The possible coexistence of ossiculum terminale and os odontoideum might also further support the congenital theory. The third ossification centre of the odontoid process, known as the ossiculum terminale, arises from the centrum of the proatlas. Usually, it appears by the age of 3 years and fuses by the age of 12 years [55]. A congenital anomaly originates when fusion does not occur, resulting in an ossiculum terminale persistens. This can be visualized as a small radiodense ossicle at the tip of the dens [1]. Fusion failure of the ossiculum terminale, also known as Bergman ossicle or « os avis », should not be confused with a type I fracture of the dens [31]. Patients with ossiculum terminale may be unaware of such anomaly since the C1–C2 junction remains stable because the transverse ligament C1–C2 complex is intact. Similar to ossiculum terminale persistens, os odontoideum has been considered as a consequence of a fusion failure between the sclerotomes proatlas and C1.

Sometimes, this anomaly consists in defects of segmentation in the midportion of the dens along with subaxial spine and skeletal anomalies. In ten children, reported in the literature, with os odontoideum, six had skeletal dysplasia, one had C5–C6 fusion and three had no other congenital anomalies [8]. Occasionally os odontoideum may fuse with the clivus. The anterior arch of C1 may be hypertrophic, whereas the posterior arch may be hypoplastic.

Os odontoideum can be isolated or can be also found in other congenital bone and connective anomalies as spondyloepiphyseal dysplasia congenita, Conradi’s disease, and metatropic dwarfism. It can be associated to chromosomal anomaly as Down’s syndrome and with metabolic disease as Morquio’s syndrome and other mucopolysaccharidoses [53].

Spondyloepiphyseal dysplasia congenita (SED) is a rare form of skeletal systemic disease, characterized by congenital dwarfism with a short trunk and epiphyseal dysplasia in the long bones and vertebral bodies. Patients also frequently suffer from atlantoaxial instability due to os odontoideum. Compression of the spinal cord caused by atlantoaxial instability is a common and serious complication in SED patients, possibly resulting in sudden death [19].

This uncommon condition is also known under a variety of names such as chondrodystrophia foetalis calcificans, stippled epiphyses, or congenital multiple epiphyseal dysplasia. Two main forms are now recognized: the classic, but uncommon, severe rhizomelic type due to an autosomal recessive gene and the more frequent Conradi-Hunermann variety due to an autosomal dominant gene or a mutant gene [53]. Conradi’s disease is characterized by the presence of stippled epiphyses at birth in association with other connective tissue disturbances. The spine shows alterations in the form of the vertebral body anomalies often with blunting of its anterior part. The cartilaginous epiphyses show patchy mucoid degeneration and cystic spaces with irregular vascularization. There is scattered calcification with poor alignment of cartilage columns and absence of a zone of provisional calcification. Juxta-articular tissues show ectopic calcification. Atlantoaxial instability due to os odontoideum has been described.

This rare genetic disorder is characterized by an extremely small stature, short limbs and skeletal abnormalities with atlantoaxial instability due to os odontoideum. It is caused by a mutation in the TRPV4 gene, although in most cases the cause has not been elucidated [53].

Down’s syndrome (trisomy 21, rarely translocation from 13 to 15 groups to chromosome 21) is the most common chromosomal abnormality occurring in 1 out of 700 live births. Anomalous metabolism of mucopolysaccharides and deficient growth hormone lead to osseous abnormalities [45]. The ligamentous laxity contributes to a high incidence of C0–C1 instability and C1–C2 luxation. The former has been assumed to occur in 10 % and the latter in 15–25 % of patients with Down’s syndrome [57]. C1–C2 luxation is « stable » in 44 % of cases whereas 60 % of these patients develop precipitous onset of cervical medullary compression. C1–C2 luxation is «unstable » in 50 % of cases, the average predental space being 8 mm in the neutral position [39]. About one-third of patients with Down’s syndrome harbour os odontoideum; 22 % present C1-arch hypoplasia and bifid anterior or posterior arches [62].

A cautious attitude towards the surgical treatment of CVJ instability in patients with Down’s syndrome is generally adopted because of a low incidence of associated neurological dysfunctions and a high surgical morbidity. Only subjects with an atlanto-dens interval exceeding 4.5 mm or with other evidence of CVJ instability are considered for operative stabilization [3]. Difficulties encountered in patients with Down’s syndrome are widely referred to the immunocompromised state (impaired monocyte and neutrophil chemotaxis, decreased phagocytoses, qualitative deficiency of T lymphocytes) which may facilitate respiratory infections and postoperative complications. Rates of bone fusion may also be lower than in other patients probably for deficient collagen synthesis, which contribute to bone graft resorption [54]. Halo immobilization is instituted early and maintained until objective evidence of bony fusion, which may take as long as 6 months [50].

Morquio’s syndrome is a genetic mucopolysaccharidosis (type IV) resulting from the absence or the reduced activity of lysosomal hydrolase leading to a pathological collection of mucopolysaccharides in the ligamentous system and consequently hyperlaxity. Children with this syndrome suffer from problems of the lower spine and, more commonly, of the CVJ especially at C1–C2 level. Dysplastic-hypoplastic dens with a detached distal portion (os odontoideum) not always ossified is present in every case. The ring of the atlas tends to be narrowed and thickened [5]. Although C1–C2 instability is mild, severe spinal cord compression can occur because of both anterior extradural soft tissue thickening and invagination of the posterior arch of C1 into the foramen magnum due to condylar hypoplasia. The anterior extradural soft tissue formation is usually reactive ligamentous tissue with or without abnormal deposition of mucopolysaccharides [5]. Decompression remains the main part of the treatment because of the narrow canal and CVJ instability; removal of the posterior arch of C1, widening of the occipital foramen and occipitocervical fusion are required [46]. Interestingly, posterior CVJ fusion results in late disappearance of the soft tissue thickening and normalization of dens development [5].

According to Crockard, every Morquio patient should be investigated for spinal cord compression between 3 and 8 years of age. Treatment aims to prevent neurological damage or, at least, to arrest the progression of the already present neurological disability. Transoral decompression and posterior fusion can be combined in cases with significant anterior compression [5].

The opponents to the congenital theory emphasize that the non-union between the dens and the body of C2 does not appear related to known ossification centres; consequently true os odontoideum due to a failure of development should be quite rare. Furthermore, such non-union may be found in patients having a normal dens before an actual diagnosis of os odontoideum [13]. According to Crockard, os odontoideum should be the product of excessive movement at the time of ossification of the cartilaginous dens [7]. Functionally, this anomaly resembles a C2 fracture at the base of the dens and may leave a weakened C1–C2 complex with a potential or real instability [62]. Thus, os odontoideum probably represents an old non-union fracture or injury to vascular supply of developing odontoid [47, 53].

To support the acquired theory, Verska and Anderson reported two identical twins, one with an os odontoideum after trauma and the other with a normal odontoid and no traumatic history [63]. In the literature, proponents of acquired aetiology also reported 11 patients who had radiographic documentation of a normal odontoid before developing an os odontoideum after a trauma. However, most of those cases of os odontoideum with a pre-traumatic normal radiograph were documented before CT was available. Obviously, only based on plain-film evaluation, the possibility of congenital anomalies cannot be totally excluded [12, 13, 16, 23, 47, 51, 73–75].

Several authors indicated disorders of blood supply as the cause of acquired os odontoideum. In such a direction, Sakaida et al. [49] suggested avascular necrosis after trauma while Hammerstein et al propounded a block of the blood supply to the proximal part of odontoid induced by a trauma as a pathogenetic factor accounting for osseous absorption, in conditions in which the ossiculum terminale continues to receive a normal blood supply, so growing into an os odontoideum [22].

In os odontoideum, hypertrophy of the C1 anterior arch frequently attracted the attention of the investigators. According to Holt et al, hypertrophy of the C1 anterior arch is always associated with os odontoideum and, consequently, it is a useful sign to distinguish os odontoideum from odontoid fracture [26]. In the literature, hypertrophy of the C1 anterior arch was related to an acquired atlantoaxial instability occurring prior to the skeletal maturity and causing the C1 anterior arch hypertrophy [59]. Other authors interpreted the hypertrophy of the C1 anterior arch as the consequence of blood supply preservation at that level after posttraumatic vascular blockage and ischemic necrosis in the watershed area [26, 49].

The stability of the atlantoaxial articulation depends fundamentally upon the integrity of the odontoid process and the ligaments [9]. Since the reserve space in the atlantoaxial articulation is relatively great in childhood, some patients may not complain of any neurological deficits even when atlantoaxial instability is present. However, if atlantal displacement exceeds the limit of reserve space, clinical symptoms will appear [44]. The neurological manifestations arise from bulbospinal compression both at rest and during motion, due to the CVJ instability itself. The impingement of bulbospinal junction due to os odontoideum may be responsible for myelopathic sensory motor and/or cardiorespiratory disturbances. Lower cranial nerve dysfunctions, bowel and bladder dysfunction, motor weakness up to para- or tetraparesis, hypo-anaesthesia, allodynia, hyperesthesia and hyperalgesia at upper and lower limbs have been described with variable onset (acute or chronic). Recently a sleep apnoea syndrome has been also described as a consequence of CVJ instability associated to os odontoideum [28].

Nevertheless, although there is insufficient evidence to support diagnostic standards/guidelines and treatment standards/guidelines, plain x-rays of the cervical spine (anterior-posterior, open-mouth odontoid and lateral) and plain dynamic lateral x-rays performed in flexion and extension are recommended. TC and/or MRI of the CVJ must be considered mandatory [73, 74].

To date the management of os odontoideum continues to be a subject of debate [4, 56, 75]. The surgical indication is based on the consideration that affected subjects may develop neurological symptoms and signs in case of a C2 shift higher than 4.5 mm, as evaluated by dynamic x-ray examination; indeed, in such a condition the CVJ is always unstable [73].

In the past three decades, increased understanding of spinal biomechanics, proliferation of sophisticated CVJ instrumentation devices, advances in bone fusion techniques, refinement of anterior approaches to the CVJ, and development of microsurgical and minimally invasive methods have made it possible to stabilize every segment of CVJ successfully, regardless of the offending pathology. Accordingly, use of spinal fusion and instrumentation has increased. The question facing the modern spine surgeon is not so much how to stabilize the spine but when to do so. Instability of the CVJ appears to be the key factor for the surgical indication.

Arvin et al recommended the patients with os odontoideum and neurological abnormality or radiological evidence of instability should be offered surgery although cervical arthrodesis in the paediatric age group has the potential for limiting growth potential and causing secondary deformity [4]. In such a condition the surgical management of os odontoideum aims to achieve neural decompression and to stabilize the CVJ [33, 36, 37].

On the other hand Fielding et al reported out of a series of 35 case series with CVJ abnormalities, eight patients without instability who had undergone conservative treatment without any neurological deterioration in a 3-year follow-up period [13].

Other authors underline that patients with os odontoideum, either with or without C1–C2 instability, are at risk for acute spinal cord injury after minor traumatic injury, although without symptoms nor neurological signs prior to the trauma. In order to prevent such an evenience, a preventing treatment fixation and fusion should be undertaken as prophylactic treatment in all cases of os odontoideum [42, 44, 45, 75]. Our personal practice fits with this philosophy.

Several surgical options have been described for os odontoideum:

Anterior transoral decompression has been suggested in children with preoperative irreducible C1–C2 dislocation harbouring os odontoideum followed by posterior instrumentation and fusion (wiring or screwing); the latter can also be considered a “stand-alone” procedure in case of reducible C1–C2 dislocation. Whereas irreducible C1–C2 dislocation seems to be associated to a significant higher incidence of C1 assimilation and C2–C3 fusion, reducible C1–C2 dislocation appears more frequently associated to os odontoideum; consequently the posterior approach is more indicated in the management of such a pathological condition [50]. The wiring technique is an old-fashioned though effective procedure for paediatric patients [21]. On the other hand, screwing may be technically demanding in this type of population where operative precision and expertise when placing screws in the transarticular C1–C2 space are particularly required [7, 8, 48, 52]. Biomechanical experimental studies have demonstrated that the immediate stability is better achieved after screwing than after wiring [32, 67]. Differences in fusion rates such as 94 % after screwing and 80 % after wiring, which were reported in the past, do not appear to influence the overall final clinical results to determine the choice of the first or the second option. In other terms, as in nearly all the surgical pathologies, the operation should have rather been tailored on the single subject. According to the literature nearly all the procedures tend to reach almost a 100 % rate of fusion [35, 37, 39].

The less invasive C1–C2 wiring instrumentation and bone fusion technique is the Gallie technique which constitutes the oldest variant and incorporates wires or cables under the posterior C1 ring and around the C2 spinous process. An onlay autogenous corticocancellous iliac crest graft is held in place by the wire. In the Brooks technique the wire passes under the C2 lamina, as well as under the C1 ring, and a tricortical bone graft is wedged between the C1 ring and the C2 lamina, blocking extension [24, 25].

While sublaminar wire passage adds surgical risk, the Brooks fusion is more rigid than the Gallie fusion.

The Sonntag technique is somewhat of a blend of the Brooks and Gallie method: one bicortical autologous bone graft is used, the wire is passed sublaminar to C1 only and the bone graft is wedged between C1 and C2 (trapping it between the loops of wire). This technique may have less tendency to pull C1 posteriorly than Gallie and yet avoids the inherent danger of C2 sublaminar wires as in Brooks fusion [17].

The Magerl technique provides posterior C1–C2 screw fixation and is a reliable means of atlantoaxial joint immobilization. While this technique does confer the most rigid fixation, it is technically demanding. The Magerl technique requires a near-anatomic reduction and normal vertebral artery anatomy [61]. Harms technique of fixation to the lateral mass of C1 and the pars or pedicle of C2 has become the most prominent means of stabilization in the adult population [60]. This technique is not as vulnerable to anatomic variability in the course of the vertebral artery as the Magerl technique [71].

Sublaminar wires with different type of autologous bone grafts to treat CVJ instabilities are less incapacitating for children, since the short fixation does not interact with the local and global paediatric spine growing pattern. The inclusion of the occipital bone to perform an occipitocervical fusion in paediatric patients better accomplishes the lever effect by producing a dynamic force pattern with a pulley-like mechanism in irreducible C1–C2 dislocations. In other words C1 is attracted by C0 and C2 by including the occiput, and, in our opinion, such a procedure remains the most appropriate treatment when the CVJ instability depends from a progressive genetic-metabolic disease predisposing to possible postoperative cranial settling after C1–C2 fixation [10, 32, 33, 40]. Other studies demonstrate that children older than 4 years, harbouring CVJ instability, can be safely managed with posterior C1–C2 transarticular screw fixation with a high fusion rate [71]. However, whichever the technique utilized, the non-union can occur, also as a consequence of postoperative CSF fistula and wound infection, and it can heal, after the resolution of the infection, with a second bone graft [68].