Pictorial review Spinal ultrasound in infants

This paper discusses the indications for spinal ultrasound, including its advantages and disadvantages compared with spinal MRI. The features and ultrasound findings both in normal infants and in those with spinal dysraphism are reviewed. Spinal ultrasound (SUS) is becoming increasingly accepted as a first line screening test in neonates suspected of spinal dysraphism (SD) [1, 2]. SUS can be useful in detecting tumours, vascular malformations and cases of trauma; however, as the indication in most cases is to exclude SD, this article will concentrate on this group of abnormalities [2–4]. The advantages of SUS are not only a diagnostic sensitivity equal to MRI [2] but that, unlike MRI, SUS can be performed portably, without the need for sedation or general anaesthesia. In addition, MRI is highly dependent on factors affecting resolution, including patient movement, physiological motion from cerebral spinal fluid (CSF) pulsation and vascular flow, factors that do not affect SUS [5]. New generation high frequency ultrasound machines with extended field of view capability now permit imaging of high diagnostic quality in young babies. SD refers to abnormalities with imperfect fusion of the midline neural and bony structures. It is the most common congenital central nervous system abnormality, with myelomeningoceles occurring in up to 2 per 1000 live births in some studies, although the incidence in the Western world is now likely to be lower [5, 6]. In infants with occult SD, early diagnosis may be useful, as SD may lead to distortion of the spinal cord and nerve roots with growth, resulting in neurological sequelae in the lower extremities, the lower urinary tract and the gastrointestinal tract. Because some paediatric neurosurgeons believe that early surgical correction of SD may avoid these sequelae, early diagnosis of SD could be important [7, 8]. How to do it and what to look for SUS was first proposed in the early 1980s but it has taken some time to enter routine clinical practice [9, 10]. The examination is performed using a 7.5–10 MHz linear probe in both the sagittal and axial plane along the entire spine. Ideally the patient is laid prone on a pillow with the neck flexed for craniocervical junction imaging, but imaging with the patient in the decubitus position is also feasible. The examination can be performed in the Ultrasound Department or on the ward with a portable machine. No sedation is necessary. SUS is possible in the neonate owing to a lack of ossification of the predominantly cartilaginous posterior arch of the spine [11, 12]. The quality of ultrasound assessment decreases after the first 3–4 months of life as posterior spinous elements ossify, and in most children SUS is not possible beyond 6 months of age. However, the persisting acoustic window in children with posterior spinal defects of SD enables ultrasound to be performed at any age [11, 12]. The normal neonatal spinal cord is displayed on ultrasound as a tubular hypoechoic structure with hyperechoic walls (Figure 1). The central canal is hyperechoic, the so-called central echo complex [13]. The subarachnoid space surrounding the cord is hypoechoic. The caudal end of the spinal cord corresponds with the conus medullaris, which continues into the filum terminale. The cauda equina is seen as echogenic linear structures surrounding a hyperechoic filum terminale. The vertebral bodies are seen as echogenic structures ventral to the spinal cord. Particular features to note are: (i) The level of the conus medullaris. In term infants the tip of the conus medullaris normally lies above the mid level of the L2 vertebral body although there is a large range of normality (from T10/11 to L2/3) [14]. In pre-term infants the tip of the conus lies between L2 and L4, i.e. the level of the conus moves proximally with age [15]. A low-lying cord may imply tethering. (ii) The position of the cord in a dorsal/ventral Received 5 January 2001 and in revised form 2 May 2001, accepted 25 June 2001. Address correspondence to Dr R de Bruyn. The British Journal of Radiology, 75 (2002), 384–392 E 2002 The British Institute of Radiology 384 The British Journal of Radiology, April 2002 or anterior/posterior orientation. In normal infants the cord lies a third to half way between the anterior and posterior walls of the spinal canal. If the cord lies more posteriorly, tethering should be suspected (Figure 2). (iii) The presence of normal pulsatile movement of the cord and nerve roots (tethering leads to an absence of pulsatility). (iv) The thickness of the filum terminale (normal 2 mm or less [16]). Viewed axially, the spinal cord is hypoechoic and round or oval-shaped, lying within the anechoic subarachnoid space (Figure 3a). The spinal cord is fixed by dentate ligaments that pass laterally from the spinal cord. Echogenic nerve roots are seen below the level of L2 (Figure 3b). Patient selection: who should be investigated for SD Spinal MRI is a limited resource and cannot be used as a screening tool. Ward et al [17] suggested that only patients with a clinically demonstrable neurological abnormality, such as increased tone or hemiplegia, should be investigated for SD using MRI, as they found that only those with clinically demonstrable neurological defects demonstrated an abnormality on MRI. Other authors concur with this suggestion. Schwend et al [18] found that all patients who needed neurosurgical intervention could be identified by abnormal neurological findings on clinical examination. SUS has the advantage over MRI that it is a less restricted resource and can therefore be used as a screening tool in all babies suspected of SD. SD may be suspected if the infant has: (i) a neurological deficit; (ii) cutaneous stigmata, such as a haemangioma, sacral pit or tuft of hair; or (iii) other abnormalities associated with SD such as cloacal exstrophy [19] or anorectal or sacral abnormalities and agenesis [12, 20]. Current Royal College of Radiologists guidelines are that all neonates with a hairy patch or sacral dimple should undergo SUS [21]. However, while more than 90% of patients with occult SD have a cutaneous abnormality over the lower spine [22], a cutaneous marker may have a low yield in predicting the presence of a clinically significant abnormality. In a recent review of 200 SUS examinations performed over an 11-year period, SD was found in less than 1% of cases when a cutaneous marker was the only clinically detected abnormality [23]. The diagnostic value of SUS has been shown to be equal to MRI [2]. Rohrschneider et al [2] found that SUS exactly correlated with MRI in 32 out of 38 cases. In five cases, SUS depicted the main abnormality but MRI gave additional information. Wherever SUS was normal, MRI was also normal, i.e. SUS had a sensitivity of 100%. Current literature therefore suggests that SUS may be used as a primary screening tool, with MRI being performed in any case where SUS revealed an abnormality [1, 2, 24].