Yinqian Wang, Tao Fan, Xingang Zhao, Cong Liang, Qifei Gai, Haijun Zhao

ABSTRACT

Objective: To describe the different pathological characteristics of congenital basilar invaginations and discuss the surgical treatment of such cases.

Methods: A total of 139 patients diagnosed with basilar invaginations underwent surgical treatment from 2008 to 2015. Based on Atul Goel’s classification and simultaneous consideration of atlantoaxial dislocation or syringomyelia, the cases were subdivided into four groups. Individualized posterior surgical decompression and/or atlantoaxial reset procedures were performed to correct atlantoaxial dislocation, decompress the brain stem, or resolve syringomyelia. The indications and critical points of each procedure were documented.

Results: All 139 patients were surgically treated; 27 patients (19.4%) had underwent at least one decompression surgery previously. On an average, there was gratifying clinical improvement based on the Japanese Orthopaedic Association score analysis. One

patient exhibited severe post-operative infection, and the fusion instrument was removed. One patient experienced fracture of internal fixation. Two patients exhibited persistent respiratory symptoms at early stages after the surgery. Four patients felt worse at the latest follow-up. There was no surgical mortality. The poor outcome/morbidity in this series was 5.8% (8/139).

Conclusion: The different pathological image characteristics of congenital basilar invaginations based on the presence or absence of syringomyelia and/or atlantoaxial dislocation, reflected the pathological features of complicated basilar invaginations more accurately. Based on these features, different posterior decompression and/or reset procedures, combined with occipitocervical fusion and C1-2 fusion, could be tailored to different patients. These individualized surgeries could reduce surgical complications, decrease morbidity and mortality, and further promote positive outcomes.

KEYWORDS basilar invagination; syringomyelia; atlantoaxial dislocation; chiari malformation; surgical treatment

1 Introduction

Basilar invagination (BI) is caused by congenital cranio-cervical junction (CCJ) that abnormally prolapses into the foramen magnum, leading to neural compression or cranio-cervical instability, and severe neurologic symptoms and deficits. The reported

etiologies include achondroplasia, clivus hypoplasia, atlantooccipital assimilation, occipital condyle hypoplasia, and incomplete ring of C1, with spreading of the lateral masses and atlas hypoplasia. Previous studies have suggested that the compression may result from the cerebellar tonsils from posterior or the odontoid process from anterior, and the instability may occur due to incompetent C1/2 joints or adjacent to Klippel-Feil syndrome of the upper cervical spine. Surgery is recommended for symptomatic patients and relief of ventral compression of the brainstem and upper spinal cord is considered to be the first choice to slow the progression of neurologic deficits. To help with the evaluation of the pathology and formulation of a rational surgical strategy, in 1998, Atul Goel presented a classification system for BI, based on the single criterion of the absence or presence of Chiari malformation. In 2004, Atul Goel divided BI into two types: Type I, distance increased from odontoid to atlas and Type II, distance not increased from odontoid to atlas. With the development of imaging techniques, patients with BI were categorized into two groups, based on the presence or absence of clinical and radiological evidence of instability of the craniovertebral junction. During our practice, we found that among the Type I patients, some exhibited severe syringomyelia and some did not. In Type II patients, some exhibited tonsillar herniation and some did not. The symptoms, signs, and surgical treatment of these patients had some differences. Thus, we subdivided these patients into four groups. These different image characteristics reflected the pathological features of BI accurately, providing a reliable basis for clinical diagnosis and rational surgical treatment. In this report, 139 patients were divided into four groups, and individualized surgical treatment was performed for each patient. The selection and the critical points of the procedure, and the prognosis of the patients have been discussed.

2 Material and methods

2.1 Patient population

The clinical data of patients who accepted treatment in Beijing Sanbo Brain Hospital were reviewed and analyzed. The study was performed after minimal risk institutional review board approval. Patients were included in the event of (1) positive diagnosis of BI, exclusion of traumatic or inflammatory factors, (2) aged from 12 to 80 years, no gender preference, and (3) no surgery contraindications. Overall, 139 patients presenting with BI were surgically treated from June 2008 to May 2015. There were 63 male patients and 76 female patients (M/F 1:1.2), aged from 13 to 68 years, with an average age of 36.5 years.

2.2 Diagnosis

Clinical diagnosis relied on radiographs of the cervical spine and measurement of the Chamberlain line, McGregor line and atlanto-odontoid interval (ADI). In this paper, BI was diagnosed when the odontoid process beyond the Chamberlain line was more than 3.0 mm,and the odontoid process beyond the McGregor line was more than 6.0 mm. Atlantoaxial dislocation was diagnosed when the distance of ADI was more than 3 mm on computed tomography (CT) scan.

2.3 Image characteristics

The MRI, MRI-Cine, and 3D-CT reconstruction of cranio-cervical junction were analyzed associated with clinical manifestation. Based on Atul Goel’s classification of Type I (unstable group) and Type II (stable group), according to whether the atlantoaxial dislocation and/or syringomyelia were present, the patients were subdivided into: (1) Type Ia: BI with ADI increase and without syringomyelia; (2) Type Ib: BI with ADI increase and with syringomyelia; (3) Type IIa: BI without ADI increase and without syringomyelia; and (4) Type IIb: BI without ADI increase but with syringomyelia (Figure 1).

Clinical examination showed that Type Ia patients predominantly exhibited brain stem compression, pyramidal sign, and CCJ instability. The Type IIa patients frequently showed cerebellum dysfunction and cervical nerve root stimulated symptoms. The Type Ib patients exhibited both brainstem compression and syringomyelia symptoms, and Type IIb patients exhibited both cerebellum dysfunction and syringo myelia symptoms (Table 1).

2.4 Surgical procedure

Informed consent was obtained from all patients (and their guardians) at an appropriate time before the surgery. Based on the clinical characteristics of the different cases and the results of monitoring cerebrospinal fluid (CSF) dynamic properties at CCJ using preoperative MRI-Cine and intraoperative ultrasound, we performed individualized surgical treatment. Twenty-seven patients were revision cases. The details of the surgical procedure for different BI types were as follows.

Figure 1 Different pathological image characteristics of congenital basilar invagination: (a) Type Ia which is BI with ADI increase and without syringomelia; (b) Type Ib which is BI with ADI increase with syringomelia; (c) Type IIa which is BI without ADI increase and without syringomelia; (d) Type IIb which is BI without ADI increase but with syringomelia.

Table 1 Preoperative neurological symptoms.

For Type Ia patients, atlantoaxial dislocation was the main cause of the pathogenesis (Figure 2). The posterior reduction of the odontoid process and maintaining the atlantoaxial joint stability were the critical aspects of the surgery. Thus, posterior reduction and fusion were performed.

For Type Ib patients, the cause of syringomyelia needs to be evaluated (Figure 3). Some patients’ CSF dynamic patterns at CCJ on MRI-Cine were obviously reduced. The reason might be associated with atlantoaxial dislocation or tonsillar herniation and arachnoid adhesion. Therefore, we performed the resetting first, and then used the intraoperative ultrasound to detect the craniocervical area. If tonsillar herniation moved up and the space of the dorsal brain stem enlarged, we only performed fusion surgery.

Figure 2 36 yrs male, presented with left finger numbness and left leg weakness for 2 months. The image characteristics of this patient is Type Ia (BI with ADI increase and without syringomelia). (a) Sagittal T2 sequence magnetic resonance imaging showing brain stem compression without syringomyelia. (b) CT scan showing odontoid process position moved upward. (c and e) Postoperattory MRI and CT showing atlantoaxial dislocation reliefed. (d) CT scan 3D reconstruction of occipito-C2 fusion. JOA score increased from preoperative value of 14 to 16 at the latest follow-up time point.

If the CSF dynamic pattern still reduced, and the tonsillar was still far below the C1 or C2 level, we performed subarachnoid manipulation, in which a small craniectomy, a standard laminectomy of C1, tonsillar electrocoagulation and/or tonsillectomy, and adhesion lysis were performed.

For Type IIa patients, there was no atlantoaxial dislocation, but their condition may be accompanied by congenital occipital cervical fusion, occipitalization of the atlas, platybasia, or lateral atlantoaxial articulation malformation. Additionally, the patients’ symptoms always related to severe cerebellar tonsillar hernia (Figure 4). Thus, we resected tonsillar herniation and performed C1-2 fusion to prevent postoperative instability.

For Type IIb patients, the same anatomic bone abnormality as Type IIa patients was observed, but these patients showed obvious syringomyelia (Figure 5). Therefore, we performed standard subarachnoid manipulation decompression procedure to restore the CSF circulation, and then performed C1-2 or occipitocervical fusion to prevent postoperative instability.

Figure 3 26 yrs female, presented with neck pain and limb num-bness for six years. Her symptom aggravated for 3 months. The image characteristics of this patient is Type Ib (BI with ADI increase and with syringomelia). (a) Sagittal T2 sequence magnetic resonance imaging showing tonsillar herniation with syringomyelia. (b) CT scan showing odontoid process position moved upward. (c) Intraoperative view of subarachnoid manuplation. (d) Intraoperative view of final construction. (e and f) Postoperattory MRI and CT showing atlantoaxial dislocation reliefed and syringomyelia significantly diminish. JOA score increased from preoperative value of 11 to 14 at the latest follow-up time point.

2.5 Outcome evaluation

All the patients were supervised in the intensive care unit until they recovered normal respiration and tracheal extubation was performed. All the patients underwent an MRI and CT scan before discharge, normally 2 weeks after surgery. The Japanese Orthopaedic Association (JOA) scoring system was used to measure the outcomes of patients with BI. Here, we documented the JOA score of each patient preoperatively, 2 weeks postoperatively, and at the latest follow-up time-point from 1 year to 8 years (average: 38 months).

Figure 4 15 yrs, female, clinical symptom includes walking unstable, intermittent headache. Physical exam reveals cerebellar ataxia. The image characteristics of this patient is Type IIa (BI without ADI increase and without syringomelia). (a) Cervical CT scan before surgery showing basilar invagination, normal atlantodental interval, without syringomyelia. (b) Sagittal T2 sequence magnetic resonance imaging showing odontoid compression over the brainstem and upper spinal cord. (c) MRI scan after surgery. (d) CT scan after surgery. 1 years after the surgery, the JOA score was 16. JOA score increased from preoperative value of 15 to 16 at the latest follow up time point.

2.6 Statistical analysis

Mean ± standard deviations of the mJOA scores pre- and post-operation, and at the latest follow-up time-point were collected and analyzed using paired t-test.

3 Results

The clinical data of 139 basilar invagination patients who underwent surgical treatment in our hospital from June 2008 to May 2015 were collected and analyzed. Among the 139 patients, 87 patients had associated syringomyelia, 76 patients had associated atlantoaxial dislocation, and 27 patients underwent at least one decompression surgery previously. The patients were subdivided into four groups: Type Ia, 31 cases (22.3%); Type Ib, 45 cases (32.4%); Type IIa, 21 cases (15.1%); and Type IIb, 42 cases (30.2%).

Neurologic dysfunction mainly observed was: (1) unstable walking, ataxia, pyramidal tract symptoms, and signs caused by the compression of the cerebellum and brain stem or instability of CCJ; and (2) asymmetric sensory disorder, muscle atrophy, and limb weakness caused by syringomyelia. Type I patients exhibited brainstem dysfunction and Type II patients exhibited cerebellar dysfunction. Type Ib patients exhibited brainstem dysfunction and symptoms of syringomyelia.The details of the neurologic dysfunction of the different groups of patients are listed in Table 2.

Figure 5 26 yrs, male. The patient underwent suboccipital de- compression surgery 2 yrs ago. And one year after his surgery, he felt progression of right hand numbness, and intermittent right upper limb pain. The image characteristics of this patient is Type IIb (BI without ADI increase but with syringomelia). (a) CT scan 3D reconstruction showing the range of resection. (b) CT scan showing basilar invagination, normal atlantodental interval. (c) MRI imaging showing tonsillar hemiation with syringomyelia. (d) 10 days after the surgery, MRI showing the syringomyelia significantly diminish. (e and f) Postoperative sagittal CT scan showing a C1-C2 fusion. JOA score increased from preoperative value of 13 to 15 at the latest follow up time point.

All the 139 patients were treated surgically; suboccipital craniotomy was performed according to the compression of the brainstem. Among the 27 revision patients who underwent at least one decompression surgery previously, 12 patients had a large bone decompression range, which further caused cerebellar ptosis. We used a titanium mesh to repair the occipital bone and performed C1-2 fusion for stability. Among the revision cases, there were 3 Type Ia patients, 10 Type Ib patients, 2 Type IIa patients, and 12 Type IIb patients. This meant that 22 patients with syringomyelia did not show a good outcome after the first surgery. This also makes the CSF dynamic pattern evaluation and subarachnoid manipulation very important for patients with syringomyelia.

We performed occipitocervical fusion surgery on 91 cases, and 48 cases were under C1-2 fusion. For the 87 patients with syringomyelia, microscopic decompression was done, including repatency of the fourth ventricle and the central canal of the spinal cord manipulation of cerebellar tonsillar or resection. Sixty-two (71.3%) patients’ syringomyelia significantly shrank (>50%) 2 weeks after the surgery. For Type Ia patients, only eight underwent a pure occipital craniotomy; the resetting of the odontoid process was the critical point of the surgery. For Type Ib patients, 42/45 underwent subarachnoid manipulation and decompression surgery. Twelve Type IIa patients underwent occipito-cervical fusion, and 9 underwent C1-2 fusion. For Type IIb patients, 36/42 underwent subarachnoid manipulation, decompression, and C1-2 or occipitocervical fusion. The outcome of the treatment was evaluated by the JOA scoring system preoperatively, 2 weeks postoperatively, and at the latest follow-up time-point (Table 3).

We performed a follow-up after 12–84 months (average: 38 months), for the 139 patients. There was no case of death in the series, and 15 patients were lost to follow-up. One patient exhibited severe post- operative infection, and the fusion instrument was removed. Two patients exhibited expiratory dyspnea at early stages after the surgery, and tracheotomy was performed. One patient experienced fracture of internal fixation. Four patients reported worsening of clinical symptoms at the latest follow-up. Neurologic improvement was observed in 123 of 139 patients (88.5%), and was stable in 7 (5.0%). The poor outcome/ morbidity in this series was 5.8% (8/139). Among these cases, the mean value of the JOA score increased from 10.97 ± 3.07 before surgery to 13.02 ± 3.46 at the latest follow-up time-point in Type Ia patients. The mean value of the JOA score was 10.78 ± 2.02 before surgery, and increased to 13.52 ± 2.58 at the latest follow-up time-point in Type Ib patients. The mean value of the JOA score was 11.71 ± 1.98 before surgery, and increased to 14.24±1.85 in Type IIa patients. Finally, the mean value of the JOA score was 10.95 ± 2.02 before surgery, and increased to 13.13 ± 2.63 in Type IIb patients.

4 Discussion

The BI is a basioccipital and C1-2 congenital abnormality. In addition to the CCJ bony abnormalities, symptoms also involve soft tissue and nervous system dysfunction, such as occipitocervical fusion, C2-3 fusion, lateral atlantoaxial joint deformity, Chiari malformation, and syringomyelia. The symptoms described above may present individually or simultaneously and are not consistent with the malformation. They usually occur in young and middle-aged adults with progressive symptoms. The symptoms are associated with the compression of medulla oblongata and upper cervical cord, mainly causing limb movement disorder, sensory disturbance, and even incontinence. Moreover, cervicothoracic syringomyelia is a common complication which induces related neurologic symptoms and deficits.

At present, many differences and disputes still exist regarding the diagnostic classification and surgical treatment of BI, and there is a lack of unified surgical strategy. The classic classification is based mainly on the bony structures and occipital cervical stability, but ignores syringomyelia and atlantoaxial dislocation, which could also significantly influence the pathological features, surgical strategy, and patient outcome.

Based on Atul Goel’s classification, we grouped the patients with BI into four categories according to the presence/absence of atlantoaxial dislocation and/or syringomyelia.

Type Ia patients always exhibited brainstem dysfunction, such as motor weakness (67.7%) and pyramidal sign (83.9%). The CT scans always showed that the odontoid process position moved upward, the medulla oblongata and cervical cord compressed, and

ADI increased. The posterior reduction of the odontoid process and maintaining the atlantoaxial joint stability were the critical aspects of the surgery. Thus, for Type Ia patients, the resetting of the odontoid process was the key point. In our series, occipital-cervical fusion or C1-2 fusion were performed for all patients, among whom only 6 cases needed a subarachnoid manipulation after craniectomy decompression because the CSF dynamic pattern evaluation preoperative at MRI-Cine was reduced, and there was severe tonsillar herniation below C2. There were three revision cases in this group, and all had a subarachnoid manipulation and C1-2 fusion. The mean value of the JOA score increased from 10.97 ± 3.07 before surgery to 12.45 ± 3.14 at 2 weeks after surgery, and then to 13.02 ± 3.46 at the latest follow-up.

For Type Ib patients, symptoms included motor weakness (71.1%), pyramidal sign (84.4%), paresthesia (91.1%), and muscular atrophy (40%). The CT scans showed similar results with Type 1a. The MRI-Cine study always showed that CSF flow in the aqueduct, the fourth ventricle, and anterior of cisterna pontis was reduced. The reason might be the atlantoaxial dislocation or tonsillar herniation. There were 10 revision patients in this group and 8 patients did not have syringomyelia. Five patients had a large bone decompression range, which further caused cerebellar ptosis. We used a titanium mesh to repair the occipital bone, and C1-2 fusion was performed for stability. For these patients, we performed the resetting first, without opening the dura. We subsequently used the intraoperative ultrasound to detect the craniectomy window. If the tonsillar herniation moved up and the space of the dorsal brain stem enlarged, we performed only fusion surgery. If the CSF dynamic pattern still reduced and the tonsillar was far below the decompression window, we performed subarachnoid manipulation. Thirty-two (71.1%) patients’ syringomyelia significantly shrank (> 50%) 2 weeks after the surgery. The mean value of the JOA score increased from 10.78±2.02 before surgery to 12.76 ± 2.09 at 2 weeks after the surgery, and to 13.52 ± 2.58 at the latest follow-up time-point.

Type IIa patients did not exhibit brainstem symptoms, but always exhibited gait ataxia (76.2%) and occipital pain (47.6%). The CT scan showed normal ADI, but the anteroposterior diameter of the spinal canal reduced and the foramen magnum narrowed; such cases were always accompanied by congenital occipital cervical fusion, occipitalization of the atlas, platybasia, and lateral atlantoaxial articulation malformation. Some patients’ symptoms always related to severe tonsillar herniation. Therefore, we resolved tonsillar herniation and performed C1-2 or occipito-cervical fusion to prevent postoperative instability. The mean value of the JOA score increased from 11.71 ± 1.98 before surgery to 13.57 ± 1.69 at 2 weeks after the surgery, and to 14.24 ± 1.85 at the latest follow-up time-point.

Type IIb patients always exhibited paresthesia (90.5%) and muscular atrophy (45.2%). The CT scans showed the same bone abnormality as Type IIa patients, but these patients showed obvious syringomyelia. Thus, we performed subarachnoid manipulation decompression to restore CSF circulation, and then also performed cervical-occipital fusion or C1-2 fusion to prevent postoperative instability. The mean value of the JOA score was increased from 10.95 ± 2.02 before surgery to 12.71 ± 2.17 at 2 weeks after the surgery, and to 13.13 ± 2.63 at the latest follow-up time-point.

In our case, based on different pathological image characteristics, the excision of the posterior arch of the atlas could not exceed the vertebral artery incisures; otherwise, it may have caused postoperative collapse of the cerebellum, ultimately resulting in spinal cord compression symptoms that cannot be effectively alleviated. For patients with severe syringomyelia, MRI-Cine shows an abnormal CSF flow at craniovertebral junction, and even severe tonsillar herniation after the resetting of odontoid; subarachnoid decompression is needed, including cerebellar tonsillar resection, releasing the opening of myelocoele, and separation of arachnoid adhesion. For Type Ia and Type Ib patients, with atlantoaxial vertebral instability, occipitocervical or C1-2 fusion was recommended. For Type IIa and Type IIb patients, there is a debate regarding whether fusion surgery should be performed, but these patients also exhibit potential atlantoaxial vertebral instability; otherwise, suboccipital decompression may aggravate the instability of the atlanto-occipital area. Therefore, we also recommend fusion surgery for Type IIa and Type IIb patients.

Previous studies on BI mainly focused on atlantoaxial instability and accompanied bone abnormalities, most surgical procedures concentrated on odontoid process resetting, including the anterior reduction of C2, transoral approach of odontoid resetting and posterior fusion. In our series, we focused on accompanied pathological characteristics, such as tonsillar herniation and syringomyelia. Consideration of these different pathological characteristics of BI may provide important clues for selection of appropriate surgical procedures.

5 Conclusions.

Successful management of congenital craniocervical anomalies requires a thorough understanding of the neural and bone anatomy. Insights on complicated craniovertebral junction abnormalities such as basilar invagination, tonsillar herniation, syringomyelia, and atlantoaxial dislocation could be evidently evaluated using cervical vertebra MRI, MRI-Cine, and 3D-CT reconstruction preoperatively. The different pathological image characteristics of congenital BI, based on the presence or absence of syringomyelia and/or atlantoaxial dislocation reflects the pathological features more accurately. Based on these features, different posterior decompression and/or resetting procedures, combined with occipitocervical fusion and C1-2 fusion, could be tailored for different patients. These individualized approaches may reduce surgical complications, decrease morbidity and mortality, and further promote positive outcomes.

Conflict of interests

All contributing authors have no conflict of interests.

Financial support

This study was funded by Construction Project of National Clinical Key Specialties of People’s Republic of China (Ministry of Health of People’s Republic of China 873 (2011)) and the Capital Health Research and Development of Special 2014-2-8011. And the corresponding author Tao Fan received the support of those funding.