Spinal abnormalities are the most common orthopedic manifestation of NF-1. It is quoted as from 2 to 36 % in the literature [10, 11]. In the NF clinic at our institution, it is 23 % [12]. In a report in 1988, Winter et al. [26]. found only 102 patients having NF-1 by clinical criteria in a pool of approximately 10,000 patients with scoliosis. Functional scoliosis resulting from limb hypertrophy or long-bone dysplasia leading to limb length inequality must be ruled out in patients with NF-1. Rarely, unrecognized extra-pleural thoracic tumors can present as focal scoliosis. These lesions are usually plexiform neurofibroma and are not visible on plain radiographs [27]. The spinal deformities tend to develop early in the life therefore, all preadolescent children with NF-1 should be evaluated by scoliosis screening or the Adam forward-bend test to rule out the presence of a spinal deformity.

It is important to emphasize that there is no standard pattern of spinal deformity in NF-1. All manner of spinal deformities in multiple planes and in any part of the spine may occur with NF-1 [28, 29]. The characteristic deformity tends to be a short-segmented, sharply angulated curvature that usually involves four to six vertebrae in the upper third of the thoracic spine [30]. We have traditionally classified the deformities into dystrophic or non-dystrophic types based on the coronal plane x-rays. The entire spine may be affected by deformity in the coronal and sagittal planes. There are nine radiographic criteria most often used to classify the deformity as dystrophic. These include rib penciling (the rib being smaller in diameter than the second rib), vertebral rotation, posterior vertebral scalloping, vertebral wedging, spindling of the transverse process, anterior vertebral scalloping, widened interpedicular distance, enlarged interverteral foramina, and lateral vertebral scalloping. Recently, two more magnetic resonance imaging (MRI) findings have been added to the criteria used to classify the deformity as dystrophic: the presence of dural ectasia and the presence of paraspinal tumors (Table 16.2) [31]. More than three of these dystrophic features are considered diagnostic of dystrophic scoliosis. Non-dystrophic curves are considered similar to idiopathic scoliosis.

Ten to thirty-three percent of children with NF-1 have spinal deformity [16]. All preadolescent children with NF-1 should be evaluated by the scoliosis screening or the Adam’s forward-bend test to rule out the presence of a spinal deformity, which usually occurs earlier in children with NF-1 [30].

The cause of spinal deformity remains unknown. Several theories including metabolic bone deficiency, osteomalacia, endocrine disturbance, and mesodermal dysplasia have been proposed and are at best inconclusive [32–36]. The dystrophic changes may be attributed to intrinsic factors or may be associated with anomalies of the spinal canal secondary to abnormalities of the spinal cord dura mater.

Pressure erosive effects of dural ectasia and paravertebral tumors have been frequently found to be adjacent to and approximated to the deformities, initiating instability and subsequent deformity. Dural ectasia, a disorder unique to certain conditions, is an expansion or dilatation of the dural sac. The changes in the spinal canal induced by dural ectasia may increase the difficulty in obtaining adequate purchase for fixation of anchors during spinal deformity correction.

Scalloping was initially thought to represent the result of erosive pressure or direct infiltration of the vertebra by adjacent neurofibroma [37–41]. A neurofibroma-derived locally active biochemical substance or hormone that triggers dystrophic features in the adjacent vertebra has also been proposed [37]. The presence of an altered response of the vertebral bone in NF-1 to a paraspinal tumor has been hypothesized. An interactive pathophysiological mechanism between a genetically compromised bone and a neuroectodermal derivative, such as a contiguous neurofibroma or an abnormal meningeal sheath, is suggested by some authors [37, 39].

The etiological theory of vertebral scalloping being a primary developmental defect was supported by the presence of scalloping without adjacent lesions [42]. This was also supported by an MRI study in patients with NF-1, in which posterior vertebral scalloping was highly associated with dural ectasia, lateral scalloping was related to dural ectasia or neurofibromas in 50 % of cases, and anterior scalloping was unrelated to dural ectasia or tumors [43]. The authors could not identify any association with dural ectasia or paraspinal tumors in more than one-third of their patients with MRI evidence of vertebral scalloping. Nevertheless, dural ectasia without associated vertebral scalloping was recorded in 10 % of the cases.

A recent study in ten monozygotic twins with NF-1 demonstrated mixed concordance and discordance for presence of scoliosis [3]. The affected twin pairs were discordant for presence of dystrophic features, degree of curvature, and need for surgery. This finding suggests that both heritable and nonheritable factors contribute to the pathogenesis of spinal deformities in NF-1 patients. Dystrophic curves most likely require a nonhereditary event, such as an adjacent tumor or dural ectasia, or a second hit event in local bone cells leading to the underlying dysplasia. If occurrence and progression of dystrophic spinal deformity is affected by adjacent neurofibromas, then therapies targeting to reduction or stabilization of paraspinal tumors could provide a promising approach to spine deformity prevention in patients with NF-1.

Apart from its tumor suppressor activities through the Ras signaling, the role of neurofibromin may converge with other bone biochemical pathways, such as bone morphogenetic protein (BMP) signal transduction [44]. This theory suggests that intrinsic bone pathology due to loss of a functional NF–1 allele with subsequent Ras deregulation may be responsible for osseous manifestations in NF-1 through altered osteoblastic/osteoprogenitor differentiation, overgrowth of cellular tissue due to preferred fibroblast differentiation of mesenchymal cells, and impaired bony callus formation. Double inactivation of NF-1 by somatic mutation of the NF–1 gene in a population of cells which depends on neurofibromin-regulated Ras signaling to maintain normal bone was suggested to contribute to the occurrence or progression of tibia pseudarthrosis [45]. Although such second hit events have been demonstrated in pathological tissue from NF-1 tibias, it is unknown whether spinal deformities of NF-1 require a second hit event.

The NF-1 heterozygous mouse has a minimal skeletal phenotype. In order to better understand the mechanism of skeletal abnormalities in NF-1, researchers have developed more complex mouse models with knock-out of the second NF–1 allele in osteoprogenitor cell lines, using a process called cre-recombination. In one such model, Col2.3Cre(+) mice showed multiple vertebral anomalies, including: short vertebral segments, reduction in cortical and trabecular bone mass of the vertebrae, increased numbers of osteoclasts, and decreased numbers of osteoblasts in vertebrae [45]. These mouse models provide additional insight to the underlying pathophysiology of spinal abnormalities in humans with NF-1.

Most often plain standing posterior–anterior and lateral radiographs are sufficient for screening the curvature. An angle of greater than 10° assigns the deformity as structural. When treatment is to be initiated, multiple planar films in supine bending modes and traction are necessary to determine flexibility. If there are adjacent structures requiring further clarification, higher levels of imaging are required, such as computed tomography (CT) for bony deformity or high-resolution contrast CT or MRI for soft tissue delineation.

Dural ectasia is a circumferential dilatation of the dural sac which is filled with proteinaceous fluid. The slow expansion of the dura results in erosion of the surrounding osseous structures resulting in widening of the spinal canal, thinning of the laminae, and ultimately destabilization of the spine. Dural expansion through the neural foramina can cause meningoceles giving the radiographic dumbbell appearance. However, enlargement of a single neural foramen on an oblique radiograph is usually caused by neurofibroma exiting from the spinal canal rather than from the dural ectasia (Fig. 16.1). Similar lesions are seen in other connective tissue disorders, e.g., Marfan’s syndrome and Ehler–Danlos syndrome, although cause of these lesions in NF-1 is not known.

During this process, the neural elements are not affected. As a result of slow nature of this process and enormous widening of the spinal canal the neural elements have adequate room for accommodation, and there may be severe angular deformity and distortion without neurological deficit. The patients remain neurologically intact until later in the course of the disease process when destabilization of the vertebral column jeopardizes the neural elements. Dislocation of the vertebral column due to dural ectasia has been reported in the literature [46]. The destabilization at the costovertebral junction can result in penetration of the rib head into the spinal canal with neurological compromise (Fig. 16.2) [47, 48]. The presence of rib head or the neurofibroma in the spinal canal can result in intraoperative neurological deficit if instrumentation is used for correction of the curve without adequate decompression.

Dural ectasia can be readily seen on high-volume CT myelography or contrast-enhanced MRI and is recommended before surgical intervention is undertaken for dystrophic curves. Higher imaging studies help to demonstrate extremely thin laminae; in which case dissection by electocautery rather than by periosteal elevators are recommended during surgical exposure to avoid direct injury to the neural elements/dura by plunging into the spinal canal. Surgical spinal stabilization and fusion does not alter the course of dural ectasia. Dural ectasia can result in failure of the primary fusion or the expanding dura ultimately can destroy a solid fusion leaving behind the instrumentation (Fig. 16.3).

Spinal affections in NF-1 can be described under following regions: cervical, thoracic/thoracolumbar, lumbosacral, and spinal canal.

This is the common variety of spinal deformity observed in NF-1. These curves behave similar to idiopathic curves with some differences [7, 9, 58]. This form usually involves 8–10 spinal segments. Most often, the deformity is convex to the right. However, these curves usually present earlier than the idiopathic curves and are more prone to progression. Furthermore, the rate of pseudoarthrosis following a fusion surgery is higher in these patients [49]. These differences can be attributed to the process of modulation and the underlying bone pathology. Compared to dystrophic curves, non-dystrophic curves tend to present in older children with less angulation and rotation of the deformity [59].

This is an uncommon but malignant form of spinal deformity. It is characterized by early onset, rapid progression and is more difficult to treat [60, 61]. Typically, the dystrophic curve is a short-segmented, sharply angulated type that includes fewer than six spinal segments. Dystrophic curves may be associated with kyphosis and have a higher incidence of neurological injury [61, 62].

Dystrophic vertebral changes develop over time (Table 16.2). Dystrophic curves are found most commonly in the thoracic region (Figs. 16.5, 16.6, 16.7, 16.8, 16.9, and 16.10) [63].