Traumatic Atlanto-Occipital Dislocation—A Comprehensive Analysis of All Case Series Found in the Spinal Trauma Literature

Search Strategy

Due to the multiple terms used to refer to TAOD, we performed multiple searches in the PubMed database with different strategies. All existing studies up until April 29, 2020, were queried and reviewed. A total of 4 different searches were performed including cross-referenced articles.

The first search was performed using the following key words: “atlanto AND occipital dislocation AND traumatic,” obtaining 122 studies, with 13 selected. The second search was performed using the following key words: “cranio AND cervical AND dislocation,” obtaining 42 studies with 1 selected. The third search was performed using the key words “occipito AND cervical AND dislocation,” obtaining 91 studies, with 3 selected. Finally, a fourth search was performed using the key words “craniocervical AND dissociation,” with an additional 4 studies obtained. Six studies were duplicates and thus excluded, for a total of 15 eligible studies. Additional cross-referencing provided 2 more clinical studies, for a total of 17 fully included. A flowchart detailing our search selection is presented in Figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flowchart diagram of our search mechanism in accordance to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).

Methodological Quality Evaluation: Potential risk for bias was assessed according to the Methodological Index for Non-Randomized Studies (MINORS) instrument (whereby fewer than 12 points suggest low-quality studies).¹⁴ All the studies were then analyzed for internal validity integrity and graded for level of evidence in accordance to the Oxford Centre for Evidence-Based Medicine.¹⁵

Eligibility Criteria: Clinical case series, retrospective or prospective, with a minimum 5 cases of TAOD were included. Studies should be written in English language; small case series or narrative and systematic reviews were excluded.

Data Collection Process

One of us (A.F.J.) independently extracted data from the included studies using a piloted data extraction form.

Data Extraction

The following data were extracted from the included studies, when described: number of patients, study design, patients' characteristics (age, Glasgow Coma Scale [GCS], sex), diagnosis and radiological findings, interventions performed, outcomes and follow-up, results, and survival rates.

Radiological Methods to Diagnose TAOD

In our review, there was a tremendous heterogeneity in the radiological methods used to diagnose TAOD. This was also consistent with the pertinent literature about challenges in the radiological diagnosis of TAOD. There were many proposed plain radiographic methods to diagnose TAOD. The most used included (1) a displacement of more than 10 mm between the basion and tip of the dens—the BDI³¹; (2) the ratio of the basion to the midpoint of the anterior cortex of the posterior arch of C1 distance divided by the ratio of the opisthion to the midpoint of the posterior cortex of the anterior arch of C1, which should be greater than 1 to diagnose—Power ratio³⁷; (3) a distance from the posterior mandible and anterior atlas of more than 13 mm or a distance from the posterior mandible and dens of more than 20 mm³⁸; (4) Failure of a line from the basion to the axis spinolaminar junction to intersect C2 or a line from the opisthion to the posterior inferior corner of the body of the axis to intersect C1—the X-line method³⁰; (5) a displacement of more than 12 mm or less than 4 mm between the basion and posterior C2 line (BAI) or a displacement of more than 12 mm from the basion to the dens (BDI).³² These measurements included studies published in 1958 and 1979, with the newest one published in 1994, which did not consider the quality of modern CT scan image and reconstructions. The limitations to proper bone and joints evaluation, as well as the low sensitivity (about 0.505) and specificity of x-rays in diagnosing TAOD (as well as other cervical injuries), are widely known.³⁶ For this reason, in the modern spinal-trauma era, we believe that plain radiographs to diagnose TAOD have only historical value. Recently, a radiological protocol proposed by the Spinal Cord Society concluded that a multidetector CT scan with reformatted images is the initial line of imaging for trauma patients and radiographs should be used only if CT is unavailable or unaffordable. In the same position statement protocol, they suggested that MRI should be indicated in patients with neurological involvement and advanced cervical degenerative changes to evaluate epidural-space compromise, disco-ligamentous injury, and also soft tissue injuries.⁷ Figures 2–4 illustrate the most important radiological measurements for diagnosing TAOD.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Sagittal midline computed tomography (CT) scan measuring the distance from the basion to the tip of the dens—the basion-dens interval. Normal values from adults' CT scans are <8.5 mm in 95% of the cases.⁴⁴

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

(A) Sagittal and (B) coronal computed tomography (CT) scan reconstructions—4 equidistant points were marked on the articulating surfaces of the occiput-C1 joint that encompassed the broadest joint surface. The joint interval at each point was recorded and the mean of these 8 measurements (from both sides and from sagittal and coronal CT scans) are the true condyle-C1 interval (CCI).³⁹

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Sagittal CT scan (A) of a normal and (B) of a patient with a traumatic atlanto-occipital dislocation (TAOD). The red line in B represents the revised condyle-C1 interval (CCI). The diagnosis of TAOD is made when distance at the point of great separation of the C1 lateral mass from the occipital condyle is greater than 2.5 mm.⁴⁴

In 2013, a guideline of the American Association of Neurological Surgeons concluded that initial lateral cervical radiographs may miss the diagnosis of TAOD, which was also observed in some cases published in our review, resulting in later neurological deficit.³⁶ By that time, they suggested that a high index of suspicion should be maintained in patients with prevertebral soft tissue swelling on “normal” lateral x-rays or in patients with craniocervical SAH on axial CT images. They also stated that there was a high level of evidence to use the condyle-C1 interval (CCI) in pediatric patients to diagnosis TAOD. This index was published by Pang et al³³ with a sensitivity and specificity of 100% compared with standard tests on plain films that had sensitivities varying from 25%–50% and specificities from 10%–60% in diagnosing TAOD. The CCI is the measurement of the interval between the occipital condyle and C1 at 4 equidistant points on the joint surface in sagittal and coronal CT reconstructions (a total of 8 measurements for each patient). The true CCI consisted in the mean of the 8 readings for both sides in coronal and sagittal images; with a cutoff of 4 mm (the CCI is highly symmetrical in children).

More recently, Gire et al³⁹ reported a radiological study designed to evaluate diagnostic measurements, with 10 cases of TAOD and 10 cases of non-TAOD. They evaluated the sensitivity, specificity, and reliability of 8 radiographic criteria applied to CT to diagnosis TAOD. They defined 2 new measurements: the revised CCI (measured at the point of greatest separation between the occipital condyle and C1 lateral mass, perpendicular from the surface of the lateral mass to the surface of the occipital condyle—TAOD was diagnosed when the left and/or right CCI was >2.5 mm) and the condylar sum (which consisted of the sum of the left and right CCI and TAOD was diagnosed when ≥ 5 mm). The revised CCI and the condylar sum were highly sensitive (100%), specific (92% for condylar sum and 84% for CCI), and reliable (intraclass correlation coefficients of 0.88 and 0.93 for CCI and the condylar sum, respectively) for detecting TAOD when using a CT scan.³⁹ The same was concluded by Dahdaleh et al²⁹ who performed another diagnostic study using CT scans comparing the BDI, the BAI, Lee X-lines, Power ratio, CCI, and revised CCI. They concluded that the revised CCI method was the most simple, sensitive, and reliable method. The high sensitivity and specificity of accessing directly the occipital-C1 joint is probably because the measurements directly assessed TAOD, whereas the other measurements were indirect evaluations. Of note, these “gold standard” CT scan measurements were not applied in any clinical series included in our systematic review, thus requiring further studies.

Martinez-del-Campo et al,⁴⁰ in 2016, tried to identify the normal CCI in healthy adults and compare these values with the CCI in adults with a known diagnosis of TAOD to establish more sensitive and specific cutoff values for diagnosing TAOD. They evaluated 59 adults without TAOD and 22 with TAOD, measuring thin-slice CT scans to assess the CCI, condylar sum, the Wholey and Harris intervals, the Power and Sun ratios, and the Wackenheim line and Lee X-lines. Patients without TAOD included 30 men (50.8%) and 29 women (49.2%) with a mean age of 42 ± 16 years (range, 19–87 years), and the group with TAOD comprised 10 men (45.5%) and 12 women (54.5%) with a mean age of 38.2 ± 9.7 years (range, 20–56 years). The mean CCI obtained for non-TAOD patients was 0.89 ± 0.12 mm, with the largest one measuring 1.4 mm. The mean condylar sum was 1.8 ± 0.2 mm, with the largest condylar sum value of 2.2 mm. In patients with TAOD, the mean CCI was 3.35 ± 0.18 mm (range, 1.5–6.4 mm). The shortest single CCI measurement in the TAOD patients was 1.1 mm. Martinez-del-Campo et al⁴⁰ reported that the mean condylar sum for all 22 TAOD patients was 6.7 ± 2.7 mm and the shortest condylar sum was 3.0 mm. After these measurements, the authors proposed new cutoff values for AOD: 1.5 mm for the CCI and 3.0 mm for the condylar sum in adults (compared with 4 mm in children for the CCI), with a sensitivity of 1 and false-negative rate of 0. The sensitivities of other measurements were 0.55, 0.46, 0.27, 0.23, 0.41, and 0.41 for Power, Wholey, Harris, Sun, Wackenheim, and Lee criteria, respectively.

The role of MRI in TAOD is still debated. The evaluation of ligamentous injury in MRI has supplanted any other image modality, but in the setting of a normal or near-normal CT, these findings may not be enough to indicate an OCF. Horn et al²¹ tried to evaluate the role of short TI inversion recovery sequences in determining cervical spine instability after trauma in 314 patients. They correlated soft tissue edema with x-rays and CT findings and concluded that when CT and x-rays did not demonstrate evident instability, MRI did not provide additional information. In the Hazboun et al²⁷ study, however, the authors reported that 7 patients (of 17 survivors, a total of 41.1%) had normal initial cervical x-rays or CT spine scans but, due to persistent neck pain, underwent an MRI with the diagnosis of a TAOD requiring an OCF. This was also proposed by a radiological study of Corcoran et al⁴¹ that reported that CCI was not enough to diagnosis some cases of TAOD. The sensitivity used by Pang et al³³ (4 mm) was low and the specificity of Gire et al³⁹ for their revised CCI (2.5 mm) was also low for less severe forms of TAOD but clinically significant. For these reasons, they stated the importance of MRI for complete evaluation of these patients.

Abouelleil et al⁴² reported an interesting case of TAOD wherein the CCI was normal, but there was severe TAOD diagnosed with an increased BDI (19 mm) and atlantoaxial vertical subluxation. TAOD was clearly confirmed using an MRI, with evident apical ligamentous disruption at the craniovertebral junction (CVJ) and C1-C2 distraction. They⁴² proposed that in the presence of C1-C2 dislocation, the CCI interval may not diagnose TAOD, and other measurements such as the BDI and MRI should be considered. It is important to note that the BDI cutoff for diagnosing TAOD on the basis of a sagittal CT scan is <8.5 mm in 95 of 100 normal adults, compared with 12 mm on plain radiographs.⁴³ We also believe that some patients with bilateral condyle avulsion may have normal CCI or revised CCI and severe forms of TAOD.

Treatment Considerations—Nonoperative Management

Horn et al²¹ described 5 cases treated nonoperatively based on abnormal findings on MRI but with mild or no changes evident on the CT scans. They proposed that external immobilization would be an option when there is a normal CT with mild findings on MRI, such as moderate to severe injury to posterior interspinous and atlanto-occipital ligaments and mild injury to the C0-C1 joints. They stated that the dilemma: to overtreat some patients with normal CT scan measurements but C0-C1 joint disruptions on MRI or not to treat them, which may lead to catastrophic neurological injuries or even death. To solve this dilemma, they proposed a classification based on 2 grades: grade 1, characterized by normal CT scan findings but moderate MRI findings (high posterior ligaments or occipitoatlantal signal) that may be treated nonoperatively; and grade 2, minimum of 1 abnormal CT scan finding of established diagnostic criteria and grossly abnormal findings in MRI, such as in the occipitoatlantal joints, tectorial membrane, alar ligaments, or cruciate ligaments, which may be surgically fixed. It seems reasonable to not treat all patients who have only MRI findings but also to fix some patients who are at risk of a severe catastrophic injury. Although the authors did not discuss this in their study, we believe that patients with grade 1 but with spinal cord injury should be aggressively managed with OCF, similar to spinal cord injuries at any other level.

Kaplan et al⁴⁴ published a literature review of nonoperative management of craniocervical ligamentous distraction injuries. They presented a case of a 27-year-old man who had severe ligamentous injury at C0-C1 and C1-C2 treated successfully with a halo for 6 months due to his poor clinical condition, with reduction and evident stability on dynamic x-rays and minimal symptoms at 41 months after his injury. They attributed their successful outcome to the fact that the patient probably had only incomplete injury to the main ligaments of the upper cervical spine. The outcomes of nonoperative management are not clear, and the risks potentially outweigh the benefits.

Child et al⁴⁵ studied the craniocervical ligament's ability to prevent craniocervical distraction in a cadaveric biomechanical model using a fluoroscopic traction test and successive sectioning of the primary ligamentous complexes. Six specimens from occiput-C3 had the following ligaments sectioned to simulate a TAOD: alar, tectorial membrane, and occiput-C1 capsules. Traction was applied using 0–20 lb with fluoroscopy, and digital images analysis quantified the displacement from C0-C1, craniocervical excursion, and under what load a 2-mm displacement could be reproducible. It is interesting that minimal dislocation was reported with sectioning 2 of 3 ligaments (<2 mm). They reported that specimens had a firm restraint to dissociation until all 3 ligaments were cut—in an “all-or-none” restraint to instability characteristic. For these reasons, all 3 ligamentous complexes (ie, alar ligaments, tectorial membrane, and atlanto-occipital capsules) may be important to maintain stability in anatomical models and may explain why some patients may survive neurologically intact.

The commonly cited classification for TAOD published by Traynelis et al³⁴ divided injuries in 3 types: type I (anterior dislocation of the occipital condyles in relationship with the atlas lateral masses), type II (longitudinal dislocation of the condyles), and type III (posterior dislocation of the condyles in relationship with the atlas lateral masses).³⁴ This classification was criticized by Horn et al²¹ due to the potential hypermobility of these injuries, which would allow all 3 categories to be displayed in the same patient. There is also a possibility that such instability could be well positioned, and this may preclude evidencing the displacement in x-rays or CT scan.²³

A significant number of studies reported other spinal injuries associated with TAOD, such as upper spinal fractures. In their series, Radcliff et al¹⁰ emphasized the high prevalence of atlantoaxial dislocations, which may or may not be associated with TAOD. In 5 cases, C1-C2 dislocations were not associated with TAOD, suggesting that separated structures stabilized the occipitoatlantal and atlantoaxial ligaments. Treatment for these associated injuries should be taken into account in a tailored approach.

Treatment Considerations—Operative Management

As already stated, the great majority of TAOD should be surgically treated as soon as possible. Although preoperative cranial traction may be used to realign the CCJ, up to 10% neurological deterioration has been reported in the literature, which may require extreme cautions and close clinical and radiological follow-up if traction is intended.³⁰

When operative treatment is chosen for TAOD, modern occipito-cervical fixation techniques are based on screw-rod hardware, with a high rate of fusion and immediate occipito-cervical stabilization.⁴⁶ Without associated subaxial injuries, an occipital C2 screw fixation is generally enough to restore CCJ stability. The C2 fixation can be achieved using different techniques, such as laminar, pars, pedicle, or transarticular screws, considering the patient's specific anatomy and the surgeon's experience.⁴⁷

Traumatic Brain Injuries

Because the CVJ is stable due to strong ligamentous structures, a high-energy impact is generally necessary to cause TAOD. For this reason, it is commonly associated with facial trauma and TBI, which may occult the spinal cord injury once patients present as comatose to the emergency department, which may also contribute to the delay in diagnosis in some patients.^12,20 In addition, some patients may die due to TBI and not from the TAOD.²² Different types of intracranial pathology were documented in our review, such as cerebral edema, brain contusion, intraventricular hemorrhage, subdural hematoma, and the most common, SAH, especially in the posterior fossa, which may lead to a high suspicion of AOD.³⁰ SAH in front of the pons, for instance, could lead to a high suspicion of CCJ pathology, especially because CT and MRI are made in a supine position and blood can flow up to the brain. On the other hand, cortical SAH is common in TBI and may not lead to a specific suspicion of TAOD.

Unfortunately, this review has several limitations: First, only English literature was reviewed, the quality of the included studies was not high, and we only used the PubMed database for screening. Finally, studies performed in the 1980s and 1990s may have different results from more recent studies, due to advances in neuroradiological exams with better anatomical details and image quality as well as improvement in surgical techniques to stabilize the occipito-cervical junction when compared with old fixation devices. We did not perform an individual analysis according to the year of the study. In spite of these limitations, we believe that our review provides a comprehensive view of TAOD that may be useful for designing future prospective studies.