Search tips
Search criteria 


Logo of eurspinejspringer.comThis journalThis journalToc AlertsSubmit OnlineOpen Choice
Eur Spine J. 2009 July; 18(7): 978–991.
Published online 2009 February 19. doi:  10.1007/s00586-009-0900-5
PMCID: PMC2899576

C2-fractures: part I. Quantitative morphology of the C2 vertebra is a prerequisite for the radiographic assessment of posttraumatic C2-alignment and the investigation of clinical outcomes


Pertinent literature exists concerning indications, techniques, complications of treatment, and risk factors for nonunion in axis and odontoid fractures; however, there are scarce data regarding the incidence and definition of malunion in these fractures. As a prerequisite for the study of anatomical alignment following surgical and nonsurgical treatment of C2-fractures, an understanding of normal C2 anatomy is essential. Therefore, the authors intended to evaluate morphometrical dimensions of the C2 vertebra. The purpose was to provide normalized quantitative data to enable assessment of malalignment following the treatment of C2-fractures within a classification system. Using digitized cervical spine lateral and transoral odontoid radiographs of 100 consecutive patients without any evidence of traumatic or neoplastic disorders, the authors performed measurements on distinct anatomical structures and investigated morphometrical dimensions of the normal axis vertebra. The incidence of atlantoaxial arthritis was also evaluated. In addition, with the assessment of twenty arbitrarily chosen sets of radiographs by three different observers we calculated the interobserver reliability in terms of intraclass correlation coefficients for each parameter. With calculation of SD and 95% confidence limits, pathological cut-offs were reconstructed from measurements performed resembling non-physiological and pathological limits. Distinct parameters were selected to form a new classification system for radiographical follow-up that focuses on the quantitative C1–2 vertebral alignment. The measurement process resulted in 2,400 data points. Distinct morphometrical parameters, such as a quantitative characterization of the sagittal atlantoaxial congruency, the lateral mass inclination and the type of degenerative changes at the atlantoaxial joint could be demonstrated to be valuable and reliably used within a proposed classification for C2-malunions following C2-fractures. The current study offers a template including recommended radiological measurements for further research on the study of clinical outcome and posttraumatic alignment following C2-fractures.

Keywords: C2-fracture, Posttreatment alignment, Morphology, Anatomy


Controversy remains over the appropriate classification and treatment of C2-fractures [23, 28, 33, 39, 45, 58, 63]. There is yet no single comprehensive treatment-related classification that stratifies C2-fractures according to their propensity to heal with or without surgery and their propensity to heal anatomically or with distortion of atlantoaxial anatomy. Previous studies concentrated on indications, complications, risk factors for and incidence of nonunion as well as the technical and biomechanical advantages of surgeries to stabilize C2-fractures [2, 3, 6, 1921, 23, 24, 29, 33, 39, 45, 49, 55, 58, 63, 64, 67]. Although a number of practice guidelines have been developed [33], outcome measures seldom included patient’s satisfaction or validated outcome measures, the incidence of long-term disability [1, 9, 19, 20, 39, 63, 67] or the quality of anatomical restoration achieved [1, 810, 37]. Following nonsurgical treatment of C2-fractures, the fragments frequently yield fusion in a position of slight to severe malunion, a fact that is rarely recognized and underreported [26, 31, 46, 49]. Accordingly, data on the incidence and definition of quality in achieving and maintaining reduction in C2-fractures are scarce and the literature lacks reports on the impact of the posttraumatic C2-alignment on the clinical outcome and the resulting C1–2 axial rotation [1, 2, 8, 9]. It remains a little-known clinical entity that, with nonsurgical treatment, a considerable number of distinct C2-fracture patterns (including multiple C2-fractures and combined C1–2 fractures, atypical hangman’s fractures, comminuted odontoid type II fractures, coronal and oblique vertebral body and type III odontoid fractures, lateral mass split and burst fractures [18, 20, 29, 3840, 43]) and particularly those involving the superior articular facets of C2 frequently heal with slight to severe distortion of the C2-anatomy resulting in atlantoaxial incongruency [18, 20]. Some of these malunions are symptomatic in regard to cervical pain and limited rotation of C1–2 [31], and can cause painful atlantoaxial osteoarthritis (AAOA) [20, 31].

To assess any distortion of the C2 anatomy and to define a ‘C2-malunion’, one needs data of normals for comparison. Therefore, as a prerequisite for the assessment of the posttreatment anatomical C2-alignment after C2-fractures, the authors intended to investigate the anatomical C2-dimensions and C1–2 relationships in a sample of 100 healthy patients. The current study yields a physiological standard for the main parameters that describe the in vivo C2 anatomy and the incidence of AAOA. In addition, based on this physiological standard our purpose was to describe a classification system that delineates the posttreatment alignment in C2-fractures. The adoption of the proposed classification might allow a more accurate assessment of posttraumatic C2-alignment. It might enable comparative studies including the investigation of C1–2 axial rotation and clinical outcome.

In part II of the current project on the outcome of C2-fractures, the classification of the C2-alignment is applied on clinical cases and the impact of the posttreatment C2-alignment on the remaining C1–2 rotation and clinical outcome is evaluated.

Materials and methods

Our study sample consisted of 100 consecutive patients who had indications for cervical spine radiographic investigation. Only radiographs showing no evidence of cervical fracture, ligamentous instability, rheumatoid arthritis, ankylosing spondylitis, DISH or neoplastic disorder were included. Lateral and transoral radiographs were selected only if there were no characteristics of a misaligned X-ray beam defined by congruency of the atlantoaxial joints on transoral odontoid views and the absence of duplication of the C1 and C2 cortical bounderies on both lateral and transoral views. In total, 54 female and 46 male patients were enrolled for anatomical measurements. Their mean age was 49.8 years (range 15–94), 44.7 years in male and 54.0 years in female patients. The radiographs were taken on a digital X-ray system (Vertix 3D-III unit, Siemens, Germany) with the patient in sitting position and stored digitally (PACS Magic View VC 42, Siemens, Germany). Using the cursor, digital measurements on the radiographs (0.1 mm increments) were performed with a commercial software programm (Escape Medical Viewer V3, Escape, Greece).

The C2 vertebra consists of a body, paired pedicles, lateral masses (superior articulating facets), odontoid, pars interarticularis, inferior articulating facets, lamina, and bifid spinous process. Each structure can be involved in C2-fractures [38, 40, 43], therefore different measurements on transoral and lateral radiographs were chosen to analyse anatomical structures of clinical interest that can be affected. Besides absolute numerical parameters, distinct ratios in the sagittal plane and side-related differences in the coronal plane were calculated from the single parameters measured. The anatomical structures evaluated, dimensions measured and ratios calculated, as well as the descriptions of measuring techniques are summarized in Tables 1, ,22 and Figs. 1, ,2.2. One measuring technique requires further explanation: in clock-wise fashion, the Harris-Ring-C2 [22, 26, 27, 48, 52, 61] resembles the radiological projection of the anterior-superior and superior parts of the C2 pedicle, the posterior vertebral border of C2, the caudad part of the transverse foramen and process, and the anterior border of the C2 superior lateral mass. Radiologically, the Harris-Ring-C2 is elliptical in shape (Fig. 1), but can be recognized in almost any lateral cervical spine radiograph. Therefore, we determined the superior depth of the vertebral body (sVBD) on a line perpendicular to the tangent of the C2 posterior wall at the level of half the diameter of the Harris-Ring-C2. Radiologically, an increased sVBD with widening of the axis body is characterized as the ‘C2 fat sign’ [52, 61]. The superior vertebral body height (sVBH) was measured from the caudad posterior corner of the axis vertebra to the upper cortical merging of the C2 pedicle, resembling the superior bow of the Harris-Ring-C2.

Table 1
Description of anatomical landmarks and measurement techniques used on lateral cervical spine radiographs
Table 2
Description of anatomical landmarks and measuring techniques used on transoral odontoid views
Fig. 1
Artist’s drawings depict landmarks at anatomical structures of interest for making the measurements. Left AP-view of axis vertebra. Right Sagittally sectioned axis vertebra at the level of the right C2 pedicle
Fig. 2
AP and lateral radiographs with drawings depict measurements performed. See Table 1 and 2 for description of abbreviations used. Dashed lines mark the accessory lines resembling the endpoints for measurement on distinct anatomical landmarks

To analyse degenerative changes of the atlantoaxial joints observed in elderly patients or following C2-fractures in younger patients [20, 31], all C1–2 joints were scored on transoral odontoid views according to Lakshamanan [44] (Table 3), and the median height of the C1–2 joints was measured. As the study showed difficulties in differentiating the subtypes 0 and 1, as well as type 2 and 3, statistical calculations were performed to stratify a group A (none or mild degenerative changes C1–2, type 0 and 1 [44]) and a group B (advanced degenerative changes C1–2, type 2 and 3 [44]). Hence, consistent with a previous statistical analysis [44], only moderate and severe changes were considered as resembling degenerated and arthritic joints, respectively. Radiographic examples for each kind of degenerative grades at the C1–2 joints are illustrated in Fig. 3.

Table 3
Grading of the severity of degenerative changes of the left and right lateral atlantoaxial joints
Fig. 3
The applied atlantoaxial osteoarthritis grading system according to Lakshamanan et al. [44] and the modified classification (in brackets). Left bilaterally Grade 0 (Type A); middle bilaterally Grade 1 (Type A); right left atlantoaxial joint shows Grade ...

The first author performed all measurements in 100 sets of radiographs. However, the value of anatomical measurements strongly depends on their reproducibility within the use of different observers. Therefore, reliability was evaluated by interobserver testing between three of the authors who assessed 20 arbitrarily chosen sets of radiographs. Interobserver differences were evaluated using the intraclass correlation coefficient (ICC). An ICC score from 0 to 0.4 was rated poor, 0.4 to 0.75 fair or moderate, and >0.75 excellent [60].

To assess the severity of any pathological alignment following fracture and union of the C2 vertebra, a classification system was constructed based on the anatomical measurements performed. The classification should equilibrate morphometrical changes measured on radiographs and the descriptive assessment of the C1–2 alignment. Measured dimensions and ratios were selected for inclusion into the classification, if standard deviations as well as upper and lower limits of the 95% ranges were shown to be small, and particularly if the ICC was shown to be ‘excellent’ with an ICC > 0.75.

Statistical analysis

In addition to overall and gender-related calculations of means, SD and ranges, upper and lower 95% ranges were calculated. Correlations among variables were analysed using Pearson’s correlation coefficient. Student t-tests were computed to analyse differences among subgroups. ICC with 95% confidence limits were used to evaluate the inter-rater reliability among the three investigators. A p-value less than 5% was considered statistically significant. All analyses were performed with SPSS 11.0 (SPSS Inc, Chicago) and Statistica 6.1 (StatSoft Inc, Tulsa).


Investigation of the in vivo C2 anatomy resulted in 2,400 data points. Means, standard deviations, ranges, and the lower and upper limits of the calculated 95% ranges are summarized in Tables 4 and and55.

Table 4
Results of anatomical measurements on lateral cervical spine radiographs, n = 100 patients (in mm and degrees)
Table 5
Results of anatomical measurements on transoral odontoid views, n = 100 patients (in mm and degrees)

Calculation of ICCs showed that all but 5 out of 22 parameters had an ICC > 0.75, judged as excellent agreement [60]. Moderate reliability was found for the assessments of the odontoid tilt angle in the coronal and sagittal planes (sA1 and sA4, cA8 and cA9).

Our analysis showed that concerning isolated anatomical dimensions wide variations exist within the axis vertebra for 95% of the population. However, we emphasize that calculations for the sagittal ‘ratios’ (sR1 to sR3), as well as for the differences between left and right sides in the coronal plane (Diff1–Diff3), showed small ranges and SDs. Inter-individually large differences were found within the numerical, but only once within the angular measurements with a few statistically significant differences between males and females. However, differences between left and right measurements in the coronal plane (parameter ‘Diff1–Diff3’) calculated for gender were found to be not statistically significant. In sagittal plane the anatomical ratios calculated showed small ranges within the axis and the C1–2 vertebrae (parameter ‘sR1–sR3’). With the latter, slight gender-related differences were statistically non-significant.

Concerning age-related differences the statistical analysis showed a significant correlation between increasing age and the measurement of iVBD (r = 0.26; P = 0.01), sVBD (r = 0.21; P = 0.04), and sR1 (r = 0.44; P < 0.001); see Fig. 4.

Fig. 4
Scatterplot with regression lines depict age-related increase of sR1

Atlantoaxial joint morphology

With the use of a modified classification that characterizes atlantoaxial joints as ‘normal/moderately degenerated’ and ‘advanced degenerated/arthritic’ according to our definition of Type A and B (see Table 3), the incidence of AAOA was 5.0% (n = 5). Three patients had bilateral joints of Type B and 2 had unilateral Type B. As concerns the influence of age on the median height of the atlantoaxial joint space (lAAJH/rAAJH), we observed that patients with Type A changes were significantly younger (left atlantoaxial joints: mean age 48.7; right atlantoaxial joints: 47.9 years) compared to those patients with Type B changes (left atlantoaxial joints: 69.8 years; right atlantoaxial joints: 84.6 years, P = 0.045; see Fig. 5). Statistical analysis revealed that at the left and right atlantoaxial joints, patients with Type A changes showed a mean lAAJH and rAAJH of 3.52 ± 0.7 and 3.63 ± 0.8 mm, respectively. In contrast, patients with Type B changes showed a mean lAAJH of 1.31 ± 0.9 mm and rAAJH of 1.60 ± 0.9 mm. The differences regarding the AAJH between Type A and B were strongly significant (lAAJH: P < 0.0001, rAAJH: P < 0.0001).

Fig. 5
Age-related decrease of rAAJH (left) and lAAJH (right), respectively

For the purpose of forming the classification system of posttreatment C2-alignment, the cut-off for Type A and B changes of the AAJ was arbitrarily set with a joint space height of ≤2.0 mm with the decision being strongly influenced by the differences of AAJH in Type A and B joints. Our evaluation process demonstrated that judging different grades of AAOA with four subtypes according to the classification of Lakshamanan et al. [44] was difficult, both for the differentiation of type 1 and 2 as well as type 3 and 4. To simplify the assessment of AAOA, we applied a modified version that showed an ICC of 1.0 and 0.85 for lAAJH and rAAJH, respectively, denoting excellent agreement. The modified classification might ease the evaluation process of AAOA, making it sufficiently accurate to assess posttraumatic degenerative changes.

Creating the classification system

The clinical relevance and reproducibility of morphological parameters measured as well as their application in a classification in a future clinical study was one of the decisive factors to perform the current study. The use of a classification without evidence of good interobserver reliability can result in inconsistent results [7]. Therefore, dimensions and ratios measured were selected for inclusion into the classification based on their SD and upper and lower limits of the 95% ranges if the ICC was shown to be ‘excellent’. Two parameters showed fair ICCs (sA4 and sA1), but were included as both measurement assess the odontoid inclination in sagittal plane. Drawing tangents along the anterior odontoid cortex was shown to be less reproducible, but significant deviations from the physiological standard that can be detected using this measurement add information to the assessment of C2-alignment following, e.g., an odontoid fracture. The parameters and morphometrical cut-offs recommended for application in future studies addressing posttraumatic C2-alignment are listed in Table 6. The proposed classification system of posttraumatic C2-alignment (CPA-C2) includes an ordinal system to grade the C2 vertebral shape and C1–2 alignment with four types (Table 7). With this classification, nine groups of morphometrical parameters are included and the evidence of any ‘pathological’ value within each group is awarded 1 or 2 points. The total point score is accumulated by grading the overall C2-alignment in the sagittal and coronal planes from a minimum of 0 points, delineating no malalignment, to 5 or more points, delineating severe malalignment. Because C1–2 spinal stenosis, which is described as ‘sR3’, and posttraumatic arthritis C1–2, described as lAAJH/rAAJH and Type B changes are supposed to represent serious sequelae following a C2-fracture, each is assigned 2 points in case of ‘pathological’ values. With this classification, different parameters are grouped, i.e., sA1 and sA4, as both assess the odontoid tilt angle. However, if the inferior endplate of C2 is altered due to a distinct fracture pattern, sA1 cannot be applied because of the distortion of its reference plane (sPEP). The same is true for distinct groups of parameters in coronal plane: loss of height of the lateral mass in the coronal plane due to collapse or fracture can be indicated by changes in several parameters (cA1 and cA2, Diff 2, cA6 and cA7, Diff 1, as well as cA5). But, for example, if cA1 and cA2 are within normal limits and only cA3 or cA4 shows pathological values, there is evidence of an isolated ‘pathological’ odontoid tilt towards the left or right side. Figure 6 illustrates the clinical application of the proposed classification in a case of C2-malunion.

Table 6
Recommended parameters for the radiological assessment of alignment following C2-fractures
Table 7
Classification of alignment following treatment of C2-fractures
Fig. 6
Illustrative case example for the application of the posttreatment C2-alignment, see text for explanation

There was a statistically significant correlation of increasing age and sR1 (r = 0.44; P < 0.001). With the proposed classification, measurements of sR1 adjusted for age according to Fig. 4 might be used to guide accurate assessment of the C2-alignment. For clinical application of the classification, the small differences associated with age-corrected values of sR1, i.e., 0.075 between a 40- and 80-year-old patient, will have marginal impact. Distinct upper and lower cut-offs were therefore selected for the sR1 parameter. Gender was shown to statistically affect the numerical dimensions of some parameters (see Tables 4, ,5),5), however, calculation of the anatomical ‘ratios’ and side-related ‘differences’ indicated no significant dependence upon gender. Therefore, with the classification separation of values according to gender was not performed.

Illustrative case example

The classification system will be explained on a case illustrated in Fig. 6. A 45-year-old patient with an odontoid type III fracture affecting both superior articular facets of C2 was treated by means of a Minerva cast for 3 months. Radiographic follow-up at 25 months depicted osseus malunion in the sagittal plane. The patient showed motion-induced pain and his total ROM for rotation in a flexed had position was 20°. The measurement of parameters and the calculations performed on the lateral radiograph (Fig. 6, left) showed that sR3 (Ratio sDSC1 : sDSC2) was 0.83, sR2 (Ratio iVBD:sVBD) was 0.88, sA1 was 72.9° and sA4 was 6°. Hence, measurements exceeded the physiological 95% ranges of a normal population. In the coronal plane, all measurements and calculations performed were within normal 95% ranges (rAAJH = 3.94 mm and lAAJH = 4.18 mm) with no changes compared to time of injury (cA6 = 40.5°; cA7 38.8°; Diff1 = 1.7°; cA1 = 21.4°; cA2 = 20.9°; Diff2 = 0.5°; cA3 = 108.5°; cA4 = 105.1°; Diff3 = 3.4°). There were no signs of advanced degenerative changes at the atlantoaxial joints at final follow-up compared to time at injury (radiographs not shown). Hence, the C1–2 joints were graded as Type A.

In the sagittal plane, according to our proposed classification system, 2 points are assigned for malalignment in terms of C1–2 subluxation with the atlas protruding anteriorly (sR3 value), 1 point for widening of vertebral body with deformity (sR2 value) and 1 for odontoid malalignment (sA1and sA4 value both assess odontoid tilt angle). In the coronal plane all measurements were within normal limits. The total score summed from single score values was 4, denoting Type 3 ‘moderate malalignment’ of the axis vertebra.


The majority of anatomical in vivo dimensions investigated in the current study have not been studied previously as a whole, any isolated dimensions in ex vivo studies, or used in the comparison of differing measuring techniques [11, 41, 65]. The authors are not aware of any study that included interobserver calculations for the purpose of assessing the reproducibility of the C2 measurements performed.

In an anatomical study of Kandiziora et al. [34], based on 50 axis vertebrae, the radiologic average of posterior body height was 18.2 mm, though this measurement was 24.6 mm in the current study using a different measuring technique. Similarly, the C2 superior body depth was 12.5 mm [34], but was 14.9 mm in our study using the Harris-Ring-C2 as the reference landmark. Monu et al. [47] also measured the odontoid tilt angle with a technique identical to ours in 175 patients with a mean age of 32.2 years. All patients had some degree of posterior tilting of the odontoid with a mean of 17.4° ± 6.1°, which showed a mean of 18.0° ± 5.7° in the current study. In another cadaver-based anatomical study, Xu et al. [70] measured the superior facet angle of C2 formed between the C2 midsagittal line and facet plane. The mean angular measurement was 66.3°/68.4° in males/females. Calculating our data according to their measurement technique gave similar results (65.9° left sides/68.6° right sides). In the study of Kandiziora et al. [34], the angle between the horizontal plane and the superior facet of C2, which was 23.4° in the current study, was 23.9° on average.

Zapletal et al. [71] evaluated the incidence of AAOA radiologically in 355 patients and considered the C1–2 joints degenerative when severe narrowing or obliteration of the joint space, subchondral sclerosis, and/or osteophytes were present. The authors found the incidence of AAOA to be 4.8%. We modified the classification of Lakshamanan et al. [44], applied it to 100 patients, and the incidence of AAOA was found to be 5.0%. In summary, the reviewed data reported in literature were similar to those of the current study, offering additional validation for the measurements we performed. Therefore, the results found in the current measuring process should be reproducible in clinical applications.

In a radiological study, Sgabanti et al. [59] evaluated the mean height of the atlantoaxial joint space in 50 males and 52 females on both sides at the medial, lateral and median levels. No statistically significant differences were found in the mean values when the genders were compared, consistent with the results of the current study. Measurement of the mean width at the median level was 2.4 mm each in males and females [59]. The same averaged measurements in our study for the left and right sides were 3.75 and 3.25 mm, respectively. Sgabanti observed that the articular space width, at the three levels, had a linear relationship with the patients’ age with progressive decrease in elderly patients. The current study confirmed these observations. However, age-related differences in joint space height are not included in the proposed classification because the acquired degenerative changes in C1–2 that appear during the clinical course following C2-fractures are supposed to be the decisive factors in clinical outcome rather than any absolute morphometrical value of joint space height.

Implications behind the assessment of posttreatment C2-alignment

Previously, outcome surveys of C2-fractures rarely included CT-assessment as endpoint anchors and held no information on the incidence and severity of malalignment and its impact on the clinical outcome [19, 20, 31, 39, 49, 58, 63]. In a study of Seybold et al. [58] the long-term outcome in patients treated with the halo-vest or cervical orthosis was not significantly worse than that after surgery, suggesting that other factors than the rate of union achieved might influence the outcome. Correspondingly, concurrent clinical results regarding similar treatment concepts for C2-fractures exist [1, 9, 42, 49, 58, 67]. There is no consens on the appropriate classification and treatment of C2-fractures because many previous studies had heterogenous samples, incomplete follow-up data, absence of validated outcome vehicles, simplified description of fracture morphology and, thus, many different C2-fracture patterns, e.g., those including the C2-facets, a burst lateral mass or a comminuted odontoid base, were stored together in groups such as ‘odontoid II or III fractures’, ‘vertebral body fractures’ and ‘hangman’s fractures’ [31, 3840]. Frequently, apples and pines were compared. Therefore, to allow for a more detailed characterization of how a C2-fracture has healed we developed a physiological standard of the morphometric dimensions that sets cut-offs for pathological limits. So, this study offers a template to assess the C2 and C1–2 alignment after fractures of the axis vertebra. Although the measurement and classification process for a single, fused C2-fracture takes less time than that of a thoracolumbar scoliosis case using the Lenke classification, the clinical application of the current classification might seem strenious. However, evaluation of the anatomical alignment following the treatment of C2-fractures will be one of the decisive factors when assessing clinical outcome and atlantoaxial rotation in comparative studies. Why? Because literature serves evidence that malaligned C2-anatomy can alter the complex C1–2 motional characteristics and damage to the articular surfaces of C2 can confer cervicocephalic pain and cause painful AAOA: The C1–2 joints and vertebrae are responsible for a complex kinematic arch of rotational motions [15, 50, 51] and contribute approximately 60% of the total neck rotation [50] that is rotation of 30°–43° to each side [15, 53, 56]. Accordingly, this was used as a strong argument in favour of motion-preserving surgical techniques, such as anterior odontoid screw fixation (AOSF) [2, 12, 19, 25, 28, 45, 54, 63]. In contrast, Jeanneret et al. [31], evaluating the postoperative C1–2 rotation following AOSF using functional CT-examination, observed a mean ROM to each side of about 25°, only 38% of patients having normal [16] C1–2 rotation. The authors emphasized that distinct fracture patterns such as an intraarticular fracture or arthritic changes at C1–2 contributed to decreased C1–2 rotation. Notably, the authors could document that intraarticular comminution of the C1–2 joints can produce painful posttraumatic malunion and AAOA. As concerns the decreased rotational ability of the C1–2 joints following non-surgical and surgical treatment of C2-fractures, several authors shared the observations of Jeanneret et al. [4, 49, 62, 67]. Müller et al. [49] reporting on 27 odontoid fractures undergoing AOSF, observed motion related pain in 29%. Forty-one percent of patients demanded pain medication. Radiographic follow-up indicated that there was no anatomic realignment in 18%. Seven patients had significant loss of ROM whereas one patient showed advanced AAOA. The authors supposed that ‘as it is with long bone surgery’ anatomical realignment should be a primary goal in upper cervical spine surgery, but it was presumed difficult in axis fractures. Tuite et al. [66] reported that almost 50% of hangman’s fractures (Hmfx) showed substantial sequelae following surgical stabilization by means of ACDF C2-3 if osseus union had occured in a position of subluxation at C2-3. As early as 1993, Heller et al. [29] reported the surgical treatment of a malunited odontoid fracture with fixed atlantoaxial subluxation requiring odontoid resection and posterior C1–2 fusion. The authors observed that odontoid malunion had received little attention in previous series, either not having occurred in the past, or to have been unrecognized. Indeed, most malunions of the axis are neglected. In a current literature review [33], the incidence of malunion reported was low in type II and III odontoid fractures following halo treatment (5 and 10%, respectively). However, with the latter the articular facets of C2 are frequently affected and symptomatic malunion or segmental C1–2 deformation after nonsurgical treatment has been observed several times [5, 10, 13, 14, 20, 24, 26, 29, 33, 35, 36, 42, 55, 57]. Clark [10] already stretched that type III odontoid fractures were ‘not benign as previously reported’. Ferrer emphasized that type III lesions involvement of the vertebral body or collapse of the lateral masses of C2 are prone to a high rate of non-anatomical osseous union [18], that is a malunion. Several reports confirmed that C2-fractures affecting the upper facets can show symptomatic atlantoaxial incongruency at follow-up [20, 32, 43] and there is evidence of AAOA following cervical trauma [17, 30, 31]. AAOA can be a distracting source of pain [17, 25, 30], and one of the sequelae following C1–2 fractures [20, 26, 31]. In a normal healthy population, AAOA was shown to be as high as 4.8% [71]. The incidence of AAOA was 5.0% in the current study in a group of healthy patients with a mean age of 49.8 years. However, the incidence of AAOA related to C2-fractures is not known. AAOA following C2-fractures remains a neglected entity [25] and demands further investigation.

A meaningful number of distinct subtypes of C2-fractures exists, in which outcome is not favourable [1, 8, 9, 29, 37, 62, 69]. However, differences in functional and clinical outcomes related to the posttreatment C2-alignment have not been studied. In this context, it is noteworthy that some of the painful C2-malunions indicate secondary anterior or posterior C1–2 fusion [20, 41, 62, 68]. Malunion in C2-fractures can produce significant sequelae and those subtypes of C2-fractures that are prone to malunion with nonsurgical therapy or motion-preserving surgery should be defined and a further stratification performed. With the current classification, the posttreatment C2-alignement can be assessed and related to the fracture morphology, the treatment applied, the C1–2 rotation and clinical outcome at follow-up. The main goal of the proposed classification system is to provide taxonomy, a tool for the assessment of malalignment following C2-fractures, and to enable comparison of like cases in follow-up studies. Further ongoing work includes its application to clinical cases. Refining the system by assessing the impact of various treatments on the clinical and morphometric outcome will be required for complete validation of the classification as a guidance used for further stratification of distinct C2-fractures that are more or less prone to clinically symptomatic malunion.

The brief review of literature shows that various C2-fracture characteristics can affect outcome. With application of the proposed classification in clinical outcome studies, guidance on when to do a C1–2 fusion can be substantiated: If there is a predictive malunion in some distinct C2-fracture pattern following the use of current motion preserving techniques or conservative treatment, then primary fusion C1–2 might be considered a sound treatment.

The authors formed an anatomical template to assess the amount of anatomical distortion of the C2 vertebra and C1–2 alignment using reproducible parameters. Whether there will be distinct relationships between degrees of malunion, a decreased C1–2 rotation and worse clinical outcome has yet to be answered and will be discussed in part II of this study project.


In the current study the authors used plain radiographs instead of CT-scans to investigate the C2 anatomy because they represent the most commonly used diagnostic tool for radiological follow-up in C2-fractures. In a study of Kandiziora et al. [34], only slight differences were observed between measurements performed using direct caliper technique, radiographs and CT-scans. To simplify the application of our radiographic results to further studies, including CT- and MRI-scans, we calculated anatomical ratios and side-related differences scaled in ‘millimeters’. With the combination of angular measurements performed, the recommended parameters included in the classification should be reproducible with CT-scans. Nonetheless, research on the reliability of parameters measured on radiographs compared to CT-scans is to be performed.

Although interobserver calculations were performed, we did not test consistency over time and intraobserver reliability. However, at first inspection, interobserver reliability was shown to be excellent for most parameters. All but two of the parameters that were included for construction of the classification demonstrated excellent reliability, with ICCs > 0.75.

All patients included in the current study were of European origin. Because differences in patients from Asian origin might be expected, the author cautions against applying our data indiscriminately in these patients.


1. Aebi M, Etter C, Coscia M, et al. Fractures of the odontoid process: treatment with anterior screw fixation. Spine. 1989;14:1066–1070. doi: 10.1097/00007632-198910000-00007. [PubMed] [Cross Ref]
2. Apfelbaum RI, Lonser RR, Veres R, Casez A. Direct anterior screw fixation for recent and remote odontoid fractures. J Neurosurg (Spine 2) 2000;93:227–236. doi: 10.3171/spi.2000.93.2.0227. [PubMed] [Cross Ref]
3. Bading S, Pape HC, Lange U, Bastian L, Krettek C. Reosteosynthesis in dens pseudoarthrosis. Unfallch. 2004;107:149–153. doi: 10.1007/s00113-003-0701-y. [PubMed] [Cross Ref]
4. Blauth M, Richter M, Lange U. Transarticular screw fixation C1/C2 in traumatic atlantoaxial instabilities. Comparison between percutaneous and open procedures. Orthopade. 1999;28:651–661. [PubMed]
5. Bohay D, Gosselin RA, Contreras D. The vertical axis fracture: a report on three cases. J Orthop Trauma. 1992;6:416–419. doi: 10.1097/00005131-199212000-00004. [PubMed] [Cross Ref]
6. Böhler JB, Poigenfürst J, Gaudernak W, Hintringer T. Odontoid screw osteosynthesis. Oper Orthop Traumatol. 1990;2:75–83. doi: 10.1007/BF02511274. [Cross Ref]
7. Burstein AH. Fracture classification systems: do they work and are they useful? J Bone Joint Surg Am. 1993;75:1743–1744. [PubMed]
8. Chang K, Liu Y, Cheng P. One Herbert double-threaded compression screw fixation of displaced type II odontoid fractures. J Spinal Disord. 1994;71:62–69. doi: 10.1097/00002517-199407010-00009. [PubMed] [Cross Ref]
9. Chiba K, Fujimura Y, Toyama Y, Takahata T, Nakanishi T, Hirabayashi K. Anterior screw fixation for odontoid fracture: clinical results in 45 cases. Eur Spine J. 1993;2:76–81. doi: 10.1007/BF00302707. [PubMed] [Cross Ref]
10. Clark CR, White AA. Fractures of the dens: a multicenter study. J Bone Joint Surg. 1985;67-A:1340–1348. [PubMed]
11. Cokluk C, Aydn K, Rakunt C, Iygiun O, Onder A. The borders of the odontoid process of C2 in adults and in children including the estimation of odontoid/body ratio. Eur Spine J. 2006;15:278–282. doi: 10.1007/s00586-005-0946-y. [PMC free article] [PubMed] [Cross Ref]
12. Cokluk C, Takazasu M, Zoshida J. Pedicle fracture of the axis: report of two cases and review of literature. Clin Neurol Neurosurg. 2005;107:136–139. doi: 10.1016/j.clineuro.2004.06.001. [PubMed] [Cross Ref]
13. Cusick JF, Yoganandan N. Biomechanics of the cervical spine 4: major injuries. Clin Biomech (Bristol, Avon) 2002;17:1–20. doi: 10.1016/S0268-0033(01)00101-2. [PubMed] [Cross Ref]
14. Dai LY, Yuan W, Ni N, Liu HK, Jia LS, Zhao DL, Xu YK. Surgical treatment of non-united fractures of the odontoid process, with special reference to occipitocervical fusion for unreducible atlantoaxial subluxation or instability. Eur Spine J. 2000;9:118–122. doi: 10.1007/s005860050221. [PubMed] [Cross Ref]
15. Dumas JL, Thoreux P, Attali P, Goldlust D, Chevrel JP. Three-dimensional CT analysis of atlantoaxial rotation: results in the normal subject. Surg Radiol Anat. 1994;16:199–204. doi: 10.1007/BF01627595. [PubMed] [Cross Ref]
16. Dvorak J, Penning L, Hayek J, Panjabi MM, Grob D, Zehnder R. Functional diagnostics of the cervical spine using computer tomography. Neurorad. 1988;30:132–137. doi: 10.1007/BF00395614. [PubMed] [Cross Ref]
17. Ehni G, Benner N. Occipital neuralgia and the C1–2 arthrosis syndrome. J Neurosurg. 1984;61:961–965. doi: 10.3171/jns.1984.61.5.0961. [PubMed] [Cross Ref]
18. Ferrer S. Letter to the editor. Spine. 2001;26:158–159.
19. Fountas KN, Kapsalaki EZ, Karampelas I, Feltes CH, Dimopoulos VG, Machinis TG, Nikolakakos LG, Boev AN, Choudhri H, Smisson HF, Robinson JS., Jr Results of long-term follow-up in patients undergoing anterior odontoid screw fixation for type II and rostral type III odontoid fractures. Spine. 2005;30:661–669. doi: 10.1097/01.brs.0000155415.89974.d3. [PubMed] [Cross Ref]
20. Fujimura Y, Nishi Y, Koyabashi K. Classification and treatment of axis body fractures. J Orthop Trauma. 1996;10:536–540. doi: 10.1097/00005131-199611000-00005. [PubMed] [Cross Ref]
21. German JW, Hart BL, Benzel EC. Nonoperative management of vertical C2 body fractures. Neursurg. 2005;56:516–520. doi: 10.1227/01.NEU.0000153908.53579.E4. [PubMed] [Cross Ref]
22. Geusens E, Pans S, Brys P, Desruelles D. The axis ring: a forgotten semiologic sign in the detection of low odontoid fractures. JBR-BTR. 2002;85:241–245. [PubMed]
23. Grauer JN, Shafi B, Hilibrand AS, Harrop JS, Kwon BK, Beiner JM, Albert TJ, Fehlings MG, Vaccaro AR. Proposal of a modified, treatment-oriented classification of odontoid fractures. Spine J. 2005;5:123–129. doi: 10.1016/j.spinee.2004.09.014. [PubMed] [Cross Ref]
24. Greene KA, Dickmann CA, Marciano FF, Drabier JB, Hadley MN, Sonntag VK. Acute axis fractures: analysis of management and outcome in 340 consequitive cases. Spine. 1997;22:1843–1852. doi: 10.1097/00007632-199708150-00009. [PubMed] [Cross Ref]
25. Grob D, Bremerich FH, Dvorak J, Mannion AF. Transarticular screw fixation for osteoarthritis of the atlantoaxial segment. Eur Spine J. 2006;15:283–291. doi: 10.1007/s00586-005-0963-x. [PMC free article] [PubMed] [Cross Ref]
26. Hähnle U, Wiesniewski TF, Craig JB. Shear fracture through the body of the axis vertebra. Spine. 1999;24:2278–2281. doi: 10.1097/00007632-199911010-00018. [PubMed] [Cross Ref]
27. Harris JH, Burke JT. Low (type III) odontoid fracture: a new radiographic sign. Radiology. 1984;153:353–356. [PubMed]
28. Harrop JS, Przybylski GJ, Vaccaor AR, Yalamanchili K. Efficacy of anterior odontoid screw fixation in elderly patients with type II odontoid fractures. Neurosurg Focus. 2000;8:e6. [PubMed]
29. Heller JG, Levz MJ, Barrow DL. Odontoid fracture malunion with fixed atlantoaxial subluxation. Spine. 1993;18:311–314. doi: 10.1097/00007632-199302000-00024. [PubMed] [Cross Ref]
30. Holly LT, Batzdorf U, Foley KT. Treatment of severe retromastoid pain secondary to C1–2 arthrosis by using cervical fusion. J Neurosurg (Spine 2) 2000;92:162–168. doi: 10.3171/spi.2000.92.2.0162. [PubMed] [Cross Ref]
31. Jeanneret B, Vernet O, Frei S, Magerl F. Atlantoaxial mobility after screw fixation of the odontoid: a computer tomographic study. J Spinal Disord Tech. 1991;4:203–211. doi: 10.1097/00002517-199106000-00011. [PubMed] [Cross Ref]
32. Jeanneret B, Magerl M. Klinische Analyse der Spätresultate nach Hangman-Frakturen. Z Orthop. 1985;123:721.
33. Julien TD, Frankel B, Traynelis VC, Ryken TC. Evidence based analysis of odontoid fracture management. Neurosurg Focus. 2000;8(Article 6):1–6. doi: 10.3171/foc.2000.8.6.2. [PubMed] [Cross Ref]
34. Kandiziora F, Schulze-Stahl N, Kodadayan-Klostermann C, Schröder R, Mittelmeier T. Screw placement in transoral atlantoaxal plate systems: an anatomical study. J Neurosurg (Spine 1) 2001;95:80–87. doi: 10.3171/spi.2001.95.1.0080. [PubMed] [Cross Ref]
35. Kerschbaumer F, Rittmeister M, Ewald W, Kandiziora F. Atlantoaxial kyphosis. Unfallch. 2001;30:919–924. [PubMed]
36. Khodadayan-Klostermann C, Kandiziora F, Schnake KJ, Hass NP, Harms J. Transoral atlanto-axial plate fixation in the treatment of malunited dens fracture and secondary atlanto-axial instability. Chirurg. 2001;72:1298–1302. doi: 10.1007/s001040170035. [PubMed] [Cross Ref]
37. Koivikko MP, Kiuru MJ, Koskinen SK, Myllynen P, Santavirta S, Kivisaari L. Factors associated with nonunion in conservatively treated type-II fractures of the odontoid process. J Bone Joint Surg. 2004;86-B:1146–1151. doi: 10.1302/0301-620X.86B8.14839. [PubMed] [Cross Ref]
38. Koller H, Assuncao A, Kammermeier V, Holz U. Simulateous anterior arthrodesis C2–3 and anterior odontoid screw fixation for stabilization of a 4-part fracture of the axis—a technical description. J Spinal Disord Tech. 2006;19:362–367. doi: 10.1097/01.bsd.0000204502.99471.9a. [PubMed] [Cross Ref]
39. Koller H, Daniaux H, Gruber H, Blauth M, Kathrein A. The unstable traumatic spondylolisthesis C2–3. Akt Traumatol. 2006;35:183–202. doi: 10.1055/s-2005-865960. [Cross Ref]
40. Koller H, Kammermeier V, Assuncao A, Holz U. Spinal stenosis C1–2 following redo surgery for failed odontoid screw fixation—scrutinizing the odontoid fracture classification. Internet J Spine Surg. 2007;3:1–11.
41. Koller H, Kammermeier V, Ulbricht D, Assuncao A, Karolus S, Berg B, Holz U. Anterior retropharyngeal fixation C1–2 for stabilization of atlantoaxial instabilities: study of feasibility, technical descriotion and preliminary results. Eur Spine J. 2006;15:1326–1338. doi: 10.1007/s00586-006-0103-2. [PMC free article] [PubMed] [Cross Ref]
42. Komadina R, Brilej D, Kosanovic M, Vlaovic M. Halo Jacket in odontoid fractures type II and III. Arch Orthop Trauma Surg. 2003;123:64–67. [PubMed]
43. Korres DS, Papagelopoulos PJ, Mavrogenis AF, Sapkas GS, Patsinevelos A, Kyriazopoulos P, Evangelopoulos D. Multiple fractures of the axis. Orthopedics. 2004;27:1096–1099. [PubMed]
44. Lakshmanan P, Jones A, Howes J, Lyons K. CT evalutation of the pattern of odontoid fractures in the elderly-relationship to upper cervical spine osteoarthrosis. Eur Spine J. 2005;14:78–83. doi: 10.1007/s00586-004-0743-z. [PMC free article] [PubMed] [Cross Ref]
45. Maak TG, Grauer JN. The contemporary treatment of odontoid injuries. Spine. 2006;31:S53–S60. doi: 10.1097/01.brs.0000217941.55817.52. [PubMed] [Cross Ref]
46. Mirvis SE, Young JWR, Lim C, Greenberg J. Hangman’s fracture: radiologic asessment in 27 cases. Radiology. 1987;163:713–717. [PubMed]
47. Monu J, Bohrer SP, Howard G. Some upper cervical spine norms. Spine. 1986;12:515–519. doi: 10.1097/00007632-198707000-00001. [PubMed] [Cross Ref]
48. Mortelsmann LJ, Geusens EAM, et al. Harris or axis ring: an aid in diagnosing low (type 3) odontoid fractures. Eur J Surg. 1999;165:1138–1141. doi: 10.1080/110241599750007658. [PubMed] [Cross Ref]
49. Müller EJ, Wick M, Russe OJ, Palta M, Muhr G. Anterior screw fixation for odontoid fractures (in German) Unfallch. 2000;103:38–43. doi: 10.1007/s001130050006. [PubMed] [Cross Ref]
50. Pang D, Li Veetai. Atlantoaxial rotatory fixation: Part 1—Biomechanics of normal rotation at the atlantoaxial joint in children. Neurosurgery. 2004;55:614–625. doi: 10.1227/01.NEU.0000134386.31806.A6. [PubMed] [Cross Ref]
51. Pang D, Li V. Atlantoaxial rotatory fixation: Part 2—New diagnostic paradigm and a new classification based on motion analysis using computed tomographic imaging. Neurosurgery. 2005;57:941–953. doi: 10.1227/01.NEU.0000181309.13211.3A. [PubMed] [Cross Ref]
52. Pellei DD. The fat C2 sign. Radiology. 2000;217:359–360. [PubMed]
53. Penning L, Wilmink JT. Rotation of the cervical spine. A CT study in normal subjects. Spine. 1987;12:732–738. doi: 10.1097/00007632-198710000-00003. [PubMed] [Cross Ref]
54. Platzer P, Thalhammer G, Ostermann R, Wieland T, Vecsei V, Gaebler C. Anterior screw fixation of odontoid fractures comparing younger and elderly patients. Spine. 2007;32:1714–1720. doi: 10.1097/BRS.0b013e3180dc9758. [PubMed] [Cross Ref]
55. Polin RS, Szabo T, Bogaev CA, Replogle RE, Jane JA. Nonoperative management of type II and III odontoid fractures: the Philadelphia collar versus the halo vest. Neurosurgery. 1996;38:450–456. doi: 10.1097/00006123-199603000-00006. [PubMed] [Cross Ref]
56. Roche CJ, King SJ, Dangerfield PH, Carty HM. The atlanto-axial joint: physiological range mof rotation on MRI and CT. Clin Radiol. 2002;57:103–108. doi: 10.1053/crad.2001.0703. [PubMed] [Cross Ref]
57. Rudzki JR, Lenke LG, Blanke K, Riew KD. Pseudoarthrosis of a thirty-nine-year old dens fracture causing myelopahty—a case report. J Bone Joint Surg. 2004;86-A:416–421. [PubMed]
58. Seybold EA, Bayley JC. Functional outcome of surgically and conservativly managed dens fractures. Spine. 1998;23:1837–1846. doi: 10.1097/00007632-199809010-00006. [PubMed] [Cross Ref]
59. Sgambati E, Capaccioli L, Biagiotti R, Giurovich E, Montigiani L, Brizzi E. Anatomical-radiological and morphometric analysis of the lateral atlantoaxial joint. Ital J Anat Embryol. 1998;103:117–123. [PubMed]
60. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420–428. doi: 10.1037/0033-2909.86.2.420. [PubMed] [Cross Ref]
61. Smoker WR, Dolan KD. The “fat” C2: a sign of fracture. Am J Radiol. 1987;148:609–614. [PubMed]
62. Stulik J, Vyskocil T, Sebesta P, Kryl J (2006) Atlantoaxial fixation using the polyaxial screw-rod system. Eur Spine J 16:479–484 (Epub) [PMC free article] [PubMed]
63. Subach BR, Morone MA, Haid RW, McLaughlin MR, Rodts GR, Comey CH. Management of acute odontoid fractures with single-screw anterior fixation. Neurosurgery. 1999;45:812–820. doi: 10.1097/00006123-199910000-00015. [PubMed] [Cross Ref]
64. Surgeons American Association of Neurological Isolated fractures of the axis in adults. Neurosurgery. 2002;50(Suppl 3):140–147.
65. Tubbs RS, Wellons JS, Blount JP, Grabb PA, Oakes WJ. Inclination of the odontoid process in the pediatric Chiari I malformation. J Neurosurg (Spine 1) 2003;98:43–49. doi: 10.3171/spi.2003.98.1.0043. [PubMed] [Cross Ref]
66. Tuite GF, Papadopoulos SM, Sonntag VK. Caspar plate fixation for treatment of complex hangman’s fracture. Neurosurgery. 1992;30:761–764. doi: 10.1097/00006123-199205000-00019. [PubMed] [Cross Ref]
67. Verheggen R, Jansen J. Fractures of the odontoid process: analysis of the functional results after surgery. Eur Spine J. 1994;3:146–150. doi: 10.1007/BF02190576. [PubMed] [Cross Ref]
68. Wang C, Zan M, Yhou HT, Wang SL, Dang GT. Open reduction of irreducible atlantoaxial dislocation by transoral anterior atlantoaxial release and posterior internal fixation. Spine. 2006;31:E30–E313. [PubMed]
69. Whitesides TE, McDonald AP. Lateral retroparyngeal approach to the upper cervical spine. Orthop Clin North Am. 1978;9:1115–1127. [PubMed]
70. Xu R, Nadaud MC, Ebraheim NA, Yeasting RA. Morphology of the second cervical vertebra and the posterior projection of the C2 pedicle axis. Spine. 1995;20:259–263. [PubMed]
71. Zapletal J, Valois JC. Radiographic prevalance of advanced lateral C1–C2 osteoarthritis. Spine. 1997;22:2511–2513. doi: 10.1097/00007632-199711010-00009. [PubMed] [Cross Ref]

Articles from European Spine Journal are provided here courtesy of Springer-Verlag