Background: The use of a transoral atlantoaxial reduction plate (TARP) system is an effective surgical approach for the treatment of basilar invagination. With the aim of improving the therapeutic efficacy of the TARP operation, we conducted a voxel-based morphometric study to quantitatively investigate the descent of the odontoid process and craniocervical volume changes.
Methods: We enrolled 20 patients with basilar invagination who underwent a TARP procedure. Craniocervical computed tomography (CT) scanning and a 3-dimensional (3-D) reconstruction of the craniocervical junction were performed. Craniocervical volumes and odontoid process descent distances were measured preoperatively and postoperatively. Individual neurological function was evaluated according to the Japanese Orthopaedic Association (JOA) scoring system for cervical disorders. Pearson correlation analysis was applied for statistical testing.
Results: Surgical efficacy (the JOA-score improvement rate) was significantly associated with the craniocervical volume improvement rate, the odontoid descent distance, and the absolute craniocervical volume changes (p < 0.01 for all), with correlation coefficients (r) of 0.83, 0.80, and 0.61, respectively. No significant correlation was noted between surgical efficacy and age, symptom duration, preoperative neurological function, odontoid process displacement, or change in clivus-odontoid angle (p > 0.05). The craniocervical volume improvement rate was significantly associated with the odontoid descent distance (r = 0.8; p < 0.01), but it was not associated with the odontoid displacement or the change in the clivus-odontoid angle (p > 0.05).
Conclusions: We found that the odontoid descent distance predicted the craniocervical volume improvement rate following TARP procedures in patients with basilar invagination, and we believe that both can serve as predictors of surgical efficacy. We believe that planning the odontoid descent distance preoperatively may help to improve the efficacy of TARP operations.
Level of Evidence: Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.
1Department of Orthopedics, Guangzhou General Hospital of Guangzhou Military Command, Guangzhou, People’s Republic of China
2No. 303 Hospital of People’s Liberation Army, Nanning, People’s Republic of China
3Southern Medical University, Guangzhou, People’s Republic of China
4Institute of Traumatic Orthopaedics of People’s Liberation Army, Guangzhou, People’s Republic of China
E-mail address for H. Xia: email@example.com
* Gejin Wei, MD, Chenglong Shi, MD, and Zhiyun Wang, MD, PhD, contributed equally to the writing of this article.
Basilar invagination is a relatively common developmental anomaly of the craniocervical junction characterized by prolapse of the odontoid process into the foramen magnum. The condition may result in a decrease of the craniocervical volume, thus leading to spinal cord and brainstem (medullary) compression and severe neurological impairment. In serious cases with medullary involvement, persistent compression can cause respiratory disturbance, with high morbidity and mortality rates1. Basilar invagination is often associated with other osseous anomalies of the craniovertebral junction, with the anatomical complexity of the area rendering treatment challenging. Current therapeutic strategies include transoral odontoidectomy2-4; anterior transoral resection of the odontoid followed by posterior fusion5-7; and use of a transoral atlantoaxial reduction plate (TARP), a recently developed method for atlantoaxial anterior fixation in patients with complicated atlantoaxial dislocation induced by congenital disease, trauma, or rheumatoid arthritis8. The aim of all of these procedures is to expand the craniocervical volume and relieve symptoms via decompression. However, in our clinical experience, a considerable proportion of patients have had unsatisfactory neurological recovery following a TARP operation, with postoperative radiographic examinations confirming an unsatisfactory location of the odontoid process. In most of these cases, the distance of odontoid descent is insufficient for spinal cord decompression.
Voxel-based morphometry is an imaging analysis technique using a statistical approach to statistical parametric mapping. With the aim of improving the therapeutic efficacy of TARP procedures in patients with basilar invagination, we conducted a voxel-based morphometric study to quantitatively investigate odontoid process descent and craniocervical volume changes.
Materials and Methods
In this respective study, we enrolled a consecutive cohort of patients treated from August 2009 to August 2012. All patients were diagnosed with basilar invagination and underwent a TARP operation. We collected preoperative and postoperative clinical data from physical examinations, radiographs, craniocervical computed tomography (CT) scans, magnetic resonance imaging (MRI), and evaluations of neurological function according to the Japanese Orthopaedic Association (JOA) scoring system (maximum, 17 points) for cervical disorders. Exclusion criteria were (1) a previous posterior operation such as posterior decompression of the foramen magnum or decompression of the posterior arch of C1; (2) a cervical intraspinal or osseous tumor; and (3) severe osseous anomalies of the craniovertebral junction, hindering measurement.
Twenty patients were included in this morphometric study. There were 7 male and 13 female patients with a mean age of 37 years (range, 6 to 71 years) and an average preoperative duration of symptoms of 7 years (range, 2 months to 30 years). The most frequent clinical symptom was numbness (13 patients, 65%), followed by dystaxia (10 patients, 50%), occipital or neck pain (8 patients, 40%), and hemiparalysis (5 patients, 25%). Clinical data are summarized in Table I. The JOA score derived with the preoperative evaluations of neurological function averaged 9.3 points (range, 4 to 15 points). Surgical efficacy was assessed according to the JOA-score improvement rate, which was calculated with the following formula: (postoperative score − preoperative score)/(maximal normal score − preoperative score) × 100%; 100% was categorized as complete remission, 60% to 99% as substantial efficacy, 25% to 59% as effective, and <25% as ineffective9-11.
The study was approved by the ethical committee of the hospital, and all participants provided informed consent.
Measurement and Computational Method
Assessment of the craniocervical volume change obtained by displacement of the odontoid process primarily consisted of 2 portions: change in odontoid volume (Vodontoid, shown as the red section in Fig. 1-A)—i.e., the volume of the odontoid process over the McRae line (the line labeled “a” in Figs. 1-A and 1-B)—and change in spinal canal volume (Vspinal canal, shown as the green section in Figs. 1-A and 1-B) between the McRae line (line “a” in Figs. 1-A and 1-B) and the line at the peak of the axis spinous process and parallel to the axis end plate (line “c” in Figs. 1-A and 1-B). On the 3-dimensional (3-D) model of the craniocervical junction, we drew 2 parallel planes: Plane α, which is perpendicular to the anatomical sagittal plane and along the McRae line (line “a” in Figs. 1-A and 1-B), and Plane β, which is perpendicular to the anatomical sagittal plane and along the inferior margin of the axis (line “b” in Figs. 1-A and 1-B). Plane β was then moved up to Plane γ, which is along the axis of the spinous process peak (line “c” in Figs. 1-A and 1-B). Using voxel-based morphometrically computational methods, software analysis was applied to calculate the preoperative and postoperative odontoid volume (Vpreop odontoid and Vpostop odontoid) and the preoperative and postoperative spinal canal volume (Vpreop spinal canal and Vpostop spinal canal). The absolute craniocervical volume change (V) and the craniocervical volume improvement rate (V%) were calculated according to the following formulas:
The odontoid process displacement, including descent distance, ante-displacement distance, and rotation angle (see below), were evaluated on the preoperative and postoperative CT scans. The distance between the top of the odontoid process and the McRae line (ML), the distance of the transverse line from the tip of the basion to the posterior axial line of C2 (BAI), and the clivus canal angle (CCA) were measured. In addition, we calculated the odontoid process descent distance (postoperative ML − preoperative ML), the odontoid process ante-displacement distance (postoperative BAI − preoperative BAI), and the odontoid process rotation angle (postoperative CCA − preoperative CCA).
Three-Dimensional Reconstruction of the Craniocervical Junction
All patients were examined with a head and cervical thin-section CT scan (SOMATOM Definition CT 2012; Siemens dual-source CT) with the head and neck in a neutral supine position12. Scanning parameters consisted of voltage of 120 kV, current of 220 mA, slice thickness of 2.9 mm, scanning width of 1500 mm, and scanning length of 450 mm.
Sequential images of thin-section CT scans were imported into the Yorktal Bone CT Image Processing Software. Software modeling identified image regions via threshold segmentation and drew an outline of the spinal canal with a region growing method. A radiologist and an orthopaedic surgeon confirmed the boundary segmentation layer by layer on the axial, sagittal, and coronal images, and completed the 3-D reconstruction13,14. The spinal canal boundary included (1) the identified spinal canal clearance; (2) the osseous vertebral canal defined as the space between the posterior margin of the vertebral body and the anterior margin of the vertebral laminae; (3) the vertebral canal defined as the space between the posterior margin of the intervertebral disc and the anterior margin of the ligamentum flavum; (4) the lateral boundaries defined as the inner margins of the vertebral pedicles or, if radiographic images revealed a lateral-open spinal canal, as the inner margins of nerve roots15; and (5) the border of the odontoid over the McRae line, which defined the osseous boundary.
All data were expressed as the mean and standard deviation. SPSS 20.0 software (IBM) was used for statistical analyses. Spearman or Pearson correlation analysis was applied to assess the association between the JOA-score improvement rate and symptoms, age, symptom duration, preoperative JOA score, absolute craniocervical volume change, craniocervical volume improvement rate, odontoid descent, odontoid ante-displacement, and clivus-odontoid angle change as well as the association between the craniocervical volume improvement rate and the odontoid descent distance. Probability (p) values of ≤0.05 were considered significant.
The preoperative and postoperative morphometric measurements are summarized in Table II. Surgical efficacy based on the JOA-score improvement rate was significantly associated with the craniocervical volume improvement rate, the odontoid descent distance, and the absolute craniocervical volume change (all p < 0.01), with correlation coefficients (r) of 0.83, 0.80, and 0.61, respectively. No significant correlation was noted between the surgical efficacy and symptoms, age, symptom duration, preoperative JOA score, odontoid ante-displacement, or clivus-odontoid angle change (all p > 0.05). These statistical results are summarized in Tables I and III. The craniocervical volume improvement rate was significantly associated with the odontoid descent distance (r = 0.80; p < 0.01) but not with the odontoid ante-displacement or the clivus-odontoid angle change (p > 0.05). These statistical results are summarized in Table IV.
The transoral atlantoaxial reduction plate (TARP) procedure is an effective surgical approach for the treatment of ventral spinal cord compression induced by ascent of the odontoid process in patients with basilar invagination as it can directly relieve the compression anterior to the spinal cord and increase the volume of the craniocervical junction8. However, in a considerable proportion of patients, neurological function does not even return to baseline. In the present study, we proposed the concept of quantitative measurement of odontoid descent and quantitatively investigated the odontoid location and craniocervical volume preoperatively and postoperatively. We retrospectively analyzed the clinical profiles of 20 patients diagnosed with basilar invagination who underwent a TARP operation. We measured the preoperative and postoperative craniocervical volume with the aim of providing some quantitative information on the efficacy of TARP surgery.
Three-Dimensional Reconstruction of the Craniocervical Junction
Direct measurement methods, such as the perfusion method and finite element models, are options for the measurement of vertebral canal volume, but they cannot be applied in the clinical setting16,17. MRI can clearly demonstrate soft-tissue structures, including the cervical spinal cord and brain stem; however, its sensitivity with regard to revealing osseous structures is limited compared with CT scanning. Furthermore, the slice thickness of MRI scanning should be 2 mm at a minimum in order to provide fine detail, making scanning time much longer18-20. Dong et al. verified the accuracy and repeatability of using CT scanning data for 3-D reconstruction of the craniocervical junction and measurement of cervical spinal canal volume15. Because the anatomical structures in the craniocervical junction are complex, volume measurement in clinical practice is difficult. In the current study, we imported craniocervical CT data into the Yorktal Bone CT Image Processing Software for 3-D reconstruction of the craniocervical junction. We attempted to identify the boundaries of the craniocervical junction and to measure the craniocervical volume.
Identification of Measurement Range
The craniocervical junction includes the skull base and upper cervical spine. Because the cross-sectional area of the skull base is several times that of the upper cervical spinal canal, the change in the volume of the craniocervical junction induced by odontoid descent is typically small. Therefore, the error of measurement can be several times that of the actual change, in which case the direct volume measurement may be substantially incorrect. The key goal of the current study was to define the boundaries of craniocervical volume changes, thereby reducing calculation errors. From our analysis of volume changes, we found that TARP operations can result in descent of the odontoid process and reduce cranial cavity invagination. Volume change occurred primarily in the vicinity of the odontoid above the McRae line. The distance between the atlas and the axis was increased and, because of the variability of the osseous vertebrae below the axis, we found that the volume change primarily occurred in the spinal canal at the level of the atlantoaxial joint.
Correlation Between Surgical Efficacy and Morphometric Parameters
We demonstrated that the correlation between surgical efficacy (JOA-score improvement rate) and the craniocervical volume improvement rate (r = 0.83) was stronger than that between surgical efficacy and the absolute craniocervical volume change (r = 0.61). The normal volume of the cervical spinal cord varies among individuals because of sex, height, and weight differences20. Therefore, surgical efficacy may also vary among individuals despite TARP operations leading to the same change in the volume of craniocervical junction. This observation may also explain the lower correlation between surgical efficacy and the absolute craniocervical volume improvement. Since the rate of craniocervical volume improvement includes the individual-specific preoperative volume as a baseline, which eliminates individual differences to a certain extent, it can reflect surgical efficacy better. We also performed a regression analysis for the JOA-score improvement rate and the craniocervical volume improvement rate of the 20 patients and achieved a regression equation of Y = 0.07465 + 0.7979X (R2 = 0.69), suggesting a significant positive correlation between surgical efficacy and the volume improvement rate.
The surgical efficacy was evaluated for the 20 patients enrolled in the present study. Eight patients had a craniocervical volume improvement rate of <60%, and surgical efficacy (based on the JOA-score improvement rate) was considered effective for all of them. Four patients had a craniocervical volume improvement rate of 60% to 85%, and all of the TARP operations in these patients were categorized as having substantial efficacy. Eight patients had a craniocervical volume improvement rate of >85%; 5 of them were categorized as having complete remission of neurological abnormalities and the operation had substantial efficacy for the other 3. A scatter diagram is shown in Figure 2. The results suggest that a craniocervical volume improvement rate of >60% may confer better surgical outcomes.
Intraoperative estimation of the craniocervical volume improvement rate is impossible. However, we found this parameter to be significantly related to the extent of odontoid descent (r = 0.80; Table IV), which can be assessed intraoperatively through fluoroscopic estimation of the odontoid location. We also performed a regression analysis for the odontoid descent distances and volume improvement rates of the 20 patients and achieved a regression equation of Y = −0.5192 + 1.386lnX (R2 = 0.63), suggesting a significant positive correlation between the odontoid descent distance and the craniocervical volume improvement rate. The scatter diagram shown in Figure 3 and the data distribution conformed to our theory that odontoid descent can lead to a significant change in craniocervical volume initially; the influence then attenuates; and, after the odontoid descends to some degree, the volume change can be negated. A quantitative assessment of odontoid descent and the craniocervical volume improvement rate may help surgeons to predict surgical efficacy. Limited by the small sample size in the present study, the quantitative relationship is not exact, necessitating further research involving a much larger cohort. The TARP system was designed to release the compression in the front of the spinal cord and to increase the volume of the craniocervical junction8. The intraoperative route of the odontoid process is a forward and downward arc—that is, the odontoid process can have morphometric changes of sagittal displacement and axial rotation; in this study, we divided the route into 3 parts: descent, ante-displacement, and anteversion. Considering that the anterior arch of the atlas can block ante-displacement and anteversion, the change in the volume of the craniocervical junction was mainly related to the change in the odontoid descent.
Additionally, we found no significant correlation between surgical efficacy (JOA-score improvement rate) and symptoms, age, symptom duration, preoperative JOA score, odontoid ante-displacement, or clivus-odontoid angle change. Due to the limited sample size, the available data did not provide a solid basis for defining risk factors related to surgical efficacy. Hence, a central register involving a much larger cohort should be established for future research.
Limitations and Outlook
Although the software that we used could identify the boundaries of the spinal canal according to the threshold values automatically, and joint efforts of a radiologist and an orthopaedic surgeon helped to confirm the accuracy, the results may still have certain limitations due to clinician experience, subjective factors, and resolution of the CT.
Because of the small sample size, we could only observe the trend line of JOA-score improvement rates with craniocervical volume improvement rates ranging from 0.3 to 1.22, as shown in Figure 2. We speculate that, if the craniocervical volume improvement rate increases further, the trend line of the curve may decline or stabilize. Furthermore, as excessive descent of the odontoid may lead to distraction injury of the spinal cord, the surgical efficacy of the TARP procedure may be limited by that factor. This needs to be studied further in relevant experiments on animals.
The odontoid descent distance can predict the craniocervical volume improvement rate following TARP operations for basilar invagination, and both can predict surgical efficacy. We speculate that a deliberate strategy of planning the degree of odontoid descent preoperatively may be helpful to improve the efficacy of TARP operations in such patients.
Investigation performed at the Department of Orthopedics, Guangzhou General Hospital of Guangzhou Military Command, Guangzhou, People’s Republic of China
Disclosure: This work was supported by the National Natural Science Foundation of China (20122057), the Key Project of the Twelfth Five-Year Project of Military Medicine (BWS11C065), and the Natural Science Foundation of Guangdong Province, People’s Republic of China (S2013010011964). The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article.
1. Simsek S, Yigitkanli K, Belen D, Bavbek M. Halo traction in basilar invagination: technical case report. Surg Neurol. 2006 ;66(3):311–4; discussion 314.
2. Hwang SW, Heilman CB, Riesenburger RI, Kryzanski J. C1-C2 arthrodesis after transoral odontoidectomy and suboccipital craniectomy for ventral brain stem compression in Chiari I patients. Eur Spine J. 2008 ;17(9):1211–7. Epub 2008 Jul 16.
3. Zileli M, Cagli S. Combined anterior and posterior approach for managing basilar invagination associated with type I Chiari malformation. J Spinal Disord Tech. 2002 ;15(4):284–9. Epub 2002 Aug 15.
4. Mummaneni PV, Haid RW. Transoral odontoidectomy. Neurosurgery. 2005 ;56(5):1045–50; discussion 1045-50. Epub 2005 Apr 28.
5. Menezes AH, VanGilder JC, Graf CJ, McDonnell DE. Craniocervical abnormalities. A comprehensive surgical approach. J Neurosurg. 1980 ;53(4):444–55. Epub 1980 Oct 1.
6. Wang C, Yan M, Zhou HT, Wang SL, Dang GT. Open reduction of irreducible atlantoaxial dislocation by transoral anterior atlantoaxial release and posterior internal fixation. Spine (Phila Pa 1976). 2006 ;31(11):E306–13. Epub 2006 May 12.
7. Finn MA, Bishop FS, Dailey AT. Surgical treatment of occipitocervical instability. Neurosurgery. 2008 ;63(5):961–8; discussion 968-9. Epub 2008 Nov 14.
8. Xia H, Yin Q, Ai F, Ma X, Wang J, Wu Z, Zhang K, Liu J, Xu J. Treatment of basilar invagination with atlantoaxial dislocation: atlantoaxial joint distraction and fixation with transoral atlantoaxial reduction plate (TARP) without odontoidectomy. Eur Spine J. 2014 ;23(8):1648–55. Epub 2014 May 18.
9. Wada E, Suzuki S, Kanazawa A, Matsuoka T, Miyamoto S, Yonenobu K. Subtotal corpectomy versus laminoplasty for multilevel cervical spondylotic myelopathy: a long-term follow-up study over 10 years. Spine (Phila Pa 1976). 2001 ;26(13):1443–7; discussion 1448. Epub 2001 Jul 18.
10. Hirabayashi K, Watanabe K, Wakano K, Suzuki N, Satomi K, Ishii Y. Expansive open-door laminoplasty for cervical spinal stenotic myelopathy. Spine (Phila Pa 1976). 1983 ;8(7):693–9. Epub 1983 Oct 1.
11. Hirabayashi K, Miyakawa J, Satomi K, Maruyama T, Wakano K. Operative results and postoperative progression of ossification among patients with ossification of cervical posterior longitudinal ligament. Spine (Phila Pa 1976). 1981 ;6(4):354–64. Epub 1981 Jul 1.
12. Schriger DL, Larmon B, LeGassick T, Blinman T. Spinal immobilization on a flat backboard: does it result in neutral position of the cervical spine? Ann Emerg Med. 1991 ;20(8):878–81. Epub 1991 Aug 1.
13. Fang CH, Xie AW, Chen ML, Huang YP, Lu CM, Li XF, Pan JH, Peng FP. Application of a visible simulation surgery technique in preoperation planning for intrahepatic calculi. World J Surg. 2010 ;34(2):327–35. Epub 2009 Dec 17.
14. Adams R, Bischof L. Seeded region growing. IEEE Transactions on Pattern Analysis Machine Intelligence. 1994 ;16(6):641–7.
15. Dong F, Shen C, Jiang S, Zhang R, Song P, Yu Y, Wang S, Li X, Zhao G, Ding C. Measurement of volume-occupying rate of cervical spinal canal and its role in cervical spondylotic myelopathy. Eur Spine J. 2013 ;22(5):1152–7. Epub 2013 Jan 6.
16. Panzer MB, Cronin DS. C4-C5 segment finite element model development, validation, and load-sharing investigation. J Biomech. 2009 ;42(4):480–90. Epub 2009 Feb 5.
17. Wheeldon J, Khouphongsy P, Kumaresan S, Yoganandan N, Pintar FA. Finite element model of human cervical spinal column. Biomed Sci Instrum. 2000;36:337–42. Epub 2000 Jun 2.
18. Tan LA, Kasliwal MK, Traynelis VC. Comparison of CT and MRI findings for cervical spine clearance in obtunded patients without high impact trauma. Clin Neurol Neurosurg. 2014 ;120:23–6. Epub 2014 Feb 25.
19. Ishii T, Mukai Y, Hosono N, Sakaura H, Nakajima Y, Sato Y, Sugamoto K, Yoshikawa H. Kinematics of the upper cervical spine in rotation: in vivo three-dimensional analysis. Spine (Phila Pa 1976). 2004 ;29(7):E139–44. Epub 2004 Apr 17.
20. Yanase M, Matsuyama Y, Hirose K, Takagi H, Yamada M, Iwata H, Ishiguro N. Measurement of the cervical spinal cord volume on MRI. J Spinal Disord Tech. 2006 ;19(2):125–9. Epub 2006 Jun 9.