Can radiologic parameters used to detect cervical spinal instability be used in patients with ankylosing spondylitis?

All patients underwent scans by multi-detector row CT of the cervical spine. Axial images were reconstructed at a section thickness of 1 mm and an increment of 0.7 mm. Axial, coronal, and sagittal reformations were obtained. Twenty-four radiologic parameters summarized in Table 1 were measured and evaluated by three independent orthopedic surgeons who were blinded to clinical data. An example of a sagittal plane CT of the cervical spine of a patient in the AS group is shown in Fig. 1.

Data were analyzed using the SPSS software package version 20.0 for Windows (SPSS Inc., Chicago, IL, USA). Descriptive statistics are presented as mean and standard error of the mean (SEM). Statistical distribution of all the data was assessed by both the Shapiro–Wilk and Kolmogorov–Smirnov tests. An independent t-test was used to calculate statistical differences between the AS group and the control group. A p-value less than 0.05 was considered statistically significant.

The BDI and BAI were both well within normal limits (< 8.5 mm) and similar in both the AS and control groups (5.1 vs. 5.3 mm, p = 0.544 and 4.9 vs. 5.6 mm, p = 0.180, respectively). Both methods of measuring Powers Ratio were also similar and within normal limits (< 0.9) in the AS and control groups (0.87 vs 0.90, p = 0.080 and 0.74 vs 0.78, p = 0.059, respectively). Eleven patients showed pathological Lee X-Line parameters in both study groups (28.2%, p = 1.00). Further, we looked at the lengths of the individual lines used to assess Powers Ratio and Lee X-Lines as well as measuring a Lee X-Line ratio to see if we could find any subtle anatomic variation between the two groups. This analysis, however, also showed no differences between the groups. The mean posterior atlanto-dens interval (PADI) was well within normal limits (> 14 mm) and nearly identical in the AS and control groups (21.0 vs. 21.3 mm, p = 0.523).

Both Chamberlain’s and McCrae’s lines showed similar results in both the AS group and healthy controls. Seven patients in the control group and 8 patients in the AS group had a pathological Chamberlain line (17.9% vs. 20.5%, respectively, p = 1.00). There were no patients with a pathological McCrae’s line in either group (0%, p = 1.00). As with the atlanto-occipital parameters above, line lengths were also compared to observe potentially subtle anatomic differences, but no differences between the groups were observed.

Endplate tilt at C2/3 (1.8° vs 1.9°, p = 0.910), C4/5 (1.6° vs − 1.4°, p = 0.082) and C5/6 (2.6° vs 1.8°, p = 0.597) were similar and stable (< 11°) between the AS and control groups. However, a consistent, but subtle trend toward more kyphotic angulation, especially at the C4/5 level, was seen in the AS group. Displacement at the level of C3 and C6 were less than 1 mm in both the AS and control patients. Finally, soft tissue swelling at C3 (4.6 vs 6.1 mm, p = 0.070) and C6 (13.8 vs 12.4 mm, p = 0.107) were within normal limits (C3 < 7 mm, C6 < 21 mm) and similar in both the AS and control groups, respectively.

BAI and BDI are helpful for the diagnosis of atlanto-occipital dissociation injuries. Normal values should be less than 12 mm on plain radiographs. An increase in this distance may indicate instability [16]. Multiple studies declared that the accepted ranges of normal values of BDI and BAI on plain radiographs cannot apply to CT images. Rojas et al. [17] argued that the BAI was difficult to reproduce on CT images; the value was found to be highly variable and a number of subjects had BAIs greater than 12 mm. They found that the distance of BDI was < 8.5 mm in the vast majority of 200 cases and the maximum distance recorded was 9.1 mm. Gonzalez et al. [18] also demonstrated a mean BDI of 4.7 mm and a maximum of 9 mm in CT images from healthy individuals. In line with the literature, the BDI and BAI in the control group of our study were well within normal limits at 5.3 (SD 1.6) and 5.6 (SD 2.2) mm, respectively. Neither BDI nor BAI differed in the AS group compared to the control group and were equally well within normal limits.

Changes to the craniovertebral junction in AS with cervical involvement has been well described and is frequently involved in severe AS [19]. All patients with AS underwent structural changes of articulation and/or ligamentous craniocervical structures [19]. The interval of the atlanto-occipital joint and the atlanto-dental joint were decreased in AS patients, but the BDI and the Power’s ratio were not changed [19]. Our results confirm these findings.

Studies of BAI measurements in AS patients are rare. Robust articular ligaments, such as the cruciform ligament, are important for atlanto-axial joint stability [20]. In the pathological process of AS, these ligaments undergo inflammation such as enthesitis, and laxity or rupture with subsequent atlanto-axial subluxation can occur [21]. Furthermore, odontoid pannus formation can lead to atlanto-axial instability [22]. We suspected that excessive kyphosis might also occur at this level, potentially altering various parameters, especially BAI. However, BAI was found within normal limits and not different to the control group. The only pathological values we found in the present study were seen with the Lee’s X-line parameters with pathological findings in 11 (28.2%) patients in both study arms. The hypothesis of altered atlanto-occipital cervical parameters on account of AS induced instability, specifically pathologic kyphosis, could not be supported in this study.

The endplate and the posterior vertebral body tangent measurement methods for C2–C3 angulation were first described by Levine and Edwards [23]. C2–C3 angulation and translation were usually used to evaluate for traumatic spondylolisthesis including Hangman’s fractures [24, 25]. The physiological cervical spine shows a slight lordotic curvature at the level C2–C3 of about -1.9 ± 5.20° [26]. Cervical lordosis in AS patients is decreased and with increasing severity, even kyphotic deformities may occur [6]. However, it is unknown if kyphotic changes in AS are focussed at specific levels, or if the kyphotic changes are equally distributed throughout the cervical spine and thus difficult to detect at individual levels. In this study, the angulation at C2–C3 as well as C4–C5 and C5–C6 were measured to assess for potential underlying spondylolisthetic changes [27]. Any bony pathologic changes of the endplate may influence the measurement of these angles. This is especially true in AS patients, where the endplate border is altered by the presence of syndesmophytes characteristic of the disease [28, 29]. In this study, no changes in single-level angulation were found between the groups. However, a subtle but statistically insignificant trend toward more kyphosis at all levels measured was seen in the AS versus control group. This was most pronounced at the C4/5 level.

Further, results did not differ regardless if measured using the endplate method or the posterior vertebral body line method [27]. We do, however, feel that the posterior vertebral line method may prove more accurate and useful in evaluating the cervical spine in AS, as it measures the angle of lines drawn perpendicular to the posterior vertebral body aspect of the two endplates. The pathological changes of the vertebral edges therefore have less impact on accurate assessment of possible spondylolisthesis. The results suggest that endplate angulation at single vertebral levels remains largely unchanged, despite overall loss of cervical lordosis in AS patients.

Additionally, no differences or pathological values in vertebral displacement (defined as > 3 mm) measured at C4–C5 and C5–C6 were found between the groups, further suggesting that spondylolisthetic degeneration was not present in the AS study group.

Parameters assessing basilar invagination were also identical between the AS and control group. While approximately 20% of patients in both the AS (n = 7) and control (n = 8) groups had a pathological Chamberlain line, none of the patients in both the AS and control group had a pathological McCrae line. While of no clinical or diagnostic significance, the line lengths were also measured as with the above parameters used to assess for atlanto-occipital dissociation. Here as well, we saw no differences between the groups, suggesting that subtle morphologic changes in the AS group were not present. While common in other arthopathies, such as rheumatoid arthritis, basilar invagination is rare in AS as pannus formation and degenerative destruction of the craniocervical junction has only been documented in late stage disease [21, 22, 30]. Our results are therefore in line with the literature and previous radiologic studies of AS.

It is important to note that with increasing severity of AS, more osteophytic and syndesmophytic changes occur, which may have influenced the accuracy of some of the measured parameters that rely on precise bony landmarks. The severity of AS was not specifically assessed for in this study and is an important study limitation, especially considering that cervical changes are usually found in late stage disease. Since the majority of the AS patients in this study population were older, however, it is likely that a significant portion of the study population was afflicted with more severe disease. Due to the high patient age, the matched study design was useful in accounting for the potentially confounding effect of general spinal degeneration. Further study limitations include the limited size of the study population, and the fact that the cervical parameters, while carefully measured by three board certified orthopedic surgeons, were not evaluated by fellowship-trained musculoskeletal radiologists, potentially affecting the reliability of measured parameters.

Further research is needed to better rapidly assess for ligamentous injuries of the cervical spine in older patients with AS or other pre-existing degenerative disease processes of the cervical spine in a polytraumatized shock room setting. A study with a larger population would be useful to confirm our results. Perhaps one area of future development could include the identification of laboratory parameters specific for ligamentous injuries of the spine as a more rapid adjunct to further imaging diagnostics, which may not be immediately attainable in the polytraumatized patient setting.