INTRODUCTION

The tethered cord syndrome is a complex developmental malformation, with the underlying pathological anomaly being a dura mater defect or dural schisis. The dural schisis may not be the only developmental defect, but it is probably the most basic one and one that occurs more commonly than is generally recognized. Establishing the diagnosis and assessing the extent of functional disability is often difficult but may be aided considerably by the use of several clinical neurophysiological studies, including somatosensory evoked potentials (SSEPs), urodynamics with sphincter and pelvic floor electromyography (EMG) and anal sphincter EMC and pressure monitoring for intraoperative use.
 
Developmental Anatomy

The spinal cord is freer within the vertebral canal than is the brain within the cranium. The spinal dura mater is composed of dense connective tissue with few elastic elements derived from paraxial mesoderm. It is separated from the vertebral internal periosteum by the epidural space, which contains fat cells, blood vessels, and loose connective tissue. The spinal cord needs to be completely free from the vertebral column during development because the rates of growth of the two structures are different. Early in development, the caudal region of the spinal cord undergoes a progressive upward displacement or retrogression relative to the caudal vertebral column. The conus medullaris, which is initially at the coccygeal level in the 30-mm embryo, ascends through the S4 level in the 67-mm embryo, to the L3 level by birth (40 weeks conceptional age), and to the adult L1-2 level by 49 to 50 weeks conceptional age.
The subarachnoid space elongates progressively to accommodate the elongating spinal nerve roots and filum terminale. The filum terminale must also elongate because the cord retains its original coccygeal attachment through this structure. Early dural schisis (below L3) through which the spinal cord comes in direct contact with subcutaneous tissue lends to tethering of the spinal cord to this tissue. Later, subcutaneous adipose tissue penetrates and expands into the intraspinal space. This results in a low conus medullaris and a short, thick filum terminale. It is possible that the adipose tissue is stimulated by its direct contact with neural elements and the abundant arachnoidal vascularity through the dural schisis.

Neuroanatomy of Somatosensory Evoked Potentials

Evoked potentials recorded from the body's surface are either near field or far field in nature, that is, the generator source is close or distant to the site of recording. The generators may be in gray matter or white matter. Generators in gray matter produce postsynaptic potentials (PSPs), which may be near field or far field. Near field PSPs are probably responsible for cortical components of SSEPs.
White matter generates compound action potentials (APs), which are propagated through fiber tracts. The latencies of the propagated APs increase proportionately to the distance from the point of stimulation and hence are dependent on the recording electrode position. These are recorded only in close proximation to the fiber tract itself and thus are termed "near field potentials" (NFPs). Because they are close to the site of origin, the amplitude is relatively large (> 1 µV). Other evoked potentials may be recorded at long distances from the point of propagation and are generated when a travelling impulse (signal) passes through a certain anatomical site or fixed point along the nerve. These are called "far field potentials" (FFPs). It was previously considered that FFPs reflected the approaching volley recorded beyond the point of termination of an active fiber. More recently, it has been suggested that FFPs are generated because of abrupt changes in the geometry of tissue surrounding the nerve, a change in the medium through which the volley is transmitted, or a change in the direction of the fibers.
In summary, NFPs have a specific distribution (topographic specificity), latencies that vary according to the recording electrode placement, amplitudes> 1 µV, and generally negative polarity. FFPs have a diffuse distribution, fixed latencies, amplitudes <1 µV, and polarity that probably reflects a volume-conducted positivity.
Potentials are labelled according to polarity and mean latency from a sample of the normal population. As one would expect, latencies change with body growth and nervous system maturation. Hence, labels differ between children and adults.

Posterior Tibial Nerve Somatosensory Evoked Potentials

There are less standard evoked potential component designations for posterior tibial nerve (PTN) SSEPs than for median nerve SSEPS. For the purpose of discussion of generators of SSEP-PTNs, adult terminology is used, with child or infant notation following in parentheses. Following PTN stimulation, electrodes over the popliteal fossa record the electronegative peripheral nerve action potential N8 (N5). Electrodes over the lower spine record two electronegative potentials: the N19 (N11) and the N22 (N14). The N19 (N11) represents the afferent volley in the cauda equina. The N22 (N14) is a stationary potential and probably reflects postsynaptic activity of internuncial neurons in the gray matter of the spinal cord. Electrodes over the cervical spine record another later stationary potential: the N29 (N20). This component may reflect postsynaptic activity in the nucleus gracilis. The P37 (P28) is the first major localized recorded component on the scalp. It reflects the ipsilaterally oriented cortical surface electropositivity, while the electronegative end of the dipole may be recorded contralaterally. There is a great deal of intersubject variability in the topography of the P37 (P28) in adults and especially in children. This is probably related to the known anatomical difference in the location of the primary sensory area for the leg. When the leg area is located at the superior edge of the interhemispheric fissure, the cortical generator for P37 (P28) is vertically oriented and its amplitude is maximally close to the vertex. When the leg area is located more deeply in the fissure, the cortical generator is more horizontally turned and the P37 (P28) projects ipsilaterally.

Pathophysiology of Tethered Cord Syndrome

Four decades ago, the usual explanation for the neurological deficit associated with tethered cord syndrome was the effect of traction in preventing the ascent of the spinal cord within the spinal canal during growth. However, Barson pointed out in 1970 that the spinal cord does not ascend significantly after birth. The incongruity in observations is due to the fact that the spine grows most rapidly during embryogenesis and after puberty (during teenage growth spurts), while symptoms of tethering most often are observed in early childhood (age 3-10 years). James and Lassman reported a clinical case in which, during a postmortem examination, a small bony septum from the midline of the laminae of L3-4 to the underlying vertebral body was found in an aged woman. She had never had any neurological deficits. Had there been significant ascension of the spinal cord after birth, she would have had to have had neurological deficits, so the authors argued. While this was not necessarily so because issues of the divided cord segments rejoining below the spur and the size of the cleft between cord segments (small or large) were not addressed, the concept that the spinal cord ascended postnatally and produced neurological deficits by traction alone in tethered cord syndrome seemed untenable.

Yamada et al examined the mitochondrial oxidative metabolic changes in the spinal cord before and after subjecting it to stretching. Using reflection spectrophotometry, they monitored in vivo changes in the reduction: oxidation (redox) ratio of cytochrome a, a3 in animal models and in human tethered spinal cords (Figure 1). They found a marked metabolic and electrophysiological susceptibility of the lumbosacral cord subjected to hypoxic conditions, especially under traction with hypoxic stress (Figure 2). They concluded that symptoms and signs of tethered spinal cord were associated with lumbosacral neuronal dysfunction and that this dysfunction is possibly due to impairment of mitochondrial oxidative metabolism. This is supported by the associated evoked potential changes (see below). Most paediatric neurosurgeons now believe that the chronic stretch on the cord produced by tethering is an essential part of the problem but that superimposed insults such as acute flexion episodes or cord hypoxia are also needed for symptoms to become manifest.
	Figure 1. Redox changes during hypoxia in one group of the human tethered cords (Type 1). No redox change is seen before untethering (dotted line), but a reduction similar to that in normal cat cords is noted after untethering (solid line). No reduction occurs while the cord is temporarily retethered (interrupted line). FIO2 = fraction of inspired oxygen. (Reproduced from Yamada et al)	Figure 2. Upper) Normal cord potentials in response to dorsa! root stimulations. IMS: from the posterior column; N1a: from the afferent terminals; N1b: from the interneurons of the first order; N2: from the interneurons of the second and third orders. Lower) Marked change in the cord with traction of 5 gm. (Reproduced from Yamada et al)

Diagnostic Clinical Neurophysiological Studies Electrophysiological studies that help with diagnostic formulation include SSEPs after peroneal nerve stimulation (SSEP-PN), after pudendal nerve stimulation (SSEP-PuN) and after posterior tibial nerve stimulation (SSEP-PTN), bulbocavernosus reflex responses (BCR), and urodynamics with sphincter and pelvic floor EMG. Somatosensory Evoked Potentials It has been three decades since SSEPs were first used to evaluate patients with occult spinal dysraphism. Cracco and Cracco recorded scalp and spinal responses after peroneal stimulation over the cauda equina and rostral spinal cord in adult and child control subjects (Figure 4). These spinal potentials consisted of low-amplitude triphasic waves over the cauda equina and larger potentials over the caudal spinal cord. Scalp potentials had latencies of 30 to 34 msec for electropositive components and 40 to 45 msec for electronegative components. In patients with sacral lipomas and no or minimal neurological findings, spinal potentials normally recorded over the lower thoracic spine (T9-12) were recorded over the lumbar spine, suggesting caudal displacement of the spinal cord (Figure 5). In children with more extensive neurological findings (foot deformities, neurogenic bladders), relatively normal potentials were recorded over the cauda equina, and cerebral potentials were absent. Others have subsequently verified the diagnostic value of SSEPs.
	Fig-4	Figure 5. Spinal responses in a 3-year-old child with thoracolumbar myelomeningocele. The large complex response that is recorded over T12 to T9 in normal children is present over L3 in this child, suggesting caudal displacement of the spinal cord.

One group reported a patient who had postoperative SSEP-PN studies that were slightly improved compared to preoperative studies, and the patient had improved clinically. The authors have systematically studied children and young adults with tethered cord syndrome using SSEP-PTNs. Because SSEP-PN scalp and spine components are lower in amplitude than those produced by PTN stimulation and because the topography of the scalp component of SSEP-PN is more variable than that of SSEP-PTN, SSEP-PTNs were used rather than SSEP-PNs for evaluation in children suspected of having tethered cord syndrome. Clinical, myelographic, and operative studies were prospectively evaluated in 22 consecutive patients, aged 18 months to 22 years, with symptoms of tethered cord syndrome. Ten had previously undergone repair of lumbosacral meningomyelocele. In 19 patients, the diagnosis was established radiologically and/ or intraoperatively. In three patients with clinical symptoms but without radiographically demonstrable lesions, SSEP-PTNs were normal.

Details of SSEP-PTN methodology have been reported elsewhere. Briefly, square wave stimuli are delivered to the PTN at the ankle, with an intensity sufficient to cause a twitch of the abductor hallucis muscle or three times the sensory threshold. If the patient were anesthetic to the stimulus, then a sensory threshold three times that of a similarly aged control subject was used. The recording montage is presented in Figure 6. The bandpass was 30 to 1500 Hz, with 40,000 amplification. Between 1000 and 2000 responses were averaged and replicated. Normative data are largely based on height. Only occasionally will age be used because generally the authors believe height is a better predictor of peak latency (Figure 7).
		Figure 7. Upper left) Relationship between stature and absolute latency of N14 in children, with height ranging from 82 to 130 cm: x = 1.84 ± 0.11 (height). Upper right) Relationship between age and absolute latency of N14 in children aged 1-8 years: x = 10.26 ± 0.74 (age). Lower left) Relationship between stature and absolute latency of N20 in children, with height ranging from 82 to 130 cm: x = 4.60 ± 0.14 (height). Lower right) Relationship between age and absolute latency of N20 in children aged 1-8 years: x= 15-51 ± 0.95 (age). (Reproduced from Gilmore et al16 with permission.)

 

Based upon the absence or presence and the latency of N22 (N14) and P37 (P28), a severity rating scale for SSEP-PTN was developed (Table 1). The generator of the N22 (N14) is the lumbar spinal cord gray matter (Table 2). Thus, the lumbosacral neuronal dysfunction reported by Yamada et al might be reflected especially in abnormalities of the N22 (N14).

Yamada developed clinical severity scales using the factors of gait, bowel/bladder continence, motor, sensation, and deep tendon reflexes (Table 3), and also an operative severity scale based on the presence of lipoma, tension on the filum terminale, and/or extent of adhesions and cord movement after lysis of adhesions (Table 4). In three patients with clinical symptoms but without radiographically demonstrable lesions, SSEP-PTNs were normal. In the 19 patients with tethered cord syndrome, the clinical score and SSEP-PTN score correlated significantly (r=.81, P<0.001). The location and direction of the tethering structures influenced the SSEP-PTN findings (Table 5). In patients with involvement primarily of the conus, the N22 (N14) was present but diminished in amplitude. In patients with extensive attachment of the spinal cord, the N22 (N14) was generally absent (Figure 8) and frequently the N19 (N11) was also absent. Patients with scalp SSEP-PTN asymmetry tended to have the more severe abnormality contralateral to cord deviation or rotation (Figure 9).

Figure 8. Abnormal SSEP-PTN with absent N14 (lumbar potential) and N20 (cervical potential) in a child with extensive attachment of the spinal cord.	Figure 9. Abnormal SSEP-PTN. This is more abnormal than the study in Figure 8 in that the lumbar, cervical, and cortical potentials are all absent. The more severe abnormality was contralateral to the spinal cord deviation, with the P28 (cortical potential) absent after left leg stimulation.

 
Postoperative SSEP-PTNs were sometimes improved (Figure 10 and Table 6). Findings associated with clinical improvement include an increase in the amplitude of N22 (N14), normalization of the P37 (P28):N22 (N14) amplitude ratio, shortening of the N22 (N14) latency, appearance of previously absent N22 (N14), and a decrease in central conduction time (latency P37 (P28) -latency N22 (N14)). Some technical points were important: children 8 years old should be tested in the waking state because latency of the scalp component of SSEP-PTN varies with that state.

The N14 in children (probably generated by structures generating N22 in adults) is considerably higher in amplitude than in adults. Hence, its absence is a more reliable indicator of dysfunction in children.

There is a small but growing literature on evaluating the spinal cord with SSEP-PuN. None of these studies have systematically evaluated the use of SSEP-PuN specifically in patients with tethered cord syndrome, but application to this clinical condition is obvious. The technique consists of stimulating any of several structures innervated by the pudendal nerve. The most accessible structure is the dorsal nerve of the penis, which can be stimulated bilaterally using ring electrodes placed at the base of the penis or unilaterally using laterally placed cup electrodes. Stimulation of the urethra or anus is possible using catheter electrodes with tip-inflatable balloons to maintain appropriate stimulus localization (Figure 11).
Square wave electrical stimuli of 0.3 msec applied at 1.7 to 3.1 Hz with an intensity 2.5 times threshold are delivered. Scalp recording electrodes are placed at Cz (2 cm behind Cz, International 10-20 Electrode Placement System) and Fpz. Spinal electrodes are placed at the spinous process of T12 or L1 with reference to electrodes at the iliac crest or T6 spinous process. A common bandpass is 10 to 500 Hz. Sampling time ranges from 100 to 200 msec. Five hundred to 1000 responses are needed for the critical response and 1000 to 2000 responses for the spinal response. Only stimulation of the dorsal nerve of the penis will elicit a spinal response; stimulation of other structures innervated by the pudendal nerve will not elicit a spinal response. The mean latencies of the cortical response of the SSEP-PuN is similar to, but slightly longer than, that of the cortical response of the SSEP-PTN: 42.3 + 1.9 msec. Amplitude varies depending upon which branch of the PuN is stimulated, The spinal response has a latency of 12.9 + 0.8 msec. This latency is much slower than that of the spinal response of the SSEP-PTN, resulting in a long central conduction time (cortical peak latency spinal peak latency) of approximately 30 msec. This has been attributed to central conduction via smaller fibers than those giving rise to the SSEP-PTN response or to a greater number of synaptic connections in the pathway. With spinal cord lesions at or above TI2, there may be absence or prolongation of the cortical potential. With spinal cord lesions at L1 or below, there will be absence or prolongation of the lumbar potential as well as the cortical potential. In actual practice, the value of this localization is limited because obesity, peripheral neuropathy, or stimulation of the PuN at sites other than the dorsal nerve of the penis will also lead to absence of the spinal potential.

Figure 11. Cortical evoked responses on stimulation of the dorsal nerve of the penis (upper), urethra (middle), and anal sphincter muscle (lower) in the same individual.	Figure 12. BCR responses recorded from different sites in the perineum on stimulation of the dorsal nerve of the penis. SER = sensory evoked response.

Bulbocavernosus Reflex

The BCR is elicited by squeezing the glans penis or glans clitoridis or pulling on an indwelling Foley catheter. The response, contraction of the external anal sphincter muscle, may be seen or palpated. The absence of a response in a man is highly suggestive of a neurological lesion. Unfortunately, it is absent in up to 20% of normal females.
The corresponding BCR electrophysiological potential is recorded over the perineum using small surface electrodes placed midway between the penis (or vagina) and anus. Stimulation electrode placements and parameters are similar to those of SSEP-PuN. Recording parameters are also similar, except that 30 to 100 responses are necessary. With a stimulator, there is often a visible contraction of the pelvic floor. The BCR potentials have a biphasic appearance (Figure 12). The latency of the potential is variable: mean = 35.9 msec, with a range of 26 to 44 msec. The BCR potential allows electrophysiological evaluation of cauda equina and conus medullaris function. That is, prolongation-or, more commonly, absence-of the BCR potential suggests dysfunction of the cauda equina and/ or conus medullaris. Thus, both SSEP and BCR studies can indicate function of some of the neural pathways traversing the cauda equina and conus medullaris, two regions frequently involved in tethered cord syndrome.

Pelvic Floor Electromyography

Another means of evaluating the child with tethered cord syndrome is EMG of the perineal floor muscles. This technique usually involves placing a concentric needle electrode into the external urethra sphincter muscle. Individual motor unit potentials are examined. 5tandard criteria for firing rates and other features are used to determine whether the muscles are normal or denervated. 5ince many pelvic floor muscles are innervated by different sacral roots, careful and thorough EMG examination of several muscles may be necessary to determine the extent of the lesion. Other muscles (in addition to the external urethra sphincter muscle) that can be sampled include the external anal sphincter muscles, the bulbocavernosus muscle, and the ischiocavernosus muscle. This sort of examination is best performed by a physician, usually a urologist, with special training in the technique. While EMG can be performed in a patient of any age, it is usually not done in infants or very young children unless there is a critically important diagnostic question regarding the integrity of the S2, S3, and S4 roots.

Intraoperative Electromyographic Monitoring for Tethered Cord Syndrome

The lumbosacral anatomy in patients with tethered cord syndrome is frequently complex. Nerve roots may be embedded in lipoma or may be visually indistinguishable from adhesions or a thickened filum terminale. The S1 and lumbar roots are recognizable by palpation of the contracting muscles after intraoperative electrode stimulation, but identification of the lower sacral roots requires some objective means of measuring sphincteric function.
James et al utilized intraoperative external anal sphincter muscle EMG in 10 patients with spinal dysraphism: four patients with tethered cord syndrome, three with lipomeningocele, and three with other miscellaneous diagnoses. These children ranged in age from 3 weeks to 15 years. Using general anesthesia with the patient in a prone position, an anal plug or catheter containing an electrode or needle electrodes was placed in the anal sphincter muscle for recording purposes. During dissection, tissues suspected of containing nerve roots close to the conus were stimulated. If contraction of the anal sphincter muscle occurred, meticulous care was undertaken to preserve these structures. No patient deteriorated postoperatively and two noted improvement.

Pang modified this technique by directly recording the "squeeze pressure" using a pressure-sensitive balloon inserted into the anal canal. There is a rationale for using "squeeze pressure": it had earlier been noted that there is a direct relationship between anal sphincter muscle integrated EMG and anal canal pressure measurements with an anal balloon. Pang used a double-lumen balloon catheter ordinarily used for intraluminal angioplasty. The balloon does not deform at high pressures so there is a high degree of sensitivity. The stimulation was done with a disposable monopolar nerve locator stimulator ring and three current intensities: 0.5 (usually sufficient for infants and small children), 1.0, and 2.0 mA. If a sacral root was stimulated and the external anal sphincter muscle contracted, the combined stimulus artefact and spike wave on the pressure tracing was easily recorded. During stimulation, the cerebrospinal fluid had to be continuously suctioned to prevent current dispersion. Pressure responses with unilateral S2, S3, or S4 root stimulation generally generated pressures >40-75 torr, even in plantar flexion of the foot without pressure change. Stimulation of the filum terminale and nonneural tissues always produced stimulus artefact without a pressure wave. Pang found this technique useful in several circumstances: 1) identifying sacral roots embedded in intradural lipoma; 2) identifying the junction between functional conus and intramedullary lipoma; 3) differentiating an elongated conus medullaris from a thickened filum terminale; and 4) identifying thickened adhesions (from previous myelomeningocele repair) and sacral roots. The more severely impaired the child or the more complex the disorder, the more the monitoring is needed to prevent nerve root injury, but also the more difficult it is to get satisfactory tracings because the sphincter muscles are paralyzed, and arachnoiditis makes the recording procedures technically much more difficult. In summary, the S2, S3, and S4 roots, and the conus can be differentiated from the S1 and lumbar roots, the filum, lipoma, fibrous adhesions, and other nonfunctional fibroneural bands.

Summary

There are several neurophysiological techniques available to the clinician to aid in the diagnosis and management of the patient with tethered cord syndrome: SSEP-PTN, SSEPPuN, urodynamics, and EMG. An understanding of the developmental anatomy, the functional anatomy of evoked potentials, EMG, and urodynamics enhance the ability to care for these patients