Fractures of the Odontoid
E. FOURNIOLS - J.Y. LAZENNEC - A. HAMMA
Pitié-Salpétrière University Hospital - F-75013 Paris, France
Odontoid fractures were first described, by Lambotte, more than a century ago (1894). Throughout the literature, attention has always been drawn to the fact that many of these fractures are detected late after the traumatic event, and that odontoid fractures are notoriously prone to non-union. They may cause neurological impairment and even death.
For a long time, odontoid fractures used to be treated nonsurgically; the first case managed with posterior surgical fixation (wiring) was reported in 1910 (Mixter and Osgood). Since 1970, pathology-based classifications have been evolved to guide the management and to help predict the outcome of these fractures. These schemes are based upon anatomical observations (Fig. 1).
Parasagittal anatomical section
Note thickness of the transverse ligament (A) and the gap behind the odontoid.
Anderson and DAlonzo classification
The earliest classification was that of De Morgues and Fischer (1972), which distinguished between fractures of the base and fractures of the neck of the odontoid process.
The system most widely used in the English-speaking countries is that of Anderson and DAlonzo, which dates from 1974 (Fig. 2). This classification recognizes three fracture types:
- Type I: Apical (tip) fracture;
- Type II: Body fracture, which passes above a horizontal line drawn through the upper border of the superior articular facets of the axis. It corresponds to the odontoid neck fracture in the De Morgues and Fischer scheme.
- Type III: Basilar fracture, with a fracture line passing into the body of the axis.
These two classification systems are considered to be of prognostic value, since neck (Type II) fractures are very prone to non-union unless sufficient stabilization of the fracture site can be provided.
Fracture site stability was analyzed in particular by Roy-Camille, who published his own classification in 1973. This system is based upon a study of the direction of the fracture line on lateral films, and on a study of dynamic (flexion-extension) films. A fracture is considered to be unstable if it is displaced at presentation, or if, while originally undisplaced, it is found to be displaced on the dynamic views taken on or about Day 10.
The Roy-Camille system comprises three fracture line patterns (Fig. 3):
- OBAV (oblique en bas et en avant), in which the fracture line slopes forwards, and where the displacement will typically be in an anterior direction (Fig. 5);
- OBAR (oblique en bas et en arrière), in which the fracture line slopes backwards, and where the displacement will normally be in a posterior direction (Fig. 4);
Figure 3: Roy-Camille classification Figure 4: Backwards sloping (OBAR) fracture, with typical posterior displacement
Figure 5: Forwards sloping (OBAV) fracture Figure 6: Typical horizontal fracture
- HTAL, in which the fracture line is horizontal; these fractures may displace either anteriorly or posteriorly.
One special odontoid fracture pattern is the to-and-fro horizontal type, which alternates between anterior and posterior displacement. Another pattern, characterized by a horizontal fracture with a wide inter-fragment gap and anterior and posterior comminution, has been termed a bobbys helmet fracture.
This classification is based upon a study of the lesions in the sagittal plane; it was devised in order to obtain prognostic criteria on the risk of non-union, and to guide treatment.
The risk of non-union is not a function of the blood supply to the odontoid, as claimed by Althoff. The decisive factor appears to be the displacement potential of the fracture, either as a gross radiographically demonstrable phenomenon, or in the form of micromotion. There is no detailed information in the literature concerning the influence of the fracture site and pattern. Roy-Camille stressed the fact that it is the combination of the direction of the fracture line and the displacement that matters. Thus, because of their displacement potential, backwards sloping (OBAR) fractures tend to heal less well than do horizontal or forwards sloping (OBAV) fractures.
There is little in the literature concerning asymmetrical fracture patterns. We performed a study with a.p. and lateral tomography, in our earlier cases, and with CT in the more recent past, which showed a large number of asymmetrical fracture patterns in the coronal plane. The fracture line may start on one side of the neck of the odontoid, and run out into the superior articular process of the axis on the other side. This pattern suggests a rotatory force vector involved in the production of the bony lesions; it also harbours an additional risk of instability, especially if the alar ligaments are no longer intact.
An asymmetrical pattern may also result in the fragment diastasis being misjudged on the lateral view, and may lead to faulty interpretation of the radiographs.
This is why we feel that it is most important to combine the Roy-Camille assessment in the sagittal plane with a study of the fracture levels in the coronal plane, in order to judge the three-dimensional stability of the lesion and to choose the correct treatment modality.
Roy-Camille did not specify the reference plane in relation to which a fracture would be considered to be oblique or horizontal. A retrospective analysis of our 225 cases has prompted us to establish this reference plane at a line drawn tangent to the endplate of the axis. The use of this landmark has made it possible for lateral radiographs to be accurately and precisely interpreted by different observers.
We have also found that a backwards sloping (OBAR) fracture does not necessarily displace posteriorly, nor a forwards sloping (OBAV) one, anteriorly. In our patient material, paradoxical displacement was seen in 5% of the cases (Fig. 8). While backwards sloping fractures are almost invariably in the neck of the odontoid, and usually symmetrical in the coronal plane, forwards sloping fractures are not synonymous with basal fractures, neither do horizontal fractures always involve the neck. Whenever the two latter types are encountered, a detailed search should be made in the coronal plane, to detect any signs of rotational instability.
Figure 7: Anderson Type II fracture, with fragment rotation Figure 8: Backwards sloping (OBAR) fracture, with paradoxical anterior displacement
There is no mention, in any of the classifications, of fractures with purely lateral displacement on the open-mouth views. In our series, we had ten such cases. Some had lateral angulation (with a slightly asymmetrical fracture line), while others had almost complete translation of the odontoid. These lesions are indicative of rupture of the alar ligament system; they are difficult to reduce, especially if the lesion is to be treated with anterior screw fixation.
Similarly, there is little mention in the literature of associated injuries of the atlas and of the remainder of the axis. With the advent of more sophisticated radiological techniques, especially of CT with 3D reconstruction, it is now obvious that the frequency of such lesions was underestimated in our earliest cases. However, these injuries have a major bearing on the treatment strategy to be adopted, since C1-C2 wiring may be ruled out if the posterior arches have been damaged, while fractures of the anterior arch of the atlas will constitute an additional - and difficult-to-manage - source of instability.
The study of odontoid fractures must involve a detailed description of the fracture lines, and requires three-dimensional reconstruction of the fracture patterns. Associated ligamentous lesions may be detected by the demonstration of odontoid angulation, asymmetrical positioning of the odontoid in relation to the anterior arch of the atlas, or of a fracture pattern that is clearly the result of a rotational force vector. The bony and ligamentous lesions will cause major instability, which is why the treatment provided should seek to reduce the fracture and to achieve perfect stability in order to prevent non-union.
From our retrospective study of 225 cases, we have learned the importance of analyzing all these lesions on a.p. and lateral films, and of studying them also in rotation, in order to be able to reduce and stabilize the fractures so as to prevent non-union. The Roy-Camille classification is of great importance, since it goes beyond the Anderson scheme by providing information on how the lesions seen should be treated. Fractures with angulation or lateral translation are in a group of their own, being extremely difficult to manage; while lesions involving a rotational force may be difficult to stabilize, especially with posterior wiring techniques.
1 - Althoff Bo. Fracture of the odontoid process. An experimental and clinical study. Acta Orthop Scand Supp, 1979, n° 177 : 5-9O.
2 - Amyes E.W AFM. Fractures of the odontoid process. Arch Surg, 1956 ; 72 : 377-393
3 - Anderson L.D DRT. Fractures of the odontoid process of the axis. J. Bone and Joint Surg. 1974, 56 A : 1663-1674
4 - Blockey N.J. POW. Fractures of the odontoid process of the axis. J. Bone and Joint Surg. 1956, 38 B : 794
5 - Bombart M. Traumatisme des deux premières vertèbres cervicales . 1972 ; 5-17
6 - Bohler J. Fractures of the odontoid process. J. Trauma 1965 ; 5: 386-391
7 - Crooks F BA. Fractures and dislocations of the cervical spine. Br. J. Surg. 1944.31:252-265
8 - Daniels DL, Williams AL, Haughton VM. Computed tomography of the articulations and ligaments at the occipito-atlanto-axial region. Radiology 1983 ; 146 : 7O9-716
9 - Dickman CA, Mamourian A, Sonntag VK, Drayer BP. Magnetic resonance imaging of the transverse atlantal ligament for the evaluation of atlantoaxial instability. Journal of neurosurgery 1991; 75 (2) : 221-227
10 - Dvorak J, Panjabi MM. Functional anatomy of the alar ligaments. Spine 1987 ; 12 : 183-187
11 - Ludwig K. Uber das ligamentum alare dentis apistrophei des Menschen. Z Anat Entwickl Geschic 1952 ; 116 : 442
12 - Maladain A. Résultats éloignés du traitement chirurgical des fractures de lodontoïde, 1972
13 - Mourgues (de) G. FLP. Fractures de lapophyse odontoïde (DENS) de laxis : 1O2 cas dont 73 fractures récentes. Rev. Chir. Orthop. 1981 ; 67 : 783-79O
14 - Ramadier J.O., AJFESJ. Les fractures de lapophyse odontoïde 94 cas dont 61 traitées par arthrodèse. Rev. Chir. Orthop 1976 ; 62:171-189
15 - Rouviere H, Delmas A. Anatomie Humaine. Desciptive, topographique et fonctionnelle. Pars : 1978 (Masson, ed ; vol I, Tête et cou)
16 - Roy-Camille, De La Caffiniere JH, Saillant G. Traumatisme du rachis cervical supérieur C1 C2. 1973, 51
17 - Ryan MD, TTKF. Fractures of the odontoid process of the axis. J. Bone and Joint Surg. 1982 ; 64-B : 417-421
18 - Saternus KS. Verletzungen der Occipito-atlanto-axis region. Z Orthop 1981 ; 119:662-664
Treatment Strategy - A Study of 225 Cases
Odontoid fractures account for 1 - 2% of all spinal fractures, and for 8 - 15% of all fractures of the cervical spine. Even nowadays, they are notoriously difficult to diagnose; displaced odontoid fractures (15 - 85%, in the different studies) are prone to result in neurological deficits and non-union, unless treated surgically. The treatment strategy to be adopted will depend on a number of factors, which have been discussed in the literature: the fracture pattern; the age and general condition of the patient; neurological deficits; and associated injuries. We performed a retrospective study of 225 cases in order to establish anatomical criteria to guide patient selection, and in order to follow the further course of patients treated with the different modalities.
The study comprised 225 odontoid fractures treated between 1972 and 1994. Twelve patients were lost to follow-up, and eleven died before union. This left 202 cases available for analysis; the mean follow-up was 14 months.
Backwards sloping (OBAR)
Forwards sloping (OBAV)
Table 1 Fracture patterns. Correspondence between Roy-Camille and Anderson & DAlonzo classifications
The mean patient age was 50 years. There was a bimodal age distribution, with the two peaks associated with two different fracture mechanisms. The first peak was in younger subjects (20 - 35 years), who had sustained violent trauma (road traffic accidents, falls from a height; sports accidents). The second peak was in elderly subjects (80 - 95 years), who had sustained the fractures in falls from a standing height.
Males were affected twice as often as were females.
In almost 20% of the cases, the diagnosis was not made in the first few days after the accident. This underlines the importance of repeat radiographs, with dynamic (flexion-extension) views, taken eight to ten days after any cervical spine trauma.
The assessment of the cases was based upon a clinical workup (neurological status) and a radiological examination (conventional radiographs; tomography, in the earlier cases; CT, once that technique had become available).
The two chief symptoms reported by the patients were neck pain (40% of the cases) and headaches (10% of the cases).
Using the Roy-Camille classification of odontoid fractures, 75 of the fractures (33%) were backwards sloping (OBAR); 76 (34%), horizontal (HTAL); 72 (31%), forwards sloping (OBAV); and 2 (1%), tip fractures. Table 1 shows the correspondence with the fracture patterns recognized in the Anderson & DAlonzo classification. Only 47 fractures were not displaced at presentation, and only 40 were not displaced on the dynamic views. Many of the patients (26% of the cases) had associated cervical spine lesions. Associated organ damage or other orthopaedic injuries (other than upper C-spine trauma) were found in 19% of the cases; in 8% of the cases, there were multiple injuries.
A number of treatment modalities were employed. Our management principles have evolved over time. The choice of how to treat a fracture was always made in the light of the particular fractures stability.
If the fracture was not displaced when first seen, and was still stable on the dynamic views taken on Day 10, nonoperative treatment would be provided. This treatment would take the form of a minerva cast to be worn for three months.
If the fracture was found to be displaced on the first films or, subsequently, on the dynamic views, the treatment would be surgical.-
Between 1972 and 1982, the choice was between posterior fusion or C1-C2 wiring without grafting (for immobilization), or wiring for reduction, in which a posterior graft might be used. Type 1 wiring (with a loop passed over the spinous process of C2) was used in the management of forwards sloping (OBAV) and of anteriorly displaced horizontal (HTAL) fractures; while Type 2 wiring (with a loop passed under the C2 spinous process) was employed in backwards sloping (OBAR) and with posteriorly displaced horizontal fractures. Grafts were used with simple wiring (Fig. 1) in a parcel pattern, and with reduction wiring.
Type 2 wiring without grafting was abandoned in 1982.
Figure 1 : Odontoid fracture treated with "parcel"-wired graft. Radiograph at four months, after union Figure 2 : Treatment of odontoid fracture using a C1-C2 spacer. Late radiograph at three months, after union Figure 3 a and b Two-screw fixation of a backwards sloping (Roy-Camille OBAR; Anderson & DAlonzo Type II) fracture. Radiographs at 8 years after trauma
- Since 1984, two new techniques have been employed, with a view to minimizing the effects of fracture treatment on the mobility of the upper cervical spine.
The first of these techniques involves the use of a polyethylene spacer inserted between C1 and C2 (Fig. 2). The spacer is placed between the underside of the posterior arch of C1 and the superior aspect of the spinous process of C2; either Type 1 or Type 2 wiring is used. The spacer prevents the anterior gaping of the fracture line that is sometimes seen with simple wiring. The device may be removed at a later stage, once the fracture has healed, in order to restore C1-C2 mobility.
Anterior screw fixation, as devised by Böhler, (Figs. 3 and 4) was used to compress the fracture site in order to obtain immediate stability and to make postoperative immobilization less cumbersome and less prolonged.
- Two other fusion techniques were occasionally employed: occipitocervical fusion, and posterior C1-C2 screw fixation, if wiring was ruled out by posterior C1-C2 lesions or if the bone stock was of poor quality.
Figure 4 : Backwards sloping fracture treated with anterior screw fixation (Herbert screw)
The technique to be used would be chosen in the light of the fracture pattern, using the classification proposed by Roy-Camille. Most of the forwards sloping (OBAV) and horizontal (HTAL) fractures were managed with simple Type 1 wiring, C1-C2 fusion with wiring, or a C1-C2 spacer. Horizontal fractures with to-and-fro displacement in the sagittal plane were managed with a parcel-wired graft. Backwards sloping (OBAR) and posteriorly displaced horizontal fractures were treated with Type 2 wiring early in the series; later on, anterior screw fixation was used in a number of cases. Wiring plus fusion was rarely employed.
Occipitocervical fusion was confined to the treatment of associated upper cervical spine lesions in very elderly patients; it was also employed as a salvage technique at revision surgery.
C2-C1 screw fixation came in later in the study; it was used for the management of old odontoid fractures associated with lesions of the posterior arch of the atlas or the axis, and as a salvage technique at revision surgery.
In cases of grossly displaced fractures, surgery was preceded by a period of halo traction. Thus, 47 fractures were put in preoperative traction; however, in only 26 cases (55.3%) did this treatment result in complete reduction. Traction on a head rest was also used to obtain reduction, in flexion for backwards sloping (OBAR) and posteriorly displaced horizontal fractures, and in extension for forwards sloping (OBAV) and anteriorly displaced horizontal fractures. Mean reduction of the displacement was 86%; however, in 10.3% of the cases, no reduction could be obtained.
Table 2 Results as a function of treatment (patients who were lost to follow-up or who had died not included in this table)
Of the 202 fractures followed up for a minimum of three months, 154 had definite union, and 9 had questionable healing, after primary treatment. There were 18 secondary displacements, of which 15 necessitated surgery; and 21 non-unions, of which 17 underwent operative treatment.
1 - Mechanical outcome
(a) Nonoperative treatment This treatment was used in 55 patients, 40 of whom had stable fractures, and 15 had unstable fractures. Of the unstable fractures, eight had shown displacement on the initial films, and seven on the dynamic views. Of the fractures in this group, 46 went on to definite union, and two to questionable union (87%); there were five non-unions (9%), and two secondary displacements (4%). Non-union occurred in four unstable fractures and in one stable fracture; in terms of fracture patterns, non-union ensued in three backwards sloping (OBAR) and in two horizontal (HTAL) fractures. Three of the five non-unions healed after C1-C2 fusion. The two secondary displacements occurred in unstable fracture patterns, both with a backwards sloping fracture line. In both cases, union was obtained only following anterior screw fixation.
The review of our results shows the efficacy (healing rate: 97.5%) of nonoperative treatment in bringing about fracture healing, providing that the fracture concerned was not displaced at presentation or on the subsequent dynamic radiographs. It also confirms the bad reputation of backwards sloping (OBAR) fractures, regardless of their displacement status: in this study, backwards sloping fractures resulted in three non-unions and two secondary displacements.
(b) Posterior wiring This technique was used in 45 patients. It resulted in definite union of the odontoid fracture in 28 cases (62%) and questionable union in five cases (11%). There were nine non-unions (20%), and three secondary displacements (7%). Of the nine non-united fractures, seven were successfully reoperated on using a different technique. Mechanical failures were most frequently observed in backwards sloping (OBAR) and in horizontal (HTAL) fractures; with posterior displacements; with initial or residual fragment rotation; and where Type 2 wiring had been used.
(c) Posterior fusion was used in 52 cases. Of these, 42 (81%) went on to definite union; two (4%) had questionable union; four (7.5%), non-union; and 4 (7.5%), secondary displacement. Secondary displacement was chiefly seen following occipitocervical fusion in very elderly patients (three out of four). In three out of four cases, non-union was attributed to technical error (insufficient freshening; poor bone stock; poor graft quality). Diastasis or an asymmetrical fracture pattern were not seen to affect fracture healing.
(d) Anterior screw fixation was used in 29 patients. Twenty (69%) went on to union; seven (24%) had secondary displacement; two (7%) failed to unite. Of these two non-unions, one involved a screw fixation of an old hypertrophic non-union, in which the fracture site was stabilized but could not be compressed. These two non-unions could be made to heal only with posterior fusion. Of the seven secondary displacements, five required reoperation using a different technique; this was then followed by four unions and one fusion failure (in a very elderly patient in whom the occipitocervical instrumentation failed shortly after surgery). In two cases, a halo was applied, which successfully limited the displacement and brought about the healing of the fractures.
The analysis of the seven secondary displacements showed that in five of the seven cases technical error was the main cause of failure. Most often, the error consisted in placing the screw entry point on the anterior rather than on the inferior surface of C2, which rapidly led to the anterior cortex being fractured and the screw, which remained fixed to the proximal fragment, being tilted.
Failure was also related to the obliqueness of the fracture line: mechanical failure was seen in 33% of forwards sloping (OBAV) fractures, and 55% of horizontal (HTAL) fractures, as against 17.5% of backwards sloping (OBAR) fractures. Unlike the other techniques, anterior screw fixation does not appear to be affected, in terms of outcome, by initial rotatory displacement or the amount of diastasis.
(e) A polyethylene spacer plus wiring was used 21 times. Of these cases, twelve were forwards sloping (OBAV) fractures, eight had horizontal (HTAL) fractures (six anterior displacements, and one lateral displacement with fragment rotation); while one was an anteriorly displace backwards sloping (OBAR) fracture. Union occurred in 18 cases (85.7%); in one of these cases, the fracture healed even though the device had slipped early on, without any ill effects. One fracture (4.7%) failed to unite; two fractures (9.5%) underwent secondary displacement. The two secondary displacements were successfully reoperated on (exchange of device, and C2-C1 screw fixation). The only non-union seen was in a patient with a very unstable horizontal fracture, which had been anteriorly displaced on the initial radiographs, with a major (3mm) diastasis. Reduction using a spacer had resulted in a posterior displacement of the odontoid.
While this technique produces a high fusion rate, it cannot always bring about a perfect reduction of anterior displacements (eight cases) and perfectly stabilize the reduction: in four of our cases, secondary anterior slippage was observed. Also, this reduction and fixation technique appears to be less effective in associated rotatory displacements.
2 - Neurological results
Eighteen per cent of the patients had neurological signs and symptoms. In one third of the cases, there had been a neurological deficit prior to the fracture event (stroke). In the cases without prior neurological impairment, clinical examination showed greater occipital neuralgia, cervicobrachial neuralgia, quadriparesis or quadriplegia. These neurological conditions were unrelated to the fracture pattern and the direction of fragment displacement. The only risk factor for quadriplegia that could be identified was the amount of displacement: quadriplegia was associated with a mean displacement of 13mm. The mean displacement in the study population overall was 3mm. The diastasis under traction did not appear to affect the risk.
Traction produced an improvement in five of nine patients with displaced fractures. Anatomical reduction with halo traction or following surgery resulted in neurological recovery in ten of the 22 patients in this study who had neurological deficits. Patients with cervicobrachial neuralgia tended to recover more rapidly and more fully than did those with quadriparesis or greater occipital neuralgia. One patient with complete quadriplegia made a full recovery ; however, three of the five patients followed up for more than a month died.
3 - Non-mechanical treatment-related complications
There were eight treatment-related deaths. Six of these fatalities occurred as a result of pressure sore complications in patients who, despite an unstable fracture pattern but because of their extremely poor general condition, were receiving nonoperative treatment. These were mainly cases treated early in the series. The two other deaths were related to surgical treatment (one Type 2 wiring, one anterior screw fixation). One major adverse factor was quadriplegia: of the six quadriplegic patients in the study (five post-trauma; one postoperative), four died.
* Neurological complications
Type 2 wiring was followed by three major neurological complications (one quadriplegia, two cases of Brown-Séquard syndrome). In another eight cases, greater occipital neuralgia and/or cervicobrachial neuralgia occurred after treatment (nonoperative treatment - two cases; wiring - four cases; screw fixation - one case; wiring plus grafting - one case).
* Other complications
These were occipital pressure sores (twelve cases); chest infections (seven cases); and postoperative infections in patients operated on via a posterior approach (five cases).
One rare complication deserves mention: pseudo-delirium tremens. We had two cases in patients with secondary fracture displacement after surgical management (one anterior screw fixation, one occipitocervical fusion).
1 - Mechanical outcome
In this study, the treatment modality to be adopted was a function of the fracture pattern (using the Roy-Camille system) and the stability of the fracture. Stability was defined as the absence of displacement on the initial radiographs and on the dynamic views taken between eight and ten days after the traumatic event.
(a) Nonoperative treatment
The mean age of the nonoperatively managed patients was 68.5 years, which was in line with the mean age of such patients in other studies, but contrasted with the mean age of 50 years in the study overall.
A stable fracture should be managed nonoperatively. In 55 patients with stable fractures, who were immobilized in a minerva cast for three months, we had an odontoid healing rate of 97.5%. However, in the 15 cases where, for cogent reasons, it was decided that an unstable fracture (usually a backwards sloping - OBAR - fracture) should be treated nonoperatively, union was obtained in only nine cases (60%). Similar non-union rates have been reported by other authors. The studies in the literature have shown three adverse factors: the level of the fracture line (Anderson & DAlonzo Type II fractures); the initial a.p. displacement and diastasis (> 2.5mm); and patient age. The failure rates reported under these conditions range from 21% to 67% [2, 3, 6, 9, 13, 16, 20, 23, 24, 32]. Also, three deaths in our study, and six of the fatalities in the study by Mourgues & Fischer , were attributable to nonoperative treatment, which had failed properly to stabilize an unstable fracture.
Halo vests do not appear to be the solution, since Anderson et al , Fujii , and Clark & White  reported failure rates in the 20% to 30% range in patients wearing these devices. Halo vests are uncomfortable to wear, which is why we used them only three times, after early failure of operative treatment in patients who were not very suitable for reoperation. We obtained one union; had one non-union; and one patient died soon afterwards.
(b) Surgical treatment
* Operative treatment and fracture healing
In unstable fractures, surgical treatment is mandatory.
Wiring produced union in 62% of our 45 cases. Wiring without grafting gives disappointing results and may, moreover, be dangerous. Type 2 wiring without grafting was abandoned by us back in 1982. Type 1 wiring without grafting has not been used since 1991. Grosse  found that three out of four cases managed with that technique failed to unite.
Posterior fusion produced the best results in our study. In our 52 cases, the healing rate was 81%, which compares favourably with the rates of between 73% and 100% published in the literature [2, 6, 9, 13, 16, 23, 26, 32]. In common with other authors [20, 23, 32], we had several cases where the posterior graft healed in, allowing an ununited odontoid to be stabilized; and of odontoid fracture healing with non-union of the graft. This accounts for our 86% union rate, which was very close to the 85.7% rate seen in the 21 cases managed with a polyethylene spacer, where the fracture site was also stabilized and healing was promoted.
The chief criticism of this technique is that it may affect the mobility of the cervical spine, especially in rotation, of which C1-C2 provides some 25° to 35° to either side, as stated by Roy-Camille . Vichard  and Ramadier  have also drawn attention to the fact that, following fusion, the range of flexion/extension movements will be limited.
The polyethylene spacer developed by Roy-Camille [28, 29] should make it possible to recover full mobility after the removal of the device a few months after surgery. In our series, only four patients had their spacers removed; the others had refused their consent. In the four patients that did not keep their spacers, a return to normal or near-normal mobility was observed.
Several authors (Cloward, Louis, and others) have advocated direct fracture stabilization via a transoral approach. However, this approach is fraught with a considerable risk of infection. We have never employed it in odontoid fracture surgery. Fujii et al  used it in nine cases, of which two failed to unite; the authors felt that the stability provided by this technique was poor, and that major adjunctive external immobilization had to be provided.
Anterior screw fixation was first suggested by Böhler  and Nakanishi et al , who both published their papers in 1982. The theoretical advantage of direct fracture stabilization is the need for less cumbersome and less prolonged postoperative immobilization. The healing rate is equal or superior to that following posterior grafting plus wiring, and the overwhelming majority of patients recover full mobility. Anterior screw fixation makes it possible to tackle odontoid fractures associated with C1 or C2 posterior arch fractures, or with a Jefferson fracture.
Using this technique, we had only 20 of our 29 cases (69%) go on to union. In the literature [1, 4, 5, 7, 8, 10, 11, 12, 13, 14, 16, 18, 19, 21, 31], the healing rates reported with this technique are very high, ranging from 80% to 100%. Failures are mainly due to forwards sloping fracture patterns and to non-unions with sclerotic fragment margins, as pointed out by Pache  and Fujii et al . Our only failure in three non-unions was in a patient with that pattern.
We had seven cases of secondary displacement. While two of these fractures healed with continued halo traction, five required early reoperation. Several authors [1, 19, 30] have reported secondary displacement and/or loosening of the screw(s) after this procedure. In all cases, this was attributed to technical error, and all authors make that point that these complications become less frequent with increasing experience.
In our study, 17 surgeons performed a total of 29 screw fixations. This fact may account for our rate of mechanical failures, since technical error was to blame in six of the seven cases. However, it should be noted that, in five of the seven failures, the fracture pattern was either horizontal (HTAL) or forwards sloping (OBAV). Failure would, therefore, appear to have been predictable, since only a backwards sloping (OBAR) pattern allows the screw to be inserted at right angles to the fracture line. With the other patterns, there is also the risk of inadvertent fracture displacement during the compression of an asymmetrical or a poorly reduced fracture pattern.
For the management of forwards sloping (OBAV) fractures, we do not use the anterior plate recommended by Böhler  and by Grosse et al ; instead, we use posterior wiring plus a polyethylene spacer, or, if required, C1-C2 fusion by grafting and wiring. This latter technique is particularly indicated in elderly subjects. Vichard et al  found simple screw fixation to be disappointing, and used a plate spanning the C2-C3 disc space for all their screw fixations. We do not feel that it is logical to sacrifice the C2-C3 disc merely to treat an odontoid fracture.
We think that anterior screw fixation is difficult to perform, and should only be considered for the management of forwards sloping (OBAR) and of some horizontal fractures associated with C1 or C2 posterior arch fractures. If reduction cannot be obtained in the sagittal and/or the coronal plane, this technique should not be performed. We would agree with other authors that it makes no difference to the fracture healing and the mechanical failure rate whether one-screw or two-screw fixation is used.
* The mobility outcome varies from study to study, because of individual variations, on the one hand; and differences in the study criteria, on the other hand. There is only one study in which mobility was rigorously analyzed using neutral and maximum right and left rotation CT scans. Jeanneret et al , who used this technique, found that 38% had normal atlantoaxial joint motion, and that only 15% had major stiffness. Our results were similar: a little more than a quarter of the patients had normal joint motion, while 40% had a slightly reduced ROM; only one patient had major stiffness. While there are several studies showing similar results [6, 16, 19], with a little over two thirds of the patients having no or only slight decrease in their pre-trauma mobility, other authors [1, 4, 7, 14, 28] are very much more optimistic, quoting normal or near-normal ROM rates of 90% to 100%. Major stiffness of the cervical spine is extremely rare in all studies, the rate being invariably below 10%.
2 - Neurological outcome
Odontoid fractures associated with neurological impairment are an interesting subset of this category of fractures. In common with Ramadier et al  and Anderson & DAlonzo , we feel that these patients must be managed with surgery. Preoperative halo traction for the reduction of the fracture should be instituted as soon as possible. Ramadier  found that 14/15 cases made a complete neurological recovery following surgical stabilization. Geisler et al  found 3 patients to be improved, while one died; and Chang et al  observed one improvement after anterior screw fixation. On the other hand, Steimle et al  and Borne et al  each had two quadriparetic or quadriplegic patients in their series, who died early on despite surgical fixation with a good anatomical result.
The onset of pseudo-delirium tremens may signal secondary displacement. In that case, emergency reduction should be performed, to save the patients life. We had two such cases in our study; Ramadier et al  described one case; while Vichard et al  had one, and Mourgues & Fischer  reported eight.
Several factors have to be taken into account in the treatment of odontoid fractures. In the first instance, the fracture pattern must be analyzed in three dimensions, to establish both the level and the direction of the fracture line, the rotation or translation of the fracture fragments, and the stability of the lesion. Equally, account has to be taken of the age and general condition of the patient; a painstaking search has to be made to detect any neurological impairment. Our retrospective study of 225 cases has reassured us that our present patient selection policy is correct.
Stable fractures need to be treated nonoperatively. Unstable fractures must be operated on after reduction by halo traction or by intraoperative reduction on a head rest. Whenever there is neurological impairment or a widely displaced horizontal fracture, halo traction must be performed as the first step in the treatment regimen.
The Roy-Camille odontoid fracture classification distinguishes among three patterns, whose management requires different surgical techniques.
Backwards sloping (OBAR) fractures and horizontal (HTAL) fractures associated with C1 or C2 posterior arch fractures should be treated with anterior screw fixation. This technique imposes certain requirements; in particular, the fragments must be meticulously reduced prior to fixation.
Forwards sloping (OBAV) fractures and anteriorly displaced horizontal fractures should be treated with wiring and a polyethylene spacer; wiring plus grafting should be used only where the overriding need is for fracture healing, and where postoperative C1-C2 mobility is a less important consideration.
Horizontal fractures with to-and-fro displacement in the sagittal plane should, preferentially, be managed with posterior fusion and C2-C1 screw fixation; parcel wiring over a graft would be used less frequently.
Occipitocervical fusion remains a rarely used salvage procedure, especially since it is of doubtful value in elderly patients. The joint use of the classifications established by Anderson & DAlonzo and by Roy-Camille has the benefit of showing the exact level of the fracture site and the course of the fracture line, which may be asymmetrical, involving different lesions in the body of C2. A careful analysis of the radiographs should provide clues to rotatory instability, which would also have to be taken into account in the planning of a treatment regimen; equally, the radiographs should reveal angulation or translation of the odontoid in the coronal plane. These lesions are very rare, but are notoriously difficult to reduce and to stabilize, especially by anterior screw fixation.
1. Aebi M, Etter C and Coscia M : Fractures of the odontoid process. Treatment with anterior screw fixation. Spine, 1989, 14(10), 1065-1070.
2. Anderson LD, dAlonzo RT : Fractures of the odontoid process of the axis. J Bone Joint Surg, 1974, 56A, 8, 1663-1674.
3. Anderson PA, Budorick TE, Easton KB, Henley MB and Salciccioli GG : Failure of halo vest to prevent in vivo motion in patients with injured cervical spines. Spine, 1991, 16, 10 suppl, 501-505.
4. Autricque A, Lesoin F, Jomin M : Le vissage de lodontoïde. Presse Med, 1988,17, 32, 1647-1649.
5. Böhler J : Anterior stabilization for acute fractures and non unions of the dens. J Bone Joint Surg, 1982, 64A, 1, 18-27.
6. Bombart M. : Traumatisme des deux premières vertèbres cervicales. Orthopédie et traumatologie : conférence 1972, 5-17.
7. Borne GM, Bedou GL, Pinaudeau M, Cristino G, Hussein A : Odontoid process fracture osteosynthesis with a direct screw fixation technique in nine consecutive cases. J Neurosurg, 1988, 68, 223-226.
8. Chang KW, Liu YW, Cheng PG, Chang L, Suen KL, Chung WL, Chen UL, Liang PL : One Herbert double threaded compression screw fixation of displaced type II odontoid fractures. J Spinal Disord, 1994, 7, 1, 62-69.
9. Clark CR, White AA : Fracture of the dens. A multicenter study. J Bone Joint Surg, 1985,67A, 9, 1340-1348
10. Esses SI, Bednar DA : Screw fixation of odontoid fractures and non unions . Spine, 1991, 16, 10, 483-485.
11. Etter C, Coscia M, Jaberg H, Aebi M : Direct anterior fixation of dens fractures with a cannulated screw system . Spine, 1991, 16, suppl 3, S25-S32.
12. Fourniols E. : Vissage antérieur des fractures de lodontoïde. Analyse dune série de 28 cas. Mémoire de D.E.S. ,1995.
13. Fujii E, Kobayashi K, Hirabayashi K : Treatment in fractures of the odontoid process. Spine, 1988, 13, 6, 604-609.
14. Geisler FH, Cheng C, Poka A, Brumback RJ : Anterior screw fixation of posteriorly displaced type II odontoid fractures. Neurosurgery, 1989, 25, 1, 30-37.
15. Graziano G, Jaggers C, Lee M, Lynch W : A comparative study of fixation techniques for type II fractures of the odontoid process. Spine, 1993, 18, 16, 2383-2387.
16. Grosse A, Bohly J, Taglang G, Dosch JC, Kempf I : Ostéosynthèse par vissage des fractures de lapophyse odontoïde. Rev Chir Orthop, 1991, 77, 425-431.
17. Jeanneret B, Vernet O, Frei S, Magerl F : Atlanto-axial mobility after screw fixation of the odontoid: a computed tomographic study. J Spinal Disord, 1991, 4, 2, 203-211.
18. Lesoin F, Autricque A, Franz K, Villette L, Jomin M : Transcervical approach and screw fixation for upper cervical spine pathology. Surg Neurol, 1987, 27, 459-465.
19. Montesano PX, Anderson PA, Schlehr F, Thalgott JS, Lowrey G : Odontoid fractures treated by anterior odontoid screw fixation. Spine, 1991, 16, suppl 3, S33-S37.
20. Mourgues de G, Fischer LP : Fractures de lapophyse odontoïde (dens) de l axis. 102 cas dont 73 fractures récentes. Rev Chir Orthop, 1981, 67, 783-790.
21. Nakanischi T, Sasaki T, Tokita N, Hirabayaschi K : Internal fixation of the odontoid fracture. Orthop Trans, 1982, 6, 176-182.
22. Pache T : Résultats et complications de 15 vissages de fracture de lodontoïde. Rev Chir Orthop, 1990, suppl I, 99.3, .
23. Ramadier JO, Aleon JF, Servant J. : Les fractures de lapophyse odontoïde. 94 cas dont 61 traités par arthrodèse. Rev Chir Orthop, 1976, 62, 171-189.
24. Roberts A, Wickstrom J. : Prognosis of odontoïd fractures. J Bone Joint Surg, 1972, 54A, 6, 1353.
25. Roy-Camille R, de la Caffinière JY, Saillant G : Les traumatismes du rachis cervical supérieur .Masson, édit, 1973.
26. Roy-Camille R, Saillant G, Lapresle Ph, Léonard P : Fractures récentes de lodontoïde: facteurs de pronostic. Presse Méd, 1983, 12, 36, 2233-2236.
27. Roy-Camille R, Saillant G, Judet Th, de Botton G, Michel G : Eléments de pronostic des fractures de lodontoïde. Rev Chir Orthop, 1980, 66, 183-186.
28. Roy-Camille R, Lapresle Ph, Mazel Ch : Les fractures de lodontoïde. Rachis cervical supérieur. Cinquièmes journées dorthopédie de la Pitié. Masson, 1986, 99-118.
29. Roy-Camille R, Henry P, Doursounian L, Ciniglio M : Anatomie fonctionnelle de la charnière cranio-rachidienne. Rachis cervical supérieur. Cinquièmes journées dorthopédie de la Pitié. Masson, 1986, 10-15.
30. Roy-Camille R, Benazet JP, Saillant G, Judet Th, Billot Ch : Tétraplégie post-traumatique haute et complète suivie de récupération. Réflexion à propos dune fracture de lodontoïde. Rev Chir Orthop, 1981, 67, 685-690.
31. Steimle R, Godard J, Jacquet G, Zaitouni M, Chico F, Orabi M, Farhat O : Fracture récente de lodontoïde. Vissage par voie antérieure à propos de 9 cas. Chirurgie,1988, 114, 585-591.
32. Vichard Ph, Dreyfus-Schmidt G, Zeil A : Réflexion à propos dune série continue de 66 observations de fracture de lapophyse odontoïde de laxis. Lintérêt dune ostéosynthèse stable sans arthrodèse C1-C2 associée. Chirurgie, 1987, 113, 427-432.
33. Vichard Ph, Gagneux E, Garbuio P. : Ostéosynthèse par plaque en trèfle des fractures de lapophyse odontoïde. Rev Chir Orthop, 1996, 82, 663-667.
| Orthopaedic |