01. Sacral monitoring

15. Neurophysiolocigal monitoring in tethered cord syndrome.

Numerous reports of neurological deterioration in untreated cases of tethered cord syndrome (TCS) have essentially silenced dispute about the need for early diagnosis and operative management of this condition. Although much remains to be learned about TCS, further work has generated new information concerning the embryogenesis, pathophysiology, radiologic diagnosis. intraopera­tive monitoring, and surgical aspects of this syndrome.

The human spinal cord is formed by two distinct processes. The cephalic cord is formed by an orderly sequence of dorsal flexion, approximation, and fusion of the neural folds to produce a neural tube, in a process called primary neurulation. The caudal cord is formed by a much less orderly process called secondary neurulation, involving aggregation of neuroepithelial cells, canalization within this cell mass, and subsequent degeneration of these cells. Primary neurulation begins with dorsal bending of the neural plate about a median hinge point (MHP) that is anchored to the underlying notochord. The neural folds thus formed, consisting of neuroectoderm still attached to the cutaneous ectoderm, elevate and then converge dorsolaterally toward the dorsal midline.

Formation of the MHP, or floor plate, depends on the inductive influence of the adjacent notochord. The MHP provides solid anchorage for the bending of the neural folds and determines the dorsal direction of the bending. This dorsal direction is fixed when the MHP transforms its columnar neuroepithelial cells (M cells) into wedge-shaped cells with a bulbous base on the side near the notochord and a tapering apical process facing the future central canal, chiefly by causing the M cell nuclei to accumulate in a basal location. The M cells also become much shorter than the cells of the lateral neural plates (L cells), and so further specify the dorsal direction of neural fold elevation.

The actual elevation and later convergence of the neural folds are driven by forces both intrinsic and extrinsic to the neuroepithelium. Some intrinsic force is provided by contractile actin-like microfilaments located in the apices of the neuroepithelial cells. Because these actin-like microfilaments are concentrated near the floor plate and at the lateral midpoint of the two halves of the neural plate, their contraction is responsible for the dorsolateral furrowing of the neural folds that leads to convergence.  Extrinsic forces are partly provided by medial sliding of the surface ectoderm, carrying with it the neural folds, and partly by expansion, condensation, and elongation of the paraxial mesoderm perpendicular to the long axis of the neural plate.

As the two neural folds approach each other, the opposing inner surfaces of the neuroepithelium show blebs and filopodia that are believed to assist in cell-cell recognition. adhesion, and inhibition, steps that culminate in structural fusion of the neural folds to form a neural tube. Surface coat material largely comprised of glycosaminoglycans (GAGs) are found in the cell membranes of the "bleb cells" of the neural folds, and are thought to be involved in cell-cell recognition and adhesion.  Disruption by tunicamysin of the incorporation of N-acetylglycosamine into the GAG molecules, exposure of mouse embryos to phospholipase C, which removes surface-coat glycoproteins, and treatment of rat embryos with enzymes that interfere with the synthesis of chondroitin sulphate (a GAG) have all been shown to produce embryos in which the neural plate completed flexion and opposition but failed to achieve fusion.

The cutaneous ectoderm remains attached to the neuroectoderm during flexion and early fusion of the neural fold. It then separates from the neuroectoderm in a critical event termed disjunction. Because disjunction normally occurs only after neural fold fusion has been completed. the medially migrating mesenchyme ventrolateral to the neural folds is entirely excluded from the neural groove and the ependymal (apical) side of the neural tube.

Primary neurulation ends with closure of the caudal neuropore at stage 12 [O'Rahilly staging, postovulatory day (POD) 25-27]. At that time, the primitive streak has almost completely regressed, and its activity is being replaced by the caudal eminence, a mass of pluripotent tissue that will ultimately give rise to all the various tissues of the caudal embryo, including the neural material for the caudal spinal cord, caudal neural crest cells, caudal notochord, somites caudal to somite 30, caudal mesenchyme, and hindgut. The production of the caudal spinal cord by secondary neurulation thus begins with the laying down of neural tissues from the caudal eminence as the solid neural cord. In the chick embryo, this results in the formation of the so-called medullary cord, which is composed of a dense outer cell layer and a loose inner cell layer. Cavitation occurs between these two layers and produces multiple lumina within the medullary cord. The inner cells are eventually resorbed, and the multiple small lumina, lined only by the outer cells, coalesce to form a single, large central lumen. This secondary neural tube is initially not continuous with the primary neural tube but lies slightly dorsal to it in an overlap zone. The tubes fuse in a final step.

Unlike the chick, the secondary neural tube in the mouse is always in direct continuity with the primary neural tube. It grows caudally from the caudal neuropore by accruing cells from the neural cord to form a medullary rosette. Moreover, the lumen of this tube in the mouse is single rather than multiple and is always continuous with the central canal of the primary neural tube. There is therefore no overlap zone or fusion site. Caudal growth of the secondary neural tube in the mouse occurs by the progressive cavitation of the medullary rosette and the recruitment of additional cells from the caudal eminence.

The characteristics of secondary neurulation and the integration of the neural cord with the primary neural tube are incompletely understood in the human. Mϋller and O'Rahilly and Schoen­wolf described human specimens containing a single lumen (ventriculus terminalis) within the secondary neural tube continuous with the central canal of the primary neural tube without an overlap zone, that is, resembling the situation of the mouse. In contrast, human secondary neurulation as described by Lemire and colleagues and Bolli  more closely resembles that of the chick. Regardless of the exact mechanism of formation (condensation) and subsequent canalization of the secondary neural tube, however, the final stage of caudal spinal cord formation is known to involve a retrogressive differentiation of structures formed during condensation and canalization, resulting in regression and complete disappearance of the embryonic tail. Atrophy of the caudal neural tube is responsible for the formation of the filum terminale, which connects the future conus medullaris (containing; for a while, the ventriculus terminalis) with the coccygeal medullary vestige, a mixed tissue containing ependymal cells, neurons, and glial cells embedded in fibrofatty tissue in the general area between the coccyx and the end of the dural sac.

There is much controversy surrounding the exact boundary on the spinal cord separating the portions formed by primary neurulation and secondary neurulation. This boundary zone has been claimed to lie almost anywhere between the L2 cord segment and the tip of the conus. The site of closure of the caudal neuropore - the event that unequivocally marks the end of primary neurulation-has been estimated to be opposite somites 30/31. This corresponds to the spinal ganglion and vertebral level of S1-S2. It is therefore logical to assume that, in the human, the spinal cord cephalad to the S2 medullary segment is formed by primary neurulation, while the lower sacral segments and filum terminale are formed by secondary neurulation.

Spinal cord lipomas have been classified into three types: dorsal, transitional, and terminal. Both the dorsal and transitional lipomas have an intradural part that blends with the substance of the spinal cord and an extramedullary part in the form of a fibrofatty stalk that fuses with the subcutaneous adipose layer through a dorsal midline defect in the dura, lumbodorsal fascia, and neural arches. In dorsal lipomas, the fibrofatty stalk inserts into the dorsal surface of the cord in one isolated, segmental region only; the cord caudal to the lipoma is completely normal and is covered by normal dura. In transitional lipomas, the rostral portion of the fibrofatty stalk inserts on the dorsal surface of the cord as in the dorsal type, but the caudal portion of the fibrofatty stalk blends with the terminal conus all the way to its junction with the filum. There is, therefore, no normal cord caudal to the transitional lipoma, and the entire distal thecal sac is penetrated by this stalk.

Terminal lipomas, .on the other hand, insert into the end of the conus without blending with the spinal cord or its root entry zones. All the sacral nerve roots leave the conus rostral to the lipoma, and in most cases the conus itself looks normal. The dural sac and the dorsal myofascial layers are intact. The lipoma either replaces the filum completely or is separated from the conus tip by a short, thickened filum.

It is clear from the above descriptions that dorsal and transitional lipomas belong in one group and share many structural similarities. The dorsal midline location of their fibrofatty stalk and the involvement of the lumbosacral spinal cord also indicate that they both result from faulty primary neurulation. In the case of the transitional lipoma, the involvement of the terminal conus by the caudal portion of the fatty stalk suggests that secondary neurulation may have been affected also. perhaps as a subsidiary target. Terminal lipomas, in contrast. appear to be an entity distinct from the other two types, and, judging from the fully neurulated lumbosacral spinal cord and intact dural sac, terminal lipomas almost certainly result from faulty secondary neurulation.

Almost all spinal cord lipomas are associated with a low-lying conus medullaris. This fact has given rise to the precept that lipomas produce symptoms by tethering the cord. In embryology parlance, this means that the early ascent of the embryonic spinal cord has been held back. In the embryo, a progressive disparity develops between the anatomic levels of the spinal cord and those of the vertebral column as a result of the faster growth rate of the latter. The caudal end of the cord therefore ascends gradually from opposite the coccyx in the 30-mm human embryo to the L3 level at the time of birth and finally comes to lie opposite the L1-L2 junction after the third postnatal year. The spinal dura more or less retains its original vertebral attachment, and rises only from the S4-S5 level in the young embryo to the S3 level in the adult. Proper ascent of the cord therefore requires a well-formed neural tube and a smooth pia-arachnoid covering. If during early development a dorsal defect exists in either the dura (duralschisis) or the neural tube (myeloschisis), mesodermal elements from the surrounding mesenchyme will enter the dural sac and form attachments with the sliding neural tube to result in its entrapment. Depending on the timing of the myeloschisis in relation to the stage of primary neurulation, ascent of the conus could be arrested anywhere between the coccyx and the L3 vertebral level. Since neural fold fusion occurs like a "closing zipper" from rostral to caudal in the caudal neural plate, the later the mesenchymal invasion, the lower the mesenchyme will attach on the spinal cord. This theory therefore features a fundamental defect in the closure of the neural tube during primary neurulation (secondary neurulation does not involve dorsal neural fold closure), and thus applies only to the dorsal and transitional lipomas. It is compatible with the observation that these two lipoma types are always associated with defective dorsal structures.

Several theories have been proposed to explain the embryologic error leading to the mesodermal invasion of the neural tube in the genesis of the lipomatous stalk. MeLone and co-workers think that the error consists in premature disjunction between the cutaneous and neural ectoderms i.e., the separation of one from the other occurs before the converging neural folds fuse with each other. This allows the paraxial mesenchyme, now being pushed dorsomedially by the enlarging lateral somitic mesoderm, to roll over the still gaping neural folds and enter the central canal. Once contact between mesenchyme and apical (ependymal) neuroectoderm is made, further closure of the neural tube is permanently prevented, and a segmental dorsal myeloschisis is created.

An alternative theory championed by Marin-Padilla claims that the embryonic error in dorsal and transitional lipomas involves instead a primary insufficiency of the paraxial mesoderm in providing the chief extrinsic forces that normally cause dorsal bending and convergence of the neural folds. This insufficiency results in a delay in fusion of the neural folds. If ectodermal disjunction is an epochal event that has its own set time independent of the state of neural tube fusion, it would then occur at its pre-set time and could precede neural fold fusion in the case of a delay in fusion. The result will be similar to premature disjunction. Alternatively, faulty fusion of the neural folds owing to metabolic disturbance of the cell membrane-bound glycosaminoglycans could like­wise reverse the temporal relationship between disjunction and neural fold fusion and expose the central canal of the neural tube to the invading mesenchyme.

Experimental studies show that mesenchymal derivatives around the neural tube form according to the inductive properties of the adjacent neuroectoderm. McLone and Naidich postulate that the inner (ependymal) surface of the myeloschistic neural tube is capable of inducing the misplaced intramedullary mesenchyme to form fat, smooth and striated muscle, and collagen. The outer (basal) surface of the neural tube, in contrast, induces the surrounding (extramedullary) mesenchyme to form meninges, but no dura can now form over the dorsal opened portion of the neural tube. The dural defect therefore begins exactly at the junction between neural tissue and intramedullary lipoma. Through this dorsal dural defect, the intramedullary lipoma links up with the extramedullary, extracanalicular adipose tissue to complete a fibrofatty stalk by which the cord is tethered to the subcutaneous tissues. In like manner, dorsal deficiencies in the overlying myofascial layer (from myotomal mesoderm) and neural arch (from scleromesoderm) also neatly surround the lipomatous stalk. Within the neural tube, the intramedullary fat and smooth muscles blend intimately with the developing neural tissues of the alar and basal plates, imparting an engorged and distorted appearance to the cord several segments cephalad to the attachment of the fibrofatty stalk.

Since the dorsal root ganglions develop from neural crest cells that are, in turn, derived from the outer surface of the neural fold lateral to its prospective fusion site, the dorsal nerve roots at the site of the lipomatous stalk grow outward ventrolateral to, and never traverse, the intracanalicular fat. The dorsal root entry zone must correspondingly be lateral to, albeit very near, the exact place where the dural margin meets the junction between functional cord and fibrofatty stalk (the "fusion line"). In the meantime, the cutaneous ectoderm, long detached from the neuroectoderm, heals over in the dorsal midline to form wholesome skin over the subcutaneous component of the lipoma.

In many ways, the dorsal lipoma is a perfect embodiment of the "mistimed disjunction" postulate, and by the theory's own logic the embryologic error can occur only during primary neurulation. Its fibrofatty stalk almost always involves the lumbar cord segments, i.e., the part of the neural tube that arises through primary neurulation. The part that forms from secondary neurulation, the terminal conus, always appears quite well developed. Furthermore, the anatomy of dorsal lipomas suggests a segmental abnormality of neural tube fusion in the sequential caudal propagation of primary neural tube closure; normal closure takes place immediately following the abnormal event to result in a normal cord caudal to the dorsal lipomatous stalk. This "square­pulse" nature of the abnormal event is illustrated by the sharply circumscribed line of fusion between dorsal lipoma, cord, and pia­arachnoid that can easily be traced circumferentially about the site of emergence of the lipomatous stalk through an equally "crisp" dorsal dural defect. It is interesting to note that dorsal lipomas represent less than 10 percent of all spinal cord lipomas, an incidence roughly similar to that of the rare segmental myelomeningocele. The latter contains a "suspended" segmental unneurulated placode but a neurulated spinal cord caudal to the defect, and thus represents the extreme form of segmental primary neurulation failure.

In transitional lipomas, the myeloschisis involves much more than an isolated segment of the primary neural tube, suggesting that the abnormal event probably does not occur in a "square­pulse" fashion, Even though its rostral part resembles the dorsal lipoma, its total involvement of the remainder of the caudal spinal cord indicates that secondary neurulation as well as the rest of primary neurulation is disturbed by the mesodermal invasion, This conclusion is supported by the observation that, while in some transitional lipomas the filum remains a distinct structure adjacent to the most distal part of the fatty stalk, in most cases the filum is incorporated into the distal lipoma, Also, empty spaces resembling the terminal ventricle of the secondary neural tube are frequently present in the caudal part of the transitional lipoma, Finally, while the fat in the rostral part of a transitional lesion is always on the dorsal aspect of the cord and respects the disjunction-myeloschisis theory of faulty primary neurulation, the distal part of the transitional lipoma sometimes involves both the dorsal and ventral aspects of the conus, a situation compatible with abnormal mesenchymal inclusion during the much less orderly beginning (condensation) phase of secondary neurulation. It is not uncommon to identify intramedullary fat three or four segments rostral to the fusion line in a dorsal or transitional lipoma, This suggests that the intramedullary mesenchyme is able to migrate along the central canal of the neural tube after having gained entrance through the dorsal myeloschisis, It is thus possible that the mesenchyme also travels caudally across the boundary zone from the primary to the secondary neural canal, especially if human secondary neurulation resembles that of the mouse, in which the two neural canals are always in continuity, In fact, the hypothesis that the rostral part of the transitional lipoma arises from aberrant primary neurulation (involving only the dorsal cord) and the caudal lipoma arises from abnormal condensation of the secondary neural cord (affecting both dorsal and ventral aspects of the conus) furnishes one explanation of the rostral-caudal obliquity of the lipoma-cord interface. 

Ample evidence suggests that terminal lipomas result from abnormal secondary rather than primary neurulation. First, the parts of the cord formed from primary neurulation, including the lumbar and upper sacral segments, are never involved, Second, a terminal lipoma is never associated with a dorsal myeloschisis or a dorsal dural defect, both hallmarks of failed (primary) neural fold fusion, Third, the lipoma itself is either part of a thickened filum or replaces the filum altogether, which places the embryogenetic abnormality within the time-frame of retrogressive differentiation of the secondary neural cord, Fourth, it is common to find fluid-filled spaces in these lipomas that resemble the ventriculus terminalis of the secondary neural cord, Fifth, scattered nests of ependymal cells, glia, neurons, and ganglion cells are frequently found throughout the lipoma, consistent with the notion that the whole mass is a "regressive" spinal -cord, Finally, large clumps of fat are frequently found in the thickened filum terminale, a lesion most likely caused by defective secondary neurulation, suggesting that the "uncomplicated" thick filum and the large tubular terminal lipoma are closely related lesions on the same pathogenetic continuum.

The normal conus and filum terminale rarely contain large amounts of mesenchymal elements, Mesenchymal derivation does not usually occur in large measure during the normal events (condensation, canalization, and retrogressive differentiation) of secondary neurulation, The fact that the distal conus remains fat-free in cases of thickened filum and terminal lipoma suggests that the pathologic mesenchymal derivation did not occur during condensation, Conversely, the abnormal filum and terminal lipoma often contain more (disorganized) cord elements and ependymal tubules than the normal filum, as if the retrogressive process in these situations has been incomplete or "ineffectual." Furthermore, excessive amounts of nonadipose mesenchymal derivatives such as cartilage, bone, and fibrous septi are found in many terminal lipomas. Thus, the evidence concerning the embryogenesis of terminal lipomas points to an abnormal retrogressive differentiation of the secondary neural cord, perhaps brought about by an aberrant accumulation of mesenchymal precursor cells from the pluripotent cell pool of the caudal eminence, Alternatively, defective retrogressive changes due to other causes could lead to abnormal proliferation of cells along mesenchymal lines.

Clinical and intraoperative evidence suggests that the neurological lesion in the TCS is within the spinal cord and not in the lumbosacral nerve roots. For example, the pain in adult TCS patients is characteristically poorly localized and dysesthetic and is frequently complicated by after-discharge phenomena long after a stretching insult has been applied to the spine. This "central" type of pain bears strong resemblance to the pain seen in intrinsic cord tumors, The L'hermitte phenomenon, which is strongly associated with dorsal column pathology, is also frequently encountered in TCS. Finally, there seems to be a correlation between the severity of the central pain and objective signs of corticospinal tract injury, such as spasticity, clonus, and the Babinski response,

Intraoperative findings also suggest that the spinal cord rather than the nerve roots is under tension in TCS. Regardless of the tethering lesion, the part of the cord adjacent to the point of entrapment is always palpably taut. For example, the conus associated with a thickened filum is invariably bowed tightly against the back of the thecal sac, and is sometimes even visibly elongated so that the normal abrupt transition between conus and filum is obscured. In dorsal and transitional lipomas, the conus is often tautly pulled toward the site of insertion of the fibrofatty stalk. In split cord malformations, tension exerted by the bony septum is easily felt at the reunion site of the hemicords, The nerve roots arising from a low-lying conus frequently course laterally or even cephalad and, indeed, appear more relaxed than usual because their points of origin are brought closer to their exit foramina.

Certain experiments suggest that impairment of oxidative metabolism as well as morphologic deformation of neural tissues is responsible for the neurological dysfunction in TCS. Metabolic activity of neural tissues can be estimated by the reduction-oxidation (redox) ratio of the terminal member of the mitochondrial respiratory chain, cytochrome aa3' Cord ischemia is reflected by an immediate shift of the redox ratio to a more reduced state. Using a noninvasive optical technique utilizing reflection spectro­photometry to measure the redox state of cytochrome aa3, Yamada and co-workers found that both acute and chronic traction of the feline conus produced ischemia in the lower cord segments. Parallel studies performed in patients during surgical release of TCS revealed remarkably similar alterations in the pattern of redox shifts before and after the untethering procedure. Subsequent determination of spinal cord blood flow in cats using the hydrogen clearance method also revealed a close temporal relationship between cytochrome aa3 redox shift and reduction of blood flow to the conus. These authors concluded that traction of the conus decreases local blood flow and reduces the cellular ATP store, which secondarily leads to neurological dysfunction. This hypothesis is supported by the observation that symptoms of TCS may be precipitated by progressive spinal stenosis, a condition known to cause cord ischemia in an otherwise uncomplicated spine.

In addition to inducing ischemia, tethering may produce morphologic deformation of neuronal membranes. Yamada and colleagues found that 5 g of traction on the feline conus overcame the viscoelastic properties of the conus and induced actual elongation of the cord. Neuronal membranes were irreversibly wrinkled and torn. Morphologic changes were maximal at the site of the traction and gradually decreased farther up the cord. These observations may explain why sphincteric disorders are such common initial complaints, whereas signs of L1-L2 sensorimotor denervation are uncommon except when a dorsal lipoma inserts directly on these segments.

Precipitating and Aggravating Factors

Children with spinal cord lipomas tend to show neurological deficits by the age of 2 years. This is slightly younger than in patients with a tight filum, who may remain asymptomatic until late childhood or early adulthood. In part this difference may be due to the often obvious subcutaneous fat lump in children with lipomas, which focuses attention on the child's spinal contour and leg function. Also, the myeloschisis and lipomatous engorgement of the cord are more likely to be associated with significant connatal "fixed" neurological deficits than in the case of a tight filum. After the first year, the deficits then tend to worsen slowly but inexorably.

Occasionally, children with a lipoma or tight filum terminale deteriorate or become symptomatic for the first time during a period of rapid growth. Presumably, the disproportionate lengthening of the vertebral column in relation to the cord accentuates the tension in the conus and precipitates neurological dysfunction. Catastrophic leg weakness and bladder dysfunction have also been reported in adolescents after specific activities that abruptly stretched the spine. These activities include ballet high-kicks; gymnastics, especially cart-wheel jumps; and exercises focusing on knee-chest bends. Medical examination and obstetric procedures in the lithotomy position, and automobile accidents in which the body is thrown into the jack-knife position, have also been implicated. Breig showed that full flexion of the head on the chin is associated with a sudden upward movement of the spinal cord by as much as 2 cm. When full neck flexion is combined with flexion of the lower spine, an already taut conus would be momentarily stretched beyond its physiologic limit and past the tissue's ability to maintain normal function. Breig also theorized that the pons and spinal cord form a single physicomechanical tract that allows for free distribution of longitudinal stresses along its length (the pons-cord tract). In a patient whose cord is already tethered from below, any bony protrusion or bulging disc on the ventral aspect of the spinal cord will deflect this tight tract back­ward and accentuate the tug on the conus, This explains why some patients with asymptomatic tethering lesions develop myelopathy rapidly with mild to moderate lumbar spondylosis. Finally, direct blunt trauma to the subcutaneous component of a lipoma can occasionally precipitate severe pain and unbearable dysesthesia down the legs, followed by accelerated neurological deterioration and a cascade of recurrent symptoms. The sudden deterioration in the tight conus is probably brought about by the shock-wave stresses directed through the subcutaneous connection.

Symptoms and Signs in Children and Adults

Pain

Unlike adults with tethered cord syndrome, who universally suffer from excruciating and unrelenting pain, children with TCS seldom complain of severe pain. The quality of the pain in children is also different from that of adults. In children, the dysesthetic, poorly localized, diffuse pain in the legs, groin, and perineum, or the cataclysmic electric shock along the spine so commonly seen in adults, is rarely encountered. The pain in children is much more confined to the low back, with only occasional radiation to the legs. Sometimes the back pain increases with prolonged bed-rest (unlike discogenic pain, which improves with rest), presumably because the intervertebral discs thicken with absorption of water and because the cervical and lumbar curvatures straighten out on recumbency, in both situations increasing the total length of the spine and placing additional tension on the conus.

The reason for the different pain pattern in children is unknown but probably is more than simply a problem of communication. However, the suffering of pain in young children does sometimes manifest in unusual ways and is expressed in unfamiliar vocabulary. For example, nocturnal pain may be interpreted as night terror when a child suddenly wakes up crying inconsolably; back pain may cause "temper tantrums" when the child is being cuddled, bathed, or generally handled; and dysthetic leg pain or paresthesia is often called "fizzing like Coca-Cola" by the child.

Cutaneous Manifestations

Cutaneous signs provide valuable clues to the underlying tethered cord before the onset of overt and often irreversible neurological deficits, and should be routinely sought in the well­child clinic. The correlation between cutaneous and intradural lesions is so good that the presence of cutaneous signs is sufficient indication for neuroimaging studies.

Almost all children with TCS have some cutaneous stigmata of underlying dysraphism, but less than 50 percent of adults do. Several helpful points can be made about the cutaneous lesions. Mid­line hairy patches are highly correlated with split cord malformations. Over 80 percent of intradural lipomas are associated with cutaneous signs; about two-thirds of these signs are subcutaneous lipomas, and one-third are capillary hemangiomas, dermal pits, or hypertrichosis without a subcutaneous lipoma. If the subcutaneous lipoma is low, the gluteal cleft sometimes veers eccentrically so that one buttock appears larger, or the cleft bifurcates into two deep furrows cradling the fat lump. Dermal dimples can lead into a dermal sinus tract but are just as likely to be associated with other tethering lesions. Features that distinguish a "dysraphic" dimple from an innocuous "terminal" dimple (remnants of the posterior neuropore) are overhanging skin edges around the opening, a high location (above the lowest sacral piece), and emanating hair tufts. Capillary hemangiomas can be midline but frequently extend far laterally. Proboscis-like skin and subcutaneous protrusions akin to residual human tails are almost always associated with cord tethering.

Sensorimotor Deficits

Motor deficits manifest in different ways depending on the age of the child. In infants, the loss of motor neurons is most often seen as atrophy of the buttocks, calves, and the hollows of the feet, although the abundant baby-fat makes this difficult to detect. The affected foot may also be smaller, with exaggerated pedal arches and/or hammer toes. Muscle weakness may be very subtle and difficult to elicit; unilateral weakness is best detected by observing how much spontaneous kicking and flexing the leg and foot occur compared with the opposite side as the infant attempts to right itself when alternatively placed in the prone and supine positions. The deep tendon reflexes are often absent on the affected side; signs of corticospinal tract involvement are extremely uncommon in infants. Insensitivity to pinprick may be confined to the perianal area or noted as an absent anal wink reflex.

In toddlers, lower extremity weakness presents as delayed motor milestones in the legs, regression in gait training, or a wide­based, divergent, wobbly gait, depending on the age of onset of symptoms. This is the age when parents will notice that the child has a clumsier leg, the turning-in or -out of one foot, foot dragging, or neglect of the affected side. Sometimes matching the left and right shoes for the child becomes a problem as the disparity of growth between the feet becomes more prominent.

Discrete muscle weakness localizable to specific cord segments is more likely to be seen in older children. Characteristically, the weakness is bilateral and involves several cord segments. With increasing age, corticospinal tract deficits become superimposed on signs of anterior horn cell injury, as when spasticity and a Babinski response accompany rapidly progressive atrophy. Barry and colleagues suggested that the susceptibility to ischemia of the large-diameter corticospinal fibers increases with the duration of tethering. Long tract signs are seen in 80 percent of adults with tethered cord but are rarely encountered in young children.

Bladder and Bowel Dysfunction

Bladder and bowel dysfunction are difficult to detect in infants. The examiner should specifically ask whether there are dry periods between diaper changes, which gives some idea about the retentive capacity of the bladder neck, and whether the infant ever urinates in a forceful arc, which is a rough measure of detrusor power. Complete absence of detrusor contraction is uncommon in infancy, since even a denervated bladder, as long as its muscular wall is sufficiently contractile and not totally fibrotic, has the ability to expel urine on a reflexive level at the right filling threshold.

In toddlers, the most common bladder symptom is delayed or unsuccessful toilet training. The normal age for full toilet training varies between individuals and cultures, but if control is still unsatisfactory by age 4 or 5 years, a neuropathic bladder should be suspected. Usually incontinence occurs during both day and night: enuresis with perfect daytime control points more to a psychogenic etiology, but exceptions have been reported. The symptoms of urgency, frequency, urge or stress incontinence, poor voluntary control, and post-void dribbling are mainly seen in older children and adults. Frequent urinary tract infections are seen at any age and more in girls than boys.

Several neural pathways are involved in the storage function of the bladder. A voluntary (somatic) pathway consists of corticospinal connections between the frontal cortex and the pudendal nucleus, which permits voluntary interruption of the urinary stream by contraction of the external urethral sphincter (composed of striated muscles). A local (sacral) pelvic-pudendal reflex pathway maintains continence at a subconscious level by conveying information regarding bladder filling through the afferent pelvic nerves to the pudendal nucleus (S2-S4), which then initiates contraction of the external sphincter. In addition, continence also involves a sympathetic pathway via the hypogastric nerves, As the bladder fills, there are increasing afferent signals along both the pelvic and hypogastric sensory fibers, but detrusor contractions prompted by the reflex discharge from the pelvic parasympathetic nuclei, in response to the bladder filling, are blocked by the efferent hypogastric nerves, which normally inhibit parasympathetic outflow at the postganglionic neurons. The hypogastric nerves also directly cause contractions and closure of the bladder neck and proximal urethra (combined to form the "internal" urethral sphincter, composed of smooth muscles) to guard further against urinary flow into the urethra.

Recent evidence suggests that the most important neural pathway concerned with voiding involves the brain stem. Afferent pelvic nerve signals carrying information on bladder filling ascend in the lateral columns of the spinal cord to reach the pontomesencephalic reticular formation ("pontine micturition center"), where they are integrated. When voiding is deemed socially acceptable, efferent discharges descend from the brain stem center to the pudendal nuclei. These signals are inhibitory and therefore result in relaxation of the external urethral sphincter. Similar inhibitory signals to the sympathetic nuclei from the pontine center permit transmission of outflow (from pre- to postganglionic stations) from the pelvic parasympathetic neurons. These neurons appear to have been well apprised of the filling status of the bladder through pelvic sensory inputs and are in a hyperactive state just prior to voiding. Once disinhibited, they initiate detrusor contractions that generate the familiar' 'micturition pressure spikes" on cystometry. Simultaneously, sympathetic inhibition also relaxes the bladder neck and proximal urethra, and urine flow is initiated.

Tethering of the conus rarely causes a pure lesion in any single neural pathway, but rather mixed abnormalities of the parasympathetic, sympathetic, and somatic pathways. Blaivas found that sympathetic innervation was often impaired first in tethered cord. The nonfunctioning internal urethral sphincter causes sagging of the proximal urethra and effacement of the bladder neck on the voiding cystourethrogram, which characteristically leads to post­void dribbling and stress incontinence, both early symptoms of neuropathic bladder. When the parasympathetic pathways are primarily injured, the detrusor mechanism is weakened and the bladder becomes hypotonic and areflexic. Poor bladder emptying leads to the subjective feeling of incomplete voiding, and when the parasympathetic and sympathetic pathways are injured together, patients have both an inability to empty the bladder and incontinence due to sphincteric incompetence. A hypotonic bladder poses little threat to the upper urinary tract, but parasympathetic denervation sometimes results in an areflexic hypertonic bladder, which maintains a high intravesicular pressure and promotes ureterovesicular reflux.

The most severe reflux is encountered with detrusor-sphincter dyssynergia. Normally, efferent discharges from the pudendal nucleus (supplying the external sphincter) and pelvic nucleus (supplying the detrusors) are mutually inhibitory via cross-linking collateral fibers, so that the two target muscles are never simultaneously activated. This delicate and polysynaptic connection is also tightly coordinated by descending inputs from the pontine micturition center. Dyssynergia is seen both with diseases involving the descending pathways and with diseases of the sacral cord itself, although more commonly with the latter. The detrusor contracts involuntarily against uncontrolled spasm or contractions of the external sphincter causing huge rises in the bladder pressure. These patients are partially obstructed by the closed sphincter (pseudocontinent), yet are intermittently incontinent owing to overriding contractions of the detrusor. They have a 50 percent likelihood of developing hydronephrosis.

Foot Deformities

Progressive talipes is one of the most common presenting features in children with TCS. The mildest form of talipes is hammer toes, that is, fixed flexion at the interphalangeal and extension at the metatarsophalangeal joints. Sometimes the entire forefoot shows a valgus or varus drift at the tarsometatarsal articulations. More profound and higher denervation causes exaggeration of the horizontal and vertical arches, hollowing of the instep, and neuromuscular imbalance at the ankle, giving rise to the various combined forms of equinus, calcaneal, varus, and valgus deformities of the foot.

Foot deformities most likely result from neuromuscular imbalance at a time when the tarsal, metatarsal, and phalangeal bones are actively growing and aligning with each other along closely set joint surfaces, that is, during childhood. The formation and orientation of these joint surfaces are affected by forces acting on their respective levers (participating bones). Confusion in this network of forces will cause malalignment to occur. If normal growth and alignment of these joints are undisturbed during the formative years, permanent deformities are unlikely to occur in later life. Adults with a tethered cord who did not have a pre-existing talipes never develop it with the onset of other symptoms.

Spinal Deformities

Progressive scoliosis or kyphosis is seen in approximately one­fourth of children with TCS. As with foot deformities, adults with tethered cord do not develop new scoliosis with onset of other neurological symptoms. However, once pre-existing scoliosis has progressed beyond a certain angle (about 40°), gravity itself will worsen the curvature regardless of age or whether the cord has been adequately untethered. When scoliosis secondary to cord tethering requires surgical correction, the tight conus must first be released before the spine is distracted to avoid catastrophic neurological deterioration.

Trophic Changes

Trophic changes of the lower extremities from sympathetic denervation - for example, smooth, shiny skin; hair loss; nail changes; and nonhealing ulcers in the toes-are seen occasionally, but only in older children. Recurrent osteomyelitis from unhealed ulcers sometimes results in successive amputation of the toes.

Progressive Neurological Deterioration due to Tethering

Both indirect and direct evidence suggests that the likelihood of neurological deterioration in TCS increases with age and ultimately becomes very high. For example, Hoffman divided the neurological status of 73 children with lipomeningocele into five grades, ranging from grade 0 for children who were neurologically normal to grade 5 for children unable to ambulate. He found that most of the infants had either grade 0 or grade I status, whereas the proportions of children with higher grades increased progressively with age. Almost all the teenagers were grade 3 or 4. This close correlation between the severity of disability and the age at diagnosis indirectly supports the argument that TCS is a progressive disease. Hoffman further reviewed 24 individuals who had undergone inadequate childhood operations for TCS and found that most of them had unequivocally deteriorated over a period of 1 to 18 years. Certainly, many adult TCS patients who had enjoyed years of perfect health deteriorate precipitously after a minor automobile accident, a fall, or even a bout of vigorous exercise. In these instances, successful untethering "after the fact" does not guarantee full neurological recovery. It is therefore advisable to treat most cases of TCS as soon as the diagnosis is made.

Plain Spine Film

Magnetic resonance imaging (MRI) is performed on every patient with TCS, plain spine films are no longer necessary in the initial workup. In fact, plain films of infants are often misleading because of poor mineralization of the bones and because large laminar defects can be obscured by rectal gas and feces. Plain films are ordered, however, for two other uses. Serial total body stand-up spine films are important for following the course of a patient's scoliosis. Also, an anteroposterior spine film with a small metal marker taped to the patient's back is helpful in identifying the intended bony levels for laminectomy.

Magnetic Resonance Imaging

In MRI, the high-intensity fat signal stands out prominently against spinal cord tissue and cerebrospinal fluid (CSF) and thus displays the extent and location of lipomas to advantage. Sagittal MRI is particularly effective in showing the suspended fibrofatty stalk of the dorsal lipoma coursing in a caudal-rostral direction across the dorsal thecal sac, with normal spinal cord caudal to the stalk. Likewise, the terminal lipoma and the low-lying, stretched­out conus are magnificently displayed. For a complex transitional lipoma, the sagittal MRI shows the longitudinal length and extent of the fat well but often not its relationship with the cord. For this, the axial image is important, to show the dorsal location of the lipoma and also which half of the cord is more involved with the fatty attachment. The fat signal is often seen to encroach more ventrally on the side of the cord that is more severely involved and to rotate the more normal side in the opposite direction.

MRI is also excellent for detecting fat within a thickened filum. The demonstration of this fat is diagnostic of a pathologic filum terminale. However, MRI often misses a dermal sinus tract and even a small dermoid cyst, and it is inadequate for displaying fine details in a complex split cord malformation.

Computed Tomographic Myelography

CTM is valuable for delineating the topographic details of dorsal and complex transitional lipomas. CT density differences allow fat to be distinguished from cord tissue, CSF, and fibromuscular elements in the extracanalicular planes. CT can thus differentiate the attachment of the fatty stalk, where there is no CSF, from the intramedullary portion of the lipoma, which is completely surrounded by CSF. This information enables the surgeon to limit bone removal to only the point where the cord is anchored and thus to minimize the risk of post-laminectomy spinal deformity.

Thin-slice CTM using 1.5 mm contiguous axial slices displayed in bone algorithms often reveals details of the nerve roots as they exit the cord. This is useful to distinguish terminal lipomas from low transitional lesions. In the latter, the caudal extreme of the lipoma lies dorsal to a small but important ventral wedge of neural placode. It is sometimes difficult to distinguish this condition from a terminal lipoma on MRI. The presence of nerve roots coursing horizontally from the ventral surface of the fatty mass on CTM indicates that part of the mass is functional spinal cord, which necessarily means that the lipoma is of the transitional type. In a terminal lipoma. the only nerve roots shown at the level of the lipoma should be those arising rostral to the fat, which therefore are displayed end-on as small dots clustered around the fatty core.

Last. the bony anatomy of the spinous processes and neural arches often helps to identify the exact level of the tethering lesion. Matching the intraoperative findings with the bony anatomy on CT allows the surgeon to determine the level and extent of bone removal. MRI is notoriously ambiguous about bony structures.

MRI is preferable as a screening test for all suspected cases of TCS. If the MRI is unequivocally normal or shows an uncomplicated lesion such as a thick filum and terminal lipoma, then no further radiologic study is ordered. If an anomaly cannot be ruled out absolutely, or if the lesion is so complicated that finer anatomic definition is needed for planning surgical strategy, a CTM with contrast is performed. Since MRI has sagittal capabilities unmatched by CTM. ideally both MRI and CTM should be used in combination when dealing with difficult dysraphic lesions.

Thickened fila and lipomas probably make up over 70 percent of all tethering lesions. Three other, less common. tethering lesions are currently being critically re-evaluated as to their embryogenesis, anatomic variations, and clinical significance. These are split­cord malformations, cervical myelomeningocele. and tethered cord with normal conus position.

Much controversy still exists concerning the nomenclature, morphology, embryogenesis, and clinical significance of double spinal cord malformations. Traditionally, the term diastematomyelia has been used for the type of double cord malformation in which the hemicords reside in separate (hence double) dural sacs transfixed by a midline bony spur. Each hemicord, being a true half-cord, is supposed to contain only lateral nerve roots and no paramedian roots. Diplomyelia, in contrast, has been used for the type in which the double cords reside in a single sac with no midline elements. Each half is thought to be a complete cord and therefore to possess paramedian nerve roots. These two forms were believed to be unrelated and to arise from different embryogenetic mechanisms; the bone spur in diastematomyelia suggests that it results from mesodermal invasion of the neural tube, whereas diplomyelia is somewhat nebulously classified as a form of segmental twinning. Most important, because the cleft cord of diastematomyelia looks transfixed by the bone spur, most agreed that it was a form of spinal cord tethering, as opposed to the two cords in diplomyelia, which appear to lie unimpaled within a capacious sac and were thus not usually thought to be tethered.

The experience with double cord malformation does not support these traditional views. Based on 50 clinical and 6 autopsy cases, a unified theory of embryogenesis has been proposed which holds that all double cord malformations result from a common ontogenetic error regardless of whether the half-cords occupy a single or double dural sac, whether paramedian nerve roots are present, or whether a midline spur is identified. The forms traditionally called "diastematomyelia,, and "diplomyelia" have been shown to be variants of the same basic malformation and not separate entities. Because these two forms share so many overlapping features, the terms diastematomyelia and diplomyelia were abandoned in favour of the term split cord malformation (SCM) to denote the entire spectrum of double cord malformations.

The unified theory proposes that all SCMs originate from one basic embryologic error: the formation of an abnormal fistula through the midline embryonic disc that maintains communication between yolk sac and amnion and makes possible continued contact between ectoderm and endoderm. This abnormal fistula necessarily causes regional "splitting" of the notochord and the overlying neural plate. The location of this abnormal fistula is variable, but it most likely lies rostral to Hensen's node, since the primitive pit marks the end of primary neurulation, and most known SCMs involve cord segments rostral to the distal conus. The origin of this abnormal fistula is uncertain but probably does not involve accessory openings from the original notochordal canal as Bremer proposed. (Bremer conceived of an "accessory neurenteric canal" arising from a perforation of the roof of the primitive neurenteric canal that normally exists within the notochordal process for only 2 to 8 days, during which time the yolk sac is in communication with the amnionic cavity. He thought that the rare dorsal intestinal fistula results from herniation of the yolk sac endoderm through this accessory neurenteric canal.) It is more likely that a developmental abnormality occurs in Hensen's node or the primitive streak during gastrulation that results in a failure of prospective notochordal cells to achieve midline integration following their ingress through the primitive pit. Doubling of the notochordal process may therefore occur over a variable length, allowing the simultaneously lengthening ectoderm and endoderm to form an adhesion across this central fenestration in the notochord. Alternatively, prospective neuroepithelial cells flanking both sides of the primitive streak may fail to integrate to form a single, mid­line neural plate. The resulting hemineural plates remain separate and develop independently into hemicords, each affected by its own set of local regulating influences.

Although the cellular mechanisms responsible for such an embryonic disorder are unknown, experimentally induced perturbations of Hensen's node and the primitive streak during gastrulation have resulted in embryos with double axial structures. Abercrombie and Bellairs removed the whole Hensen's node, including the primitive pit, at the stage when the primitive streak is at full length, and inserted a graft from the posterior third of the primitive streak into the defect. More than half the embryos showed doubling of the axis, with two neural plates and two notochords, complete with clusters of mesoderm in between the double notochords. However, these double-axis embryos differed from human split cord malformations in that all of them were essentially duplicitas anterior monsters with two brains and two upper spinal cords. In none was there a segmental splitting of the more caudal neural tube. Also, none of the double neural plates in the experimental embryos neurulated. Finally, the doubled axes were extremely malformed, with chaotically arranged paraxial mesoderm and poorly regulated anterior neural plates. Despite these differences, such experiments suggest that doubling of axial structures can result from tampering with primitive streak blastoderm.

The formation of the abnormal fistula (the basic error) is the committal step common to all double cord malformations, but further developments around the fistula must occur to explain all the other features of SCMs. As the abnormal adhesion between ectoderm and endoderm is forming, surrounding multi potential mesenchyme condenses around it to form an endomesenchymal tract. This endomesenchymal tract not only permanently bisects the notochord but also forces each overlying hemineural plate to neurulate against its own heminotochord in a severely compromised manner. The basic malformation therefore consists of two heminotochords and two hemineural plates separated by a midline tract containing ectoderm, mesenchyme, and endoderm. Further evolution of this basic form into the full-grown malformation depends on four factors: (I) the ability of the heminotochords and hemicords to heal around the endomesenchymal tract; (2) the developmental fates of the three germ elements within the tract; (3) the variable extent to which the endomesenchymal tract persists; and (4) the interaction between the heminotochord and the hemineural plates during neurulation.

Healing of the notochord and neural plate

Because the neural tube lies dorsal to the notochord, the unified theory predicts that in all cases of split spinal cord, the notochord must also have been traversed at some time by the abnormal fistula. The appearance of the full-grown vertebral body therefore depends on the ability of the bisected notochord to heal around the fistula. If midline healing is inhibited by persistence of the fistula, the body will remain completely bifid. If partial healing has taken place, the vertebral body will be a single block but will show a sagittal tract on CT outlining the fistula remnant. When no bifid body or sagittal tract is identified, the vertebral bodies at the level of the SCM are sometimes broader transversely than at adjacent levels, suggesting that the notochord was once wedged as under by the fistula even though later invasion by sclerotomal mesenchyme filled in the central void. Eighty-one percent of the patients in one SCM series have vertebral body abnormalities. Partial healing of the hemineural plates probably results in the so-called cleft spinal cord: a single bilobed structure with double central canals and a deeply indented midsection.

Developmental fates of the endomesenchymal tract and classification of SCM

The usual fate of the endoderm within the endomesenchymal tract is attrition, probably owing to the absence of specific inducer molecules for intestinal epithelia near the notochord and neural tube. In the rare instance when the endoderm does undergo differentiation, a cyst lined with gut or respiratory epithelium (neurenteric cyst) will form in close association with the split spinal cord. The cyst is most often found loosely attached to the hemicords within the median cleft, but it may be almost entirely intramedullary within one of the hemicords. In either case, part of the cyst wall maintains continuity with other derivatives of the endomesenchymal tract, that is, with the median septum.

Unlike the case of the endoderm, some derivative of the mesenchymal component almost always persists in the midline cleft of the mature SCM. The manner in which the mesenchymal derivatives form meninges around the hemicords conveniently classifies SCMs into two major types. According to Sensenig, the normal spinal cord dura mater develops from the meninx primitiva. This is a loose cluster of mesodermal cells first seen as a distinct zone ventrolateral to the neural tube in the stage XV embryo (Streeter's horizon; 30 days gestational age). It supplies meninx precursor cells that first line up ventral to the neural tube and then gradually migrate dorsally over the neural tube, completely enveloping it at stage XXIII (Streeter's horizon). Since the abnormal midline fistula forms during the early lengthening of the notochordal process, around the third embryonic week, the mesenchyme condensing around it shortly afterward may well pick up precursor cells from the meninx primitiva. The final transition from meninx primitiva to dura mater depends on induction by the adjacent neural tube. The precursor cells within the midline tract, lying next to the medial aspect of the hemicords, will develop into a median layer indistinguishable from the normal dura growing around the lateral aspects of the hemicords, thereby completing a separate dural tube for each hemicord. In addition, since the meninx precursors are also thought to form the neural arch of the normal vertebra and therefore possess dual fibrogenic­sclerogenic capability, they are the likely source of the midline bone spur.

In addition to blood vessels, muscle tissue has also been found either within the dural sleeve of a type I lesion or embedded in the median fibrous septum of a type II SCM. Lipomas may also be found in the median cleft or just dorsal to the hemicords. Unlike the usual dysraphic lipoma, which is associated with incomplete neurulation and therefore blends intimately with spinal cord tissue, the lipomas found in SCMs are usually loosely attached to the pial surface of the hemicords.

Paramedian nerve roots, myelomeningocele manqué

Much has been made of the presence of paramedian nerve roots as definitive evidence of true cord duplication and therefore the distinguishing feature between diastematomyelia and diplomyelia. The unified theory proposes that the existence of paramedian dorsal roots depends not on twinning but on the involvement of the neural crest cells in the early phases of endomesenchymal tract formation. They should be, and are in fact, found in equal frequency in both types of SCM.

In the normal human embryo, neural crest cells occupy a mid­line position between the dorsal aspect of the neural tube and the surface ectoderm as early as POD 32. Beginning around day 28, these neural crest cells migrate ventrally to reach the intervertebral foramina, where they congregate into spinal ganglia. Thus, the endomesenchymal tract forms at a time (around day 20) when the neural crest cells still occupy a dorsal midline location. With splitting of the neural plate, it is conceivable that some of the neural crest cells may be entangled in the endomesenchymal tract and may subsequently separate from the main cluster. The central processes of neural crest cells are "obligated" to interface with the developing cell groups within the spinal cord, regardless of whether the cell bodies occupy an aberrant location. With the entrapped neural crest cells now clinging to the endomesenchymal tract, their central processes must stretch between the midline mesenchymal structures and the dorsomedial surface of the hemicords. These paramedian dorsal roots are therefore "extra" to the normal set of lateral dorsal roots from the spinal ganglion.

Paramedian dorsal roots are found about equally often in the two types of SCM (in about 75 percent); it is reasonable to expect type I and type II endomesenchymal tracts to have about an equal chance of entrapping neural crest cells. In type I SCMs, the usually bilateral paramedian dorsal roots stretch out horizontally from the medial aspect of the hemicords and end blindly within the median dural sleeve. Mature ganglion cells are seen buried in the dense fibrous tissues of the median dural sleeves, and they are directly traceable to the paramedian dorsal roots. In type II SCMs, the paramedian dorsal roots usually project dorsally away from the dorsal surfaces of the hemicords before merging with the thin median fibrous septum just before it attaches to the dorsal dura. The ganglion cells are found within the median septum near the dura.

In 1972, James and Lassman coined the term myelomeningocele manqué to describe a mixed bundle of nerve roots, fibrous bands, and blood vessels arising from an aberrant dorsal location of the spinal cord and attaching itself to the dorsal dura. Such fibroneurovascular bundles are found in over 60 percent of both types of SCM. There are usually multiple bundles per split cord, all originating from the paramedian dorsal surface of the hemicords, projecting dorsally, and attaching to or penetrating the dorsal dura at a point caudal to their cord attachment. In rare instances, the myelomeningocele manqué represents the only remnant of the endomesenchymal tract and the main tethering bands in a type II lesion. The tuft of extradural tissue attached to the intradural part of the myelomeningocele manqué characteristically contains nests of ganglion cells and tangles of large blood vessels.

The frequent association of myelomeningocele manqué with SCM is not surprising if one recalls that the original endomesenchymal tract reaches beyond the level of the dorsal dura. The neural crest cells in the tract may also be brought into the extradural space where they form ganglion cells. Their central processes then join the fibrous bands and blood vessels derived from the midline mesenchyme to form the fibroneurovascular stalk of the myelomeningocele manqué. These nerve bundles are in fact paramedian dorsal nerve roots that had been translocated extradurally. One reason for the abnormal dorsal migration of neural crest cells in SCM may be the presence of the double notochords. Neural crest cells are known to migrate from their origin along the outside of the neural tube through a fibronectin-rich extracellular matrix, where they avoid chondroitin sulphate. Since chondroitin sulphate is secreted by the notochord in large amounts, the wayward paramedian neural crest cells are likely to be driven in the opposite direction from the ventrally located paired notochords. Their dorsal shuffling with the endomesenchymal tract will thus be greatly facilitated.

While paramedian dorsal nerve roots are commonplace in SCM, paramedian ventral nerve roots are extraordinarily rare. One possible explanation for this great discrepancy is that the hemicord is more likely to "acquire" an extra dorsal horn than an extra ventral horn at the time of the split. The study by Herren and Edwards does suggest that definable medial dorsal horns are found more commonly than medial ventral horns. However, in several of their hemicord sections, there were clearly two ventral gray columns, of which only the lateral one had a nerve root; the medial ventral horn contained large neurons but no outgoing axons. Thus, the extreme rarity of paramedian ventral roots cannot be explained solely by the absence of paramedian ventral horn neurons. An alternate explanation may be the different peripheral factors guiding the outgrowth of motor versus sensory axons. Keynes and Stern have shown that surgical removal of the rostral half-sclerotome in chick embryos at one or two adjacent segments causes absence of motor axon outgrowth from the corresponding segments of the neural tube. Following the initial sprouting from the neural tube, motor axons are highly attracted to and guided by developing myotubes from specific segmental myotomes. When two or more dermomyotomes are removed in the chick embryo, the corresponding motor rami do not form. The initial growth cones of the emerging motor axons may be responding to specific chemotactic cues from the myoblasts and rostral sclerotome, or to direct cell contact with the rostral sclerotome or the extracellular matrix of the migrating myotome.

With either mechanism, the distance between the growth cones of the ventral roots and the axon guidance signal is critical for promoting successful axonal growth. In the case of a SCM, the medial ventral horn. if present. is almost always rotated somewhat dorsally so that it is not only more medially located than the lateral ventral horn but actually dorsal to it, facing deep into the median cleft. The motor neurons within the medial ventral horn are therefore sequestered from the rostral sclerotome and dermomyotome and may not receive adequate induction cues.

Growth of the peripheral process of the dorsal ganglion cells likewise responds to chemotactic cues from the developing dermatomes. but their central processes are obligated to interface with the internal cell groups of the neural tube regardless of the location of the cell bodies, For this reason. paramedian dorsal roots of SCMs, which are in fact central processes of misplaced ganglion cells, are much more likely to be found than are paramedian ventral roots.

Variable retention of the endomesenchymal tract and association with other anomalies          

If the ventral portion of the endomesenchymal tract remains, a structure containing mesenchymal and/or endodermal derivatives will persist, which connects the gut with the vertebral body and spinal cord. Differentiation of the endoderm in the ventral tract into gut epithelium will result in a duplicated intestine whose apex is connected to the vertebral body or split cord by a fibrous band. This band may persist even if there is no endodermal differentiation, and will then restrict the normal counter-clockwise rotation of the foregut and midgut, resulting in intestinal malrotation.

If the ectoendodermal fistula remains patent, the cutaneous ectoderm around it cannot heal over, and a permanent dermal opening will persist. In this situation, the dorsal portion of the endomesenchymal tract also persists, so that its subsequent derivative will be in continuity with the dermal opening. In the case of a type I SCM, the opening connects via a dermal sinus tract with the median bone spur; the entire dermal sinus tract therefore remains extradural. In the case of a type II SCM, the sinus tract necessarily penetrates the dorsal dura before reaching the median fibrous septum. The dorsal part of the sinus tract is usually lined with stratified squamous epithelium, but the ventral part is a solid fibrous cord that contains ganglion cells, nerve, roots, and glial tissue, typical of an endomesenchymal tract remnant. The intraspinal part of this sinus tract may become encysted to form a dermoid cyst, which lies dorsal to the hemicords but maintains a physical connection with the midline mesenchymal elements . Dermal sinus tracts are reported in 15 to 40 percent of SCMS.

Abnormal neurulation of the hemineural plates; association with open myelomeningocele

In some cases of SCM, the persistence of the dorsal endomesenchymal tract interferes with normal neurulation of the hemineural plate or the adjacent neural plate. The cutaneous ectoderm as well as the paraxial mesoderm fail to close over the non-neurulated cord, and an open dysraphism forms at the site of the split cord. The association of open dysraphism and SCM is well known.

During neurulation, if splitting of the neural plate occurs after the formation of the median hinge point (MHP) and divides the MHP in halves, each hemineural plate will have an eccentrically placed hinge point. The unpaired lateral neural plate still elevates around this eccentric hinge point, but elevation of the lateral neural plate can now only occur from one side, lateral to medial, and the dorsolateral in-bending of the neural fold will also occur only on one side. The result is therefore a very misshapen neural tube with an eccentric central canal and an emphatically larger lateral half. If splitting occurs prior to MHP formation, the heminotochord on each side should be capable of inducing an MHP in the new midline of the hemineural plate, and each hemineural plate should have its own midline floor plate and two lateral plates. The initial furrowing can now occur at the true median hinge point of each hemineural plate. Normal individual neurulation might then take place except for the fact that the lateral plate of each hemineural plate is now subjected to vastly different forces relative to the medial plate, the latter being neither attached to cutaneous ectoderm nor driven by paraxial mesoderm. The result will likewise be a grossly misshapen hemicord, which may also be rotated toward the midline, as has been described.

It is conceivable for the splitting of the neural plate and noto­chord to so greatly disturb the MHP and dislocate the relationship between paraxial mesoderm and lateral neural plate, that neurulation does not take place at all. If this happens to both hemineural plates, the result will be two plaque-like hemicords (without central canals) associated with an overlying open myelomeningocele. If neurulation takes place in one but not the other hemineural plate, then a hemimyelocele results. This may happen if, during the split, one hemineural plate remains close to the notochord and neurulates as usual, whereas the other is dislocated dorsally so that it neither receives the inductive effect of the notochord nor is subject to the proper neurulating forces provided by the mesoderm. The neurulated hemicord will thus have a central canal and lie deeper toward the notochord (vertebral centrum), whereas the un­neurulated hemineural plate will remain dorsal, plaque-like, and without a central canal, and will maintain connections with the opening of the hemimyelocele sac.

Clinical Features

In 13 percent of children with SCM in Pang.D. series, the diagnosis was made from the cutaneous stigmata alone, most notably hypertrichosis and capillary hemangioma. In almost one-third of the children, the asymptomatic and unsuspected SCM was detected on screening MRI ordered because of a previously treated open neural tube defect. The association between SCM and open myelo­meningocele has been shown to be between 26 and 80 percent, and the wider use of screening MRI on children with myelomeningocele will likely turn up even more SCMs before they become symptomatic. Considering the high likelihood of progressive and irreversible neurological damage, it is arguably justified to institute surgical treatment as soon as the diagnosis is made.

About 40 percent of patients with SCM in Pang D. series presented with symptoms and signs of neurological deterioration. In children, the onset of symptoms is usually insidious; only rarely is a definite precipitating event recognized. Pain and other irritative sensory phenomena are prominent symptoms in less than half the paediatric cases. The pain is usually sharp and well localized to the spinal segments around the SCM. Hyperpathia and dysesthesia in a dermatomal pattern are only occasionally seen. Much more frequently, children with SCM present with painless deterioration in sensorimotor functions. Depending on the age of the child and the location of the split-cord lesion, the signs to look for are definite leg or arm weakness, deteriorating gait or regression in gait training in toddlers, and decreased spontaneous movements in the lower extremities in infants. Children with cervical lesions sometimes show unexplained worsening in athletic abilities owing to progressive leg spasticity, and infants with similar lesions may even exhibit poor truncal tone in the sitting position. Sphincter dysfunction is not a common presenting complaint in children with SCM. Among children without associated dysraphism, the most common urinary symptom is recurrent urinary tract infections, with or without frank incontinence. Without the dramatic signs of an infection, it is amazing how many parents regard intermittent incontinence and incomplete toilet training in children as "normal" phases of child development. Increasing thoracolumbar scoliosis is seen in 10 to 20 percent of children with SCM. Slow but inexorable progression of talipes is noted in another 10 to 20 percent of children, typically with high vertical pedal arches and hammer toes. Fifteen percent of children in this series also had florid symptoms of chronic sympathetic dystrophy in the lower extremities.

Surgical Treatment

The aim of surgical treatment is to eliminate the septum. In type I SCMs, the bony septum is always enclosed within a dural sleeve. The bone is frequently fused with the neural arch dorsally. A laminectomy is performed carefully around the attachment of the septum until only a small island of lamina is left attached to the dorsal end of the septum. This permits subperiosteal dissection of the septum from its dural sleeve deep within the median cleft. Once the dorsal attachment of the septum is eliminated, it is no longer rigidly anchored at both ends. Excessive lateral movement of the septum may injure the hemicords and must be avoided. For most septa, the ventral attachment with the vertebral body is narrow or fibrocartilaginous and can easily be avulsed from the depth. If the ventral attachment is broad and bony, the stub of the septum can be removed by a small pituitary rongeur or microdrill. Time is well spent on this manoeuvre, since extradural removal of the bony spur greatly facilitates subsequent resection of the dural sleeve. Whether bony or cartilaginous, the septum always contains one to several large blood vessels, which can give rise to brisk bleeding when torn.

The dura is opened on both sides of the dural cleft to isolate the sagittal dural sleeve. The medial aspect of each hemicord often adheres tightly to the dural sleeve by fibrous bands that must be cut. Paramedian dorsal nerve roots and the fibroneurovascular bundles of myelomeningocele manqué are divided from their dural connections. The dural sleeve is always wedged against the most caudal reunion site of the hemicords. In a widely split cord, a considerable length of the rostral split is free from septal attachment. This free part offers a safe area to begin resection of the dural sleeve. Proceeding caudally from here, the base of the sleeve is cauterized to seal the central vessels and then cut flush with the ventral dural wall. The most hazardous part of this undertaking is at the caudal end, where the reuniting hemicords tightly hug the caudal margin of the sleeve.

Complete resection of the dural sleeve exposes the ventral extradural space, but closure of this midline anterior dural defect is neither necessary, because the abundant adhesions of the ventral dura to the posterior longitudinal ligament naturally prevent CSF leakage, nor desirable, because it potentially increases the likelihood of anterior retethering of the hemicords. Posterior dural closure ultimately converts the double dural tubes into a single one.

In type II SCM, three kinds of non-rigid median septa are seen. The rarest kind is a complete fibrous septum stretching between the ventral and dorsal surfaces of the dural sac, firmly transfixing the hemicords to the surrounding dura in the same manner as the type I bone spur. The second kind of fibrous septum is purely ventral, anchoring the cord to the ventral dura. The third kind, the most prevalent, is the purely dorsal septum, attaching the dorsomedial aspect of the hemicords to the dorsal dura.

Hypertrophic and fused laminae are rarely found in type II SCMs. Laminectomy for type II lesions is technically easy. A mid­line dural opening immediately exposes the purely dorsal and complete septa, but purely ventral septa have to be sought for, either between the hemicords or by gently rotating the hemicords to one side. Like the bony septa, all fibrous septa are found near the caudal end of the split. The length of splitting is much shorter in type II than in type I lesions, and the split has very little "free" part. The point of attachment between the hemicord and septum is always rostral to the point of attachment between dura and septum. Release of tethering is accomplished by cauterizing the central vessels and excising the median fibrous septum and the nonfunctional paramedian roots.

The great majority of reported cases of tethered cord syndrome involve tethering at the lumbosacral spinal cord. Few neurosurgeons would think of tethering at the cervical and upper thoracic cord because the slight ascent of the cervical cord with growth seemingly argues against much transmitted tension at the site of the tethering. However, several recent reports suggest that tethering at the cervical segments does produce neurological deterioration. Eller and colleagues reported a patient with progressive dorsal column dysfunction who had a dorsally cleft cord tethered to the dorsal dura by taut fibroneural bands. Simpson and Rose, Gower, and Rawanduzy and Murali have all reported examples of cervical SCM with progressive neurological deficits. Vogter and colleagues described an infant with a skin-covered, pedunculated cervical myelomeningocele (CMMC) who deteriorated at 8 months of age, as did most of the patients of Steinbok and Cochrane. In a  series of nine patients with CMMC included five who developed progressive spasticity and worsening of hand function after an initial unsuccessful operation to untether the cord. The fact that cervical tethering in incompletely treated CMMCs causes deterioration of hand function and not just leg function makes CMMC a potentially more incapacitating lesion than the familiar varieties of conus tethering lesions.

Cervical myelomeningoceles differ from the common variety of thoracolumbar and lumbosacral myelomeningoceles in several respects. First, the thoracolumbar and lumbosacral sacs are usually fragile, covered only by a thin, delicate arachnoid that is not infrequently tom during delivery. In contrast, the CMMC sac is a sturdy, tubular protuberance covered by full-thickness skin at the base and by tough, opaque squamous epithelium at the dome; CSF leaks are highly unusual. Second, the neural contents of a lumbo­sacral myelomeningocele usually consist of a flattened, exposed neural placode "floating" on top of the dome of the sac; well-defined ventral and dorsal nerve roots stream from its ventral surface toward the exit foramina. This neural placode is most often a terminal placode, that is, the unneurulated caudal end of the spinal cord. In contrast, the neural content of a CMMC usually consists of a basal neural nodule of highly disorganized glioneuronal tissues and nerves. This neural nodule is always connected with the dorso­medial surface of the "parent" spinal cord (the part within the spinal canal) by a fibroneurovascular band through a dorsal dural fistula; it is, in essence, a vertical outgrowth from a functioning spinal cord in the form of a limited dorsal myeloschisis (LDM) and not a terminal placode. Third, in keeping with the respective neural contents of these lesions, the patient with a lumbosacral myelomyocele usually has no neurological function below the level of the terminal placode, whereas in a patient with CMMC the neurological function caudal to the level of the lesion is normal or near-normal.