AUTHOR(S): Pang, Dachling, M.D., F.R.C.S.(C), F.A.C.S. Pediatric Neurosurgery, Children's Hospital of Pittsburgh, and The University of Pittsburgh, School of Medicine, Pittsburgh, Pennsylvania Neurosurgery 31; 481-500, 1992 ABSTRACT: THIRTY-NINE PATIENTS with split cord malformations (SCM) were studied in detail with respect to their clinical, radiographic, and surgical findings as well as their outcome data. Eight patients were adults and 31 patients were children. According to the classification endorsed by Part I of the SCM study, 19 patients had Type I SCM (6 adults and 13 children), 18 patients had Type II SCM (2 adults and 16 children), and 2 patients had composite SCM with both lesion types situated in tandem. Six SCMs were cervical, 2 were thoracic, and 31 were in the lumbar region. All 8 adults had pain and progressive sensorimotor deficits at diagnosis. Only 16 of the 31 children had symptoms, and among these, 14 had progressive sensorimotor deficits, but only 6 had pain. The difference in the clinical picture between adults and children is similar to that described in the tethered cord syndrome, except for left-right functional discrepancy, which was prominent in 8 children with SCM but rarely seen in tethered cord syndrome due to other causes. Cutaneous manifestations of either occult or open dysraphic states were present in all but 3 patients; hypertrichosis was by far the best predictor of an underlying SCM, being found in 56% in the series. Neurological deterioration in SCM was independent of the lesion type: the Type I:Type II ratio for symptomatic progression was 13:11. It was also independent of the location of the lesion: 67% of patients with cervical SCMs had symptomatic progression versus 64% of patients with thoracolumbar lesions. High-resolution, thin cut, axial computed tomographic myelography using bone algorithms was more sensitive than magnetic resonance imaging in defining the anatomical details of the SCM. Radiographic classifications of the SCM, using the nature of the median septum and the number of dural tubes as criteria, was always possible without ambiguity. However, whereas every Type I bone septum was identified preoperatively, only 5 Type II fibrous septa were revealed by preoperative imaging, even though a fibrous septum and/or other fibroneurovascular bands were found tethering the hemicords in every Type II case at surgery. Complete imaging studies also showed that all lumbar SCMs had low-lying coni and at least one additional tethering lesion besides the split cords, whereas only 1 of 7 cervical and high thoracic SCMs had a low conus and a second tethering lesion. The surgical

goal for SCM was release of the tethered hemicords by eliminating the bone spurs, dural sleeves, fibrous septa, or any fibroneurovascular bands (myelomeningoceles manqué) that might be transfixing the split cord. Type I cases were technically more difficult and had a slightly higher surgical morbidity than Type II cases, especially if an oblique bone septum had asymmetrically divided the cord into one larger hemicord and one smaller, hence, very delicate, hemicord. Overall, 89% of patients surgically treated experienced either improvement or stabilization of their preoperative neurological status. Thus, this study strongly argues that both types of SCM are cord tethering lesions likely to cause neurological damage, and both should be treated. All Type II SCMs should be explored, even if a definite median septum was not revealed by imaging studies. The entire neuraxis should be studied to look for other tethering lesions, which should also be treated. Surgery is excellent for improving or stabilizing the neurological status. KEY WORDS: Double spinal cord malformation; Median septum; Neural imaging studies; Neurological deterioration; Split cord malformation; Surgical techniques; Tethered cord syndrome In Part I of the study on split cord malformations (SCM) (44), the author proposed a unified theory of embryogenesis for all double spinal cord malformations, based on a series of 39 patients and two autopsy cases with this anomaly. The radiographic and surgical findings were presented in detail, particularly in relation to different facets of the proposed theory. A new classification of SCM based on easily identifiable imaging characteristics was also endorsed. In reviewing the clinical features, the radiographic techniques used in making the diagnosis, the surgical findings, and the treatment outcome of the 39 patients, the authors recognized several previously unknown or unemphasized facts about SCM that are of great clinical significance. In Part II of the study on SCM, these clinical data and recommendations for investigation and management are presented. PATIENTS, MATERIALS, AND METHODS Age and sex The clinical series consists of 38 patients treated at the University of Pittsburgh Health Science Center hospitals between 1979 and 1990 and one additional patient treated at the University of British Columbia. The author is indebted to Dr. Paul Steinbok for supplying the case material of this patient. Thirty-one cases are of children ranging in age from birth to 17 years, with a mean age of 5.7 years. Eight cases are of adults ranging in age from 21 years to 60 years, with a mean age of 36.9 years. Among the children, the female to male ratio is 19:12. Among the adults, the female to male ratio is 5:3. The female to male ratio for the entire group is 24:15. Reasons leading to diagnosis

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Neurosurgery 1992-98 September 1992, Volume 31, Number 3 481 Split Cord Malformation: Part II: Clinical Syndrome Clinical Study

Radiological evaluations Before 1984, all patients underwent metrizamide (Amipaque, Nyegaard & Co., AS, Oslo, Norway) or iohexol (Omnipaque, Winthrop-Breon Lab, DesPlaines, IL) myelography, using plain films and contiguous nonoverlapping axial computed tomography (CT). The initial plain myelographic film marked the approximate locations of the double cord lesions and the conus. Three-mm thin-section axial CTs with conventional soft tissue and bone algorithms were then performed to evaluate these two sites as well as the craniovertebral junction to rule out associated hindbrain anomalies. Occasionally, ultrathin sections measuring 1.5 mm in thickness were necessary to define details of the split cord and any additional abnormality suggested by the scout film. Since 1984, magnetic resonance imaging (MRI) has also been used in addition to computed tomographic myelography (CTM). The MRI unit used had a magnetic field strength of 1.5 tesla. The purpose of obtaining both CTM and MRI on the same patient was to specifically compare the respective sensitivities of these two modalities in evaluating different aspects of the SCM. In addition to MRI and CTM, pre- and postoperative spine films were obtained on all patients to evaluate preoperative spinal alignment and to detect postlaminectomy spinal deformity. Physiological evaluations All patients underwent somatosensory evoked potentials measurement, using standard surface stimulating electrodes overlying the common peroneal and median nerves. If the SCM was at the lumbosacral region, pudendal somatosensory evoked potentials were also obtained, using pairs of disc electrodes on either side of the penis in the male and on the preclitoral skin and the labia majora in the female. Patients with symptoms of neurogenic bladder were studied with a urodynamic protocol that included a cystometrogram, urethral pressure profile, external urethral sphincter electromyography, and external anal sphincter pressure measurement. A renal ultrasound was obtained to visualize the size of the ureters and renal pelvis, and a voiding cystourethrogram was used to detect ureteral reflux. Urine bacteriological studies were performed for each patient, and intravenous pyelogram and creatinine clearance measurements were sometimes used to evaluate renal function. Treatment

All patients with SCM were explored surgically. The aim was to untether the hemicords by resecting the midline bony or cartilaginous septum, the dural sleeves ensheathing the septum, and any soft tissue band or septum attaching any part of the hemicords to the surrounding dura or bone. All other tethering lesions were also treated. If a thickened filum terminale, lipoma, or dermal sinus tract not associated with the split cord lesion was also shown on preoperative MRI or CTM, it was always eliminated at the time when the split cord was dealt with. If the SCM was in the vicinity of a previous myelomeningocele repair, the neural placode was also detached from the overlying dura. Follow-up All patients were followed in the neurosurgical clinic after surgery. The follow-up period ranged from 6 months to 10 years, with a mean of 3.9 years. Statistical analysis Student's t test was used to compare the means of the groups; Levene's F-statistics was used to compare the variances of the groups (35). This was done because for groups whose variances are different, the t-statistics and corresponding confidence in the result may be inflated. To equalize the variances, a second ttest was performed in each case on data transformed by the square root. Although this tended to equalize the variances, it did not reduce the statistical significance of the t-test, suggesting that unequal variances did not produce erroneously significant results. RESULTS Clinical features Pediatric patients The 31 children in this series can be divided into three groups, depending on the reason for which the diagnostic study was ordered. Group I (Figure 1) In Group I (13 patients), the children had no history of spinal dysraphism but presented with symptoms and progressive neurological deficits suggestive of tethered cord syndrome. The onset of symptoms was precipitated by a definite incident in only 1 child. This 16-year-old boy suffered acute low back pain and unilateral quadriceps weakness after a severe flexion injury to the lumbar region while playing football. In the other 12 patients in this group, the onset of symptoms was insidious. Pain and other irritative sensory phenomena were prominent symptoms in 6 children. In 4, the pain was sharp and well localized to the spinal segments around the SCM. Hyperpathia and dysesthesia in a dermatomal pattern compatible with the segmental locations of the SCMs were present in 3 children. The absence of pain in the other 7 Group I children did not appear to be due to young age and difficulty in communication because only 3 of them were younger than 3 years. Progressive sensorimotor deficits were found in 11

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

In the children group, radiographic examinations leading to the diagnosis of SCM were obtained for one of three reasons: signs and symptoms suggestive of tethered cord syndrome; cutaneous stigmata of occult spinal dysraphism without neurogenic symptoms; or a history of an open myelomeningocele repaired at birth with or without the recent progression of neurological deficits. All 8 adults were radiographically studied because of signs and symptoms suggestive of tethered cord syndrome.

Group II In Group II (five patients), the children were asymptomatic but radiographic studies were ordered because of cutaneous signs of occult spinal dysraphism. Hypertrichosis and capillary hemangioma were the most common skin lesions and usually found together, the hemangioma always underlay the hairy patch but tended to extend over a much larger area (Fig. 2). Two children had dermal sinus openings over the spinal segments bearing the SCM, typically showing overhanging skin edges and protruding hair tufts from the opening. Although these 5 children did not have progressive signs or symptoms of cord dysfunction, several had subtle neurological deficits such as pathological reflexes, mild atrophy of leg muscles, and dulling of superficial sensations. One child also had a slight nonprogressive lumbar scoliosis. Combining Group I and II patients (Table 1) confirms that hypertrichosis was the most common cutaneous manifestation of SCM, being present in 11 of 18 children. Capillary hemangioma was the next most common skin lesion in SCM, being found in 8 children, among whom only 2 were without an overlying hair patch. Dermal sinuses were found in 4 children and subcutaneous lipomas in two. Only 2 of these 18 children had no cutaneous abnormality. Group III The 13 children belonging to Group III all had

routine closure of an open "uncomplicated" myelomeningocele at birth. Three children (Group IIIa) had deterioration of neurological function 16 months to 5 years after the initial closure, and their radiographic studies revealed the unsuspected split cord lesion. The youngest in this subgroup was a 16month-old girl who showed rapid ascension of sensorimotor paralysis of the legs from an original L4 functional level. Another 3-year-old girl with a closed cervicothoracic myelomeningocele had progressive hand weakness and spastic gait. Finally, the third patient with a myelomeningocele at T10, but an L3 sensorimotor level on the right, an S2 level on the left, and nearly normal bladder function presented at age 5 with urinary incontinence and severe constipation. None of the patients in Group IIIa had pain. Seven children (Group IIIb) had no change in their neurological status at the time of radiographic diagnosis. Since 1984, it has been our policy to perform an MRI on spina bifida children who exhibited the following unusual features: a discrepancy between the spinal level of the myelomeningocele sac and the neurological level (3 patients, all had much better neurological function than expected from the high location of their sacs); a gross difference between the neurological functions of the left and right legs (2 patients); and the discovery of a segmental neural placode during closure in which the cord caudal to the placode appeared neurulated (2 patients). Finally, in 3 children (Group IIIc), the original myelomeningocele sacs were well covered by skin and thickened membranes so that detailed radiographic studies were done before closure. In 2 children with a "covered" lumbar sac, the split cord lesion thus discovered was dealt with at the time of the sac closure. In the other child with a "covered" cervical myelomeningocele, the SCM was electively treated 17 months after resection of the sac. It is significant to note that among the 13 Group III children with associated myelomeningoceles, 6 had a marked difference between the neurological function of the right and left legs (Table 2). All 6 were found to have a hemimyelocele (44); in 3, the hemicord involved in the myelocele sac was on the side of the poorer neurological function, but in the other 3, there was also asymmetrical splitting of the cord (44) and the smaller hemicord was on the side of the poorer function. Among the 18 patients belonging to Group I and Group II combined, only 2 had a left-right functional discrepancy, and their imaging studies both revealed asymmetric splitting with the smaller hemicord on the more paralyzed side. Adult patients All eight adult patients had neurological signs and symptoms at the time of diagnosis (Fig. 3). In three patients, the onset was abrupt, after a fall on the buttocks or the low back. The onsets were insidious in the other five patients. The interval between symptom onset and diagnosis varied between 1 month to 3 years. Pain was the predominant symptom in all eight

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

of the 13 Group I children. Definite leg or arm weakness was noted in 8 children, but two other toddlers had deteriorating gait and one 8-month-old infant with a cervical SCM exhibited poor truncal tone in the sitting position. Increasing pedal numbness without associated weakness was seen in only 1 patient. Only one 16-year-old boy had a new onset of bowel incontinence and decreasing anal tone. None of the children reported recent loss of bladder function, but one child had an ilioconduit and another had a neurogenic bladder from birth and was never toilet trained. In addition, 5 children were below toilet training age and were therefore difficult to evaluate. Increasing thoracolumbar scoliosis was noted in 2 children. In addition, worsening head tilt and torticollis was seen in 2 infants with a cervical SCM. Rapid progression of talipes deformity was noted in 2 children, typically with high vertical arches and hammer toeing. However, some degree of fixed fibroosseous foot deformity was noted in 6 other children, even though no recent worsening was reported. Two children had florid signs of sympathetic dystrophy in the lower extremities; in one, these trophic changes were the sole manifestation of her disease. Both had a long history of nonhealing ulcers, osteomyelitis of the phalanges, and serial toe amputations. Both had thin, shiny skin, hairlessness, anhidrosis, dependent rubor, and thickening and fissuring of the toe nails. Mild degrees of these were seen in 4 other children. Cutaneous signs of occult dysraphism were present in 11 of the 13 Group I patients (Table 1).

Radiographic diagnosis The SCMs could be classified as Type I or Type II lesions according to the definitions given in Part I of the SCM study (44). Twelve children had single Type I SCMs and 16 children had single Type II lesions. One child had a double Type I lesion and 2 children had composite lesions. Among the adults, six had single Type I SCMs and two had single Type II lesions. Single Type I SCM The two dominant radiographic characteristics of a Type I lesion are the double dural tubes and the osseous or osseocartilaginous midline septa. Among the 18 patients with single Type I lesions, 12 septa were of pure bone and 6 had a bony main part and a ventral cartilaginous part; none was made of pure cartilage. The Type I septa always reached from the posterior surface of the vertebral body to the corresponding neural arches. In 16 patients, the septa were oriented vertically in the midline, thereby bisecting the spinal canals into two equal sagittal compartments; the corresponding hemicords were of equal size. However, in 2 patients, the bony septa

were obliquely oriented so that the spinal canals were divided into two unequal diagonal compartments. The hemicord within the larger diagonal was correspondingly much larger (major hemicord) than its partner (minor hemicord) within the smaller diagonal. The Type I septum was always located at the caudal end of the split cord, so that the two hemicords hugged the caudal margin of the bone spur tightly before reforming a single cord. Rostrally, the hemicords stretched a variable distance (sometimes as much as 7 vertebral levels) before rejoining. Here the two hemicords were rostral to the median dural sleeve and therefore resided in a single thecal sac. Lateral ventral and dorsal nerve roots were frequently seen on the axial CTM and less so on the axial MRI. Occasionally, paramedian dorsal roots were also discernible on the CTM or MRI as fine lines stretching between the medial aspect of the hemicords and the median dural sleeve. Single Type II SCM The two dominant radiographic characteristics of a Type II SCM are the single thecal sac housing the full length of the split segments and the nonrigid (nonosseous and noncartilaginous) nature of the median septum. Because there is no crossover regarding these two features between Type I and Type II lesions, the SCM could always be classified radiographically before surgery. Whereas an osseous septum was always seen on CTM in a Type I SCM, a fibrous median septum was identified on CTM in only 5 of 18 Type II SCM patients, even though one was found in every case at surgery. Only 16 of these patients underwent MRIs, and none of these suggested a median septum. When discernible, the Type II septum was always at the caudal extreme of the split cord where the hemicords appeared to be closely apposed with the septum and with each other before rejoining into a single structure. In general, the hemicords in a Type II lesion were much closer together and the split segment was much shorter than in a Type I lesion. In 15 patients with single Type II lesions, the hemicords were of equal size and were symmetrical side by side. In two patients, the fibrous septum was diagonally oriented and associated with obvious size disparity between the hemicords. In another patient, the hemicords were of equal size but were arranged in a directly anteroposterior relationship with each other (44) . Paramedian dorsal nerve roots were identified near the median septum on seven CTMs, suggesting either that they ended in the septum or continued on as a myelomeningocele manqué (44). Paramedian dorsal roots were identified as separate entities from the septum in 3 CTMs, suggesting they were part of a myelomeningocele manqué. Composite SCMs and multiple median septa In 2 children who underwent resection at birth of what was thought to be a routine myelomeningocele, subsequent imaging studies showed two typical Type I SCMs separated by a typical Type II SCM, all three

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

patients. The initial dull pain was always localized to the spine roughly at the site of the split cord. The one patient with a cervical lesion had a head tilt and nuchal spasm. Six patients also had ill-localized, dysesthetic pain diffusely involving both legs, usually aggravated by lifting and forward bending. In four, the pain was perceived as "boring" or "searing" in the perineal and anal regions. All eight adult patients had some degree of leg weakness or spasticity. In one patient with a cervical split cord, the spasticity involved both upper and lower extremities. Overall, the characteristic motor pattern among adults was a mixture of upper and lower motor neuron findings in the same limb. All eight patients also had objective decrease in pain and temperature appreciation in either the legs or saddle area. Proprioception was also affected in five patients. Two patients described bouts of electric shock sensation propagating down the legs and up the spine with trunk movements. All eight patients had symptoms of neurogenic bladder; in three, these were static since childhood. In the other five patients, the symptoms were recent and progressive. These included urinary frequency and urgency, feeling of incomplete voiding, postvoid dribbling, poor voluntary control, and urge or stress incontinence. Frequent urinary tract infection was a prominent feature in six patients, and in one it led ultimately to renal transplantation. A history of renal colic was noted in four patients, presumably related to recurrent bacteriuria and ureteral reflux. Chronic constipation was a universal complaint in this group, but fecal soiling only occurred in one patient who had a precipitous onset of other neurological symptoms. Two patients had rectal prolapse, presumably as a result of poor tone in the muscular pelvic floor and repeated straining due to constipation. Two female patients also had a history of ineffective labor during childbirth.

SCM associated with myelomeningocele Thirteen children with SCM had an associated open myelomeningocele repaired at birth. Three configurations of this association were noted. In 5 patients, the neural placode (the nonneurulated part of the spinal cord) involved both hemicords and the adjacent rejoined cord either near the rostral or caudal extent of the split. The placode was recognized on the axial CTM or MRI by the adherence of both hemicords to the dorsal dura at the original repair site and by the adjacent extradural scarring. In another 6 patients, only one hemicord participated in the original myelomeningocele sac as a segmental hemimyelocele. Only the involved hemicord was adherent to the dorsal dura, whereas the other hemicord was more ventrally located and unattached to the dura. In 2 patients with "closed" cervical myelomeningoceles covered by skin and thickened membranes, strands of neural tissues arose from the dorsomedial aspects of both hemicords and entered the narrow neck of the myelomeningocele sac, forming a limited dorsal myeloschisis involving both hemicords. Location of the split cord Because in many cases the split portion of the cord spanned many vertebral levels, it would be unfeasible to assign a location for the SCM if the entire split segment was considered. However, because the median septum is usually narrow and occupies the most caudal part of the median cleft, its location may be considered representative of the lesion. Moreover, because it is the remnant of the endomesenchymal tract (44), its location is also the locus of the original ontogenetic error and can justifiably be the focal point of the malformation. Figure 4 shows that the majority of median septa are found in the lumbar segments. Twenty-five out of the total 39 septum (counting only the most caudal septum of the 3 cases of multiple septa) are found between the L1 and L5 vertebral levels. In fact, there appears to be a bimodal distribution, with a small peak at the cervical and cervicothoracic junction and a much taller one at the lumbar and thoracolumbar junction. No septum was found below S1, and only two were in the midthoracic segments.

Separating these septa by type (Fig. 5) reveals that the Type I (bony) septa (including the two composite lesions) are exclusively located in the lumbar and thoracolumbar region, whereas the Type II (fibrous) septa are distributed in both the cervical and lumbar ends of the spine. Also, correlating the location of the septum with age groups shows comparable distributions among the adult and pediatric patients (Fig. 6). Correlation between location of septum and symptoms Of six patients with cervical SCM, four had progressive neurological symptoms. One child with a T5-T6 lesion also had progressive neurological deficits. Of 32 patients with SCM in the lumbar and thoracolumbar regions, 20 had progressive neurological symptoms. Thus, the propensities to cause neurological deterioration are the same between cervical and lumbar lesions. The neurological syndromes caused by these two groups of lesions are, however, very different. The cervical lesions are invariably associated with paresthesia and clumsiness of the hands and fingers, with or without leg spasticity. The lumbar and thoracolumbar lesions usually caused a mixed picture of lower and upper motor neuron findings in the legs and feet. Position of the conus and other tethering lesions In only six patients was the conus at or above the lower border of the L1 vertebral body, a level that sets the lowest limit for a normal "untethered" conus (6) . The other 33 patients had coni ending between L2 and S2, with the majority in the lower lumbar and upper sacral levels (Fig. 7). None of the coni ending above the lower border of L1 had a secondary tethering lesion (in addition to the SCM). Without exception, every low lying conus (below the lower border of L1) was associated with at least one tethering lesion that was not connected with the SCM. The most common lesion was a taut filum thicker than 2 mm. In addition, there were six terminal lipomas, one dorsal lipoma inserted on the cord caudal to the SCM, six dermal sinus tracts, and two cases of limited dorsal myeloschisis within the stalk of a skin-covered myelomeningocele. The role played by the nonneurulated placode among the 13 Group III patients with treated myelomeningocele is difficult to ascertain. There is also an interesting correlation between the level of the median septum and the level of the conus (Fig. 8). Among the six patients with cervical SCM, five had a conus ending at L1; none had any secondary tethering lesions. The one exception was a child with a C7 Type II SCM and a conus tethered by a terminal lipoma at L2. The one case of upper thoracic SCM (above T6) was associated with a normal conus. Without exception, all SCMs situated below T7 were associated with a low lying conus complicated by at least one additional caudal tethering lesion, such as a thick filum or terminal lipoma. As shown in Table 3, the frequencies and variety of secondary tethering lesions are comparable among patients with Type I and Type II SCMs.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

lesions being in longitudinal tandem. In one case, the bone spur of each Type I lesion was diagonally oriented in the same angle, dividing the neural structure into a slightly dorsal, larger hemicord and a slightly ventral, smaller hemicord. The Type II lesion in between had a complete fibrous septum lying coplanar with the two bone spurs, suggesting that all three structures were one continuous sheet. The situation was similar in the other composite SCM except that the septa were all vertical. The most peculiar case was that of a 2-year-old girl whose CTM at a glance showed a tripartite cord created by two parallel bone spurs. Careful reconstruction of the images in fact showed two unconnected slightly oblique bone spurs rostrocaudally arranged along the length of the involved cord segment (44).

Abnormalities of the adjacent vertebral bodies Four types of vertebral body abnormalities were identified at the approximate level of the SCM: complete bifid body, widened body, widened body with a midsagittal tract of variable CT density, and fusion of two or more adjacent bodies. Some of these

abnormalities occurred alone, but more often they occurred at contiguous levels in the same patient. Altogether, there were 13 patients with complete bifid bodies (Fig. 11); in 12, the bifid state was either found in multiple levels or coexisted with adjacent bodies with midline tract. Twenty-two patients had bodies with midline tract, and plain widened bodies were found in 26 patients. Fusion of adjacent vertebral bodies was found in 4 patients. In only 6 patients was there no vertebral body abnormality at any level. Types I and II SCM have slightly different patterns of vertebral body abnormalities. Of 19 Type I patients, 17 had severe abnormalities such as complete bifid state, body fusion, and widened bodies with midline tract; another 2 had some body widening, and none had no abnormality. In contrast, only 6 of the 18 Type II patients had severe body abnormalities, and 6 Type II patients had no abnormality. Abnormalities of the neural arch Hypertrophic laminae and bulbous spinous processes at the level of the midline septum were common findings among patients with Type I SCM; they were present in 16 of 19 cases. Occasionally, two or three adjacent laminae were fused into a massive gnarl. In contrast, exuberant neural arches were seen in only two of the 18 Type II cases. Both cases of composite lesions had hypertrophic neural arches related only to their Type I SCMs. Incomplete neural arches (posterior spina bifida) were present in the 13 cases of SCM with associated myelomeningocele and in 14 of the other 26 cases of SCM. Abnormality in spinal alignment Scoliosis was noted in 21 of 39 patients at the time of diagnosis. The location of the major curvature corresponded well with the site of the SCM. Scoliosis was progressive in 5 of the 15 pediatric patients. In contrast, none of the six adult patients with scoliosis had any worsening of the spinal curvature. Osseous abnormalities of the vertebral elements such as hemivertebra, asymmetric segmentation defects, and unilateral pedicular or transverse process fusions contributed to the development of the spinal curvature in 4 patients. In 17 patients, no such anomalies were noted and the scoliosis was attributed to neuromuscular imbalance. Hyperlordosis of the lumbar spine was seen in 6 children with associated myelomeningocele. Other congenital anomalies One child with a Type II cervical SCM was also found to have complete malrotation of the midgut and partial obstruction of the distal duodenum. There was also eventration of the right hemidiaphragm, suggesting that the original endomesenchymal tract remained as a band that linked the distal duodenum with the lower cervical neural tube and notochord through the right hemidiaphragm and cervical prevertebral fascia (44).

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Length of the split segment The length of the split segment was measured by the number of vertebral bodies it spanned. This unit is used instead of the absolute length of the split segment because it takes into account the wide range of ages in this series and automatically adjusts for the increase in the vertical dimension of the spinal cord with growth. This method is not foolproof because the cord and vertebral column lengthen at different rates during the first year of life (6), but beyond that the segmental relationship between the vertebral column and the spinal cord remains constant. The CTM and not the MRI data were used for the computation of split length because CTM was performed on every patient, its collimator width was always 3 mm or less, and it displayed finer details than the MRI. The rostral end of the split was taken where the first CT cut showed two distinct hemicords, and the caudal end where the first CT cut showed the hemicords reunited as one. Thus, the maximum error was 1.5 mm on top and 1.5 mm on the bottom, or 3 mm total, which was still less than half of the vertebral body height of an infant. The split lengths of the 19 Type I lesions (including the one case of double bone spurs and tripartite cord, in which the two split lengths were identical) varied between 1 and 7 vertebral levels, with a mean of 2.97. The split lengths of the 18 single Type II lesions varied between 0.5 and 2 vertebral levels, with a mean of 1.17. The difference is statistically significant (t = 4.50; P = 0.0001; comparison of variances: F (1,35) = 15.14; P = 0.0004). Figure 9 shows the correlation between the split length and the age of the patient. It shows that the mean split length in 10 infants younger than 1 year was 1.25 vertebral levels, whereas the mean split length of those patients diagnosed after 1 year was 2.62 vertebral levels. The difference is statistically significant (t = 2.33; P = 0.025; comparison of variance: F (1,37) = 10.24; P = 0.0028). Figure 9 also suggests that the split length remains relatively constant from age 1 year onward. There were 10 patients between the ages of 1 to 3 years, with a mean split length of 2.40 levels. This is compared with 19 patients between the ages of 3 to 60 years, with a mean split length of 2.73 levels. The difference is not statistically significant (t = 0.47; P = 0.64, comparison of variance: F (1,27) = 3.97; P = 0.06). Figure 10 correlates the split length with the level of the midline septum. The mean split length of the seven cases of high SCM (C1-T6) was 1.14 levels, compared with a mean split length of 2.52 levels in the 32 cases of low SCM (T7-L5). The difference is statistically significant (t = 2.02; P = 0.05, comparison of variance: F (1,37) = 9.08; P = 0.0046).

Operative technique The neurological deterioration seen so frequently in SCM is due to the tethering effect exerted on the spinal cord by the median mesenchymal septum. The aim of the surgical treatment is therefore to relieve the cord of this tethering effect by eliminating the median septum. Because the relationship between septum and spinal cord is so drastically different between Type I and Type II SCMs, the surgical techniques for the two types also vary considerably. Type I SCM In a Type I SCM, the bony septum was always extradural, being completely excluded from cerebrospinal fluid (CSF) by the medial walls of the dual dural tubes which formed a sagittal sleeve for the bone. The septum was frequently fused with the neural arches dorsally. After subperiosteal exposure of the spinous processes and laminae, the septum was usually not visible at first but could reliably be located where the spinal canal was widest. Laminectomy was performed carefully around the attachment of the septa until only a small island of lamina was left attached to the dorsal end of the septum. This permitted subperiosteal dissection of the septum from its dural sleeve deep within the dural cleft. Once the dorsal attachment of the septum was eliminated, it was no longer rigidly anchored at both ends and might be pushed from side to side depending on its ventral anchorage (Fig 12A). Excessive lateral movement of the septum may injure the adjacent hemicords and must be avoided. In the majority of cases, the broadest attachment of the septum was dorsal, its ventral junction with the vertebral body was usually narrow or even fibrocartilaginous, which made it easy to avulse the septum from its deep attachment. If the ventral attachment was broad and bony, the ventral stub of the septum could be removed by a small pituitary rongeur or microdrill. Time was well spent during this part since extradural removal of the bony spur greatly facilitated later resection of the dural sleeve. In all cases of Type I lesions, the rigid septa were either purely bony or part bony and part fibrocartilaginous. Bony or cartilaginous, the septum always enclosed one to several prominent central blood vessels which could give rise to brisk bleeding when torn by the unwary. The dura was opened on both sides of the dural cleft to isolate the sagittal dural sleeve (Fig 12B). The medial aspect of each hemicord was often tightly adherent to the dural sleeve by fibrous bands that must be cut. Paramedian dorsal nerve roots, when present in a Type I lesion, typically stretched from the dorsomedial aspect of the hemicords to end blindly within the median dural sleeve. These were nonfunctional and must be cut prior to resection of the dural sleeve. The dural sleeve was always wedged between the most caudal reunion site of the hemicords, and any "free" part of the hemicords not closely apposed to median mesenchymal structures would be rostral to the septum. In a widely split cord,

this free area was considerable and constituted a safe area to begin resection of the dural sleeve (Fig 12C). Proceeding caudally from the rostral free margin of the sleeve, the ventral attachment of the sleeve was cauterized to seal the central vessels and then cut flush with the ventral dural wall. The most hazardous part of this undertaking was at the caudal end where the hemicords reunited and hugged tightly against the caudal margin of the sleeve, where the taut downward pressure on the cord was readily felt and where slight upward migration of the cord might sometimes be seen right after the whole sleeve had been resected. Complete resection of the dural sleeve exposed the ventral extradural space in the sagittal midline, but closure of this anterior dural defect was unnecessary because of the abundant adhesions of the ventral dura to the posterior longitudinal ligament that would naturally prevent CSF leak (Fig. 12D). Anterior dural closure was undesirable for it would potentially increase the likelihood of anterior retethering of the hemicords. Posterior dural closure ultimately converted the double dural tubes into a single one. Type II SCM All 18 cases of Type II SCMs were surgically explored, and in all 18, some form of fibrous (mesenchymal) septum was found within the midline cleft. Three patterns of such nonrigid median septa were documented. In the first kind, found in only four patients, there was a complete fibrous septum stretching between the ventral and dorsal surfaces of the dural sac. The septum was completely intrathecal and, like both hemicords, was contained within a single dural tube. Thus, except for this feature and the fact that they were nonosseous, the complete fibrous septa transfixed the hemicords to the surrounding dura in the same manner as the Type I bony septum. In the second kind, found in three patients, the fibrous septum was purely ventral. Its intimate adherence to the ventromedial aspects of the hemicords in effect anchored the cord ventrally to where the incomplete septum fused with or penetrated the ventral dura. The third kind was the most prevalent, being present in 11 patients. This was the purely dorsal septum attaching the dorsomedial aspect of the hemicords to the dorsal dura. In several cases, the septum penetrated the dorsal dura and blended with a tuft of fibrovascular tissues in the extradural space. Hypertrophic and fused laminae and spinous processes so common in Type I lesions were found in only two cases of Type II SCM. In fact, more often the neural arches of Type II lesions were attenuated or even bifid. Laminectomy for Type II lesions was technically easy but the exact location or extent of the Type II fibrous septum was usually not apparent after dural exposure, unless a tuft of adherent extradural fibrovascular tissue marked the site of penetration of an underlying dorsal septum. Midline dural opening immediately exposed the purely dorsal and complete septa. Purely ventral fibrous septa had to be sought for either in between the hemicords or by gently rotating the hemicords to one side. Like the Type I bony septa, all Type II

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Surgical findings and treatment

Composite SCMs In the two composite SCMs, the split cord was coextensive with the median mesenchymal septum so that the rostral and caudal reunion sites of the hemicords were wedged against the two bone spurs. Central vessels and paramedian dorsal roots were scattered throughout the extent of these two long midline septa, stretching as much as 7 vertebral segments in one case. Release of the cord was accomplished with total excision of all three midline septa. SCM associated with open myelomeningoceles In 11 of the 13 patients with open myelomeningoceles, the operation for the SCM was performed several years after the initial closure of the dysraphic sac. In two patients with hemimyeloceles, the diagnosis of SCM was made at birth and both lesions were treated during the primary operation. Because the myelomeningocele placodes or hemiplacodes were always in the vicinity of the SCM and might well be contributing to the tethering, they were always freed from the dura at the time when the

SCM was treated. For all patients with myelomeningoceles, the most caudal set of intact laminae above the open spinal defect were removed to expose normal, nonadherent dura. Depending on whether it was rostral or caudal to the SCM, the neural placode was carefully detached from the dorsal dura using sharp dissection either before or after the median septum was excised. In four cases of hemimyelocele, only the hemicord that was the original hemiplacode needed to be detached from the dura; the other hemicord was shielded by the median septum and was never involved in the myelomeningocele sac. In the fifth and sixth cases of hemimyelocele, a stalk of central neural tissue arose from one hemicord and entered the neck of a lumbar myelomeningocele sac just rostral to the Type I bone spur (44). This neural stalk was simply incised flush with the surface of the hemicord and the sac was removed. In the two cases of cervical myelomeningocele and limited dorsal myeloschiosis (44) , the tethering elements of the dysraphic sac consisted of taut fibrous bands, blood vessels, dorsal nerve roots and strands of central neural tissue stretching from the dorsomedial aspect of both hemicords into the neck of the myelomeningocele sac. These elements were simply cut flush with the hemicords. SCM associated with dermal sinus tracts and dermoid cyst If the endomesenchymal tract responsible for the splitting of the neural tube maintains connection with the surface ectoderm, the midline mesenchymal septum of the SCM will be in continuity with a dermal sinus tract or a dermoid cyst if the squamous lining of the tract becomes encysted (44). There were three examples of dermal sinus related to a Type I septum in which the tract ended in the median bone spur (44). This type of dermal sinus did not contribute to the tethering and was simply excised with the bone spur. There were also three examples of dermal sinus tract and associated dermoid cyst associated with Type II SCMs. In two cases, the dermoids were small but because they were tautly inserted on the fibrous median septum, they significantly contributed to the tension on the hemicords. In this situation, it was essential that the entire sinus tract was excised with the fibrous septum for remnants of squamous cells could regrow within the median cleft. In the other case, the dermoid cyst was enormous and besides being inserted tautly on the median fibrous septum and producing tethering, its massive size also compressed the hemicords and possibly also impeded normal ascension of the developing neural tube (44). The entire cyst was excised with its sinus tract. SCM associated with other tethering lesions The two most common coexisting tethering lesions were myelomeningocele manqué and thickened filum. As described (44), myelomeningoceles manqué consisted of taut fibrous bands admixed with dorsal paramedian nerve roots and blood vessels that tethered the hemicords to where the

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

fibrous septa were found near the caudal end of the split. The length of split in Type II lesions were, by comparison, much shorter than that of Type I lesions, and, because all fibrous septa were thin, the hemicords in Type II SCM were apposed much closer together, with very little "free" part. The shape of the Type II septa varied from a broad rectangular or trapezoid sheet to a narrow triangular sail, but one invariant feature was that the point of attachment between the hemicord and septum was always rostral to the point of attachment between dura and septum. This was true of all three kinds of fibrous septa, giving the appearance that the fibrous septum was dragged upward by rostral movement of the cord occurring after formation of the primordium of the septum. This upward dragging converted the incomplete septa into backward-pointing oblique sheets and the complete septa into v-shape structures with the apex pointing rostrally. Central blood vessels were always seen with the Type II septa, either as a marginal artery skirting the rostral or caudal edge of the septum, or as a leash of parasagittal vessels loosely incorporated with the septum on either side (44). A true arteriovenous malformation was seen within the median cleft of one case (44). In two cases, the vessels and ventral septum entered the dural sac through an obvious ventral dural defect (44). Paramedian dorsal nerve roots were noted in 13 Type II SCMs. They coursed dorsally after emerging from the dorsomedial aspect of the hemicords, and either ended blindly within the septum or penetrated the dorsal dura with the septum (44). In spite of the common occurrence of paramedian dorsal roots, paramedian ventral roots were seen only in two patients (44). Release of tethering was simply accomplished in each Type II lesion by cauterizing the central vessels and excising the median fibrous septum and the nonfunctional paramedian roots.

Correlation between neural imaging and surgical findings Table 4 shows the relative sensitivity of MRI and CTM in predicting the existence of various anatomical structures of SCM that were later verified at surgery. In general, CTM was superior to MRI in almost all categories. In some categories, the superiority of CTM over MRI was obvious. For example, CTM delineated the size, shape, and configurations of bony (Type I) septa much better than MRI. If a typical bone spur contained cortical bone and a fatty marrow core, the MRI displayed the high signal intensity of fat bordered by lines of signal void in the T1-weighted image, but some bone spurs did not have a marrow core, and the signal void of bone was sometimes lost among a confusing mosaic of neural tissue, fibrous tissue, blood vessels, and the uncertain signal intensity of flowing CSF. In contrast, the bone algorithms of CTM displayed the bone spur to great advantage. By varying the window width, CTM also gave excellent details of the hemicord outline, the shape of the thecal sac(s), and the orientation of the hemicords in relation to the bony septum (Fig. 13). Even the fine sagittal sliver of iohexol between the hemicords seen in some Type II lesions at the two extreme ends of the split was well shown by CTM, so that the exact length of the split could be computed. Although MRI revealed intramedullary syrinx better than CTM, it was often unsatisfactory in defining the interphase between CSF and cord and in estimating the exact length of the split. For similar reasons, CTM was far superior than MRI in delineating the hypertrophic neural arches and the vertebral body abnormalities. In other categories, CTM and MRI had comparable sensitivities. For example, they were equally satisfactory in displaying associated tethering lesions such as thickened fila, lipomas, and the dural

adhesion of a previously treated myelomeningocele or hemimyelocele placode. With the dermal sinus tract and dermoid cyst, CTM and not MRI predicted one case of Type II SCM and a small dermoid by revealing a cylindrical mass in addition to the thickened filum within the thecal sac. In another case of Type II SCM and a large dermoid cyst, both CTM and MRI revealed the large intrathecal mass, but only MRI clearly showed that it was not lipomatous and thereby inferred the identity of a dermoid. In the three cases of Type I SCM and dermal sinus tract, neither CTM nor MRI predicted the tract that lay exclusively extrathecal. The most problematic diagnostic categories were the fibrous septa of Type II lesions and other median cleft contents such as myelomeningocele manqué and paramedian dorsal roots. Of 18 cases of pure Type II lesions, CTM only revealed three fibrous septa for certain (Fig. 14) and suggested the presence of two others. A particularly well-displayed Type II fibrous septum disclosed by CTM was that belonging to the middle section of one of the two composite split cord lesions, in which the complete fibrous sheet followed the same oblique orientation of the two bone spurs on each end of the split (44). No fibrous septa were ever identified for certain by MRI. This leaves 13 cases of Type II lesions without preoperative visualization of any "tethering" median sagittal lesion. Moreover, only three CTMs and none of the MRI revealed the single or multiple myelomeningoceles manqué found in 24 cases of SCM (Fig. 15). Because paramedian dorsal nerve roots most often took a caudal or rostral course to reach the median structures, they were seldom "caught" in a horizontal plane by the axial CTM, and their presence was predicted only by seven CTMs of 30 patients found at surgery to have these roots. Only 1 of the 21 MRIs vaguely suggested their presence. Lipomas within the median cleft were shown on both CTM and MRI, but central vessels were not predicted except for the one case of Type II SCM with an arteriovenous malformation between the hemicords (44). Treatment outcome Response of neurological symptoms to treatment The response of different preoperative symptoms to surgery varied considerably. In Table 5, the preand postoperative status of the 39 patients are listed according to their preoperative symptoms. The postoperative status denotes the patient's neurological examination 3 months after surgery. Transient compromise of neurological function was not counted as "worsened" unless it lasted longer than 3 months. Among the 24 symptomatic patients, including all 8 adults and 16 children, pain was a dominant complaint in 14 and was also the symptom that responded most dramatically to treatment. In 13 patients, pain either completely abated or was substantially reduced. The hyperpathia and dyesthetic component always improved, and a mild form of localized backache that was thought to be tolerable remained in 3 adults. In most cases, the improvement

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

fibroneurovascular stalk penetrated the dura. These bands arose from the cord or hemicords just rostral to, caudal to, or at the midline septum but traveled away from the septum in a caudal direction toward the dorsal dura. Frequently, multiple stalks were found in association with a single SCM. Their distinct course made it easy for excision flush with the surface of the hemicords. Twenty-five patients had thickened filum, and these were all incised after treatment of the SCM. None of the cervical or high thoracic lesions had an abnormal filum. Only seven cases of lipomas could be considered as a co-existing tethering lesion: six were terminal lipomas and were simply excised, and one was a dorsal lipoma tethering the cord rostral to the split. There were six instances of small lipomas within the median cleft of the SCM either rostral or caudal to the septum; three of these were Type I SCMs and the other three were Type II lesions. These lipomas were not attached to the surrounding dura and were therefore "nontethering" to the hemicords. They may simply represent fatty differentiation of the pluripotential mesenchyme within the original endomesenchymal tract (44).

Overall outcome and type of SCM Table 6 compares the overall outcome between the two types of SCM. A patient is said to have improved if one or more of the preoperative symptoms responded favorably to surgery and none has been made worse. A patient was "stable" when there was no change in the pre- and postoperative status. A patient was designated worsened if any one of the six categories of symptoms listed in Table 5 had progressed or was made worse by surgery. If a patient experienced improvement of one symptom but worsened in other categories, the overall well-being and functional status of the individual had to be

assessed to determine the overall outcome. Overall, 35 patients either improved or remained stable after surgery. Among the 19 stable patients were 5 patients who had progressive deterioration before surgery, and cessation of progression clearly represented benefit. Among the 4 patients that had worsened, 1 with a Type II SCM experienced the progression of a foot deformity in spite of improvement in other areas, and thus only 3 patients were actually made worse by the operation. Among these 3 were 1 child with worsened leg function and 2 others with worsened bladder function. Table 6 also shows that the treatment outcome between Types I and II SCM did not differ significantly. Complications Two children with previously treated myelodysplasia had CSF leakage and shunt infections. In both cases, the operative site was through the old incision and the soft tissue coverage was tenuous. Postoperative urinary retention and worsening of urinary incontinence were noted in 9 patients but in only two were these permanent. Seven of these 9 patients had Type I lesions, and 1 had a composite lesion. DISCUSSION The results of this study may be discussed from two different standpoints: clinical significance and embryogenesis. Clinical significance Clinical features Cutaneous lesions are exceedingly common in SCM, among both adults and children. Excluding the 13 cases of open myelomeningoceles (Group III children), 23 patients had "closed" skin abnormalities. Thus, in the whole group, only 3 patients truly had no discernible cutaneous clue of either previously treated open neural tube defect or underlying occult spinal dysraphism. Among these "closed" skin abnormalities, hypertrichosis was by far the most common, being present in 56% of all patients. This agrees with the consensus opinion that hypertrichosis is the most frequent cutaneous marker of SCM, with incidence varying between 20 to 55% (11,13,19,20,26-28,36,39,46,50) . This correlation between hypertrichosis and SCM is probably higher than any other combination of cutaneous abnormality and underlying spinal cord lesion. Capillary hemangioma was the next most common skin abnormality, but it seemed to be part of the hairy patch complex because only three were without overlying hair. The correlation of capillary hemangiomas with SCM is probably no higher than with other occult dysraphic lesions (1,7,9,24,25,32). There were six cases of dermal sinus opening, three of which were connected with an intraspinal dermoid cyst. This is less prevalent than the reported incidence of 15 to 40% in other series (13,26,29-31,36,46) and is also not a reliable surface indicator for SCM. It is interesting that only 5 of 13 children with open myelomeningoceles had a

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

occurred within 1 month. In 1 adult patient receiving chronic disability compensation, the pain became generalized from back and leg to upper spine and shoulders. His postoperative neural imaging studies did not reveal remaining or additional pathology. Twenty-nine patients had some degree of sensorimotor deficits preoperatively, but in only 19 patients were the deficits progressive. Surgery improved the deficits or halted their progression in these 19 patients, the completeness of recovery being inversely proportional to the duration of symptoms before surgery. Nine of the 10 patients with static preoperative deficits had no alteration of status after surgery, but 1 child with myelomeningocele experienced worsening of motor function. Nineteen patients had bladder and bowel symptoms preoperatively. In 9 patients, the symptoms were progressive: of these, 4 improved, 4 remained stable, and 1 worsened after surgery. Of the 10 patients with long-standing sphincter dysfunction, 9 were unchanged and 1 was worsened by surgery. Thus, only those patients with sphincter dysfunction of relatively recent onset had any hope of recovery, and among those that improved, none had complete normalization of function. No patient with flaccid external anal sphincter or large capacity bladder improved. All 4 patients that improved had preoperative small capacity bladders, uninhibited detrusor contractions, and intermittent incontinence because of high vesicular pressures and incompetent sphincters. In the other 4 patients, designated as "stable," their cystometrogram and sometimes their bladder symptoms had actually altered to some extent, but because they required as much bladder and bowel care as before, their lifestyles could not be considered improved. Among the five patients with progressive scoliosis before surgery, four had no change in their spinal alignment after surgery, but one patient's spinal deformity continued to worsen and ultimately required instrumentation. Among the 16 patients with nonprogressive spinal deformity preoperatively, 14 remained stable but two worsened after surgery and subsequently required surgical fusion. Two of three patients with progressive foot deformity had cessation of fibro-osseous changes, but one patient continued to worsen and required tendon transfer and partial arthrodesis of the ankle. All five patients with sympathetic dystrophy of the feet improved after surgery.

common among children. In fact, the overall clinical spectrum of SCM is somewhat similar to that described for the adult and childhood counterparts of tethered cord syndrome (41-43,45) , which is not surprising because SCM is a form of cord traction and it also most commonly involves the lumbosacral cord. There is, however, one notable exception: the group of patients with SCM who showed a great discrepancy between the neurological functions of the left and right legs, which can be as dramatic as a flaccid limb opposite a perfectly normal mate. This neurological pattern is seldom seen in other tethered cord lesions. Even among SCMs, it is only seen in two specific circumstances. It is most often associated with a hemimyelocele in which the hemicord involved in the myelocele corresponds to the side of the poor neurological function. This was previously described by Duckworth et al. (12) and Gilmor and Batnitzky (17) , but in Duckworth et al.'s cases (12), the hemineural placodes were terminal and ended within the myelocele sac. In our cases, the hemiplacodes were segmental, with a variable amount of neurulated cord distal to the hemimyelocele sac (44). The neurological function below the level of the hemimyelocele is roughly proportional to the size of this distal hemicord, which suggests that the left-right functional discrepancy is caused by the unilateral neurulation error and not by the SCM itself (44). Less often, bilateral functional discrepancy is associated with asymmetric splitting of the spinal cord in which an oblique median septum divides the neural plate into a large "major" hemicord and a small "minor" hemicord, the latter corresponding to the side of the poor function. In these cases, the oblique course of the endomesenchymal tract has not only divided the population of segmental neurons unevenly for the two halves of the body but also seems to have deprived the minor cord of ascending and descending axons. The pathological specimens reported by Rokos (47) and Emery and Lendon (15) showed that not only were the anterior and posterior gray horns rudimentary in the minor hemicords, their white funiculi were also thin and disorganized. The major hemicords, on the other hand, were frequently "overendowed" with three gray columns of robust neurons. There is no clinical correlate of this entity in the literature, but judging from the 5 patients reported here, the "over-endowed" major hemicord does not cross over to innervate the side of the minor hemicord. The clinical significance of all this lies in the fact that both circumstances of left-right functional discrepancy are associated with distinct pathological entities that are eminently demonstrable by imaging studies. Any child with a previously treated "routine" myelomeningocele who presents with functional discrepancy should be studied with MRI and CTM to rule out a SCM and a "missed" hemimyelocele. Just as important, any child with a known SCM who presents with functional discrepancy should also be carefully studied to display the "minor" hemicord, for this delicate structure is exquisitely prone to injury at the time of resection of the median septum, and a

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

cutaneous lesion distinct from the sac, whereas 23 of 26 patients without myelomeningocele did. This disparity suggests that hypertrichosis, capillary hemangiomas, and dermal sinus tracts all represent minor aberrations in the development of the surface ectoderm brought about by the adverse influence of a dorsal endomesenchymal tract, but these aberrations might be largely overshadowed by the chaotic changes in the surface ectoderm occasioned by nonneurulation of the underlying neural plate in the case of an associated myelomeningocele (44). In this series, all the adult patients had progressive neurological deficits, but just over half of the children with SCM had neurological symptoms. This discrepancy must be carefully interpreted and not be too readily attributed to a difference in pathophysiology between the two age groups. First, children designated as asymptomatic (Groups II, IIIb, and IIIc) were not neurologically normal; in fact, most had obvious sensorimotor and urological deficits. They were "asymptomatic" only because they themselves or their parents observed no frank deterioration or complaints. Slow deterioration might well have been compensated for by the children or overlooked by the parents. Second, more than half of these "asymptomatic" children were younger than 2 years. Subtle changes in function at such a young age are notoriously difficult to detect, and the true incidence of progressive deficits may be higher than reported here. Finally, referral bias may account for some of this discrepancy. Most physicians are well aware of the ample data that equate tethered cord in childhood to a progressive spinal cord disease (23,41) and rightly refer asymptomatic children for prophylactic surgery. In contrast, the evidence to support prophylactic surgery in asymptomatic adults with tethered cord is much less convincing (42,45), and most adults were referred for treatment only after the onset of symptoms. There is currently no published data on the natural history of tethered cord syndrome in asymptomatic adults. It is perhaps advisable to operate on asymptomatic and otherwise healthy adults with SCM if they lead a physically active lifestyle, because trauma has been known to precipitate neurological deterioration (42,45). Older asymptomatic adults leading sedentary lifestyles are probably best managed expectantly. Among symptomatic patients, the neurological picture differs considerably between adults and children. Pain is a universal and dominant feature among adults, and most frequently it has a dysesthetic quality and involves the perineal or perianal region. Pain was only noted in 6 of 16 children with symptoms, and none described dysesthesia or involvement of the perineal region. All eight adult patients also had either new or worsening bladder symptoms. None of the 16 symptomatic children reported deterioration of bladder functions, although 5 of these were below toilet training age and 2 others had a complete absence of bladder function since birth. That still leaves over half of the symptomatic children with normal preoperative bladder function. Also, progressive kyphoscoliosis and foot deformities were not seen in adults but

bone spur of a Type I lesion. For the group of 24 symptomatic patients, the Type I:Type II ratio is an even 13:11. The second debated issue concerns tethering of the cervical cord. Tethering at this level has not received much attention, partly because high dysraphic lesions are rare and partly because the normally slight ascent of the cervical cord seemingly argues against much transmitted tension at the site of the tethering. However, several recent reports indicate that tethering at the cervical segments does indeed produce neurological deterioration. Eller et al. (14) reported a patient with progressive dorsal column dysfunction who had a dorsally cleft cord at C1 tethered to the dorsal dura by taut fibroneural bands. Vogter et al. (51) described an infant with a skin covered pedunculated cervical myelomeningocele who deteriorated neurologically during the first year of life. Simpson and Rose (49) and Gower (18) also reported two other examples of cervical split cords with progressive neurological deficits. In this series, 4 of the 6 (67%) cervical SCMs had definite neurological deterioration, and 21 of the 33 patients with thoracolumbar lesions had symptoms (64%). In fact, because cervical SCMs frequently cause deterioration of hand functions in addition to spastic legs, they are potentially more disabling lesions that ought to command more instead of less attention. One helpful diagnostic hint for cervical SCMs is their high association with the Klippel-Feil syndrome in the cervical spine, so that the child often has fixed torticollis, a short neck, raised and webbed trapezial borders, and a low hair line. Children with this appearance should receive an MRI to rule out an SCM and other related neuropathological anomalies.

Pathophysiology The results of this study also cast light on two much debated issues concerning the pathophysiology of double cord malformations. The first issue concerns the question of tethering in diplomyelia. A number of authors remarked that double cord malformations that do not possess median bone spurs and reside within single dural sacs (i.e., diplomyelias or Type II SCMs) do not present as obvious a picture of cord tethering as the Type I SCM where the cord is solidly transfixed by a rigid spur (5,10,22,37,38,48). In this study, a stiff fibrous septum was found in the median cleft of each of the 18 Type II lesions explored. By virtue of their adhesions to this fibrous septum, the hemicords were tautly tethered to the adjacent dura where the septum penetrated the dural sac. Except for the septum being completely intrathecal and nonosseous, the impression of cord transfixation was, to the author, as vivid as if a dura-wrapped bone spur had been present. Owing to the belief that cord duplication was more an embryological curiosity akin to aborted twinning than an actual lesion that could cause neurological deficits (2,20,38,39,48), many still favored a nonoperative approach for diplomyelia (40,48). In this study, the large number of Type II lesions among the symptomatic patients strongly suggests that the fibrous septum is just as damaging to the cord as the

Correlation between radiological and surgical findings The new classification of SCMs based on the number of dural tubes and the nature of the midline septum allows all double cord lesions to be typed using preoperative imaging studies. These two radiographic features do not overlap between the two types of SCM, and, in each case, these features were confirmed at surgery. Preoperative typing mentally prepares the surgeon for the degree of technical complexity that could be encountered, for a Type I lesion is technically much more demanding than a Type II lesion. For example, the Type I bone spur is always attached dorsally to the neural arches, which are frequently hypertrophic and fused with neighboring laminae. This bony mass can be cumbersome to remove; it is usually vascular, and it also hides the attachment of the bone spur from view. If the exact location of this attachment is not appreciated while using the osteotomy instrument, the surgeon could inadvertently knock the bone spur about and cause injury to the hemicords. Extradural removal of the bone spur also has to be done with minimum manipulation to the median dural sleeve because the underlying hemicords are often adherent to it. During subsequent resection of the dural sleeve, its caudal margin must be carefully freed to avoid injury to the delicate crotch of the reunited cord. In a

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

timely recognition of the minor hemicord prepares the surgeon for any extraordinary strategy that might be helpful in preserving function. Previous pathological studies of children with open myelomeningoceles revealed a high incidence of associated SCMs. Campbell et al. (8) found 36 cases of SCM in 100 infants with open dysraphism; and Emery and Lendon (15) reported 78 cases among 100 dysraphic patients. Conversely, clinical series of patients with SCM also showed a frequent association with myelomeningoceles, with incidences varying between 26 to 80% (21,38,39,50). The 13 children in this study with associated myelomeningoceles represent approximately 5% of the total population of children in our own spina bifida clinic, but most of these children had not yet been screened for SCM. Our current policy of ordering MRI only on spina bifida children with some unusual features such as left-right functional discrepancy, discrepancy between sac location and neurological level, and the finding of a segmental placode is obviously inadequate in accounting for the true incidence of SCM in open dysraphism. If 13 children turned up with unsuspected SCM through this limited protocol within a short period of 2 years, the incidence of SCM in children with open neural tube defect must ultimately prove to be much higher with more vigorous MRI screening. Considering the high likelihood of progressive and irreversible neurological deterioration if an SCM is unrecognized, it is arguably justified to begin screening every spina bifida child with an MRI. If an SCM is found, it should be treated, because its deleterious effect on the cord is independent of the state of the treated neural placode.

Treatment outcome In general, pain responds best to surgical therapy and still remains the best indication for surgery among adult patients. Sensorimotor deficits are the next best responder, but the trend is clear that only deficits of recent onset improve after treatment. This trend is also true for bladder and bowel dysfunction that overall carries a 40% improvement rate, although stabilization of previously progressive bladder symptoms must also be counted as a major benefit. There were altogether one case of worsening foot deformity and three cases of worsening scoliosis after

surgery. Progressive foot deformity after successful release of tethered cords has been reported and was thought to be due to irreversible osseo ligamentous changes that were no longer influenced by the neurological condition (7,9,28,34,41-43,45). Continued progression of severe scoliosis after cord untethering is likewise thought to be due to the effect of gravity after the scoliotic curve had exceeded a certain critical angle (34). Thus, these osseous deformities in SCM patients may require orthopedic treatment in spite of the improved neurological status, and their continued progression after operative treatment of the SCM should not necessarily be regarded as adverse effects of surgery. The higher surgical morbidity of Type I SCM is probably related to the removal of the bone spur. The one instance of worsened leg function occurred in a child with a Type I SCM in which the cord was split into a large major hemicord and a small minor hemicord by an oblique bone septum. The exposure of the minor hemicord was hampered because it was partly sheltered by the overhanging oblique bone spur as well as being ventrally rotated away from the surgeon's view. It was inadvertently injured during removal of the bone spur. This unusual pattern of asymmetric splitting must be recognized before surgery as a signature of heightened risk, so that every effort can be made to avoid jarring the delicate minor hemicord while dealing with the blind underside of the oblique bone spur. Children with previously treated myelodysplastic lesions also tend to have more wound complications and CSF leakage because of the tenuous skin over the operative site. These patients are best managed in the full-prone position after surgery similar to the way newborns with myelomeningoceles are positioned after sac closure. Only one patient had late deterioration 4 years after a Type II septum was resected from the cervical region. An associated limited dorsal myeloschisis not accessible through the original exposure was not initially treated. At reoperation, there was no evidence of reattachment of the hemicords to the dura, and after simple excision of the fibroneural bands of the limited dorsal myeloschisis, the patient improved neurologically. This case underscores the importance of comprehensive diagnosis and treatment of all tethering lesions in patients with complex dysraphic anomalies. Embryological significance In Part I of the SCM study, it was suggested that several constant features in the median septum and hemicords of SCMs are explainable by the continued ascension, relative to the vertebral column, of the neural tube after its early transfixation by the endomesenchymal tract (44). This ascension was in turn attributable to the disparate growth between the neural tube and the developing vertebral column (6). The constant features mentioned include the invariable location of the median septum at the caudal extreme of the split cord, the oblique orientation of the Type II fibrous septum with its neural attachment always more rostral than its dural

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Type II SCM, all these precautions are unnecessary because there is no bone spur or dural sleeve. Release of a Type II lesion simply involves incising the fibrous bands or septa in full view of the hemicords and related vasculatures, which makes inadvertent neural injury unlikely. Table 4 clearly shows the superiority of CTM over MRI in defining the features of SCMs. CTM is much more accurate in delineating the size, shape, and orientation of the Type I bone spur and in displaying the topographic relationship of the bone or fibrous septum with the neural and vascular structures. It is also slightly better than MRI in detecting unexpected entities around the hemicords, such as dermal sinus tract and paramedian fibroneural bands. The only clear advantage MRI has over CTM is in identifying and sizing a syrinx near or within the hemicords. However, CTM does involve radiation; it necessitates a spinal puncture; and general anesthesia is often required in young children. Currently at our institutions, MRI is used as a screening test for any suspected case of SCM or other tethering lesions. If an SCM is found, an appropriate site for the spinal needle is determined on the MRI and a CTM is performed to display the fine details and to serve as a "surgical road map." Also apparent from the correlative chart is the fact that the fibrous septum of many Type II SCMs will be missed by both MRI and CTM. Many such septa and bands and almost all the myelomeningoceles manqué are beyond the resolution of today's imaging technology. Despite previous protestations that nonvisualization of a septum in a diplomyelia meant no tethering, this study shows that many Type II SCMs without radiographic evidence of a median septum turned out at surgery to have taut fibrous bands that were unquestionably tethering the hemicords. Because exploring a Type II SCM carries a low surgical morbidity and, conversely, not recognizing a tethering band may cost the patient irreversible neural damage, I recommend exploring all Type II SCMs regardless of whether imaging studies display a definite median cleft septum. Finally, the obligatory finding of at least one tethering lesion unrelated to and sometimes remote from the split cord in all SCMs found below the midthoracic level mandates including the whole neuraxis in the screening MRI. Only cervical SCMs are unlikely to be associated with a second tethering lesion, but even among the six cases reported here, one had a terminal lipoma.

Summary 1. The signs and symptoms of SCM are similar to

those described for the tethered cord syndrome. Adult patients present with severe dysesthetic pain in the legs and perineum, and sensorimotor findings. Children commonly present with gait disorder and less so with pain and progressive spinal and foot deformities. 2. The incidence of SCM among children with open myelodysplastic lesions is probably higher than currently recognized. This incidence is significantly increased if the children show an obvious left-right functional discrepancy of the legs or if their neurological level does not correspond with the level of the myelomeningocele sac. These latter children should be screened with MRI. 3. MRI is a useful screening test for SCM, but CTM is superior in delineating anatomical details of the split cord lesion and in predicting associated lesions. With these two tests, all SCMs can be accurately classified into Type I or II lesions preoperatively, with no crossover ambiguities. 4. Both Types I and II SCMs are equally likely to cause neurological deficits. 5. Even with the most sophisticated CTM and MRI, many Type II SCMs will not be shown radiographically to possess a definite fibrous septum even though one is always present to tether the hemicords. As with Type I lesions, all Type II SCMs should be surgically explored regardless of whether a definite septum is found on neural imaging. 6. At least one unrelated tethering lesion is found in all lumbosacral and lower thoracic SCMs and in a much smaller number of cervical SCMs. The entire neuraxis should be studied radiographically and if a second tethering lesion is found, it should be treated. 7. The surgical outcome is excellent for SCM. The overall surgical morbidity is low but slightly higher with Type I lesions. 8. The radiographic data on the split lengths when correlated with age, lesion type, and level of the septum are compatible with the hypothesis of an endomesenchymal tract transfixing the neural plate during early development of the neural tube. ACKNOWLEDGMENT The author wishes to thank Donald N. Krieger, Ph.D., for his assistance with statistical methods. Received, November 23, 1991. Accepted, March 3, 1992. Reprint requests: Dachling Pang, M.D., F.R.C.S.(C), F.A.C.S., Associate Professor of Neurosurgery, Chief, Pediatric Neurosurgery, Children's Hospital of Pittsburgh, 3705 Fifth Ave., Pittsburgh, PA 15213. REFERENCES: (1-51) 1. 2. 3.

Anderson FM: Occult spinal dysraphism: A series of 73 cases. Pediatrics 55:826-835, 1975. Andersson H, Sullivan L: Diastematomyelia. Acta Orthop Scand 36:257-264, 1954. Bardeen CT, Lewis WH: Development of the limbs, body-wall and back in man. Am J Anat

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

attachment, and a similar orientation of the associated myelomeningoceles manqué. The radiographic data from the present study on the split length lend further support to this assertion. First, the mean split length of the 19 cases of Type I lesions is significantly greater than the mean split length of the 18 cases of single Type II lesions. Because the cephalo-caudal width of the septum is usually very small, this essentially means that the "free" part of the split cord rostral to a Type I bone spur is longer than that of a Type II (fibrous) septum. It is also true that even though a bone spur may occasionally be obliquely oriented in the axial (horizontal) plane, it almost always takes the shortest and straightest reach across the spinal canal in the sagittal plane. The fibrous septum, on the other hand, is very commonly oblique upward in the sagittal plane. These contrasting features between the split lengths and the sagittal orientation of the septa of the two SCM types suggest that when the neural tube was transfixed firmly by a rigid (bony) derivative of the Type I endomesenchymal tract, its subsequent ascent results in a long cleavage in the cord but no alteration in the noncompliant shape of the median bony septum. If, instead, the neural tube is transfixed by a nonrigid fibrous derivative of the Type II endomesenchymal tract, its subsequent ascent drags with it the semicompliant fibrous septum and results in a lesser split length. Second, the mean split length of the 10 patients younger than 1 year is significantly shorter than the mean split length of those patients diagnosed after 1 year. Barson (6) and others (3-5,16,33) have shown that normal ascension of the spinal cord continues after birth for at least 1 year. It would therefore not be unexpected for the split length to increase further during the first year of postnatal life. Third, Barson (6) also showed that the difference in growth rates between the vertebral column and spinal cord gradually approaches zero by age 1 year, after which the relative position between the spinal cord and vertebrae at any given segment of the spinal column remains constant during subsequent lengthening of the trunk. The mean split lengths of any two age populations after age 1 should therefore be comparable. This is borne out by comparing the mean split lengths of the 10 patients between the ages of 1 to 3 years (2.4 levels) with that of the 19 patients between the ages of 3 years and 60 years (2.71 levels). Finally, Bardeen and Lewis (3) and Lassek and Rasmussen (33) have shown that the ascension of the cervical cord segments relative to the adjacent vertebra is less than that found in the thoracolumbar cord segments. Thus, an endomesenchymal tract transfixing the cervical neural tube should therefore result in a shorter split length than if the transfixation were at the lumbar level. This is indeed borne out by comparing the mean split length of the seven cases of high SCM (1.14 levels) with that of the 32 cases of low SCM (2.52 levels).

5. 6. 7. 8.

9. 10. 11. 12. 13.

14. 15.

16. 17. 18. 19. 20.

21. 22. 23.

24.

1:1-35, 1901. Barry A: A quantitive study of the prenatal changes in angulation of the spinal nerves. Anat Rec 126:97-110, 1956. Barry JF, Harwood-Nash DC, Fitz CT, Byrd SE, Boldt DW: Metrizamide in pediatric myelography. Radiology 124:409-418, 1977. Barson AJ: The vertebral level of termination of the spinal cord during normal and abnormal development. J Anat 106:489-497, 1970. Bruce DA, Schut L: Spinal lipomas in infancy and childhood. Childs Brain 5:192-203, 1979. Campbell LR, Dayton D, Sohal GS: Neural tube defects: A review of human and animal studies on the etiology of neural tube defects. Teratology 34:171-187, 1986. Chapman PH: Congenital intraspinal lipomas: Anatomic considerations and surgical treatment. Childs Brain 9:37-42, 1982. Cohen J, Sledge CB: Diastematomyelia. Am J Dis Child 100:257- 263, 1960. Dale AJD: Diastematomyelia. Arch Neurol 20:309-317, 1969. Duckworth T, Sharrard WJ, Lister J, Seymour N: Hemimyelocele. Dev Med Child Neurol (Suppl) 16:69-75, 1968. Eid K, Hochberg J, Saunders DE: Skin abnormalities of the back in diastematomyelia. Plast Reconstr Surg 63:534549, 1979. Eller TW, Bernstein LP, Rosenberg RS, McLone DG: Tethered cervical spinal cord: Case report. J Neurosurg:67:600-602, 1987. Emery JL, Lendon RG: The local cord lesion in neurospinal dysraphism (meningomyelocele) J Pathol 110:83-96, 1973. Fitz CR, Harwood-Nash DC: The tethered conus. AJR 125:515- 523, 1975. Gilmor RL, Batnitzky S: Diastematomyelia-Rare and unusual features. Neuroradiology 16:87-88, 1978. Gower DJ: Diastematomyelia--A 40 year experience. Pediatr Neurosci 1 988; 14:90-96. Guthkelch AN: Diastematomyelia with median septum. Brain 97:729-742, 1974. Guthkelch AN: Diastematomyelia, in Wilkins RH, Rangachary SS (eds): Developmental Anomalies and Neurosurgical Diseases in Childhood. New York, McGraw-Hill Book Company, 1985, pp 2058-2061. Hendrick EB: On diastematomyelia. Prog Neurol Surg 4:277-288, 1971. Herren RY, Edwards JE: Diplomyelia (duplication of the spinal cord) Arch Pathol 30:1203-1213, 1940. Hoffman HJ: The tethered spinal cord, in Holtzman RNN, Stein BM (eds): The Tethered Spinal Cord. New York, Thieme-Stratton, 1985, pp 91-98. Hoffman, HJ, Hendrick EB, Humphreys RP: The tethered spinal cord: Its protean manifestations, diagnosis and surgical correction. Childs Brain 2:145-155, 1976.

25.

26.

27.

28.

29. 30. 31. 32. 33. 34.

35.

36.

37.

38. 39. 40.

41. 42.

Hoffman, HJ, Taecholarn C, Hendrick EB, Humphreys RP: Management of lipomyelomeningocele. Experience at The Hospital for Sick Children, Toronto. J Neurosurg 62:1-8, 1985. Hood RW, Riseborough EJ, Nehme AM, Micheli LJ, Strand RD, Neuhauser EB: Diastematomyelia and structural spinal deformities. J Bone Joint Surg 62A:520-528, 1980. Humphreys RP, Hendrick E, Hoffman HJ: Diastematomyelia, in Weiss MH (ed): Clinical Neurosurgery, Baltimore, Williams & Wilkins, 1982, vol 23, pp 436-456. James CCM, Lassman LP: Spinal dysraphism: The diagnosis and treatment of progressive lesions in spina bifida occulta. J Bone Joint Surg 44B:828-840, 1962. James CCM, Lassman LP: Diastematomyelia. Arch Dis Child 39:125-130, 1964. James CCM, Lassman LP: Diastematomyelia in spina bifida occulta. Psychiatr Neurol Neurochir 70:453-457, 1967. James CCM, Lassman LP: Diastematomyelia and the tight filum terminale. J Neurol Sci 10:193-196, 1970. James CCM, Lassman LP: Spinal Dysraphism: Spina Bifida Occulta. London, Butterworths, 1972, pp 25-28, 89-97. Lassek AM, Rasmussen GL: A quantitative study of the newborn and adult spinal cords of man. J Comp Neurol 69:371-379, 1938. Lassman LP, James CCM: Lumbosacral lipomas: Critical survey of 26 cases submitted to laminectomy. J Neurol Neurosurg Psychiatry 30:174-181, 1967. Levene H: Robust tests for equality of variance, in Olkin, I (ed): Contributions to Probability and Statistics. Palo Alto, California, Stanford University Press, 1960, pp 278-292. Matson DD, Woods RP, Campbell JB, Ingraham FD: Diastematomyelia (congenital clefts of the spinal cord). Diagnosis and surgical treatments. Pediatrics 6:98-112, 1950. McCaig CD: Myoblasts and myoblastconditioned medium attract the earliest spinal neurites from frog embryos. J Physiol 375:3954, 1986. Meacham WF: Surgical treatment of diastematomyelia. J Neurosurg 27:78-85, 1967. Moes CAF, Hendrick EB: Diastematomyelia. J Pediatr 63:238-248, 1963. Naidich TP, Harwood-Nash DC: Diastematomyelia: Hemicord and meningeal sheaths; single and double arachnoid and dural tubes. AJNR 4:633-636. 1983. Pang D: Tethered cord syndrome. Neurosurgery: State of the Art Reviews 1:4579, 1986. Pang D: Tethered cord syndrome in adults, in Holtzman, R, Stein B (eds): The Tethered Spine. Thieme-Stratton, New York, 1985, pp

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

4.

44.

45. 46. 47. 48.

49.

50. 51.

99-115. Pang D: Tethered cord syndrome: Newer concepts, in Wilkins RH, Rangachary SS (eds): Neurosurgery Update II. New York, McGraw-Hill, 1991, pp 336-344. Pang D, Dias M, Barmada-Abdah M: Split cord malformation. Part 1: A unified theory of embryogenesis for double cord malformations. Neurosurgery 31:000-000. Pang D, Wilberger JE: Tethered cord syndrome in adults. J Neurosurg 57:32-47, 1982. Perret G: Diagnosis and treatment of diastematomyelia. Surg Gyn Obstet 105:6983, 1957. Rokos J: Pathogenesis of diastematomyelia and spina bifida. J Pathol 117:155-161, 1975. Scotti G, Musgrove MA, Harwood-Nash DC, Fitz FR, Chuang SH: Diastematomyelia in children: Metrizamide and CT metrizamide myelography. AJR 135:1225-1232, 1980. Simpson RK, Rose JE: Cervical diastematomyelia: Report of a case and review of a rare congenital anomaly. Arch Neurol 44:331-336, 1986. Till K: Spinal dysraphism: A study of congenital malformations of the back. Dev Med Child Neurol 10:470-477, 1968. Vogter DM, Culberson JL, Schochet SS, Gabriele OF, Kaufman HH: High spinal dysrhaphism: Case report of a complex cervical meningocele. Acta Neurochir (Wein) 84:136-139, 1987.

COMMENTS The authors have clarified the clinical syndromes associated with the split cord syndrome (SCM). Their categorization of the SCM into a Type 1 and Type II malformation is certainly superior to labeling them all as diastematomyelia. In this article, the authors convincingly demonstrate that neuroimaging frequently fails to show the fibrous bands present in SCM II, even though all of these Type II SCMs do have such fibrous bands and require exploration at the level of the split regardless of whether neuroimaging shows such a band. This concept is a radical departure from our thoughts about this condition in the past. We have long been aware that patients with an SCM II have tethering at the distal end of the spinal cord, which results in progressive symptomatology. The treatment has generally been to untether the cord at the point of tethering at the distal end of the dural sac. The authors have provided us with convincing evidence that in addition to untethering the cord distally in such patients, one must consider exploring the area of the split, even though current imaging studies will show no evidence of any bands or fibrous septa at the level of the split.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

43.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 1. Presenting signs and symptoms of 13 Group I children with SCM.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 2. Cutaneous stigmata of a patient with Type II SCM: A, hypertrichosis B, capillary hemangioma beneath the hypertrichotic patch.

Figure 4. Location of the most caudal extent of the median septum of all 39 SCM patients. The majority of septa are found in the lumbar region, most frequently at L3. There is a much smaller peak at the cervical region.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 3. Presenting signs and symptoms of 8 adult patients with SCM. The hatched bars represent current complaints or progressive deficits. The solid bars represent static problems.

Figure 6. Location of the median septum in children and adults with SCM showing comparable distributions.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 5. Location of the caudal extent of the median septum of all 39 patients correlated with SCM types (the two composite SCMs were counted as Type I in this graph). All Type I septa are in the low thoracic or lumbar segments, whereas Type II septa are distributed over both cervical and thoracolumbar regions.

Figure 8. Correlation between the location of the median septum (diamond, the most caudal extent of the median septum) and that of the conus (dot). Cases are arranged with decreasing level of the septum from left to right. Six of seven cervical and upper thoracic SCMs have normally located coni; whereas all SCMs with septa below T7 are associated with a low-lying conus. The two dotted lines represent the distance between the rostral extent of the upper septum and the caudal extent of the lower septum in the two composite SCMs.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 7. Location of conus. Only 6 coni were in normal locations, i.e., above the lower border of the L1 vertebral body. Thirty-three others were abnormally low.

Figure 10. Correlation between split length and location of median septum. The mean split length of the 7 cases of cervical and high thoracic SCMs is significantly shorter than that of the low thoracic and lumbar SCMs.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 9. Correlations between split length and age. The shaded part and the hollow bars highlight the 10 infants and their split lengths, which are mostly shorter than those of patients diagnosed after 1 year of age. The difference is statistically significant.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 11. Vertebral body abnormalities in SCM, arranged by SCM types. A horizontal cross-hatched bar across two or more abnormalities means a body with two or more abnormalities identified by imaging studies. The number inside each bar denotes the number of patients with similar abnormality(ies).

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 12. Surgical treatment of a Type I SCM. A, partial resection of the bone septum (arrowheads) and its subperiosteal dissection from the investing dural sleeve. B, after dural opening along both sides of the bone septum, the medial aspects of each hemicord (hc) are adherent to the dural sleeve (DS), and the caudal extent of the sleeve is tightly wedged against the reunion of the hemicords. C, dural sleeve (DS) retracted caudally to show its rostral base. The safest area to start the dural sleeve resection is at its rostral end where there is a "free area" between the hemicords (hc). D, after complete resection of the dural sleeve flush with the ventral dural wall.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 13. Comparative reliability between MRI and CTM in preoperative SCM typing. Two-year-old child with a Type I SCM. A, MRI was interpreted as showing a Type II SCM with a single thecal sac and no midline bone septum. B, CTM shows the double thecal sacs and a thin midline bone septum (arrowhead), identifying it as a Type I lesion. C, at surgery, the small bone septum (held by forceps) is seen in the caudal extent of the split, surrounded by a complete median dural sleeve.

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 14. Comparative sensitivity between MRI and CTM in detecting Type II septa. A child with a Type II SCM proven at surgery. A, axial CTM at the caudal extent of the split shows a definite dorsal midline septum and a single thecal sac. B, MRI at the equivalent level does not show a septum.

Table 1. Cutaneous Manifestations of Split Cord Malformation

Table 2. Eight Pediatric Patients with Left-Right Functional Discrepancy

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Figure 15. CTM of a Type II SCM shows the single thecal sac, a right "minor" hemicord, a left "major" hemicord, and a separate dorsal, round, filling defect (arrow). At surgery, the round dorsal lesion proves to be the fibroneurovascular bundle of a myelomeningocele manqué.

Table 4. Sensitivities of Computed Tomographic Myelography and Magnetic Resonance Imaging in Predicting Anatomical Structures of Split Cord Malformations

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Table 3. Associated Secondary Tethering Lesions in Split Cord Malformation (SCM)

Table 6. Overall Outcome of Split Cord Malformation (SCM)

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.

Table 5. Outcome of Individual Symptoms and Neurological Deficits

Split cord malformation: Part II: Clinical syndrome.

Thirty-nine patients with split cord malformations (SCM) were studied in detail with respect to their clinical, radiographic, and surgical findings as...
2MB Sizes 0 Downloads 0 Views