Diastematomyelia

Patients and Methods

Retrospective data were collected on 138 patients treated at the Second Affiliated Hospital, Xi’an Medical University, between May 1978 and April 2010. Ethical approval for this investigation was obtained. Data on symptoms, differential diagnosis, type and location of the spinal deformity, treatment, length of follow-up, and outcome were collected from a central database and analysed.

All patients underwent a whole spine radiograph. Imaging using CT, MRI and CT myelography (CTM) was available in our institution from 1983, 1992, and 1995, respectively. A number of patients had these latter examinations when the radiological image failed to give a clear diagnosis. All patients were reviewed at three and six months and then yearly after operation except for one Type 1 diastematomyelia patient who died from an anaesthetic complication during the surgery.

After epidural anesthesia, Type 1 diastematomyelia patients were placed prone. A posterior midline approach was used. The affected lamina was removed. A paramidline incision was made in the dura and it was found that the spinal cord was often adherent to the dura in the midline. Bone, cartilage and the fibrous band were identified microsurgically and local adhesions between the vertebral body and dura were divided and the fibrous band, if present, was resected. The spinal evoked potential of the patients was monitored intra-operatively in order to avoid damage to the spinal cord. Any malformations inside or outside the spinal canal, for example lipoma, hemangioma, skin sinus, tethered terminal filament were also resected. Local reconstruction was undertaken using a “spindle shaped resection” in order to change the double dural sac to a single dural tube and allow the double spinal cord to move freely inside the single dural tube. This involved a lateral incision in the dura which was left unsutured while the dorsal side of the dura was sutured tightly. A drainage tube was placed in the epidural space after surgery.

The surgical procedure for Type 2 differs as the two separate spinal cords are located in one dural tube, meaning that spindle-shaped resection of the dura was not required after resection of the fibrous septum and the malformations inside or outside the spinal canal, and only a tight suturing of the dorsal dura was needed.

Results

Basic demographic data of the 138 patients are shown in Table I. The male-to-female ratio was 1:3. A total of 18 patients (13%) were aged ≥ 16 years. The clinical manifestations are summarised in Table II. Only Type 1 patients presented with unilateral lower-limb muscle atrophy and malformations of the lower limb.

The results of imaging studies are shown in Table III. An important radiological finding seen on plain X-rays was a longitudinal shadow at the centre of the spinal canal (Fig. 1a).

Table III

Summary of imaging results from radiography, myelography, CT scan, and MRI

Imaging type and definitions (n, %)
Radiography (n = 138)
Lateral deformation of spine (scoliosis)	116 (84.1)
Longitudinal ridge shadow at the centre of the spinal canal (longitudinal bony spur)	75 (54.3)
Myelography (n = 60)
Typical ‘island-like‘ filling defect	54 (90.0)
Dense ridge shadow at the defect	46 (76.7)
CT scan (n = 116)
Spinal canal and epidural mater separated into two parts	90 (77.6)
Hemicords asymmetry	78 (67.2)
Bony septum*
Flattened cylindrical-like	9 (7.8)
Branch-like	19 (16.4)
Triangular-shaped	30 (25.9)
Irregular shaped	25 (21.6)
Cartilaginous septum	7 (6.0)
CT myelography (n = 14)
Fibrous septum	9 (64.3)
MRI (n = 114)
A single dural sac	32 (28.1)
Spinal canal separated into two parts	82 (71.9)
Spinal cord distribution in hemicords
Symmetrical	20 (24.4)
Asymmetrical	62 (75.6)
Spina bifida occulta (segments)
1 to 3	38 (46.3)
4 to 6	30 (36.6)
7 to 9	14 (17.1)
Meningocele	9 (11.0)
Sacrococcygeal tumours	13 (15.9)

1. ^* the classification of bony septum was based on the shape observed in image

Figs. 1a - 1b

Figure 1a – anteroposterior radiograph of a 3.75-year-old female with a Type 1 diastematomyelia, with the arrow indicating the characteristic high-density shadow of the longitudinal bony spur at L2-L3. Figure 1b – CT scan of the same patient, with the arrow indicating the longitudinal bony crest that has split the spinal cord into two parts.

Of the 116 patients who underwent investigation by CT, the spinal canal and dura mater were seen to be separated into two parts in 90 patients (77.6%) (Fig. 1b), and the spinal canal was not split in the remaining 26 patients (22.4%) (Fig. 2). A total of 25 patients (21.6%) underwent three-dimensional CT reconstruction (CT-3D), and the local anatomy was identified so that surgery could be planned.

Fig. 2

CT scans of a six-year-old female with a Type 2 diastematomyelia, showing a single dural sac and abnormal development of the lamina and spinous process (arrows).

MRI was available in the hospital from 1992. Of the 114 patients who underwent routine MRI, the spinal canal was seen to be separated into two parts in 82 patients (71.9%). Of these, the two halves of the spinal cord were symmetrical in 20 patients (24.4%) and asymmetrical in 62 (75.6%) (Fig. 3). The longitudinal bony spur split could be seen in most patients (Fig. 4). For the patients with Type 2 diastematomyelia, frontal MRI T₁-weighted images showed a single dural sac (Fig. 5).

Fig. 3

Transverse view MRI scan from a six-year-old male with a Type 1 diastematomyelia at L2-3 (arrow); the spinal cord is split into two parts.

Fig. 4

MRI scan sagittal section from the patient with a Type 1 diastematomyelia in Figure 1. The arrow indicates the bony spur at L2-3, which runs through the vertebral body and the lamina.

Fig. 5

MRI frontal sections of the patient with a Type 2 diastematomyelia in Figure 2, showing a single dural sac (arrows).

Of the 106 patients with Type 1 diastematomyelia, a total of 96 (89.6%) underwent removal of the ridge and dural reconstruction to transform the double tubes into a single dural tube. At the same time amputation of the sacrococcygeal fibrous terminal filament was performed in 54 patients (56.3%). A sacral dermoid cyst or lipoma was removed in 13 patients (13.5%) and spinal surgery was carried out in 35 patients (36.5%). A total of 32 patients had scoliosis with a mean Cobb angle of 54.3° (25° to 93°). Posterior spinal correction, intertransverse fusion with bone grafting, and pedicle screw fixation were performed. There were four patients with hemivertebrae, one in T11, one in T12, and 2 in L4 who underwent hemivertebral resection, intertransverse fusion with bone grafting and pedicle screw fixation. Ten patients refused surgery and were given symptomatic treatment (including acupuncture, electrostimulation and traditional Chinese manipulation) and followed closely.

The findings at surgery in the 96 patients were that the septum separated the spinal cord or the cauda equina into two parts and each half had ipsilateral nerve roots. In patients with type 1 diastematomyelia it was found that each half of the cord was completely wrapped with dura and arachnoid mater and the dura close to the septum was thickened and adherent to the arachnoid. In 78 patients (81.3%), the septum inclined to one side of the spinal canal and the two halves of the spinal cord were not equal in thickness. Only 11 patients (11.5%) had both an intramedullary dermoid cyst and a lipoma (L3, L1) and two patients (2.1%) had combined nerve root deformities (T11 and L2).

A total of 16 Type 2 patients (50%) underwent surgery. Of these 16 patients, spinal exploration was performed in nine (56.3%) with neurological symptoms in order to assess the spinal cord after resection of the fibrous band and release of adhesions. Correction of spinal deformities was performed on seven patients (43.8%) and consisted of posterior spinal correction and intertransverse fusion with bone grafting. In six patients (37.5%) internal fixation using a pedicle screw-rod system was used. Hemivertebral resection, intertransverse fusion with bone grafting and internal fixation using pedicle screw-rod system was used for one patient. The remaining 16 patients elected not to undergo surgery and received symptomatic treatment as described above.

In Type 2 patients the two halves of the spinal cord were located within the same dural sac and were symmetrical. A fibrous septum was present in three patients, attached to the anterior and the posterior dura in two patients and to both ends of the split in the other patient. Fibrous adhesions were seen in four patients. The adhesion was located in the dorsal midline in one patient, making the spinal cord and dura mater immobile. In one patient, the adhesion was present throughout the entire anterior-posterior diameter and adherent to the dura. A lipoma was found in the spinal canal in one patient.

A number of pathological changes were seen intra-operatively including a hypoplastic spinous process found at the site of the lesion, bifurcation of a spinous process, a short or absent spinous process or fusion of several spinous processes. In 21 patients, the vertebral plate at the site of the diastematomyelia was widened and thickened and was associated with dural compression. In 30 patients there was partial absence of the vertebral plate and in 21 there was thinning of the vertebral plate. In seven patients, one side of the vertebral plate consisted of cartilage and the other side of bone. One patient had a bony septum at three sites (T5, T6, and L3), and three other patients had bony septa at two sites (T11 and T12, T3 and L4, and T12 and L2, respectively). A cartilaginous septum was found in seven patients, a fibrous septum was found in nine patients and no septum was found in nine patients. There was a bony septum in the remaining 83 patients but it was not connected with the vertebral plate in seven patients. The morphology and the thickness of the bony septa varied; they had a mean transverse diameter of 2.8 mm (2 to 6).

Of the original number of 138 patients who were included in the review, one died during surgery, 26 patients did not have an operation, leaving 111 patients for review post-operatively (Table IV).

Table IV

Prognosis of the 111 patients who were treated surgically

Prognosis (n, %)	Total (n = 111)	Type 1 (n = 95*)	Type 2 (n = 16)	p-value
Recovered from lumbodorsal pain, sphincteric dysfunction, and lower extremity sensory-motor disorders	78 (70.3)	76 (80.0)	2 (12.5)	< 0.001
Partial improvement	17 (15.3)	11 (11.6)	6 (37.5)
No improvement	13 (11.7)	5 (5.3)	8 (50.0)
Deterioration in neurological signs	3 (2.7)	3 (3.2)	0
New neurological symptoms occurred	0	0	0

1. * although 96 Type 1 patients underwent surgery, one died intra-operatively

The prognosis was significantly better for patients with a Type 1 diastematomyelia after surgery (p < 0.001).

A total of 26 patients underwent non-surgical treatment, of these 13 patients (50%) showed no improvement, in ten patients (38.5%) the neurological symptoms deteriorated and three patients (11.5%) had new neurological symptoms.

Discussion

The cause of diastematomyelia remains unclear. It has been proposed that the disorder arises from adhesions between the embryonic ecto- and endoderm, leading to the formation of an accessory neurenteric canal, which is then filled with mesenchyme.¹ Studies have shown that increased maternal plasma homocysteine concentrations and the methionine synthase reductase (MTRR) gene polymorphism are related to the occurrence of neural tube deformity.² Pang, Dias and Ahab-Barmada² have suggested that the disorder is related to recessive heredity. Gardner⁷ reported that diastematomyelia occurred in triplets, and Dikov, Chenrnv and Ivanov⁸ showed that folic acid supplementation can reduce the incidence of neural tube defects.

The neurological deficits which are associated with diastomatomyelia are related to two factors. First, the septum fixes the spinal cord at a low anatomical position, so that the normal upward movement of the spinal cord during growth is limited and the spinal cord is stretched, interfering with its blood supply or that of the nerve roots. This is then followed by ischaemia and tissue necrosis as well as unilateral spinal dysplasia at the site of diastematomyelia.^9,10

In our patients, the septa within the spinal canal in patients with Type 1 diastematomyelia were mostly bony structures (83 of 106, 78.3%), and a few were cartilaginous (seven, 6.6%) or fibrous (nine, 8.5%). The bony ridge often gradually extended to the vertebral body from the base of the broad deformed vertebral plate and fused to the posterior edge of the vertebral body. This ridge may also form a ‘joint’ through a cartilage cap. We also found that in patients with Type I diastematomyelia, the fusion and thickening of the dura and the periosteum of the ridge could be as thick as 1 mm to 2 mm. The thickened dura also limited the upward movement of the spinal cord, causing traction on it.

Fibrous septa were often seen in patients with Type 2 diastematomyelia, adhesion formation between the septum and the dura was relatively minor, and pathological factors causing tethering of the cord were fewer compared with cases with Type I cases, which may explain the relatively mild neurological symptoms in Type 2 patients.

Congenital spinal deformity is the most common clinical manifestation in these patients.^7,8 In this study 116 (84%) patients had a spinal deformity, which is higher than has been reported in most studies.¹ Patients are usually diagnosed as having diastematomyelia when seeking treatment for spinal deformity or for other conditions such as a varus foot deformity. Some patients are found to have diastematomyelia on physical examination. Neurodevelopmental defects of diastematomyelia usually occur in one lower limb, although they may also occur in both lower limbs with varying degree of severity. The reason why the incidence of symptoms and signs is significantly higher in Type 1 than in Type 2 is not clear. It may be that there is a greater degree of tethering in Type 1 than Type 2 diastematomyelia and also that the bony, cartilaginous, or fibrous septa of Type 1 diastematomyelia are located at the bifurcation of dual dural tubes, thus limiting the superior displacement of the spinal cord during the development and compressing the spinal cord and causing local damage to neural tissue. Autopsies of children showed disappearance of the grey matter within the spinal cord at the bifurcation of dual dural tubes, impaired development of neural cells and nerve fibres and dysplasia of motor cells in the anterior horn.¹¹ On the other hand Type 2 diastematomyelia has fewer associated pathological features such as spinal cord tethering and nerve compression and patients have fewer neurological symptoms.

The existence of diastematomyelia should be suspected in patients with scoliosis or kyphosis when there are broadened and flattened gaps between the pedicles in association with a hemivertebra. In the present study there were 116 patients (84.1%) with scoliosis and kyphosis and 128 patients (92.8%) with a broadened and flattened gap between the pedicles. The characteristic bony ridge appeared at the centre of the spinal canal in 75 (54.3%) of these patients.

Radiographs may not reveal the presence of a diastematomyelia in patients with cartilaginous or fibrous septum or in adult patients. In such situations the diagnosis may require a combination of myelography, CT and/or MRI. The typical manifestation during myelography is the ‘island-like’ filling defect formed by the contrast medium around the septum, but can produce false positives.³

On the other hand CT and MRI can clearly show the true state of the spinal canal; the type, travelling direction and shape of the septum; the condition of the separated spinal cord and other concurrent local abnormalities. Spinal rotation can be detected on reconstruction of CT images and MRI scans. In most patients in our study, CT and MRI examination identified the starting point, the range, and exact location of the bony septum thus determining the scope of surgery needed. More advanced CT, CT-3D and MRI techniques have improved the diagnosis of diastematomyelia.¹² Prenatal ultrasound has been used recently for detecting fetal diastematomyelia.^13,14

In patients with Type 1 diastematomyelia neurological symptoms will gradually progress, particularly in patients who are younger than 18 years of age. This is because of the presence of many spinal septa or adhesion factors, which can be bony or non-bony, and also because of the increasing degree of scoliosis. Early surgical intervention is recommended for these patients. Neurological symptoms improved in 87 patients (91.6%) who underwent surgery in our series.

In Type 2 diastematomyelia, neurological symptoms do not change significantly during the patient’s life unless there are associated abnormalities such as a scoliosis with Cobb angle < 40° or a hemivertebra. Some pathological features are absent such as the spinal septum and adhesion, which explains why many adult patients with Type 2 diastematomyelia remain very active and early surgical treatment is not needed. Instead, these patients require rehabilitation with the proviso that surgical exploration could be considered if the neurological symptoms or signs change. This contrasts with the approach of some authors, who advocate surgery for all patients with diastematomyelia, except those with meningomylocele who have non-progressive disability and those with Type 2 diastematomyelia.¹²

The principles of treatment for congenital spinal deformity should be followed with the aim of preventing an increase of the spinal curvature.^14,15 Surgical treatment should be considered when patients with Type 1 or 2 diastematomyelia meet the following criteria: 1) there are progressive neurological changes; 2) before correction of spinal deformity the septum of the diastematomyelia is removed to prevent spinal cord injury whether or not there is evidence of neurological impairment; 3) Type I diastematomyelia is treated by preventive surgery in childhood; 4) back pain and pain radiating to a lower limb in adults.

For patients with intracranial lesions, prophylactic surgery is appropriate for children with Type I diastematomyelia, and adults with back and leg pain.¹⁶

Selecting the appropriate surgical approach is the key to successful treatment of these patients. Laminectomy and resection of the septum are the basic methods.¹ The surgical area should be fully exposed, and septum running through the spinal cord should be resected completely at the base, leaving no residue. The split head of the dura mater and/or abnormal fibre bundles should be resected, so that the separated spinal halves can move freely and can reach the normal position as they are no longer fixed in the deformed area.

We do not advocate extensive laminectomy. Bone and soft-tissue structures should be retained as much as possible while removing the lesions. The bilateral upper and lower articular processes should be retained to maintain spinal stability. Multiple-stage surgery can be adopted for patients with lesions at multiple locations. Concurrent abnormalities within the canal should be resected completely.

There was a significant difference between the two types of diastematomyelia with regard to surgical outcomes (p < 0.05). The efficacy of surgery can be judged by relief of back pain and improvement of lower limb sensorimotor, urinatory and defecatory dysfunction.¹⁷ By these criteria, the efficacy was 95% in patients with Type I diastematomyelia and the neurological function improved significantly after surgery. The efficacy was only 50% in Type 2 diastematomyelia and the difference between pre-operative and post-operative symptoms did not reach statistical significance. One explanation for this is that the septum of Type 2 diastematomyelia consists mostly of fibrous tissue, there is little local adhesion formation and relatively minor symptoms in most patients. Hence, rehabilitation, including muscle function exercise, acupuncture, massage and electrical nerve stimulation, can achieve a good outcome. Rehabilitation alone in patients with a Type 1 diastematomyelia cannot achieve a good outcome. Surgery is needed to remove the abnormal structures and release the spinal cord.

In summary, diastematomyelia is rare and affects females more often than males. The incidence of symptoms including characteristic changes of the skin, neurological disorders, congenital spinal deformity and varus of the foot deformity are significantly higher in Type 1 than in Type 2. Diagnosis relies on radiological results, CT, MRI and CTM scans. In our study, surgery was effective for patients with Type 1 diastematomyelia, but was less effective for patients with Type 2 diastematomyelia.