Patterns of congenital bony spinal deformity and associated neural anomalies on X-ray and magnetic resonance imaging

The vertebral column and spinal cord are closely related from an anatomical and developmental perspective. During the first 8 weeks of development, bony elements of the spine form in coordination with the infolding and closing of the neural tube [1]. Therefore, congenital malformations of the bony structure of the vertebral column, particularly those associated with scoliosis and kyphosis, are often accompanied by abnormalities of the spinal cord [1‐4].

Congenital vertebral deformities can be classified as defects of segmentation or defects of formation; however, a mixture of both is most common [2]. Defects of formation include hemivertebrae, anteriorly wedged, and butterfly vertebrae, while defects of segmentation include unsegmented bars and block vertebrae [2‐4]. This morphologic classification has prognostic value in that spinal column deformity associated with unilateral segmentation defects (bars), contiguous ipsilateral hemivertebra, or bars opposite hemivertebra are more likely to progress. Additionally, congenital deformities can be classified by location (cervical, thoracic, lumbar, sacrococcygeal) or curve pattern (scoliosis, kyphosis, lordoscoliosis) [2]. However, the three dimensional nature of spinal deformity makes the distinction between scoliosis and kyphosis arbitrary and often both sagittal and coronal deformity exists [1, 2].

Common associated neural axis anomalies include Chiari malformations, tethered cord, diastematomyelia, lipomas, and a variety of intraspinal cysts [2‐4]. The associations between bone and neural abnormalities have been refined with the advent and improvement of sophisticated imaging modalities, such as computed tomography (CT) and magnetic resonance imaging (MRI).

McMaster [3] used radiographic evidence to observe the relationship between congenital bony abnormalities and intraspinal neural abnormalities. In a cohort of 251 children presenting with congenital scoliosis, McMaster found the incidence of neural axis malformation to be 18.3 % [3]. A myelogram was used to identify intraspinal anomalies without external signs of neural tube defect or neurologic symptoms [3]. Because neural axis abnormalities are often initially asymptomatic and unobservable, the use of intraspinal imaging with MRI has become the standard for patients with congenital spinal deformity, although precise indications are controversial [3, 5‐8]. MRI is also useful in providing details of the vertebral anatomy, including the morphology of the disc and endplate, which has prognostic and surgical implications.

Several studies have recommended ordering MRI of the spine in all patients with congenital scoliosis due to the high incidence of occult neurologic pathology [9‐11]. More recent recommendations suggest obtaining MRI only in the presence of neurologic deficit, progressive spinal deformity, or for preoperative planning [7, 8]. Based on the widespread use of MRI today and its increased sensitivity over the myelogram, the incidence of associated neural axis abnormalities with congenital bony spinal malformation has been suggested to be as high as 30–53 % [4, 11‐18]. However, authors do not stratify their analysis by the type of malformation (failures of formation or segmentation) [19] nor do they make a distinction between the complexity of the bony malformation (single or multiple affected vertebrae) [11, 15, 18]. The purpose of our investigation was to determine the overall relationships between congenital bony vertebral abnormalities and spinal cord pathology, and to analyze if some subtypes (simple vs complex anomalies, segmentation vs formation defects, with or without associated syndromes) have a higher incidence of neural anomalies.

Furthermore, congenital deformities of the bony spine often present in the context of defined congenital syndromes and associations such as those of genetic origin (Jarcho−Levin, multiple pterygium), those with unconfirmed etiology (Goldenhar syndrome), and VACTERL constellation (vertebral anomalies, anal atresia, cardiac defects, tracheoesophageal fistula, renal anomalies, and limb deformities) [20]. To our knowledge, there are few studies looking at the incidence of spinal cord anomalies in specific syndromic populations [15, 21] and none comparing neural axis abnormalities in patients with congenital syndromes to those in which congenital vertebral deformities present as an isolated condition. A second purpose of our study was to make this comparison.

Our hypotheses were twofold—as the number and complexity of bony vertebral abnormalities increases, there will be a corresponding increase in the incidence of spinal cord anomalies. Patients with associated syndromes are more likely to have spinal cord anomalies than patients without associated syndromes.

The study was approved by our institutional review board. All patients aged ≤13 years with congenital spinal deformity who attended our institution, a tertiary referral center, were identified using ICD-9 754.2 (congenital musculoskeletal deformities of the spine) and CPT billing codes for MRI of the cervical, thoracic, and lumbar spine from 2003−2013. Two hundred and eighty-two patients with congenital vertebral malformations were identified, and 75 aged ≤13 years with full length spinal X-ray and MRI were reviewed including the radiographic and MRI images, as well as the medical records. We compared patients with and without spinal cord anomaly using the following variables—complexity of bony malformation (simple: 1–2 affected vertebrae vs complex: >2 affected vertebrae), malformation of segmentation and/or formation (mixed failures of segmentation and formation vs failures of segmentation alone vs failures of formation alone), spinal level (cervical, thoracic, lumbar, sacral, and/or coccygeal), number of affected vertebrae, associated syndromes, and gender.

Radiographs were evaluated for bony vertebral abnormalities, and MRI images were then reviewed for anomalies of the spinal cord and meninges and for bony anomalies missed on plain radiographs. Radiology reports, patient medical records, and CT were used to establish the diagnoses. To gain insight into the outcome of the neural abnormalities, the follow-up period was defined as the time between the diagnosis of spinal cord pathology (date of MRI) and the last orthopedic or neurosurgical clinic note.

Chi-squared analysis was performed to compare groups for all categorical variables (spinal level, complexity of malformation, malformations of segmentation and/or formation, associated syndromes, and gender) and independent t tests were used to compare groups for continuous variables (number of abnormal vertebra). The level of significance was set a priori at p < 0.05.

Of the 282 patients initially screened, we excluded patients with myelomeningocele, patients determined to have idiopathic scoliosis (spinal deformity in the absence of congenital bony deformity on X-ray), or congenital bony deformity that could not be determined from X-ray alone (patients for which congenital spinal deformity was diagnosed only after advanced imaging such as CT or MRI). We did not exclude patients based on treatment or previous surgery.

Of 75 patients with congenital bony spinal deformities and MRI data meeting inclusion criteria, 51 % were male (n = 38) and 49 % were female (n = 37). There were 30 patients without associated syndromes and 45 with associated syndromes. The most common syndromes included VACTERL (n = 30), neurofibromatosis type 1 (n = 2), and Goldenhar syndrome (n = 2). The average age at the time of MRI was 5.4 years (range 3 days–14.2 years). Fourteen patients underwent MRI when aged <12 months, 13 patients between 12 and 23 months of age, and 48 patients when aged ≥24 months. MRI was ordered in all patients, regardless of symptoms, based on previous reports of high concurrence of neural abnormalities with congenital spinal anomalies and imaging recommendations [9‐11]. All patients undergoing spinal surgery underwent MRI in advance of surgery.

Patients with congenital malformations of bony vertebrae on X-ray had an overall incidence of brainstem and spinal cord anomalies on MRI of 55 % (n = 41). Tethered cord (a visible strand of tissue attaching the conus medullaris to the coccyx, or a conus ending below L2) (n = 25, 33 %) and syrinx (a dilation of the central canal of the spinal cord) (n = 8, 11 %) were the most common spinal cord findings. Eight percent (n = 2) of patients with tethered cord were found to also have a Chiari 1 malformation (tonsils >5 mm below the foramen). Table 1 lists all patients with pathologic findings on MRI (n = 41) along with their bony deformities, demographics, and surgical outcomes.

The primary intent of the study was to determine the incidence of neural axis anomalies associated with congenital bony spinal anomalies and to determine if complex bony deformities were more likely to have positive MRI findings in the neural axis. As hypothesized, we found a higher incidence (67 %) of spinal cord anomalies in patients with complex bony malformations (>2 vertebra involved) when compared to patients with simple malformations (37 %). An added benefit of MRI was the uncovering of 124 vertebral anomalies not seen on radiographs in 47 % of patients, as the MRI helped to overcome the obscuring effects of the skull and pelvic viscera and improved characterization of abnormal vertebrae, showing the segmentation characteristics. With regard to our second hypothesis, we did not find a higher incidence of neural pathology in those patients with defined syndromes.

This study confirms a high prevalence of cord abnormities associated with congenital spinal anomalies, with 55 % of children with congenital bony spinal deformities having associated intraspinal neural axis defects, supporting recommendations to order MRIs of the brainstem and spinal cord for all patients with congenital vertebral malformation [9‐11, 17]. However, we acknowledge that the timing of ordering the MRI remains controversial and our data represent a highly selective cohort of children referred to a tertiary center and, as such, our results may not apply to all patients with less severe congenital curves. The average age of MRI for surgical patients was 7.1 years, yet the average age for surgical intervention was 7.9 years, frequently for tethered cord and syringomyelia. Therefore, many children had waited a considerable amount of time before MRI. Thus, recent guidelines for timing of MRI seem reasonable and include the presence of neurologic symptoms or signs such as bowel or bladder dysfunction, spasticity (upper motor neuron findings) or brainstem findings (swallowing difficulties), rapidly progressing spinal deformity (curvature), and for preoperative planning [10, 11]. We found in this study that MRI provided additional insights into the structure of the spine and may influence surgical decision-making. Fourteen patients underwent MRI before 12 months of age, with the youngest being 3 days old; all for assessment of multiple congenital defects including obvious motor defects, deformities of the neck or lower extremities, and congenital defects of other organ systems. We conclude that clinical judgment should be the basis for the timing of obtaining an MRI. However, due to the high incidence of underlying neural defects, all patients with congenital spinal deformities should have regular examinations including a neurological survey.

We also investigated the records to gain insight into the rate at which patients progress to require neurosurgical or orthopedic surgical interventions. Within the time frame of our review, 56 % of patients with spinal cord anomalies eventually underwent neurosurgical (15 %) or orthopedic (37 %) spine surgery, or both (5 %), at a mean age of 7.9 years. The remaining 44 % of patients with MRI findings did not have surgery during our limited follow-up period; the majority of these patients had low-lying or blunted spinal cord without associated clinical signs or symptoms. Because we had a median follow-up of 4.2 years (7 days–11.4 years) after the diagnosis of spinal cord anomaly on MRI, we cannot state definitively that surgery was not performed later.

Basu et al. reported that patients with scoliosis due to mixed malformations of the vertebrae were found to have a 40 % incidence of spinal cord anomaly, while failures of formation and segmentation alone had incidences of 30 and 40 %, respectively, with highest incidence of spinal cord anomaly in patients with kyphosis (56 %) [15]. While statistically non-significant (p = 0.065), the incidence of spinal cord anomalies in patients with mixed failures of segmentation and formation was high (73 %). However, due to the small sample size, particularly of the segmentation-only group (n = 2), this particular analysis was underpowered to demonstrate the true relationship between failures of segmentation, formation, or mixed deformities. We also did not stratify patients as ‘kyphosis’ versus ‘scoliosis’ because we found patients commonly have both deformities, making the distinction imprecise.

The incidence of cord anomalies was high in patients with bony defects throughout the spinal column, with the highest for those with sacrococcygeal deformity (13 of 15 patients, 87 %). Basu et al. found the highest incidence of spinal cord anomalies associated with spinal column deformities of the cervical and thoracic spines (37 %) [15]; however, their study looked only at patients with hemivertebrae in the cervical, thoracic, and lumbar spine [15]. It should also be noted that in our study, 70 % of cervical, 61 % of lumbar, and 54 % of thoracic bony anomalies had spinal cord anomalies. Sacrococcygeal deformities included hypoplasia or aplasia of vertebrae, including four patients with caudal regression. Additionally, lumbarization of the sacrum was found on X-rays in five of 13 patients with MRI findings in the sacrococcygeal group. Lumbosacral transitional vertebrae (LSTVs), including sacralization of the lumbar and lumbarization of the sacrum, are congenital spinal anomalies, although due to their high prevalence some may consider them to be a variant of normal [22, 23]. However, all patients with LSTVs had other congenital bony deformities in the cervical, thoracic or lumbar spines, and thus would meet inclusion criteria even if one were to consider LSTVs as a variant of normal. Moreover, all patients with LSTVs on X-ray and MRI findings demonstrated some level of cord tethering/filum terminale thickening, suggesting a possible link between LSTVs and distal cord pathology.

Several shortcomings of this study must be acknowledged. First, the data are from a regional referral center, and persons with minor congenital vertebral malformations may never seek medical attention or are asymptomatic and never referred to our center. Thus, the incidence of cord anomalies in the context of vertebral malformation found in this study is likely to be higher than in the overall pediatric population of individuals with congenital bony deformities. Furthermore, the retrospective nature of this study makes it difficult to determine whether the diagnosis of spinal cord pathology could have been made by physical examination prior to ordering MRI. Nevertheless, clinically important neural axis abnormalities can exist with absent or subtle clinical signs and symptoms [7]. Additionally, our small sample size increases the risk for type II statistical error. The inter-relationship of variables, such as syndromes and complex deformity, make the relative importance of each variable difficult to interpret. Finally, the inherent limitations of a retrospective study and our limited follow-up prevent our data from establishing definitive progression of spinal deformity and clinical implications of the cord pathology.

In summary, we found a 55 % overall incidence of underlying cord anomalies in association with bony spinal column malformations. This incidence was highest when complex bony deformities were involved (67 %) and when the bony deformity was located in the sacrococcygeal region (87 %). Nevertheless, cord and meninges anomalies were found in association with simple bony abnormalities located in other regions of the spine. Fifty-six percent of patients ultimately underwent surgery during the 4.2 year follow-up.