Assessment of atlantoaxial rotation: how accurate is clinical measurement? a comparative study of cervical range of motion using MRI and standard orthopedic techniques

Measurement of neck rotation is currently reliant on radiologic imaging. Given the radiation exposure for CT imaging and the additional inconvenience for the patients, an alternative assessment is needed. Goniometers are comfortably to use and easy to access, also for private consulting. The aim of this study was the assessment of whether a handheld goniometer can be used for accurately measuring the rotation of C1-C2. Clinical measurement of rotation was taken in flexed position of the neck. As comparison functional MRI was used. The measured rotation of C1-C2 was compared to identify the accuracy of the goniometer, in comparison to functional MRI scan. Analysis of accuracy using a goniometer and dynamic MRI to assess C1-2 axial rotation showed significant differences for absolute values, but not regarding the percentage of rotation compared to total neck rotation. The goniometer is exact to impartially determine the percentage contribution of C1-2 rotation to total neck rotation.


Introduction
The physiologic range of motion (ROM) of the human atlantoaxial joints enables axial rotation of 70° and flexion-extension of 20°. With initiation of head rotation, C1-2 starts to rotate, and C1-2 rotation accounts for up to 60-70% of overall cervical axial rotation [1][2][3].
Atlantoaxial pathologies and different treatment strategies lead to a variable loss of both total cervical and C1-2 rotation. There is, however, insufficient knowledge of the impact a specific treatment has on rotation C1-2 [4][5][6][7][8]. Some studies have investigated ROM using functional CT scanning, following the treatment of C1-2 injuries [9][10][11], or the surgical treatment of degenerative disorders of the cervical spine [12].
The estimated loss of C1-2 rotation compared to total neck rotation has been investigated in previous studies. Jeanneret et al. [9] measured C1-2 rotation after AOSF using functional CT scanning in 14 patients, observing a mean of 25° to the left and 24° to the right side, and only a few patients had normal rotation of C1-2. Koller et al. [10] studied C1-2 rotation using functional CT scanning in 35 patients with C2 fractures. Total neck rotation averaged 103° and C1-2 axial rotation averaged 40°. Using dynamic CT, the authors showed that with C2 malunion the ability to perform C1-2 rotation was reduced, as were the clinical outcomes. Using dynamic CT scans, Guo et al. [11], demonstrated better C1-2 axial rotation in 20 patients who had anterior odontoid screw fixation (av. 40°), than 20 patients who had temporary fixation of C1-2 (av. 32°), for the treatment of odontoid Type-II fractures. These numbers are suitable as clinical reference. Nevertheless, all of these studies relied on CT scanning for measurement, which involves radiation exposure, and can delay of diagnosis because CT scanning is not always readily available in the outpatient clinics or in practice.
Standardizing functional results after various treatment options, in evaluating conservative vs. surgical and motion sparing vs. fusion surgery, is an important research tool that needs validated tools that are easy-to-access [9][10][11][12].
A clinical handheld goniometer is a standard instrument that physicians are comfortable with. Recent studies have shown that measurement accuracy using a handheld goniometer is comparable with that using smartphone technology [14]. The accuracy of clinical measurements to access C1-2 rotation, however, has not yet been evaluated in depth. A previous study by Koller et al. [13] reported a poor correlation between C1-C2 axial rotation measured with functional CT scans and the clinical measurement of overall cervical axial rotation using a handheld goniometer. Total neck rotation showed the ability to mask actual C1-2 rotation. According to this study the accuracy of measurements of total cervical rotation using a handheld goniometer is not predictive of C1-2 rotation. That study, however, did not analyze the accuracy of clinical measurements of C1-2 axial rotation in the flexed head position, which is a posture that locks the subaxial neck rotation [1].
Therefore, the aim of this study was to analyze the accuracy of a common handheld goniometer for the measurement of total cervical rotation and C1-2 rotation. The accuracy of the assessment of the clinical measurements was compared to measurements of functional MRI.

Materials and methods
This study is an in vivo comparative analysis of the total cervical axial range of motion and atlantoaxial rotation.

Study sample and inclusion criteria
Twenty healthy volunteers, all medical staff, including 10 males and 10 females, were recruited for this study. The inclusion criteria were age above 18 years, no previous cervical spine surgery or diagnosed pathology, and no history of chronic neck pain or shoulder-related disorder. The age of the participants was 34 ± 6 years, and the body-mass-index was 27 ± 4. The technique was explained to the volunteers, and informed consent was obtained from each participant.

Clinical measurements
Active ROM was assessed using a common orthopedic handheld goniometer with 1° increments. ROM was measured in a sitting position, and subjects were sitting comfortably, with their feet flat on the floor and hands relaxed by their sides. They were asked to sit up straight and position their head in a neutral position, as described elsewhere [15]. Clinical measurements were performed for each direction of rotation, and the sum is reported as total clinical neck rotation (cROT neck ). Measurements of atlantoaxial rotation, namely, C1-2 rotation, during left and right head rotation were performed in terms of the flexion-rotation test (FRT). The sum of the left and right rotation is reported as cROT C1−2 .
The FRT is a commonly used test to analyze the ability of atlantoaxial joints in axial rotation [16]. It was shown to be reliable to analyze C1-2 ROM by Hall et al. [17]. With the subjects sitting in the upright position, the head was actively flexed, followed by active left and right rotation to take the measurements.
Landmarks for clinical measurements included the vertex and a line imagined from the top view connecting the nose to the opisthion (Fig. 1). For the statistical analysis, ROM in the left and right directions were summed and reported as total ROM values.
Clinical measurement of total cervical rotation and C1-2 axial rotation in the head-flexion position was performed by an experienced orthopedic surgeon. Measurements were performed three times for each direction of rotation, and the average was taken for statistical analysis. To benchmark the clinical measurements with a standard of reference, cROT neck and cROT C1−2 were studied in detail using functional MRI as described below.

Functional MRI
ROM was studied using MRI with the subjects in a comfortable supine position. For the study, a 0.5 Tesla Philips Gyroscan NT 5 Magnetic Resonance scanner (Philips Medical Systems, Reigate, U.K.) was used. Axial T1-weighted images from the base of the skull to T1 were obtained. The image slices were 4 mm thick with a 0.4 mm interslice gap. At first, a set of images was acquired with the subject supine and the head in a neutral position. Next, the head was turned to the individual maximum left and right positions, and the imaging was repeated. The subject was asked to remain in the rotated position without movement for the duration of the investigation. Because axial rotation of the head is accompanied by some tilting of the vertebrae relative to the axial plane, the initial‚ scout or planning MRI was repeated after each change in head position, as recommended by Roche et al. [2].
Analysis of the MRI slices was performed by three observers, and the means were used for statistical analysis. Digital measurements with 0.1-mm increments were performed on the axial MRI slices using commercially available software (Escape Medical Viewer V3, Escape Thessaloniki, Greece). The measurements were performed as shown in Fig. 2, similar to a technique applied for C1-2 ROM measurements in three previous studies using functional CT scanning [12,13,18]. The C1 angle and the C2 angle were measured in the left and right head-rotation positions.
Maximum rotation of C1 to either side (mROT C1 ) was defined by the angle formed between the midsagittal line and the C1 bisector line, which connects the anterior and posterior C1 tubercle, crossing the tangent line of the posterior surface of the C1 lateral masses.
The rotation of the axis vertebra was assessed by drawing a bisector line for C2 that crossed the posterior cortical wall tangent at its widest level. The midsagittal line of C2 was calculated to the vertical axis delineating the maximum rotation of C2 to either side (mROM C2 ). Axial rotation of the subaxial spine is expressed by mROT C2 . Left and right mROT C1 and mROT C2 were calculated, and the separation angle between C1 and C2 was defined. This is the algebraic difference between the angle subtended by C1 and the angle subtended by C2.
Because the contribution of occipitoatlantal rotation to total neck rotation is negligible [1][2][3], C0 angles were not created, and maximum C1 rotation was defined as maximum head and neck rotation in this MRI study (mROT C1 ). Maximum C1 rotation is the ability of the total cervical spine to rotate to the left (mROT neck−left ) and the right (mROT neck−right ). For the statistical analysis, total left and right neck rotation (mROT C1 ) was summed and reported as total neck rotation (mROT neck ). The sum of the separation angle C1-2 in left and right neck rotation was reported as the ability of C1-2 to perform axial rotation (mROT C1−2 ).

Statistical analysis
Statistical analysis included a direct comparison of cROT neck and mROT neck as well as of cROT C1−2 and mROT C1−2 . To study the amount of C1-2 axial rotation contributing to total cervical rotation using clinical and MRI-based measurements, C1-2 axial rotation was expressed as the percentage of C1-2 rotation for each modality studied (%cROT C1−2 and %mROT C1−2 ). Data were checked for consistency and normality. ANO-VAs with and without the assumption of homogeneity of Fig. 2 The technique for measuring the C1-2 angle on MRI slices. Green-line: midsagittal plane. α = right mROT C1 ; α′ = right mROT C2 ; β = left mROT C1 ; β′ = left mROT C2 ; Separation angle C1-2 right side = α-α′. Vice versa for the left side. Total axial rotation of C1-2 is defined by the net sum of both separation angles variance were used. Levene's test was used to assess variance homogeneity. Paired and unpaired Student's t tests were used for pairwise comparisons. Pearson and Spearman correlation coefficients were computed to analyze correlations. Intraclass correlation coefficients (absolute agreement, ICCs) were computed. All reported tests were two-sided, and p-values < 0.05 were considered statistically significant. All statistical analyses in this report were performed using NCSS (NCSS 10, NCSS, LLC. Kaysville, UT) and STATISTICA 13 (StatSoft, Tulsa, OK).

Results
The results of all clinical and MRI measurements are summarized in Table 1, and differences are emphasized in Fig. 3. Analysis of intraobserver agreement for MRI measurements showed high intraobserver accuracy, reflected by an ICC of 0.97 (95% CI: 0.985-0.997). Side-related significant differences among the 20 healthy volunteers did not exist.
The statistical correlation between cROT C1−2 and cROT neck was high for the clinical measurements using the handheld goniometer (p < 0.0001, r = 0.71) and between mROT C1−2 and mROT neck for the MRI measurements  . 3 Results of the motion analysis (p = 0.001, r = 0.67). Age, BMI and sex had no significant correlation with neck rotation in the current study. Statistical analysis revealed significant differences for both the total neck rotation and atlantoaxial rotation between the clinical measurements using the handheld goniometer and the MRI-based measurements ( Table 1). The differences between the absolute values of the clinical and MRI measurements were all significant. However, comparing the clinical (59%) and MRI measurements (55%), statistical analyses revealed no statistically significant difference between the percentage of C1-2 rotation compared to total neck rotation.

Discussion
This is the first study that analyzed the accuracy of clinical assessment of C1-2 rotation using a standard handheld goniometer and the FRT in comparison to MRI, the latter being the reference standard. We did observe significant differences in absolute values of ROM for total cervical neck rotation and atlantoaxial cervical rotation. The differences can be explained by the ligamentous coupling of the cervical segments during cervical motion. With the head in flexion, the lower cervical spine is assumed to be locked, and total rotation is anticipated to occur in the C1-2 joints, while the axial rotation of C0-C1 is minimal (Av. 1-3°) [1]. In the flexed head position, Takasaki et al. [19] noted a 16% reduction in rotation range at C1-2 but a 70-80% reduction at the levels below C1-2. C1-2 provided 74% of the total rotation in the flexed head position. Hence, measurements in the current study comparing clinical methods and MRI methods show substantial differences for mROT neck (120°) and cROT neck (145°) because not all subaxial motion is locked with the FRT. However, differences in mROT C1−2 (66°) and cROT C1−2 (85°) reflected this trend and the relation between mROT neck and cROT neck in the increased measurements using the handheld goniometer method. Notably, using functional MRI in a study of 29 normal individuals, Roche et al. [2] measured atlantoaxial rotation of 69 ± 14°. This value is remarkably close to the 66 ± 10° in our study. Normal total axial neck rotation using the clinical handheld goniometer in our sample was 145 ± 15°, which compares fairly well to the 150 ± 15.1° in females and 155.5 ± 31.8° in males in the 30 to 39-year age group in a study by Castro et al. [20] using an ultrasound-based measuring system.
It is important to stress that we did not find statistically significant differences for %mROT C1−2 (55%) and %cROT C1−2 (59%). This finding indicates that this clinical measurement technique yields a valuable estimate of the percentage distribution of the upper cervical spine contributing to total cervical rotation.
The loss of C1-2 rotation compared to total neck rotation can be different according to a patient's individual ability to rotate in the subaxial spine. This estimate is suitable for use as a clinical rationale for patient counseling and for comparison of study results [9][10][11].
For a surgeon consulting with patients, it is important to give individualized estimates of the expected loss of cervical rotation following a specific intended treatment, particularly after fusion of C1-2. This loss varies from patient to patient given the divergent degree of the subaxial degenerative pattern, the neck dimension or the history of prior cervical injury and surgery [13]. The individual measurements during a consulting process should be exact, but not depend on functional CT scans. Similarly, from the perspective of medico-legal aspects and in expert opinion cases and insurance claims, it is of relevance to define the degree of disability based on objective findings such as function and restriction of motion. The measured motion must be impartial, and the measurement tool should be readily available in a physician's office and on an outpatient basis.
This study offers evidence that the use of the percentage of C1-2 rotation is a useful alternative to the more sophisticated use of functional MRI/CT.
The authors acknowledge the limitations of this study. Among them is the issue related to a small sample size and the difficulty of generalizing results for older age groups. Future studies might improve the generalizability of data with larger samples and other age groups, which might enable creation of age-related percentage motion scales for the cervical spine. Nevertheless, this is the first study to compare clinical handheld goniometer measurements with MRI findings. CT landmarks might be more exact in defining angulations. However, MRI was preferred to avoid radiation exposure for the volunteers. Inclusion of the landmarks that delineate head position in MRI and enable conducting reliable measurements of C0 angles would have increased the time that the volunteers were actively rotating their head to the maximum possible position during MRI, and this is uncomfortable. Because there is only minimum motion between C0 and C1 in healthy subjects, this study used C1 angles as surrogate measures of total neck rotation to either side.

Conclusion
Objective analysis of the assessment of total neck rotation, particularly C1-2 rotation, is an important estimate in clinical practice and when consulting with patients with upper cervical disorders. Instruments to assess upper cervical axial rotation should be readily available, exact and not expose the patients to avoidable radiation, use medical resources or have high costs.
This study showed that clinical assessment of total cervical neck rotation in the neutral sitting position and C1-2 axial rotation in the flexed head position and expression of C1-2 rotation as a percentage of total neck rotation serve as useful estimates of the ability of the C1-2 joints to take part in overall neck rotation.
Funding Open access funding provided by Paracelsus Medical University.