Neurophysiological monitoring during endovascular procedures on the spine and the spinal cord

29.1.1

Vascular anatomy of the human spinal cord

Before describing the role of neurophysiological monitoring during endovascular procedures aimed to treat spinal hypervascular lesions, an overview on the vascular anatomy of the normal spinal cord is mandatory. While we refer the reader to classical textbooks and articles for a detailed analysis of the vascular anatomy and the wide range of variations, here we will concisely describe those aspects relevant to the discussion of intraoperative neurophysiological techniques.

During early embrional development, each somite receives one pair of so-called segmental arteries, arising from the dorsal aorta. Blood supply to the neural crest is then provided by a dorsomedial division of the ipsilateral segmental artery, the dorsospinal artery. This vessel supplies the neural tube via paired ventral longitudinal arteries. Dorsally oriented branches of these ventral arteries deeply penetrate the ipsilateral half of the neural tube. A network of capillaries around the neural tube then organizes into longitudinal arterial axes. Between the sixth week and the fourth month of uterine life, the cranio-caudal formation of a more mature vascular pattern is characterized by the ventral migration and then fusion of the ventral longitudinal arterial axes, to form the anterior spinal axis (ASA). On the posterior aspect of the cord, the pial network organizes into paramedian dominant axes, which will give rise to two posterolateral spinal axes (PSAs). Like the embryonal circulation of other systems, a multistage rearrangement of the original radicular feeders to the ventral and posterior axes also occurs. As a result of this process, no more than 4–8 anterior and 10–20 posterior radicular arteries remain by the end of development.

In the adult, we observe three main longitudinal arterial systems. The ASA extends almost uninterrupted from the medulla to the filum terminale. At the cervico-medullary junction, it originates from the two vertebral arteries near the vertebrobasilar junction. Caudally, the major blood supply comes from dorsal branches of intercostals and lumbar arteries. The major radiculomedullary arteries arise, at the level of the cervical enlargement, from vertebral, deep cervical or ascending cervical arteries. The thoracolumbar territory is supplied mainly by the arteria radicularis anterior magna or “artery of Adamkiewicz.” This usually rises from the 9th to the 12th intercostal artery, on the left side in approximately 80% of the cases. It gives off a small ascending branch and a large descending branch, which anastomoses with the posterior spinal arteries to configurate the anastomotic basket surrounding the conus medullaris .

Because of its segmental vascularization, each major arterial group (cervical, upper thoracic, and Adamkiewicz) irrigates its own portion of the cord without significant anastomoses among them. Consequently, the spinal cord is typically vulnerable to hypoperfusion at the middle thoracic level.

The paired posterior spinal arteries arise, at the cervical level, either from the vertebral arteries or, less frequently, from the posteroinferior cerebellar arteries. Caudally, these paired posterior spinal axes receive radiculopial feeders also from the vertebral, the intercostal, and lumbar arteries and are located on the posterolateral surface of the cord adjacent to the dorsal root entry zone. The numerous anastomoses in this posterior system decrease the risk of ischemia for the posterior spinal cord.

From a neurophysiological perspective, it is important to bear in mind that the direction of spinal cord blood flow (SCBF) at any level in the cord cannot be easily predicted because it depends on the location of the dominant anterior or posterior spinal artery for that segment of the cord . Nevertheless—as summarized in Fig. 29.1 —the ASA, through perforating sulcocommissural arteries, is assumed to supply the anterior two-thirds to four-fifths of the cord, including the anterior column of the central gray matter, the anterior and lateral corticospinal tracts (CSTs) and the anterior and lateral spinothalamic tracts. The ASA, therefore, accounts for vascularization of those structures involved in the propagation of motor-evoked potentials (MEPs) from their cortical generators to the α-motoneurons: anterior and lateral CSTs and, to a lesser extent, the propriospinal system. Conversely, the PSAs supply the posterior horns of the central gray matter and the dorsal columns; although the debate is still open, these posterior columns are usually considered the main tracts for central propagation of somatosensory-evoked potentials (SEPs), after peripheral stimulation .

Circumferential vessels from the ASA anastomose with the PSAs through a complex pial network, the so-called vasa corona , which supplies the peripheral rim of the white matter and represents a functionally relevant dorsoventral connection. Therefore in the axial plane, the watershed zone of the cord is located in the anterior two-thirds of the cord, in the white matter adjacent to the anterior horn cells, where penetrating branches from the ASA and PSA meet at the circumferential pial network .

The complexity of both longitudinal and axial angioarchitecture of the vascular supply to the spinal cord accounts for the unpredictability of hemodynamic patterns in the spinal cord; the direction of SCBF becomes even more bizarre in the presence of a vascular malformation, which interferes with normal patterns.

Although studies on spinal cord vascular anatomy have mainly focused on the arterial circulation, it has to be emphasized that venous anatomy is equally essential since it is dramatically involved in the pathophysiology of most vascular malformations. Although venous anatomy is even more unpredictable than its arterial counterpart, two drainage pathways are usually considered.

Sulcal veins drain blood from the central portion of the cord, through the anterior median fissure into the anterior median spinal vein; this receives blood from tributaries of central veins, which drain the central gray matter, including anterior horns. A dorsal spinal vein, often larger than the anterior one, drains the postero-central portion of the cord. The radial or coronal veins originate from capillaries at the gray–white junction, coursing centrifugally and draining the anterolateral and dorsal regions of the spinal cord.

The final common pathway of spinal cord venous drainage is through the radicular veins, which pierce the dura to drain into the epidural veins; these radicular veins lack valves but typically narrow at the dural penetration to prevent retrograde venous flow .

29.1.2

Primers on pathophysiology of spinal cord ischemia secondary to spinal cord vascular malformations

More recent classifications of spinal cord vascular malformations include a number of lesions, the characteristics of which will be described later in this chapter. However, regardless of their specific hemodynamics, the final common pathway in the pathophysiology of these lesions is spinal cord ischemia or hemorrhage. Basically, in both the arteriovenous malformations (AVMs) and in the arteriovenous fistulas (AVFs), the main mechanism is the lack of a capillary bed and a direct shunt of the arterial blood into the venous compartment. This arteriovenous shunt leads to vascular steal phenomena from the adjacent normal vasculature; the more the malformation shares its arterial supply with the normal spinal cord, the more the cord will be exposed to a vascular steal, and therefore to an ischemic injury. On the venous side a multifactorial phenomenon leads to venous hypertension and thrombosis. Venous inflow is increased because of direct arterial feeders; venous outflow is sometimes compromised by a malfunctioning of the valve system of radicular veins when these pierce the dura (which contributes to venous engorgement). A subacute necrotizing myelopathy resulting from thrombosis of a spinal AVM has been described as the Foix–Alajouanine syndrome . The coexistence of venous or arterial aneurysms increases the risk for subarachnoid and parenchymal hemorrhage, which accounts for the acute onset of symptoms in intradural AVMs. Arachnoiditis may evolve from repeated hemorrhages.

29.2.1

Evoked potentials in spinal cord ischemia: experimental and clinical studies

Most data on the role of neurophysiological techniques in decreasing the incidence of spinal cord ischemia come from thoracoabdominal aneurysm surgery. The role of SEPs and MEPs in detecting cord ischemia and preventing irreversible neurological deficits has been investigated both in experimental animal models as well as in clinical studies.

Konrad et al. tested the sensitivity of MEPs recorded epidurally from the spinal cord as well as from the peripheral nerve after direct cortical stimulation of the motor cortex in dogs. The peripheral nerve response appeared to be very sensitive to cord ischemia after cardiac arrest compared to the late disappearance of the spinal cord response .

Similarly, Kai et al. concluded that peripheral neurogenic MEPs provide a better warning system for spinal cord ischemia than spinal MEPs and SEPs recorded from peripheral nerves after stimulation of the spinal cord; unfortunately the spine-to-spine response is not specific since it activates neural pathways both orthodromically and antidromically .

Laschinger et al. investigated spinal cord ischemia after thoracic aortic cross-clamping in dogs. He documented a time- and level-dependent deterioration and loss of the spinal cord response recorded from subcutaneously inserted spinal electrodes after spinal cord stimulation at T3–T4. This suggests that ischemia begins in the most distal cord, makes progress upwardly, and can be prevented by maintaining an adequate distal aortic perfusion .

Similar results on the higher vulnerability of the lower spinal cord were reported by Reuter et al. who correlated MEPs to ischemic spinal damage after aortic occlusion in dogs; the greater portion of cord damage was confined to the gray matter of the caudal segments of the cord . This might be related to a discontinuous ASA; if the lumbar cord relies only on the Adamkiewicz’ artery for its blood supply, occlusion of this artery would not be corrected by collateral feeders and perfusion would be inadequate. These authors also confirmed the early disappearance of the peripheral nerve MEP, which was considered even too sensitive as an indicator of spinal cord damage. Conversely, they found a clear correlation between spinal MEPs after brain stimulation, spinal cord perfusion, and histopathologic findings. In the same study, SEPs appeared to be more sensitive than spinal MEPs to ischemia .

Concerning the different role of spinal MEPs as compared to peripheral nerve MEPs, it is noteworthy that Reuter et al. observed the presence of spinal MEPs but the absence of peripheral nerve MEP 24 hours after cord ischemia, when animals were paraplegic. This was explained on the basis of histological findings, since the damage was primarily confined to the gray matter but did not significantly affect the white matter where propagation of the descending volleys was preserved . To elicit peripheral nerve MEPs after brain stimulation, conversely, requires the functional integrity of the anterior horns.

A similar comparison of the sensitivity between spinal and muscle MEPs (mMEPs) after transcranial electrical stimulation in the detection of spinal cord ischemia was performed by de Haan et al. . They concluded that mMEPs disappear earlier and are therefore more sensitive than epidural MEP, suggesting their clinical use to assess spinal cord perfusion during surgery at risk for ischemia. Early disappearance of mMEP is secondary to the polysynaptic transmission of this potential, so that a reduction in SCBF, which affects functional integrity of anterior horns, will switch off neural transmission at that level. Epidurally recorded MEPs, conversely, are more robust since no synapses are involved and white matter is more resistant to ischemia than gray matter .

In the past few years, along with experimental work, there has been increasing clinical evidence on the usefulness of neurophysiological monitoring during thoracic aorta surgery. For a long time, SEPs have been used to assess the functional integrity of the spinal cord . Unfortunately, SEPs are aimed at monitoring the dorsal column and the posterior spinal cord, but they do not reflect the functional integrity of motor pathways. Dorsal column response to ischemia, moreover, is relatively slow and SEP monitoring could not detect ischemia in time to revert the injury before irreversible neuronal damage occurred.

Machida et al. described a dissociation of mMEPs (after spinal cord stimulation) and SEPs (also following spinal cord stimulation) resulting from ischemic damage to the spinal cord in both an experimental setting as well as during spinal fusion with Cotrel–Dubousset instrumentation; because of the greater vulnerability of mMEPs to ischemia when elicited by this method, they suggested the use of mMEPs as a sensitive measure of anterior cord function. Similarly, de Haan et al. proposed mMEPs after transcranial stimulation as optimal tools for assessing the status of motor pathways from the cortex to the muscle . In their experience, mMEPs turned out to be sensitive and specific since they correctly predicted motor outcome in all patients with no false-negative (postoperative motor deficits despite unchanged motor-evoked response) or false-positive results (significant changes in intraoperative MEPs despite unchanged motor outcome).

It is less likely that spinal cord surgery could damage the anterior horn gray matter while leaving the white matter intact. The opposite is true during spinal cord embolization due to the selectivity of spinal cord vascularization. For generation of epidural MEPs after transcranial electrical stimulation, only intact long motor tracts are necessary. The generation of mMEPs after transcranial electrical stimulation, however, depends on long motor tracts and the segmental level anterior horn gray matter. Therefore mMEPs should be a better monitoring tool for endovascular embolization of the spinal cord vessels.

29.2.2

Clinical application of neurophysiological monitoring for endovascular treatment of spine and spinal cord vascular lesions

Endovascular techniques are increasingly used in the treatment of hypervascularized lesions in the spinal cord and surrounding structures. The injection of embolizing materials has proven useful in the devascularization of spinal cord tumors and occlusion of intramedullary, dural or spinal AVMs or AVFs . The occurrence of vasospasm or the unrecognized obliteration of vessels feeding the normal spinal cord, however, puts the spinal cord at risk for ischemia. If ischemia is not detected in time before irreversible damage has occurred, patients can suffer from permanent neurological deficits .

Since a detailed angiographic study and the following embolization can last for several hours, these procedures are often performed under general anesthesia, which also allows for an optimal angiographic study as discussed later. In the past, a so-called wake-up test was performed, to assess the neurological status of the patient immediately after any critical maneuver. These tests, however, prolong the procedure, carry discomfort to the patient, and might take too much time before protective measures are readily available. The greater reliability of neurophysiological monitoring in detecting spinal cord ischemia, when compared to the wake-up test, has been established for spine surgery .

Through the years, intraoperative neurophysiological monitoring (IOM) has been proposed as a valid alternative to the wake-up test to assess the functional integrity of neural pathways during endovascular procedures. SEPs have been used since the mid-1980s based on the clinical evidence that SEPs were sensitive to compromises in anterior spinal artery circulation . Concerns about the reliability of SEPs in evaluating the integrity of descending motor tracts during spine and spinal cord surgery , as well as during aortic surgery, have then appeared in the literature . As previously discussed in this chapter, the occurrence of motor deficit despite intraoperatively unchanged SEPs is explained by the limited ability of SEPs to assess the functional integrity of CSTs. Nevertheless, despite the advent of reliable techniques to elicit MEPs under general anesthesia , reports on the use of MEPs during endovascular procedures in the spinal cord remain anecdotal .

In the following sections, we describe the protocol we currently use at the Institute for Neurology and Neurosurgery to perform multimodal neurophysiological monitoring during endovascular treatments.