Diagnostic Benefit of MRI for Exclusion of Ligamentous Injury in Patients with Lateral Atlantodental Interval Asymmetry at Initial Trauma CT

Introduction

Up to 2%–6% of patients show acute cervical spine injuries (CSIs) after blunt trauma (1,2). Most CSIs are reported to be fractures; ligamentous injuries and combined injuries are far less common (2), and spinal cord injuries occur rarely (0.07%–0.26%) (3). However, if such lesions are missed during the initial clinical examination, serious neurologic deficits and cervical spine instability may arise. Previous studies have shown that about half of all CSIs are not suspected at clinical examination. Thus, criteria suggesting CSIs have been established as a decision aid for the use of further radiologic evaluation (eg, the National Emergency X-Radiography Utilization Study Group, or NEXUS, criteria and the Canadian C-Spine Rule) (4). On the basis of those criteria, guidelines such as the American College of Radiology Appropriateness Criteria (5) and the Eastern Association for the Surgery of Trauma Practice Management Guidelines (6) recommend further radiologic evaluation following adequate cervical trauma. CT has replaced conventional radiography as the modality of choice because of its superiority in the detection of skeletal injuries (7). Hence, CT is frequently used for initial evaluation of trauma patients in the emergency department because it is fast and accurate, has widespread availability, and is cost-effective (8).

The exclusion of ligamentous injury at cervical spine CT (CSCT) may cause diagnostic difficulties. Although an increased anterior atlantodental interval (AADI) is accepted as a sign of underlying pathologic abnormality (9), the clinical relevance of an asymmetric lateral atlantodental interval (LADI) remains unclear. Besides ligamentous injury or atlantoaxial rotational fixation and/or subluxation at the C1/C2 level (10,11), anatomic variations, head rotation, and muscle spasm may also affect LADI. Hence, LADI asymmetry may also be observed in healthy individuals as a normal variant (12,13). Trauma patients with LADI asymmetry at radiography or CT are frequently admitted for subsequent MRI to rule out CSI. This leads to a tremendous amount of MRI examinations that occupy MRI resources and may also increase treatment costs. Previous studies have shown ambiguous results concerning the benefit of additional MRI after negative findings at CSCT (14–18).

The purpose of this study was to assess the additional diagnostic value of cervical spine MRI in trauma patients with detected LADI asymmetry as the only abnormality at preceding CSCT.

Materials and Methods

The study was approved by the institutional review board, and the need for written informed consent was waived because of the retrospective study design.

Results

Patient Characteristics

A total of 1553 trauma patients underwent CSCT in the emergency department of our hospital between March 2017 and October 2019. Among these, 1407 patients were excluded for lack of LADI asymmetry of 1 mm or greater, leaving 146 of the 1553 patients (9%) in the study sample. Among those 146 patients, 62 (43%) underwent supplementary MRI of the cervical spine to exclude CSI. Among those, another 16 patients had to be excluded because there were pathologic findings on the initial CT scan (n = 13) or the patient had previously undergone cervical surgery (n = 3). Thus, 46 of the 62 patients (74%) with asymmetry of the LADI as the only radiologic sign of potential pathologic abnormality at CT were included (Fig 1). The median time between CT and MRI was 1 day (IQR, 1–3.3 days). Twenty-eight of the 46 patients (61%) underwent 1.5-T imaging and 18 (39%) underwent 3.0-T imaging (P = .06). The mean patient age was 39 years ± 22 (standard deviation); 28 patients were men (mean age, 42 years ± 23) and 18 were women (mean age, 32 years ± 20) (P = .14). Forty-two of the 46 patients (91%) presented to the emergency department with a pre–clinically applied hard collar. The primary indication for MRI in these patients was exclusion of ligamentous injuries due to an asymmetry of the LADI at trauma CT. A more detailed overview of the patient characteristics and indications is given in Table 1.

Figure 1:

Download as PowerPoint

Table 1: Patient Characteristics, Mechanism of Injury, and Posttraumatic Diagnoses

CT Results

The median AADI was 1.3 mm (IQR, 1.0–1.5 mm). The median right-sided LADI was 3.7 mm (IQR, 2.8–5.1 mm), and the median left-sided LADI was 4.2 mm (IQR, 2.7–5.8 mm). The median LADI asymmetry was 2.4 mm (IQR, 1.8–3.1 mm); asymmetry was right-sided in 27 of the 46 patients (59%) (see Tables 2 and 3 for detailed magnitudes of LADI asymmetry). The median head rotation was 2.3° (IQR, 1.4°–5.1°), mostly with rotation toward the right side (24 of 46 patients [52%]). The median lateral tilt was 1.8° (IQR, 0.6°–3.3°). No correlation was observed between LADI asymmetry and head rotation (r = –0.13,; P = .38) or lateral head tilt (r = 0.07; P = .66) (Fig 2). The lateral margins of C1 on C2 showed a median lateral translation of 2.7 mm (IQR, 2.4–3.0 mm) on the right side and 2.7 mm (IQR, 2.4–2.9 mm) on the left side. No patients showed a loss of C1/C2 facet overlap less than 20% as an indicator of rotatory subluxation.

Table 2: Measurements of Cervical Spine CT in 46 Patients

Table 3: Magnitude with Frequencies of LADI Asymmetry in 46 Patients

Figure 2:

Download as PowerPoint

MRI Results

Six of the 46 patients with LADI asymmetry and lack of cervical injuries at initial CSCT (13%) showed CSI at MRI (Table 1). Ligamentous injuries of the cervical spine at the C1/C2 level were verified in two of the 46 patients (4%); one patient had a complete rupture of the right alar ligament (LADI asymmetry, 5.1 mm; Fig 3), and the other had a partial rupture of the right alar ligament (LADI asymmetry, 3.4 mm; Fig 4). The diagnoses influenced the therapeutic regimen in both patients: surgical fixation of the C1/C2 joint was needed for the first patient, and a 6-week extension of cervical collar treatment was needed for the other patient. Apart from those ligamentous injuries, an interspinous ligament strain at the C6/C7 level (n = 1, symptomatic patient), mild bone bruises (n = 2, both patients asymptomatic), and a prevertebral hematoma (n = 1, symptomatic patient) were diagnosed in the remaining four patients with CSI. These findings did not necessitate any treatment. Signal intensity alterations of the alar ligaments without ruptures or with partial ruptures were observed in 13 of the 46 patients (28%) (Fig 5). In four of the 46 patients (9%), signal intensity alterations were considered as ligamentous injuries at the initial clinical reading but not at consensus review by the two expert readers. This initial evaluation led to an adjustment of treatment regimen by extending the use of a cervical collar for 6 weeks. These patients were asymptomatic. The κ coefficient of the two expert readers regarding the evaluation of spine MRI before consensus reading was good (κ = 0.79; 95% CI: 0.39, 1.0).

Figure 3:

Download as PowerPoint

Figure 4:

Download as PowerPoint

Figure 5:

Download as PowerPoint

Discussion

Until now, studies investigating the utility of MRI in the detection of cervical spine injuries after a negative CT scan have shown ambiguous results (14–18). Trauma society guidelines recommend cervical collar removal after a negative CT scan alone, even in adults with obtunded blunt trauma (19). Because lateral atlantodental interval asymmetry in trauma patients can be caused by normal variants, ligamentous injuries, and incorrect positioning, such a finding may lead to diagnostic difficulties at cervical spine CT.

Our study aimed to clarify whether trauma patients with LADI asymmetry and lack of cervical injuries at initial CT benefit from complementary MRI. We found an LADI asymmetry of 1 mm or greater in 146 of 1553 trauma patients (9%), with no correlation to head rotation (r = –0.13; P = .38) or lateral head tilt (r = 0.07; P = .66). Six of the 46 patients (13%) with supplementary MRI had CSI; however, only two of these patients (4%) had unilateral alar ligament injuries that resulted in a therapeutic consequence. Both patients were symptomatic, which already clinically suggested a ligamentous injury. No other patients had ligamentous injuries, even if larger LADI asymmetry (up to 7.1 mm) was present. Most patients (36 of 46 [78%]) were asymptomatic at the cervical level. Thirteen of the 46 patients (28%) showed at least one-sided signal intensity alterations of the alar ligaments without ruptures as a normal variant; only two patients were symptomatic. These signal intensity alterations were initially misinterpreted as ligamentous injury in four of the 46 patients (9%), all of them asymptomatic, and led to unnecessary treatment.

Following Billmann et al (13), we used a threshold value of 1.0 mm for LADI asymmetry. The incidence was lower in our posttraumatic study sample (9%) than in a different posttraumatic sample (29%) (13) and a healthy Chinese sample (31%) (20).

Unilateral alar ligament injuries seem to be rare, which aligns with our study findings (two patients [4%]). Although a biomechanical study even argues against an involvement of alar ligaments in whiplash injuries (21), Unal et al (22) suspected these injuries to be more common than diagnosed in a recent small study with only six patients.

Neck pain aggravated with rotation and dens lateralization at CT are the most significant findings for alar ligament injuries during clinical and radiologic examination (22). The discrepancy between clinical symptoms and morphologic diagnosis is a challenge in evaluating trauma patients: Patients with severe clinical symptoms may not show any cervical injuries (23,24). In addition, healthy individuals serving as control participants show signal intensity alterations and anatomic variations of alar ligaments (25–27). Chen et al (28) underlined the difficulty of differentiating between signal intensity changes due to physiologic evolution and chronic physical damage in older patients. Because it is impossible to differentiate between injuries and normal variants, Muhle et al (29) proposed that ligament injuries should be diagnosed only if tears or avulsion fractures are visible. Vetti et al (25) even concluded that additional posttraumatic cervical MRI should not be recommended because of a lack of clinical consequence. Small amounts of fluid are evident in 8% of the joints of the craniocervical junction and in 56% in the C1/C2 joints; thus, they should not be overinterpreted regarding potential CSI (30). Review of these studies illustrates that even in the acute trauma setting, differentiation between traumatic and nontraumatic ligament signal intensity changes is nearly impossible, especially if secondary signs of injury, such as hemorrhage, are missing. Therefore, we did not interpret sole ligament signal intensity alterations or adjacent asymmetric fluid collections as injuries. In that aspect, our results differ from those of other studies with a higher incidence of detected CSI at MRI. In a recent study, 92.6% of 149 “positive MRI” results were classified as “stable injuries” and consisted mainly of ligament signal intensity changes (14).

Several investigators have postulated that head rotation (13,31) and even lateral head tilt (32) result in LADI asymmetry. We found no correlation of LADI asymmetry to either of these, supporting the theory that LADI asymmetries are mostly normal variants regardless of head positioning. Nevertheless, the missing correlation between LADI asymmetry and head rotation could also be explained by the low individual head rotation (median rotation, 2.3°) in our study sample, perhaps due to the emphasis on correct head positioning in our institution.

The patient with complete rupture of the right alar ligament underwent surgery because of atlantoaxial instability, whereas the patient with partial rupture of the right alar ligament received conservative treatment with a cervical collar for 6 weeks. Four of 13 (31%) asymptomatic patients with signal intensity alterations of alar ligaments received conservative treatment with a cervical collar for 6 weeks. These treatments followed initial radiologic misinterpretation of the ligamental signal intensity alterations as injuries. Because these signal intensity alterations—according to the above-mentioned studies—should primarily not be interpreted as injuries, overtreatments have probably taken place. The use of cervical collars carries the rare risk of complications, such as pressure ulcers (19). Thus, overtreatment should be avoided. Early mobilization of the cervical spine may even be considered beneficial in stable injuries compared with continued immobilization (33,34).

The reliability and benefits of MRI in excluding alar ligament injuries are disputable. With use of intraoperative correlation, Dyas et al (35) revealed that MRI is imperfect in depicting upper cervical ligament injuries in patients with atlanto-occipital dissociation due to inconsistent and unsatisfactory inter- and intraobserver reliability. They concluded that standardized algorithms are needed for the use of MRI and interpretation of MRI findings. Moreover, additional MRI might increase the number of low-value diagnoses, risks unnecessary treatment plans, puts patients with multiple injuries at risk by necessitating a move to the MRI suite, and, at best, results in the same clinical action of collar removal (19). Some authors conclude that additional MRI after CSCT may have utility only in certain patients with persistent abnormal findings at neurologic examination (14–16,18). Our results also indicate that MRI may be beneficial only in symptomatic patients with LADI asymmetry.

Our study had limitations. Because of the retrospective design, we could not determine why some patients with LADI asymmetry and persisting symptoms were not admitted for subsequent MRI. Most patients in our study sample (36 of 46 [78%]) were asymptomatic at the cervical spine, lowering the probability for CSI. However, by strictly applying the National Emergency X-Radiography Utilization Study Group criteria and/or the Canadian C-Spine Rule, all of these patients required further CSCT, mostly because of intense trauma mechanism or other “distracting injuries” independent of the cervical spine. Moreover, the final number of patients included in our study is low. At the same time, the ratio between “true-positive” and “false-positive” cases is very low. This is most likely related to the particular scenario evaluated; mere ligamentous injuries of the upper cervical joints without concomitant fractures in general are rare. In addition, according to the Injury Severity Score and Abbreviated Injury Scale, our study sample consisted of only a few patients with serious injuries.

In conclusion, our results indicate that additional MRI for exclusion of ligamentous injuries in trauma patients with lateral atlantodental interval asymmetry may be considered reasonable only in symptomatic patients. The additional diagnostic value in clinically asymptomatic patients without proven cervical injuries seems low and may contribute to overtreatment. Larger study samples and prospective study designs are needed to eventually verify our results.