Interobserver reliability of the Gehweiler classification and treatment strategies of isolated atlas fractures: an internet-based multicenter survey among spine surgeons

Isolated atlas (C1) fractures account for 2.0‒13.0% of cervical spine injuries and 1.3‒2.0% of all spinal injuries [1, 2]. Individuals in their mid-20s and over 80 years of age are particularly at risk of sustaining C1-fractures, and the mean age of diagnosis in case of trauma is 64 years [3]. While low-energy trauma causes about 50.0% of all cervical fractures [4], atlas fractures are typically seen after an impact to the vertex of the skull [5]. Therefore, most of the C1-fractures occur during falls, motor vehicle accidents, and when diving into shallow water [6]. The craniovertebral junction is especially vulnerable to spinal instability, because the majority of the cervical spine’s axial rotation occurs at the atlantoaxial complex [7], and flexion and extension are greatest at the occiput-atlas joint [8].

In Europe, the Gehweiler classification [9] is widespread and commonly used to grade atlas fractures [10]. Despite the paucity of literature regarding its reliability, it has formed the basis for recent treatment algorithms for the management of atlas fractures [10, 11]. Furthermore, although the incidence of atlas fractures has nearly doubled during the last two decennia [2], there is only a small amount of evidence regarding the most appropriate treatment for such injuries [12]. This may be since upper cervical spine injuries depend greatly on the conjuncture of concomitant bony and ligamentous injuries [13]. Therefore, assessing fracture stability and choosing the most appropriate treatment is complex. An unstable atlas fracture has been defined as a Jefferson fracture with a lesion of the transverse atlantal ligament (TAL) with an incongruence of the atlanto-occipital joint that results in atlanto-occipital instability or displaced atlantoaxial joint facets [3]. It is widely accepted that surgical treatment is indicated for these unstable and displaced atlas fractures [14]. However, there is a paucity of internationally accepted diagnostic and treatment algorithms [12]. In addition, to the best of our knowledge, no studies have systematically investigated the reliability of the Gehweiler classification, which is fundamental to establish a classification-based treatment algorithm.

Based on ratings of experienced spine surgeons, the aims of this study were to evaluate (1) the interobserver reliability of the Gehweiler classification and (2) the C1-fracture stability as well as (3) to assess treatment strategies according to their prior fracture classification.

This internet-based, multicenter survey was performed by presenting case-specific information and radiographic data regarding isolated atlas fractures to spine surgeons affiliated with various Spine Centers throughout Germany. For this purpose, a website was created to make the clinical and radiological data, derived from two centers, available to the spine surgeons. The treatment choices of each participating surgeon were documented using an anonymised questionnaire. The study was approved by the local Ethics Committee (EK 221/18).

Cases rated by the six board-certified spine surgeons had a mean age of 64.0 years (± 17.7), 52.9% were men, and none had pathologic neurological findings according to the American Spinal Cord Injury Association Impairment Scale (ASIA E). Most often (64.8%) the origin of the injury was a fall, either from the same (32.4%) or an elevated (32.4%) level (Table 1). In total, 204 case-specific ratings of individual atlas fractures according to the Gehweiler classification (Table 2) and consecutive treatment strategies were obtained.

Reliable fracture classification systems are crucial, as they form the basis for treatment algorithms and help clinicians understand and communicate injury pathomechanism. Our main findings were as follows:

Previous studies on the reliability of spine fracture classification systems have reported inferior interobserver results for the Thoracolumbar Injury Classification and Severity Scale (κ = 0.23–0.29) [18, 19]. However, studies have shown slightly superior interobserver reliability for the AOSpine Thoracolumbar Spine Injury Classification System (κ = 0.55–0.59) [18, 20, 21], and the AOSpine Subaxial Cervical Injury Classification (κ = 0.56) [22]. In a small sub-sample of a study investigating the interobserver reliability of the Swedish Fracture Register for Vertebral Fractures, atlas fractures (N = 8) were rated according to the Jackson classification [23] by four orthopaedic surgeons and one student. Substantial interobserver reliability was reported (κ = 0.785) [4]. However, the interpretation and generalizability of findings from this study were limited due to the small number and homogenous morphology of atlas fractures, which consisted of two posterior arch fractures, five burst fractures, and one lateral mass fracture. Therefore, by applying the Gehweiler classification, we demonstrate, for the first time, a statistically valid interobserver reliability for the classification of isolated atlas fractures.

Characteristics of the 34 cases in the present study appropriately reflected the gender distribution, mean age, and injury mechanism of patients suffering from isolated atlas fractures [2, 6] as well as the fact that neurological deficits are very rarely observed [24, 25]. Pathological neurological findings were not observed in any cases, so in line with the literature, surgical intervention with early decompression was not indicated, and non-emergency surgery was justified [26]. However, moderate interobserver reliability for the evaluation of fracture stability and the need for surgical or non-surgical treatment, which was assessed independent of the Gehweiler rating, suggested a relevant discrepancy in the assessment of fracture stability. Among the observers, there was a strong consensus on treatment strategy for Gehweiler type 1, 2, 3a, 4, and 5 fractures; however, a lack of consensus was observed for Gehweiler type 3b fractures. These challenges were in line with the reported highly demanding clinical and biomechanical assessment of isolated C1 fracture stability [27, 28].

Except for one case, Gehweiler type 1, 2, and 5 fractures were rated as stable, and non-surgical treatment with a cervical collar was recommended for these cases. Using this conservative treatment as the method of choice was in line with recommendations for the treatment of atlas fractures of the spine section of the German Society for Orthopedics and Trauma [10]. In line with recommendations for the diagnostic testing of Gehweiler type 5 atlas fractures, additional CT-A was required due to their anatomical vicinity to the vertebral artery [11]. Interestingly, although the vast majority of the observers rated Gehweiler type 1 and 2 fractures as stable and recommended non-surgical therapy, they recommended additional radiographic diagnostics (CT-A and MRI) in half the cases. For economic reasons and in view of the limited availability of radiological resources, in the opinion of the authors, the indication for additional imaging in Gehweiler type 1 and 2 fractures should be critically questioned.

In cases rated as Gehweiler type 3a fractures, 35.0% of the observers could not assess stability and the lowest interobserver reliability of all types of atlas fractures is observed. However, non-surgical treatment was chosen in all cases except one. An intact TAL is part of the definition of a Gehweiler type 3a fracture [14]. Due to the stability inherent in cases with an intact TAL, non-surgical treatment of this type of fracture is widely accepted. Nevertheless, careful follow-up is necessary to identify further dislocation, non-union, and/or signals of atlantoaxial instability as early as possible [29]. In 75.0% of cases rated as Gehweiler type 3a fractures, additional MRI was ordered, reflecting the need for practitioners to visualise this type of fracture using imaging that is capable of analysing the ligamentous status of the craniocervical junction and to rule out hidden type 3b fractures after an upper cervical spine injury [30].

For Gehweiler type 3b fractures, additional CT-A was recommended in more than two-thirds of cases and is most likely to be interpreted in the context of preoperative preparation. Due to the heterogeneous shape of the C1 posterior arch and the vertebral artery, preoperative CT-A is crucial to identify anatomical variations, such as an anomalous vertebral artery, and reduce the risk of intraoperative vertebral artery injury [31] and difficulties with screw placement [32]. Gehweiler type 3b fractures were rated unstable in 86.7% of cases, and surgery was recommended in 90%. Isolated C1 fixation using posterior stabilisation was most frequently recommended. In comparison to atlantoaxial or occipito-cervical fusion, which was the treatment methods of choice for unstable atlas fractures for many decades, atlas osteosynthesis provides inherent stabilisation limited to the atlas, which may preserve motion [29]. Consequently, isolated atlas osteosynthesis may be associated with shorter surgery durations, lessened surgical trauma, and fewer complications.

In cases in which the observers assumed an intraligamentous lesion (Dickman type 1) or when the status of the TAL was unclear, additional MRI was requested. In addition, in these cases, the observers suggested the fixation of multiple segments (for example, C1–C2 or C0–C2) rather than an isolated C1 posterior fixation more often than in the surgical treatment of type 3b fractures, which had bony avulsions of the TAL. This was in line with the assumptions that an intraligamentous lesion of the TAL results in translational atlantoaxial instability and that these Dickman type 1 lesions require C1‒2 fusion [15]. However, in line with the fact that this paradigm has been increasingly questioned, nearly half of our cases of Dickman type 1 rupture and those with unclear TAL status were treated using isolated atlas osteosynthesis. The rationale for this surgical treatment may be related to the fact that isolated osteosynthesis of the atlas in cases of violated TAL integrity was not associated with atlantoaxial stability on flexion/extension films in Dickman type 1 and type 2 injuries. This was reported by Shatsky et al. [28] in their retrospective case-controlled study (N = 12). Historically, the necessity of C1‒2 fixation for intraligamentous TAL lesions originated from ligamentous translational atlantoaxial instability of affected TALs in patients suffering from rheumatoid arthritis who had rheumatoid processes involving multiple ligaments of the upper cervical spine. Isolated atlas fractures result from traumatic compression injuries and, therefore, are commonly associated with intact secondary stabilisers, including further ligaments of the upper cervical spine. In addition, Koller et al. [27] reported in their biomechanical investigation of displaced Jefferson burst fractures appropriate C1‒2 stability by conducting an isolated atlas osteosynthesis.

Nearly two-thirds of Gehweiler type 4 fractures were rated stable, and surgery was favoured in one-quarter of the cases, while occipito-cervical instrumentation was suggested for all cases except two. Thus, surgical treatment was recommended for Gehweiler type 4 fractures rated as unstable. The instability of type 4 fractures may occur as a result of a fractured lateral mass that is primarily or secondarily significantly dislocated or in cases of a sagittal split fracture of the lateral mass [33]. In those cases, the treatment guidelines of the German Society for Orthopaedics and Trauma recommend halo traction for 6‒12 weeks or temporary occipito-cervical stabilisation without fusion to avoid the development of persistent pain caused by post-traumatic arthritis [10]. However, especially in the elderly, problems caused by the invasiveness of this conservative form of halo traction might outweigh potential surgical complications [34, 35]. During upper cervical trauma, many vascular structures are at risk, especially the vertebral artery, which is also endangered in Gehweiler type 4 lesions when the foramen of the transverse process is involved. Therefore, in line with the recommendation for additional CT-A to exclude lesions of the vertebral artery [36], the observers in this study called for CT-A in 68.2% of Gehweiler type 4 fractures. Blunt injury to the vertebral artery may result in a stroke, a loss of posterior circulation to the brain, and neurologic sequelae that have potentially devastating consequences [37].

Postoperative cervical immobilisation using a soft or hard collar was recommended in most cases, and the rationale for this was most likely the well-known reduction in pain and spinal stability it provides [38, 39]. However, in particular the indication for postoperative immobilisation in cases of surgical treatment by occipito-cervical instrumentations should be critically questioned due to the appropriate stabilisation and minor residual range of motion that is inherent with this surgery. Therefore, in cases of occipito-cervical instrumentations, the risk of complications associated with postoperative immobilisation may outweigh the benefit of such an immobilisation.

The present study had some limitations. First, it can be argued that a sample size of 34 isolated atlas fractures is too small to provide an appropriate understanding of interobserver reliability. This could be the focus for future studies. However, finding a larger number of isolated atlas fractures would be difficult, and the requirement to evaluate a larger sample size might negatively influence the response rate of experienced spine surgeons. Additionally, the small number of the observers may limit our reliability assessment [40]. However, experienced spine surgeons from Level-I Trauma Centers guide the management of atlas fractures, and the role of training on the size and precision of reliability estimates is crucial [41]. Furthermore, the Gehweiler classification is commonly applied in Europe, but in North America and Asia, the Jefferson [42] classification is widely used. Therefore, the generalizability of interobserver agreement might differ depending on the working location of individual spine surgeons. In contrast to the Jefferson classification, the Gehweiler classification differs in the mandatory distinction between type 3 fractures as stable (3a) or unstable (3b). This distinction in particular forms the basis for the decision for a non-surgical or surgical therapy. In the opinion of the authors, the use of the Gehweiler classification is therefore especially beneficial in the assessment and treatment of atlas fractures. Evaluating the integrity of the TAL, and thus the possibility of distinguishing a stable type 3a and an unstable type 3b fracture, might have been easier with the additional radiographic information provided by an MRI scan. This assumption is supported by our findings of only fair interobserver reliability for the subtype distinction between type 3a or type 3b fractures (κ = 0.35) (Fig. 2 illustrates an exemplary case). Further on, in these cases stability assessment is even more impaired (κ = 0.28). Therefore, a higher interobserver reliability might have been achieved with the availability of MRI scans. However, testing the reliability of a global classification system should not rely on an imaging tool that has limited availability. For this reason, we did not provide MRI scans of the cases. Nonetheless, since the treatment strategy is decisively guided by the distinction between type 3a and type 3b fractures, the performance of MRI diagnostics is mandatory in the case of Gehweiler type 3 fractures [13].

The Gehweiler classification and stability assessments of isolated atlas fractures had moderate interobserver reliability in this study of experienced spine surgeons. For stability assessment and treatment recommendation high level of interobserver agreement was observed for C1-fractures rated as Gehweiler type 1, type 2 and type 5 but consensus was considerably less for type 3a, type 3b and type 4 fractures. Therefore, further biomechanical and clinical investigations are necessary to broaden our understanding of atlas fractures and refine treatment guidelines that integrate the therapeutic and prognostic implications of the Gehweiler classification system. Future studies may also want to determine whether atlas osteosynthesis in unstable C1-fractures with intraligamentous rupture or displaced bony avulsion of the TAL is reliably associated with atlantoaxial stability and good outcomes.