Exploring the Pathogenesis of Atlanto-Occipital Instability in Chiari Malformation With Type II Basilar Invagination: A Systematic Morphological Study

Abstract

BACKGROUND: 

Our previous study suggested that atlanto-occipital instability (AOI) is common in patients with type II basilar invagination (II-BI).

OBJECTIVE: 

To further understand the pathogenesis of AOI in Chiari malformations (CM) and CM + II-BI through systematic measurements of the bone structure surrounding the craniocervical junction.

METHODS: 

Computed tomography data from 185 adults (80 controls, 63 CM, and 42 CM + II-BI) were collected, and geometric models were established for parameter measurement. Canonical correlation analysis was used to evaluate the morphological and positional relationships of the atlanto-occipital joint (AOJ).

RESULTS: 

Among the 3 groups, the length and height of the condyle and superior portion of the lateral masses of the atlas (C1-LM) were smallest in CM + II-BI cases; the AOJ had the shallowest depth and the lowest curvature in the same group. AOJs were divided into 3 morphological types: type I, the typical ball-and-socket joint, mainly in the control group (100%); type II, the shallower joint, mainly in the CM group (92.9%); and type III, the abnormal flat-tilt joint, mainly in the CM + II-BI group (89.3%). Kinematic computed tomography revealed AOI in all III-AOJs (100%) and some II-AOJs (1.5%) but not in type I-AOJs (0%). Morphological parameters of the superior portion of C1-LM positively correlated with those of C0 and the clivus and significantly correlated with AOI.

CONCLUSION: 

Dysplasia of the condyle and superior portion of C1-LM exists in both CM and II-BI cases yet is more obvious in type II-BI. Unstable movement caused by AOJ deformation is another pathogenic factor in patients with CM + II-BI.

ABBREVIATIONS:

AAD

atlantoaxial dislocation

AOI

atlanto-occipital instability

AOJ

atlanto-occipital joint

BI

basilar invagination

C0

occipital condyle

C1-LM

lateral masses of the atlas

CCA

canonical correlation analysis

CM

Chiari malformation

CVJ

craniocervical junction

CXA

clivo-axial angle

IAS

inferior articular surface

PCF

posterior cranial fossa

SAS

superior articular surface.

Basilar invagination (BI) and Chiari malformations (CM) are the main malformations of the craniocervical junction (CVJ) in adults.¹ In Goel classification, the presence (type I, I-BI) or absence (type II, II-BI) of mechanical instability of the atlantoaxial dislocation (AAD) was used to divide all cases with BI into 2 groups.^2-4 Cases of CM coexisting with type II-BI (CM + II-BI) are generally considered to be unrelated to mechanical instability and attributed to overcrowding of nerve structures in the small posterior cranial fossa (PCF). Foramen magnum decompression is the primary treatment.^2,3,5-7 However, in 2015, Goel proposed the concept of “central AAD” and defined it as a common feature in all patients with CM or type II-BI; thus, fusion surgery should be performed.^8-10 Nevertheless, many authors argue that such AAD cannot be identified using standard radiological parameters^7,11 and that shallow PCFs are well-documented in patients with CM.^12,13 The fact that no treatment strategy has found universal acceptance indicates that the pathogenesis of CM and II-BI, and whether CVJ instability exists, is still unclear.

Our initial study¹⁴ found that CVJ instability was common in patients with CM + II-BI, mainly manifesting as excessive movement of the atlanto-occipital joint (AOJ) and asymmetrical movement of the atlantoaxial joint. Without exception, all previous publications focused on the pathogenesis of atlantoaxial instability,^8,15 while the main reason for atlanto-occipital instability (AOI) remains unknown. Therefore, in this study, we performed 3-dimensional geometric reconstruction and systematic measurements to compare the structural differences of the PCF, occipital condyle (C0), lateral masses of the atlas (C1-LM), and axis in patients with CM alone and CM + II-BI. Changes in these structures were analyzed to explore the pathogenesis of the disease and to better guide the choice of surgical methods.

METHODS

Patient Enrollment

The inclusion criteria for the controls and patients with CM were in accordance with our previous article.¹⁴ The patient flowchart is presented in Supplemental Figure 1, https://links.lww.com/NEU/D534. This retrospective study was approved by the Institutional Review Board of Nanfang Hospital, Southern Medical University. Kinematic computed tomography (CT) data were prospectively collected with the patients' informed consent and ethical approval (NFEC-202006-K3).

Geometrical Model

GE Revolution CT machine (GE Healthcare; slice thickness, 0.65 mm) was used to obtain the cervical CT and kinematic data using the methods previously described.¹⁴ The data were saved in Digital Imaging and Communications in Medicine format and imported into Mimics software (v.20.0, Materialize) for morphometric measurements. We established clivus, supraocciput, C0, C1-LM, and axis measurements through 3-dimensional geometric modeling. To ensure comparability among individuals, all samples were reconstructed and resliced in the Frankfort horizontal plane.

Systematic Morphological Measurement

Systematic morphological measurements were performed on clivus, supraocciput, C0, C1-LM, and axis. The C1-LM was divided into superior and inferior portions according to their embryological origin (Figures 1 and 2).¹⁶ The AOI-related parameters, such as C0-1 distance, C0-1 tilt angle, and clivo-axial (CXA) angle, were measured using kinematic CT (Supplemental Figure 2, https://links.lww.com/NEU/D535), as previously reported.¹⁴ All measurements were performed separately by 2 blinded raters and 2 blinded reviewers and repeated after 1 month. Disagreements were resolved by consensus.

Systematic measurement of the morphology of the occipital condyle. A and B, Occipital condyle measurement points. Points a and b correspond to the anterior and posterior end points of the longest axis in the anteroposterior orientation, respectively. Points c and d indicate the anterior and posterior intersections of the vertex line of the superior articular surface on the lateral masses of the atlas and occipital condyle, respectively. Points e and f are the upper and lower end points of the longest axis in the vertical orientation, respectively. The following point distances were measured: a–b (superior length), c–d (inferior length), e–f (middle height), a–c (anterior height), b–d (posterior height), a–d (front hypotenuse), and b–c (back hypotenuse). C-I, Morphometric parameters of the occipital condyle in the 3 types of AOJ. The length, height, and hypotenuse of the condyle in CM + II-BI cases are significantly lower than those in the other 2 groups. Contrastingly, there is no significant difference between types II-AOJ and I-AOJ. ***P < .001; **P < .01; *P < .05; “#” indicate P > .05. AOJ, atlanto-occipital joint; CM + II-BI, Chiari malformations which coexistence with type II basilar invagination.

Systematic measurement of the morphology of the superior and inferior portions of the lateral masses of the atlas. A-I, Systematic measurement and morphometric parameters of the morphology of the superior portion of the lateral masses of the atlas. A-C, Lateral masses of the atlas measurement points. The g and h points are the anterior and posterior edges of the SAS on the lateral masses of the atlas; i and j are the anterior and posterior intersections of the extension line of the upper edge of the posterior arch and the lateral masses, respectively; k and l are the anterior and posterior edges of the IAS, respectively; m and n are the lowest point of the SAS and the midpoint of the IAS, respectively; o is the intersection of the g–h line and g–h vertical line passing through point m. Based on Menezes report,¹⁶ the blue and green areas in pattern diagrams of the atlanto-occipital joint represent different embryonic origins. The lateral masses of the atlas are divided into superior (green) and inferior (blue) portions. The following point distances are measured to describe the morphological characteristics of the superior portion of the lateral masses of the atlas: g–h (superior length), i–j (inferior length), g–i (anterior height), h–j (posterior height), and m–o (depth of the SAS). The atlas tilt angle is the angle between the g–h extension line and the k–i extension line in the neutral position. D-I, The curvature of the atlanto-occipital joint is calculated by dividing the depth (m-o) by the length (g-h). In Chiari malformation coexistence with type II basilar invagination (CM + II-BI) cases, the length and height of the superior portion of the lateral masses of the atlas are smaller than those of the other 2 types, and the atlas tilt angle is larger. There are significant differences in curvature among the 3 groups (control > CM > CM + II-BI). J-N, Systematic measurement and morphometric parameters of the morphology of the inferior portion of the lateral masses of the atlas. J, The following point distances are measured to describe the morphological characteristics of the inferior portion of the lateral masses of the atlas: i–j (superior length), k–l (inferior length), i–k (anterior height), and j–l (posterior height). K-N, Morphological parameters of the inferior portion of the lateral masses of the atlas are not significantly different among the 3 types, except for the superior length. ***P < .001; **P < .01; *P < .05; “#” indicate P > .05. CM + II-BI, Chiari malformation coexistence with type II basilar invagination; IAS, inferior articular surface; SAS, superior articular surfaces.

Cluster Analysis of the AOJ Morphological Data

To classify AOJ morphology, Z-score normalization was performed on the major measurement data of 370 AOJs in 185 subjects, followed by cluster visualization analysis using “ComplexHeatmap” and “clusterProfiler” software packages in R (v.4.2.0, R Foundation for Statistical Computing). Major AOJ-related morphometric data were compared between groups. Significant differences indicated that the classification was feasible. Receiver operating characteristic curves were prepared to assess the usefulness of target parameters as predictors of AOJ morphological typing.

Statistical Analysis

Statistical analysis was performed using the SPSS software (v.26, IBM Corp.). Craniometric data are presented as mean ± SD. Age was compared using the Wilcoxon rank-sum test, and sex was compared using the χ² test. After the analysis of normal distribution, a 1-way analysis of variance and a least significant difference were used to compare the parameters among the groups. A paired t test (2-tailed) was used to compare the parameters between the left and right sides. In addition, we conducted an intraclass correlation coefficient analysis of the 2 measurers and the same measurer at different time points (interobserver and intraobserver reliability).

We used the CANONICAL CORRELATION (CANCORR) program in SAS 9.4 software (SAS Institute Inc.) to perform canonical correlation analysis (CCA) to determine the strength of association between the morphological and positional relationship of the AOJ.¹⁷ For the output results, the first canonical correlation variable was taken and the value of the first canonical correlation coefficient was discussed to judge the correlation.

RESULTS

Study Population

The medical records of 105 adult patients with CM from January 2016 to May 2022 (63 without BI [CM group] and 42 with type II BI [CM + II-BI group]) were analyzed. In addition, 80 adult patients without CM who underwent cervical CT scans for mild cervical spine diseases were enrolled in the control group (Supplemental Figure 1, https://links.lww.com/NEU/D534). Notably, the series of kinematic CT data (extension, neutral, and flexion positions) of 112 cases (48 control, 37 CM, and 27 CM + II-BI) among the total patients were also analyzed.

Baseline Characteristics

The proportion of female patients in the control group, CM group, and CM + II-BI group was 48.8%, 63.5%, and 59.5%, respectively (χ², P = .19). The mean age of patients in the control group, CM group, and CM + II-BI group was 45.0 ± 13.3, 43.1 ± 11.8, and 41.0 ± 12.7 years, respectively (Wilcoxon rank-sum test, P = .26).

Intraobserver and Interobserver Reliabilities

The intraclass correlation coefficient of 2 measurers at all data ranged from 0.93 to 0.98 (P < .01), while the intraclass correlation coefficient of the same measurer at 2 time points ranged from 0.92 to 0.96 (P < .01), indicating good data consistency.

Atlanto-Condyle Morphological Diversity

The length and height of C0 and C1-LM in the CM + II-BI group were significantly lower than those in the other 2 groups (Figure 1 and Supplemental Figure 3, https://links.lww.com/NEU/D536). In particular, the superior portion of C1-LM showed significant morphological differences in patients with CM + II-BI, who had the shallowest depth and the lowest curvature of the AOJ, followed by the CM and control groups. No significant differences were noted between parameters on the left and right sides. However, the parameters of the inferior portion of the C1-LM were similar among the 3 groups (Figure2). Other parameters, such as clivus length, supraocciput length, and CXA, were significantly different (Supplemental Table 1, https://links.lww.com/NEU/D537).

Classification of AOJ

The heat map shows that all AOJ-related morphological data can be divided into 3 clusters, which indicated 3 different shape types. Most parameters of type III (III-AOJ) differed from those of the other 2 types, but only the curvature and depth of the AOJ were different among the 3 types, suggesting that they may be used as potential indicators to predict the AOJ types. Further testing revealed that the best practical cut-off points for AOJ depth and curvature were 3.69 mm and 0.22, respectively, in type II (II-AOJ). In type III-AOJ, these 2 points were 2.71 mm and 0.13, respectively (Figure 3).

Classification of the AOJ based on morphological features. A, Visualization of main morphological measurement data of the 3 types of AOJ. Heat map showing that all morphological data can be divided into 3 clusters corresponding to 3 morphological types. The depth and curvature of the bilateral atlanto-occipital joint are different in the 3 types (red dotted box). B-F, Comparison of the morphological measurement data of the 3 types of AOJ. The curvature and depth of AOJ are different among the 3 types. G-J, ROC curve of AOJ curvature and length. G and H, ROC curve of AOJ curvature for the prediction of types II-AOJ and III-AOJ. The best cut-off point of AOJ curvature for type II-AOJ is 0.22, with a sensitivity of 78.0%, specificity of 78.1%, and AUC of 0.80. The best cut-off point of AOJ curvature for type III-AOJ is 0.13, with a sensitivity of 81.3%, specificity of 76.9%, and AUC of 0.84. I and J, ROC curve of AOJ depth for the prediction of types II-AOJ and III-AOJ. The best cut-off point of AOJ depth for type II-AOJ is 3.69 mm, with a sensitivity of 79.3%, specificity of 70.3%, and AUC of 0.79. The best cut-off point of AOJ depth for type III-AOJ is 2.71 mm, with a sensitivity of 62.5%, specificity of 76.9%, and AUC of 0.75. AOJ, atlanto-occipital joint; AUC, area under the curve; ROC, receiver operating characteristic.

By comparing all AOJ geometric models, the morphology of the 3 types could be visually distinguished (Figure 4). Type I (I-AOJ) showed a typical ball-and-socket structure. In type II-AOJ, the superior articular surface (SAS) of C1-LM was slightly abnormal, but the ball-and-socket shape was still identifiable. Type III-AOJ showed a more severe deformity; the ball-and-socket shape disappeared completely. The anterior and especially the posterior protuberances in the AOJ became flatter. The facet tilt angle markedly increased, and a distinctly flat joint was identified. Among the 3 groups, type I-AOJ was the main shape of the AOJ in the control group (100.0%), II-AOJ in the CM group (92.9%), and III-AOJ in the CM + II-BI group (89.3%) (Supplemental Table 2, https://links.lww.com/NEU/D538).

Three-dimensional geometric models of the occipital condyle and atlas lateral masses. A, Three-dimensional model of the I-AOJ, which has a strong left-right symmetry (i and ii). Viewed from the top and side, the SAS of the lateral masses of the atlas (C1-LM) has obvious anterior and posterior protuberances (red dotted line), which correspond to the anterior and posterior edges of the SAS in the sagittal computed tomography image (points g and h in Figure 2). The middle region (yellow dotted line) of the SAS is obviously depressed. Therefore, the entire articular surface exhibits a concave structure and forms a ball-and-socket joint with C0, which is approximately spherical in shape (ii and iii). Lateral view, three-dimensional model of C1-LM. The area is divided into superior and inferior portions according to their embryological origin (iv and v).¹⁶B, Three-dimensional model of type II-AOJ. C0 is basically normal in shape and resembles a spherical protuberance (i). The SAS is slightly abnormal, showing that the middle depression (yellow dotted line) is slightly shallow and the articular surface is slightly flat (ii and iii). However, the anterior and posterior protrusions have developed normally (red dotted line), and the difference in height between the 2 protrusions and the atlas tilt angle has not significantly changed (iv and v). Generally, a ball-and-socket joint can still be recognized. C, Three-dimensional model of type III-AOJ. Compared with type II-AOJ, type III-AOJ shows severe deformity. C0 is abnormally thin and the spherical protuberance at the bottom becomes flat (i). C1-LM is more deformed, and the depth and curvature of the SAS are further reduced (ii). The anterior and posterior protrusions are dysplastic, and the posterior edge of the SAS has almost disappeared (iii). The height difference between these 2 protrusions and the C1 tilt angle has increased. The atlas tilt angle in the type III-AOJ group is significantly higher than that in the type I-AOJ and II-AOJ groups (iv). The classic ball-and-socket joint has become distinctly flat. Among the 3 types, the shape of the inferior portion of the lateral masses of the atlas is basically similar, but the shape of the superior portions of the lateral masses of the atlas in type III-AOJ is abnormal, especially the atlas tilt angle, which is III-AOJ > II-AOJ ≈ I-AOJ (v). I-AOJ, type I atlanto-occipital joint; SAS, superior articular surfaces.

Atlanto-Occipital Stability Analysis

As reported by Klimo et al,¹⁸ dynamic CT imaging in flexion and extension positions showed that the change in the C0-1 distance was more than 1.0 mm, indicating the presence of AOI. AOI was found in 6 patients (6/37, 16.2%) with CM alone and 26 patients (26/27, 96.3%) with CM + II-BI. The former was unilateral instability, whereas the latter was mostly bilateral instability (25/27, 92.6%). None of the control participants exhibited instability (Table 1).

TABLE 1. - AOJ Stability in the 3 Groups and the 3 Morphological Types (n/%) Based on the Kinematic CT Data

Side (Number)	Control (n = 48)	CM (n = 37)			CM + II-BI (n = 27)
	I-AOJ	I-AOJ	II-AOJ	III-AOJ	I-AOJ	II-AOJ	III-AOJ
Left side
Number a	48 (100%)	1 (2.7%)	33 (89.2%)	3 (8.1%)	1 (3.7%)	0 (0%)	26 (96.3%)
Unstable case b	0 (0%)	0 (0%)	0 (0%)	3 (100%)	0 (0%)	0 (0%)	24 (100%)
Right side
Number a	48 (100%)	3 (8.1%)	32 (86.5%)	2 (5.4%)	0 (0%)	2 (7.4%)	25 (92.6%)
Unstable case b	0 (0%)	0 (0%)	1 (3.1%)	2 (100%)	0 (0%)	0 (0%)	25 (100%)
Both sides
Number a	48 (100%)	0 (0%)	28 (75.7%)	0 (0%)	0 (0%)	0 (0%)	25 (92.6%)
Unstable case b	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	25 (100%)
Total
Number a	96 (100%)	4 (5.4%)	65 (87.8%)	5 (6.8%)	1 (1.9%)	2 (3.7%)	51 (94.4%)
Unstable case b	0 (0%)	0 (0%)	1 (1.4%)	5 (100%)	0 (0%)	0 (0%)	51 (100%)

AOJ, atlanto-occipital joint; CT, computed tomography.

^aRepresents the number and proportion of each AOJ type in the 3 groups (Control, CM, and CM + II-BI).

^bRepresents the number and proportion of unstable AOJ in different AOJ morphologic types.

Note: As reported by Klimo et al,¹⁸ dynamic CT imaging in flexion and extension position shows that the change value of C0-1 distance is more than 1.0 mm, indicating that there is pathological translation activity between the atlanto-occipital joint (AOJ) and the loss of normal contraposition relationship of articular surface, which is the natural outcome and evidence of potential AOJ instability. Based on this criterion, as long as there is a C0-1 distance change value of more than 1.0 mm, it is considered that there is an unstable AOJ.

Among the parameters related to AOI (CXA change, C0-1 tilt angle change, and C0-1 distance change), type III-AOJ was significantly higher than the other 2 types (Supplemental Table 3, https://links.lww.com/NEU/D539). The geometric model also showed that patients with type III-AOJ had unstable movement. AOI existed in all type III-AOJs (56/56, 100%) and rarely occurred in type II-AOJ (1/67, 1.5%) and never in type I-AOJ (0/101, 0%). Notably, such unstable movements in the type III-AOJ can occur in either head flexion or extension, but mainly in the former (Figure 5 and Table 2).

Kinematic CT and corresponding geometric modeling images of the 3 AOJ morphological types. A and B, CT (i) and corresponding geometric modeling image (ii–iv) of type I-AOJ in the head flexion position A and head extension position B. To better display the positional relationship between the occipital condyle and the lateral masses of the atlas, we moved the occipital condyle upward along the vertical direction of the AOJ (iii). The anterior margins of the occipital condyle and the lateral masses of the atlas overlap, and there is no obvious displacement between them (green dotted line). In the midsagittal section of the AOJ, the stability can be identified more clearly (iv). C and D, CT (i) and corresponding geometric modeling image (ii-iv) of a type II-AOJ in the head flexion position C and head extension position D. As indicated by three-dimensional reconstruction of the AOJ (ii and iii), the anterior margins of the occipital condyle and the lateral masses of the atlas overlap with no obvious displacement between them (green dotted line). It can also be seen in the midsagittal profile of the AOJ (iv). E-H, CT (i) and corresponding geometric modeling image (ii-iv) of type III-AOJ in the head flexion position E and G and head extension position F and H. The unstable movement can be recognized in both head flexion E and F and head extension G and H, but mainly in the former (54/56, 96.4%) and only a small part in the latter (2/56, 3.6%). A three-dimensional model (ii and iii) indicates that the atlanto-occipital joint facet tilts backward and the anterior margin of the occipital condyle moves backward significantly relative to the anterior margins of the lateral masses of the atlas (green dotted line). Such ultramovement can be identified more clearly in the midsagittal profile of the AOJ (iv). AOJ, atlanto-occipital joint; CT, computed tomography.

TABLE 2. - AOJ Stability of the 3 Morphological Types in All Cases (n/%) Based on the Kinematic CT Data

Summary	I-AOJ	II-AOJ	III-AOJ
			Flexion position c	Extension position d
Number a	101 (101/224, 45.1%)	67 (67/224, 29.9%)	54 (54/224, 24.1%)	2 (2/224, 0.9%)
Unstable case b	0 (0/101, 0%)	1 (1/67, 1.5%)	54 (54/54, 100%)	2 (2/2, 100%)

AOJ, atlanto-occipital joint; CT, computed tomography.

^aRepresents the number and proportion of each AOJ type in all cases.

^bRepresents the number and proportion of unstable AOJ in different AOJ morphologic types.

^cRepresents that the unstable movement of the AOJ mainly occurs in the head flexion position.

^dRepresents that the unstable movement of the AOJ mainly occurs in the head extension position.

Canonical Correlation Analysis

CCA analysis suggested a positive correlation between the parameters of C0 and the superior portion of C1-LM and clivus length (Table 3). The bone morphology of the superior portion of the C1-LM is related to AOI. However, the parameters of the inferior portion and condyle did not affect such instability, although obvious dysplasia of the condyle was found. Further analysis showed that the tilt angle and curvature of the SAS of C1-LM were closely related to AOI, and the correlation of the former was stronger.

TABLE 3. - Results of Canonical Correlation Analysis

First set of variables	Second set of variables	First pair of canonical correlation coefficients	Adjusted first pair of canonical correlation coefficients	Correlation strength	Cumulative contribution rate	Significance
C0 morphology	Clivus length	0.666	0.631	Strong	1.000	P < .0001
C1 morphology	Clivus length	0.654	0.596	Moderate	1.000	P < .0001
C0 morphology	C1 morphology	0.880	0.851	Very strong	0.539	P < .0001
C0 morphology	C1 superior portion	0.833	0.809	Very strong	0.724	P < .0001
C1 superior portion	Clivus length	0.575	0.554	Moderate	1.000	P < .0001
C1 inferior portion	Clivus length	0.271	0.224	Weak	1.000	P = .035
C1 inferior portion	C0 morphology	0.380	0.222	Absent a	0.765	P = .661
Curvature of AOJ	Clivus length	0.498	0.495	Moderate	1.000	P < .0001
Curvature of AOJ	C0 morphology	0.619	0.581	Moderate	0.874	P < .0001
C0 morphology	AOJ stability	0.566	0.418	Absent a	0.406	P = .182
C1 superior portion	AOJ stability	0.616	0.555	Moderate	0.613	P = .0003
C1 inferior portion	AOJ stability	0.399	0.239	Absent a	0.419	P = .092
C1 TR	AOJ stability	0.705	0.680	Strong	0.768	P < .0001
Curvature of AOJ	AOJ stability	0.488	0.470	Moderate	0.852	P < .0001

AOJ, Atlanto-condyle joint; C0, Occipital bone; C1, Atlas; C2, Axis; C1 TR, atlas tilt angle, the angle between the superior articular surfaces and the inferior articular surfaces of atlas at neutral position; Curvature of AOJ, which was calculated by dividing the depth of AOJ by the length of AOJ.

^aAs the P value is greater than 0.05, the 2 variables are not considered to be correlated.

Variables of C0 morphology include all the parameters presented in Figure 1.

Variables of C1 morphology include all the parameters presented in Supplemental Figure 3, https://links.lww.com/NEU/D536.

Variables of C1 superior portion include superior length, inferior length, anterior height, and posterior height presented in Figure 3B.

Variables of C1 inferior portion include superior length, inferior length, anterior height, and posterior height presented in Figure 3J.

Variables of AOJ stability include all the parameters listed in Table 1.

Supplementary note: This study systematically measured the morphological data of the AOJ, and each variable (such as C0 morphology and C1 morphology) contained multiple original data. Canonical correlation analysis can be used to extract typical variables that can effectively reflect the original data information. Finally, a number of canonical variables that effectively reflect the morphological characteristics of the C0 and C1 lateral masses were extracted to analyze their correlations.

DISCUSSION

Surgical treatment of CM has always been controversial, and no treatment strategy has been universally accepted. In particular, it remains unclear whether CVJ instability truly exists in these patients. AAD has been identified in type I-BI, while little attention has been paid to the morphological characteristics of the CVJ in type II-BI. Our previous study showed that excessive movement of AOJ was common in patients with CM + II-BI.¹⁴ In this study, we further investigated the pathogenesis of such instability and found that patients with CM with or without type II-BI had bony dysplasia at the CVJ, but to different degrees. After systematic measurement, the length and height of C0 and the curvature of AOJ in normal subjects were within the previously reported range.^19-22 On the other hand, most patients with CM alone have deformities, including clivus dysplasia, a short supraoccipital bone, and flat AOJ (type II-AOJ), although a ball-and-socket joint was still identified. However, when CM coexisted with type II-BI, the degree of CVJ deformity was significantly aggravated, which was reflected in the flat articular surface of the AOJ (type III-AOJ). Moreover, type III-AOJ is closely related to AOI as indicated by the statistical results (Table 1). Conversely, although atlantoaxial instability does exist in patients with CM + II-BI, as reported in our initial study,¹⁴ it mainly manifests as left-right asymmetric motion and rotational instability. These patients do not have abnormalities of atlanto-dental interval and axis bone structure (Supplemental Figure 4, https://links.lww.com/NEU/D540). In addition, the atlantoaxial joint is known to be the most active joint in the human body; we believe that this kind of atlantoaxial asymmetric movement is likely to be secondary to the AOI. The bone malformations in CVJ are quite complicated, such as unilateral or bilateral atlas assimilation, or Klippel-Feil syndrome, which may have individually varying bone development. It is difficult to quantitatively analyze uniformly; hence, such cases of malformation were excluded from this study.

Embryological Development of AOJ in Patients With CM and Type II-BI

Embryologically, the clivus and supraoccipital bone are derived from the axial component of the first 3 occipital sclerotomes, and C0 is formed from the lateral component of the proatlas.^16,23 However, controversy exists regarding the embryonic origin of the C1-LM. Pang and Thompson²³ suggested that the lateral component of the first cervical sclerotome eventually forms the entire C1-LM; Menezes¹⁶ believed that the C1-LM is divided into 2 parts (bounded by the posterior arch of C1), with the superior portion originating from the lateral part of the proatlas and the inferior portion developing from the lateral component of the first cervical sclerotome. However, neither author has provided strong evidence. In this article, we intentionally divided C1-LM into superior and inferior portions, as previously suggested,¹⁶ and performed separate morphological measurements. The superior portion of C1-LM showed different degrees of morphological deformity in the patients with CM and CM + II-BI. According to CCA, C0 morphology strongly correlates with the clivus and the superior portion of C1-LM, but not with the inferior portion. To some extent, this indicates that the C0 and the superior portion of C1-LM may have the same developmental pattern. In both CM and type II-BI, we believe that abnormal embryonic development of the axial components of the first 3 occipital sclerotomes and the lateral component of the proatlas coexist and a continuum of mesodermal malformations exists between them. Type II-BI has more intense mesodermal involvement.

Possible Pathogenesis of CM and Type II-BI and Its Proper Surgical Strategy

The stability of the AOJ is mainly dependent on the geometry of the osseous structure.^16,24 According to kinematic CT imaging, AOI exists in all type III-AOJs and rarely occurs in type II-AOJ and never in type I-AOJ. CCA revealed that AOI was mainly related to the bone deformity of the superior portion of the C1-LM, especially abnormalities in its tilt angle (main factor) and C1 curvature (secondary factor). Therefore, it can be inferred that despite the flat articular surface of the CM (type II-AOJ), the C1 tilt angle did not increase and the anterior and posterior protrusions could still limit the anterior and posterior translation of C0, preventing AOI. In patients with CM + II-B, AOJ morphology changed from type II to III, increasing the C1 tilt angle and dysplastic anterior and posterior protrusions (especially when posterior protrusions were absent), ultimately promoting AOI (Figure 6). The literature suggests that the morphological characteristics of type III-AOJ are common in children with immature bones. As bones develop, C0 volume gradually increases and the joint fossa becomes deeper, resulting in a stable ball-and-socket joint.^18,25,26 Accordingly, it is speculated that patients with CM + II-BI might have gradually developed malformations during growth. At the initial stage, the flat deformity of the AOJ and the abnormal increase in the C1 tilt angle gradually cause the static C0 to slide posterior and inferior on the SAS of C1-LM. The joints are constantly displaced, and the ligaments gradually become tired, damaged, and denatured, eventually resulting in AOI. The characteristics of long-term and slow progression also explain the delay in the appearance of neurological symptoms and the age heterogeneity in patients with type II-BI.

FIGURE 6.: Mechanism of AOI caused by bone malformations. A-C, In the AOJ, the SAS of the lateral masses of the atlas (C1-LM) presents concave structures and form a “ball-and-socket” joint with the occipital condyles, which are approximately spherical. Their unique articular structure determines that the major movements of the AOJ are flexion and extension and a smaller range of rotation and lateral flexion. Because the anterior and posterior protuberances on the AOJ prevent the occipital condyle from moving forward and backward horizontally, the AOJ is almost unable to perform translational activities. D-F, Although the articular surface of the type II-AOJ becomes flat, the C1 tilt angle does not increase. The anterior and posterior protrusions still exist and can play the function of limiting the anterior and posterior translations of C0. Therefore, the AOJ is still unable to perform the translational activity nor is there AOI. G-I, In type III-AOJ, the AOJ morphology is abnormal, the posterior protuberances of the SAS on the lateral masses of the atlas almost disappear, the atlas tilt angle increases, and the occipital condyle slides backward and downward on the SAS of the atlas. At this point, there is pathological translation activity between the AOJ and loss of the normal contraposition relationship of the articular surface, which is the natural outcome and evidence of potential AOI. AOI, atlanto-occipital instability; AOJ, atlanto-occipital joint; SAS, superior articular surfaces.

Regarding contributors of the disease, CM + II-BI may be derived from the integration of a small PCF and an unstable AOJ. Moreover, the underlying cause is bony malformations of the CVJ. Shortened supraoccipital length causes inadequate posterior fossa volume and contributes to the occurrence of CMs. Combined with the thin C0 and short clivus, the opisthion of the skull is extended inferiorly to the axis vertebra, eventually forming a basilar depression. The relative superior movement of the dens axis further compromises the space available for the CVJ and aggravates the existing cerebellar tonsil herniation. Notably, with flexion and extension of the head, the secondary AOI results in the progressive retroinferior deviation of the C0 on the SAS of the C1-LM, a process similar to spondylolisthesis in the lumbosacral spine. This reduces the space of the PCF, resulting in further compression of the spinal cord by the cerebellar tonsil hernia. Therefore, treatment approaches should pay equal attention to decompression of the PCF and reconstruction of the atlanto-occipital stability. Previously, either method alone has often produced unsatisfactory results, highlighting the need to apply them in combination.^27-30 We have retrospectively analyzed the surgical outcomes of 17 cases of CM + II-BI, and the results confirm that foramen magnum decompression, atlantoaxial facet distraction conducted with angled spacers (high before low) implantation, and occipitocervical fusion performed with cantilever techniques were shown to be beneficial for such patients.¹⁴

Limitations

This study had several limitations. First, this was a single-center study, and different ethnic groups may have different CVJ bone structures. Second, undoubtedly ligaments, muscles, and other surrounding tissues still play an indispensable role in maintaining the stability of the AOJ. This study only evaluated the influence of the atlanto-occipital bone structure on stability, ignoring the role of the above tissues. In future, more experimental techniques, such as cadaver studies and high-quality dynamic finite element models, can be used to further verify the mechanism of AOI and the role of each tissue.

CONCLUSION

Accompanying the underdevelopment of the clivus and supraocciput, dysplasia of the condyle, and superior portion of C1-LM exists in both CM and type II-BI cases; however, it is more obvious in type II-BI as the anatomic morphology of the AOJ changes from a classic ball-and-socket joint to a flat-tilt joint. Consequently, the AOJ becomes unstable because of the increased tilt angle and decreased curvature of the joint, which is another pathogenic factor in patients with CM + II-BI in addition to the reduction of posterior fossa volume.