checklist for responding to intraoperative neuromonitoring alerts in spine surgery (153).

The most common pediatric procedure in which intraoperative somatosensory-evoked potential monitoring is used is the surgical correction of scoliosis (44). In one series, intraoperative somatosensory-evoked potential monitoring was used to monitor orthopedic procedures in 326 children (46). Among the cases undergoing the latter procedure, 63.7% had idiopathic scoliosis, and 31.2% had neuromuscular scoliosis (46; 44). The presence of cerebral palsy was common among these patients; thus, a complete preoperative baseline neurologic examination is essential for the comparison with the patient’s status after the surgery. As follows, the same is also true for most of the other conditions for which this type of surgery is done, such as myelodysplasia, neuromuscular disorders, and brain and spinal cord malformations.

In the Boston Children's Hospital study, nine of 326 cases had acute changes in the intraoperative somatosensory-evoked potentials during surgery; in eight cases the surgical manipulation (traction-distraction) was reversed sufficiently to prevent new postoperative neurologic deficit (46; 44). In one case the patient had postoperative paraplegia, most likely due to the need for resection of a spinal cord tumor. The same study noticed three cases (0.9%) of new neurologic deficits that were not detected by the intraoperative somatosensory-evoked potential monitoring. Two of these cases were focal L4 to L5 radiculopathies (one resolved spontaneously and the other required surgical correction of a loose lamina). One patient had a new onset of urinary retention that resolved in one week. Harper also found that radiculopathies, which were difficult to detect by intraoperative somatosensory-evoked potential monitoring, occurred in four of 184 cases after surgery for scoliosis repair (43). Earlier work by Wilber and colleagues showed that transient postoperative neurologic deficits were more difficult to predict with intraoperative somatosensory-evoked potential monitoring (144). Dermatomal somatosensory-evoked potentials would theoretically be ideal to monitor radicular injury. However, dermatomal responses do not have good reproducibility, and this technique has not been utilized in children (44).

The likelihood of neurologic deterioration after surgical repair of scoliosis is greater in patients with neuromuscular spinal curvature or severe scoliosis (77). Postoperative deterioration is more likely to be seen in patients with pre-existing neurologic deficits, such as patients with meningomyelocele (44). The risk of neurologic injury is also higher in patients who need skeletal traction, Harrington rod instrumentation, or sublaminar wire placement, and after a second spinal surgery (144; 44). These high-risk patients are more likely to benefit from intraoperative somatosensory-evoked potential monitoring during these neuro-orthopedic procedures (44). Intraoperative neuromonitoring can be challenging – indeterminate or unreliable – in young patients with immature neural pathways or underlying preexistent malacia (37). The group at Seoul National University Hospital affirms that EMG, MEP, SEP, and BCR (bulbocavernosus) reflex are essential modalities in intraoperative neurophysiological monitoring for untethering of tethered cord in spinal dysraphism (60). They contend that early intervention is better than waiting for onset of neurologic deficits. Findings include that pathophysiology of the sacral nervous system is both first in onset and principally foremost. McKinney and Islam stress that when evaluating a patient with pediatric scoliosis – as a potential surgical candidate with intraoperative monitoring – be attentive to possible concurrent polyneuropathy; the presence of otherwise unrecognized polyneuropathy could prompt a change in recording parameters (85).

Table 6. Risk Factors for Postoperative Neurologic Deficit After Spinal Surgery

Intraoperative somatosensory-evoked potentials, as a procedure, is rather benign as long as the technologist performing it complies with the usual safety rules of bioelectrical and EEG equipment. However, burns do occur occasionally (126). The most critical causes of these injuries include improperly grounded electrocautery, defective equipment, and the placement of recording electrodes too near sources of high voltage.

Allergic and contactant-related skin changes, such as contact dermatitis, are uncommon problems with EEG lead placement. Skin abrasion (to reduce impedance) and the various chemical components of the leads (silver, gold, and copper) as well as the paste or collodion used may contribute to these dermatological complications.

One report called for an electrocardiogram artifact that can be produced by intraoperative somatosensory-evoked potential monitoring (21). The report describes a 3-year-old girl with Goldenhar syndrome in whom the somatosensory-evoked potentials produced an artifact in the electrocardiogram resembling supraventricular tachycardia leading to inappropriate treatment. Another rather odd occurrence was a case report revealing the onset of an acute postprocedural compartment syndrome as a complication of the use of intraoperative neuromonitoring needle electrodes in the arm, prompting multiple emergent surgical fasciotomies (31).

Communication has been an issue. There needs to be familiarity and trust amongst the neurophysiologist, the surgeon, and the anesthesiologist. The “interventional cascade” should follow: test, then interpretation, then communication, then intervention, then outcome (121).

Special considerations

Anesthesia

Because the most disabling postoperative deficits are motor deficits, one should be aware of the drugs that can alter TcMEPs. GABAergic drugs, such as barbiturates, benzodiazepines, and propofol, are the intravenous anesthetics that are more likely to decrease the amplitude and increase latency of the TcMEPs (55). As a general rule, barbiturate anesthesia precludes TcMEP monitoring and should not be used unless the possible benefits of neuroprotection outweigh the risks of a lack of motor monitoring.

Inhalational anesthesia with either halogenated or nitrous oxide may also decrease the amplitude in a concentration dependent manner TcMEPs (55). In general, even concentrations 0.5 MAC may affect TcMEPs by halogenated agents with preferential suppression of the motor tracts at the level proximal to the anterior horn cells.

A few combinations have been used including a “nitrous-narcotic” combo remifentanil infusion of 0.2 to 0.5 microgram/kg/minute with 60% nitrous oxide although nitrous oxide also has significant deleterious effects on the Tc-MEPs. Other combinations are variations of total intravenous anesthesia (TIVA) using some propofol infusion associated with ketamine or etomidate (55). Total intravenous anesthesia using ketamine or etomidate is especially interesting due to low potential of these two drugs to depress either SSEPs or TcMEPs. In fact, etomidate may even increase the SSEP and TcMEP amplitudes. Nonetheless, etomidate may be a proconvulsant and may cause adrenal suppression (55).

High doses of boluses or inhalational anesthetics may produce a decrease in the amplitude of the cortical potentials during intraoperative somatosensory-evoked potential monitoring (40). Increasing concentrations of halothane can quickly produce a decrease in the amplitude of the cortical potentials, which is directly proportional to the end tidal concentration of that gas (148). Patients less than 10 years of age are particularly susceptible to the effects of high concentration boluses of general anesthetics, which may produce attenuation of the cortical potentials (46). Cortical intraoperative somatosensory-evoked potentials in children are also more likely to be attenuated by a combination of anesthetics, such as isoflurane and nitrous oxide (46). The most optimal scalp recordings are obtained by avoiding the combination of anesthetics and by, as stated, keeping the concentration of nitrous oxide less than 50% and the concentration of isoflurane less than 0.6% (46). The greater instability of the cortical responses in children is thought to be due to the lack of symmetry and synchrony in the myelination process (36). Due to the unreliability of the cortical response in children, the use of recording sites over the cervical spine has been recommended to monitor the cervical potentials (46). Cervical potentials are more resistant to the effects of general anesthetics and can be used to monitor the spinal cord integrity above the surgical level when the cortical potentials are absent (57; 39).

Overall, the most commonly used and validated anesthetic protocol with Tc-MEP and SSEP recording is now total intravenous anesthesia with propofol. It was determined that methadone has a statistically significant effect on SSEPs (latency and amplitude) but not TcMEPs; regardless, the difference did not translate into clinical significance (47). Biscevic and colleagues state, emphatically, that the monitoring of motor pathways with transcranial electric motor-evoked potentials requires the avoidance of halogenated anesthetics (halothane, sevoflurane, isoflurane, etc.) and neuromuscular blockade (vecuronium, rocuronium, etc). Further, they state that ketamine-based anesthesia allows for appropriate MEP recordings but that total intravenous anesthesia with propofol is preferred (13).

As stated previously, the brainstem auditory-evoked potentials are fairly resistant to the effects of medications including benzodiazepines, barbiturates, narcotics, and nitrous oxide. Inhalational general anesthetics (eg, isoflurane, halothane, enflurane) produce only a mild latency delay and decrease in amplitude. Wave V is the most sensitive to the effects of these drugs (64).

Further various applications. Intraoperative monitoring is broadening in its indications and strengthening as an important element toward optimal patient care. It is accepted for, inclusively now, the monitoring of pelvic autonomic nerves during laparoscopic low anterior resection of rectal cancer, for example (152). Multimodality intraoperative neurophysiological monitoring is utilized in anterior hip arthroscopic repair surgeries (100). It is employed for brachial plexus neurolysis during delayed fixation of a clavicular fracture (08). Its use – and clinical benefit with optimal resection and safe outcome – in thyroid surgery is well accepted (146; 22; 90). The Japanese group designed an electromyography endotracheal tube for successful intraoperative neurophysiological monitoring dentification and preservation of an extralaryngeal bifurcation of the recurrent laryngeal nerve (147). Schneider and colleagues state that, in experienced hands, continuous intraoperative neural monitoring in thyroid surgery can diminish permanent vocal fold palsy, a devastating complication in terms of compromise in quality of life, rates to 0% (113). As a powerful risk minimization tool, it can offer some protection in a medicolegal litigious environment (151). Pelvic intraoperative neural monitoring has been associated with a significantly lower rate of fecal incontinence after total mesorectal excision for rectal cancer (58). The utility of intraoperative neural monitoring has been evaluated, by cardiothoracic (aortic arch) surgery, and was found to have both a high sensitivity and specificity, but also a high negative predictive value is reassuring for low risk of stroke in the absence of alerts (35). Intraoperative neural monitoring of the facial nerve has been applied, by EMG, in cases of tympanomastoidectomy for chronic ear disease (96); it was found to be both feasible and effective for facial nerve stimulation and identification. The facial nerve is also monitored during surgery at the CP or cerebellopontine angle, along with the cochlear nerve. As stated, intraoperative neurophysiological monitoring is essentially the mainstay (‘standard of care’) in various spine surgeries, including adult and pediatric procedures (33; 50; 53; 130; 149). Controversy remains within certain neurosurgical circles whether intraoperative neurophysiological monitoring (SSEP/TcMEP) is key in elective microsurgical clipping of unruptured intracranial aneurysms (38).

The Journal of Clinical Monitoring and Computing has released provocative dialogue regarding the “new” American Society of Neuromonitoring supervision guideline and is under scrutiny or attack, principally by SA Skinner from Northwestern and colleagues. He contends that leniency exists in the new guidelines in favor of the telemedicine industry. He favors, as proper science and proper care, a personal-in-room approach toward quality intraoperative neurophysiological monitoring that enhances communication (122). The response – by Gertsch and associates from UCSD – applauded Dr. Skinner’s pursuit of important concepts such as teambuilding, collaboration, and effective communication (34). The resources toward personal-in-room intraoperative neurophysiological monitoring are sparse.