Clivopalate angle: a new diagnostic method for basilar invagination at magnetic resonance imaging

Basilar invagination (BI) is a common deformity of the craniovertebral junction (CVJ), characterized by flattening of the base of the skull and upward displacement (impression) of the basilar and condylar portions of the occipital bone and odontoid process, and is often accompanied by other complications such as other skeletal deformities, tonsillar herniation, and cervical syringomyelia [1‐5]. Magnetic resonance imaging (MRI) plays a vital role in the diagnosis and management of BI.

Although a series of imaging indicators is available, clinicians still desire a more reliable MRI index that significantly correlates with clinical findings of cervical medullary syndrome and accurately monitors treatment efficacy in patients with BI [6, 7]. The distance from the odontoid tip to the Chamberlain line (CL), the line drawn from the posterior margin of the hard palate to the dorsal margin of the foramen magnum, is the most commonly used index for diagnosing BI [8, 9]. However, a major challenge of this method on MRI is accurate identification of the cortical bone of the odontoid tip, the posterior margin of the hard palate, and opisthion [8]. Furthermore, the anatomical variations of the odontoid process or changes in its position due to trauma may also significantly interfere with these measurements [10]. Many studies demonstrated that clivoaxial angle (CXA) or clivocanal angle, defined as the intersection of the Wackenheim clivus baseline with a line drawn along the posterior surface of the axis body and odontoid process, is a practical indicator for BI with high diagnostic value [11]. However, several factors, such as subjects’ cervical curvature, which might change significantly after surgery, as well as hyperostosis, may affect CXA measurement and thus compromise its diagnostic value for BI [6, 12]. Recently, Xu and Gong [13] proposed a new index developed from CXA, the clivodens angle (CDA), to overcome the above-stated shortcomings, whereas its diagnostic value for BI was similar to that of CXA.

We noticed that clivopalate angle (CPA), formed at the intersection of the Wackenheim line and a line along the hard palate plane, was different between patients with and without BI. Moreover, this angle is easy to measure because these two lines naturally exist and are clearly visible in the midsagittal MR image (refer to “Materials and methods” section). Thus, we hypothesized that CPA may be a useful indicator for BI. In this study, we measured CPA in 112 patients with BI and 200 control subjects to investigate its diagnostic value for BI at MRI.

This retrospective study received approval from the institutional review board and was performed with a waiver of informed consent.

No significant difference was observed between the study and control subjects in sex (p = 0.185) or age (p = 0.117). Chief complaints from patients in the study group are summarized in Table 1. Accompanied deformities in both groups are summarized in Table 2. The odontoid process showed five types of morphological variations (Fig. 3) in 50 cases in the BI group and in 18 cases in the control. No structural abnormality of the hard palate was identified in either group.

Our present study demonstrated that measuring CPA on midsagittal MRI exhibited high inter-observer agreement. With an optimal cutoff value of 53.5°, CPA exhibited a high AUC value similar to that of CXA and CDA. Additionally, CPA provided complementary information to CXA or CDA for BI and improved the diagnostic performance.

The distance from the odontoid tip to the CL is the most widely used index for diagnosing BI [11, 14]. Generally, a CL value ≥ 5 mm is suggestive of BI [5, 12, 15, 16]. In contrast, using a CL value of 2.5 mm as the criterion may produce a relatively high false positive rate [13]. Thus, the criterion was set at 5 mm in this study, which resulted in a low false positive rate (5/200).

CXA, also called the clivocanal angle, reflects the angle between the clivus plane and C2 vertebrae, ranging from 150° in flexion and 180° in extension in a normal adult population [6]. In our present study, CXA ranged from 130° to 172° in the control group and using the traditional criterion (150°) would result in a very high false positive rate (36.5%, 73/200 cases) [6]. This variation might reflect the confounding effect caused by the subjects’ cervical curvature and hyperostosis [6, 12]. This might also be partly due to the newer method we used to measure CXA. The Wackenheim line used in this study ran along the lower third of the clivus, whereas traditionally the line extends from the top of the dorsum sellae to the basion (Fig. 2d) [12, 13, 17]. The newer method was recently recommended to standardize the measurement method for CVJ evaluation, which can be used to analyze both neck and head MRI data [6]. Many patients with BI complain of symptoms similar to those of cervical radiculopathy and undergo cervical MRI examination clinically. In the present data, most cases were cervical spine images; thus, the clivus was not completely covered and traditional measurement of CXA was impossible.

CDA was a new BI indicator proposed by Xu and Gong [13], who claimed that it could evade the influence from the posterior surfaces of the clivus and the axis that are not perfectly flat planes. However, CDA showed lower AUC than CXA in the present study, which was not in agreement with the previously reported [10]. Moreover, we found that the deformities or morphological variants of the odontoid process were rather common, especially in patients with BI, which may have an adverse effect on accurate measurement of CDA, as supported by the lowest inter-observer agreement in the present study.

The CPA proposed in this study is a naturally existing angle formed by the hard palate line and the Wackenheim clivus line. Interestingly, although the hard palate line has been used in the Bull angle and modified Omega angle and the Wackenheim line has been widely used to evaluate CVJ abnormalities, the CPA, formed at the intersection of these two lines, has not been proposed for use in this regard before [9, 15, 18‐21]. Our study demonstrated that CPA could accurately reflect the changes of craniovertebral angle. Of note, the AUC of CPA in diagnosing BI was similar to that of CXA or CDA. Furthermore, CPA exhibited the following peculiar advantages. First, both the hard palate line and Wackenheim line are clearly depicted as a fine line on midsagittal MRI T1-weighted image (T1WI) or T2-weighted image (T2WI), which facilitates identification and measurement and thus leads to higher inter-observer agreement. Second, congenital malformations (e.g., os odontoideum), fracture or dislocation of the dens, osteosclerosis of the posterior margin of the axis, and ossification of ligaments can affect measurements of CXA and CDA [8, 10]. Also, CXA or CDA was affected by the curvature of cervical spine [6, 11]. In contrast, the osseous basis of the hard palate is the palatine process of maxilla and the horizontal plate of the palatine bone, which are stable and unaffected by the aforementioned factors. Moreover, facial hypoplasia or hard palate abnormality is rare and will not affect the measurements [19]. In short, CPA exhibited less variation when compared with CXA and CDA.

Moreover, any single parameter varies within a normal range, which necessitates assessing at least two different measures for patients with suspected BI [12]. The present study demonstrated that sensitivity, specificity, and accuracy in diagnosing BI were significantly improved when CPA + CDA or CPA + CXA was used. The mechanism underlying this phenomenon is unclear. CPA may reflect BI differently from CDA or CXA. CPA seems to mainly reflect the degree of clivus tilt toward the horizontal plane because the hard palate is stable and unaffected by the relative movement of the head and the cervical spine in the setting of BI [19]. In contrast, CXA (or CDA) is influenced by a great variety of accompanied anomalies affecting this region, such as C2–3 assimilation, atlas-axis subluxation, and atlas occipitalization [6, 11, 13, 22].

Our study has several limitations. This is a retrospective study based on MRI, where CT data were not available in many cases. This may affect the measurement of CL, CDA, and CXA because MRI is not the best choice to depict bone structure [23]. However, we identified bone cortex as best as we could and our results are highly comparable to other published reports [13, 17]. Further, this study did not examine the relations of CPA with other indicators (e.g., McRae line, Klaus index, and Welcher basal angle) or with clinical manifestations, which shall be addressed in future studies [11, 20]. In addition, future studies should address the value of CPA in evaluating treatment efficacy of BI.

In conclusion, on midsagittal MRI T1WI or T2WI of the head or neck, CPA accurately reflected the changes of craniovertebral angle with higher inter-observer agreement than that of CXA or CDA. Patients with BI had a sharper CPA compared to control subjects. CPA showed similar diagnostic performance with CXA or CDA with the cutoff value being 53.5° and provided additional diagnostic value to CXA and CDA for BI. Therefore, CPA can serve as a useful sentinel to alert the radiologist and surgeon to the possibility of CVJ deformity.