Journal of Orthopaedics

Study design: A retrospective review of image data was performed in 123 patients with congenital scolisis.

Objective: To assess the vertebral deformity and intraspinal anomaly and try to establish a classification for genetic study of congenital scoliosis.

Summary of Backgroud Data: Three-dimensional reconstructions facilitate visualization and evaluation of vertebral anomaly in congenital scoliosis. Up to date, an appropriate classification of congenital scoliosis is still unavailable for genetic study.

Method: We examined all the components of vertebrae to find the entire anomalies by three-dimensional CT. We classified these anomalies according to formation failure, segmentation failure or mixed type, solitary malformation or multiple malformation, and anterior malformation, posterior malformation or anteroposterior malformation. The intraspinal anomaly was detected by myelography and CT myelography.

Results: 123 patients consisted of 28 patients with formation failure, 43 patients with segmentation failure, and 52 patients with combination of formation failure and segmentation failure. The incidence of intraspinal anomaly was 31.7%.

Conclusion: The combination of three-dimension reconstructions with CT myelography can provide a complete understanding of both vertebral deformity and intraspinal anomaly before surgery. The classification, which based on three-dimensional reconstructions, is all-round and reliable.

Congenital scoliosis is due to the presence of vertebral anomalies that cause an imbalance in the longitudinal growth of the spine. The classification of congenital scoliosis proposed by Moe and Winter et al [1,2]classified the vertebral anomalies according to morphologic characteristics on AP and lateral plain images into formation failure, segmentation failure, and a mixed type, and this classification has been widely accepted. However, this classification can be evaluated only by the vertebral body and the pedicle of the vertebral arch in the malformed vertebrae, and it is impossible to perform an all-round evaluation of the morphology of the malformed vertebrae including the posterior components only based on the plain images. Furthermore, the vertebral rotation associated with scoliosis and the vertebral overlapping of severe kyphoscoliosis on plain images make it difficult to identify the vertebral pedicles, which is important to discriminate between hemivertebrae and wedge vertebrae, and evaluate the morphology of vertebral bodies . So the classification system needs to be revised to provide more detailed information of the anomalies.

Three-dimension CT scans are playing a more and more important role in scoliosis diagnosis and classification. The detailed observation of the posterior and anterior components of malformed vertebrae, which can not be obtained from plain images and only be viewed during the operation before, has become possible with the aid of three-dimension CT scans. Many studies have shown the advantages of CT application in the diagnosis and treatment of congenital scoliosis, but it is a pity that no new classification based on the detailed CT data was put forth. Since CT myelography (CTM) and three-dimension CT reconstructions have been routinely applied in the preoperative evaluation of scoliosis in our hospital, so we did some work to classify the congenital scoliosis on the basis of combination of plain images and 3D CT reconstructions

A retrospective review was performed of patients with congenital scoliosis cared for at Peiking Union Medical College Hospital from 2005 through 2007. Institutional Review Board approval was obtained. 123 patients with congenital scoliosis were diagnosed and treated in the spine center. The initial diagnosis was made based on the vertebral anomalies on the plain images and was further verified after the CTM and 3D reconstructions were performed. The subjects that are diagnosed as the known syndromes, such as hemifacial microsomia, Alagille, Jarcho-Levin, Klippel-Feil, Goldenhar, Joubert, basal cell nevus, trisomy 18, diabetic embryopathy, and VACTERL (vertebral, anal, cardiac, tracheal, esophageal, renal, and limb) syndromes, were excluded so that the data become relative simple to be analyzed. In this group, the subjects consisted of 51 males and 72 females, and their age at the time of admission to the hospital was between 2 and 25 years (mean, 13.5 years).

All the patients had standing anteroposterior and lateral views of the spine from C1 to S1. The patients were subjected to myelography followed by computerized axial tomography to detect the associated intraspinal anomalies. The dye flowed up to the occipitovertebral junction to rule out Arnold-Chiari malformation, syringomyelia, etc.

Both the malformed vertebra (e) and its (their) adjacent vertebrae were observed in detail by 3D CT. The morphology of each of components, including vertebral body, vertebral pedicles, neural arch, and the relationship between the anterior and posterior components were evaluated.

The CT scanner is a helical scanner (Somatom Plus S or Somatom Plus 4; Siemens; Erlangen, Germany). The slice thickness was 3 mm, and 3D images were displayed by volume rendering. Slices over the entire area of the scoliosis, including the thorax cage, were obtained. High quality three-dimensional CT images were then generated by the CT technician for every 30° of rotation about the longitudinal axis of the spine.

Although the classification of congenital scoliosis proposed by Moe and Winter et al was simple and primitive, it was very helpful to define the feature of the vertebral anomaly. Philip and Robert et al had proposed some genes related to the vertebral anomaly, which is the direction of our following research. So here we made some modification of the classification for the further genetic study. First, the patients were classified into 3 groups: formation failure group (FF), segmentation failure group (SF) and mixed type group (MT). Anomaly of any part of the vertebra was determined to be abnormal vertebra. The formation failure group was further subdivided into the solitary malformation group (SM), which exhibited only a single malformed vertebra in the entire spine; and the multiple malformations group (MM), which exhibited multiple malformed vertebrae. The segmentation group was also subdivided into the solitary malformation group (SM), in which only the consecutive two vertebrae were involved; and the multiple malformations group (MM), in which more than two vertebrae were involved. In MT group, solitary malformation meant segmentation failure between the malformed vertebra and one of the adjacent vertebrae, and the rest were multiple malformations. Then each group was subdivided into anterior malformation subgroup (AM), posterior malformation subgroup (PM) and anteroposterior malformation subgroup (APM).

Congenital scoliosis was more frequent in girls (male:female ratio 1:1.41). Associated kyphosis was seen in 61 (49.6%) patients. The distribution of all the cases in three groups is showed in Table 1. Of all the 123 patients, 29 cases were formation failure,; 43 cases were segmentation failure, and 51 cases were mixed type. The frequency of each involved segment was exhibited in Fig. 1. In the FF group, most of the involved segments were located between T8-L4 (76.4%), and there was no obvious peak. In the SF group, 91.6% of the involved levels were located in the thoracic segments. In the MT group, T7 was the most frequently involved segment and the wave centered on T7 and decreased caudally and rostrally. Intriguing, in our series C1-5 were free of anomaly. The average lengths of involved vertebrae were 1.5, 5.0, and 5.3 in FF group, SF group and MT group respectively.

Nineteen-nine patients (67.9%) were solitary malformation, and nine patients (32.1%) were multiple malformations in the twenty-eight patients of FF group. The multiple malformations were consecutive in six patients and skipping in three patients. In the forty-three patients of SF group, nine patients (20.9%) were solitary malformation, and thirty-four patients (79.1%) were multiple malformations. The thirty-four MM patients included 27 patients of consecutive malformations, 2 patients of skipping malformations, and 5 patients of consecutive and skipping malformations. In MT group, six patients (11.5%) were solitary malformation and 46 patients (88.5%) were multiple malformations. These 46 MM patients included 24 cases of consecutive malformations, 4 cases of skipping malformations, and 18 cases of consecutive and skipping malformations.

Fig. 1. The frequencies of each involved segment. If the total segments were more than the normal, the malformed segment was named by the adjacent two vertebrae. For example, the hemivertebra between T4 and T5 was named T4/5.

The anomalies of both the anterior component and the posterior components were the commonest pattern in vertebral deformities. The proportions of AP malformation were 32.1%, 72.1%, 88.5%, and 69.9% in FF group, SF group, MT group and the series respectively. The solitary posterior malformation was rare in our series and all the 3 patients were in SF group. The majority in FF group was the solitary anterior malformation and the ratio was 67.9%.

Formation failure included hemivertebra, wedge vertebra, and butterfly vertebra. In our series, hemivertebra was found in 51 patients (30 in MT group and 21 in FF group), wedge vertebra in 24 patients (16 in MT group and 8 in FF group), and butterfly vertebra in 9 patients (all in MT group).

In FF group, anomaly of posterior components was not found in 8 patients with wedge vertebra and 11 patients with hemivertebra. Malformation of posterior

components was found in only 9 of 21 (42.9%) patients with hemivertebra. In SF group, segmentation failure of posterior components was found in 33 of 43 (76.7%) patients. In MT group, formation failures of posterior components were observed in 21 patients (40.4%) and segmentation failure of posterior components were found in 38 patients (73.1%).

Intraspinal abnormality was found in 31.7% (n=39) of these patients, including diastematomyelia in 12 patients, tethered cord in 1 patient, combination of diastematomyelia and tethered cord in 21 patients, syringomyelia in 2 patients, combination of diastematomyelia and syringomyelia in 2 patient and combination of syringomyelia and tethered cord in 1 patient. There were 26 females and 13 males (female: male ratio 2:1). Intraspinal abnormality was present in 3.6% (n=1) of the patients with formation failure (n=28), in 51.2% (n=22) of the patients with segmentation failure (n=43), and 30.8% (n=16) of the patients of mixed type (n=52) (Table 2).

Anomaly	Total	Intraspinal anomaly
FF	28	1(3.6%)
SF	43	22(51.2%)
MT	52	16(30.8%)
Total	123	39(31.7%)

Diastematomyelia was present in total 35 patients. The levels of diastematomyelia could be T2-S2 and was showed in detail in Table 3. The length of diastematomyelia ranged from 1 to 14 vertebrae (mean 5.3 vertebrae). The levels of diastematomyelia were thoracic segments in 5 patients, lumbar segments in 8 patients, thoracic and lumbar segments in 21 patients, and lumbar and sacral segments in 1 patient. In these patients the diastematomyelia was not always at the site of spinal deformity, whereas in some patients it was in the region distant from the site of the malformed vertebrae. The majority of diastematomyelia were consecutive in our series and the diastematomyelia was found to be skipping in only two cases (Table 3, No.17 and 19). Osseous septum was identified in 6 of 35 patients (17.1%).

Table 3. The levels of diastematomyelia and malformed vertebrae in 35 cases.

T1-2，T3-5，T12* means there are segmentation failures between T1-2, and between T3-5 and there is not segmentation failure between T1-2. T12 is formation failure

Most of the time, the diagnosis of congenital scoliosis through the AP and lateral plain films is not difficult. However, to describe the anomaly accurately is not so easy only based on the plain films. The low sensitivity of plain films to identify the fusion and hypoplasia of posterior element makes it impossible to obtain a all-round view of the anomaly. Plain films alone do not necessarily give the surgeon a clear understanding of the malformed anatomy, though they remain the standard for diagnosis and follow-up of these deformities. Plain films are difficult to interpret because of the complex nature of the deformity, superimposed structures obscuring visualization of the anomaly, and spine rotation. Three-dimension CT scans can showed additional abnormalities not appreciated on plain films, as reported by Newton [3], so the advantages of three-dimension CT scans over the conventional radiography are obvious. We do not have the evidence that three-dimension CT scans gained before surgery improved the surgical outcome or decreased the occurrence of complications, but they can help the surgeons to have an intimate knowledge of the anomaly during the intraoperative localization and avoid being confused by the abnormal anatomy during the procedures. Hedequist and Emans [4] described that three-dimension CT scans accurately predicted anterior posterior vertebral anomalies in all cases. Nakajima and Kawakami et al [5] even put forth a new classification of hemi-lamina type for formation failure with the aid of three-dimension reconstructions.

Congenital vertebral anomalies encompass a wide variety of defects. Moe and co-workers use two basic concepts of pathogenesis in defining most of the anomalies: defects of segmentation and defects of formation. Van Schrick and MacEwen modified the classification on this basis. Tsou et al [6] raised a different classification system. Their unique classification system was very inclusive and there were two major types and 27 subtypes. All these classifications were based on either the anatomic feature of deformity or the classical embryogenesis and prenatal developmental pattern of the vertebra. However, it was not helpful to understand the etiology of congenital vertebral anomaly if the classification was too simple or too complex. Vertebral development is a complex process and dozens of genes have been linked to this process [7.8, 9]. The different components of vertebra were controlled by different genes during vertebral development. So formation failure or segmentation failure, anterior deformity or posterior deformity, and solitary malformation or multiple malformations are taken into consideration in our classification for further genetic study.

In this study, all of the patients underwent myelography followed by CTM and three-dimension reconstructions. The aim of myelography and CTM was to find the intraspinal deformity. Myelography and CTM are sensitive to diastematomyelia and tethered cord. In our series, the diagnosis of syringomyelia was made based on MRI examination, which was done in only 12 patients before they presented to our hospital. The incidence of intraspinal anomaly in our series was 31.7%, which was close to the literature reports. Bradford et al [10] reported an incidence of 38%, using MRI, in a series of 42 patients. Prahinski et al[11]found a 30% incidence of intraspinal anomaly in a series of 30 patients. Belmont et al [12] reported a 35% incidence of intraspinal anomaly in 106 patients with isolated congenital hemivertebra with MRI detection. In their report, these abnormalities included diastematomyelia, cord tethering, Chiari malformations, and intradural lipomas. Basu et al [13] reported a 37% incidence of intraspinal abnormality of 126 patients with congenital scoliosis with MRI examination. However, McMaster [14] reported that the incidence of intraspinal deformity was 18.3%. In his report, 106 patients underwent myelography, and diastematomyelia was diagnosed in 8 patients on the basis of midline bony spur on plain radiograph. Blake et al [15] performed myelography on a series of 108 patients with congenital spinal deformity. They reported an incidence of 58%. Both of these two reports were based on myelography and the disparity was great compared with other reports. One possible explanation was the differences of sensitivity and specificity between myelography and MRI.

Three-dimensional reconstructions of computed tomography scans are helpful to provide all-round view of vertebral anomaly and reliable foundation of classification in congenital scoliosis. CTM is useful and sensitive to detect the intraspinal deformity, especially diastematomyelia and cord tethering. The application of CTM and three-dimensional reconstructions is necessary to obtain an understanding of both vertebral anomaly and intraspinal abnormality before the surgery.

The authors are grateful to Lijuan Zhao for her assistance of collecting the data in this study.

1. McMaster MJ, Ohtsuka K. The natural history of congenital scoliosis: a study of two hundred and fifty-one patients. Journal of Bone And Joint Surgery Am 1982;64:1128-47.

2. Winter RB, Moe JH, Eilers VE. Congenital scoliosis: a study of 234 patients treated and untreated. Journal of Bone And Joint Surgery Am 1968;50:1-47.

3. Newton PO, Hahn GW, Fricka KB, Wenger DR. Utility of three-dimensional and multiplanar reformatted computed tomography for evaluation of pediatric congenital spine abnormalities. Spine 2002;27:844-50.

4. Hedequist DJ, Emans JB. The correlation of preoperative three-dimensional computed tomography reconstructions with operative findings in congenital scoliosis. Spine 2003;28:2531-4.

6. Tsou P, M’Yau A, Hodgson AR. Embryogenesis and prenatal development of congenital vertebral anomalies and their classification. Clinical Orthopaedics And Related Research 1980;152:211-31.

10. Bradford DS, Heithoff KB, Cohen M. Intraspinal abnormalities and congenital spine deformities: A radiographic and MRI study. Journal of Paediatric Orthoppaedics 1991;11:36-41.

11. Prahinski JR, Polly DW, McHale KA, Ellenbogen RG. Occult intraspianl anomalies in congenital scoliosis. Journal of Paediatric Orthoppaedics 2003;20:59-63.

12. Belmont DJ, Kuklo TR, Taylor KF, Freedman BA, Prahinski JR, Kruse RW. Intraspinal anomalies associated with isolated congenital hemivertebra: the role of routine magnetic resonance imaging. Journal of Bone And Joint Surgery Am 2004;86-A:1704-1710.

14. McMaster MJ. Occult intraspinal anomalies and congenital scoliosis. Journal of Bone And Joint Surgery Am 1984;66:588-601.

15. Blake NS, Lynch AS, Dowling FE. Spinal cord abnormalities in congenital scoliosis. Annual Radiology 1986;29:377-9.