Management of malignant atlanto-axial tumours

INTRODUCTION

Malignant tumours involving atlanto-axial spine are uncommon and mostly due to metastases.1,8 The most common primary sources of malignancy are from the breast, lung, and prostate.5,9,10 Atlanto-axial tumours generally involve elements of the anterior spine, i.e. the lateral articular mass of C1, the dens, and the body of C2.1,2,6,10 Pathological fracture and progressive subluxation are common and may result in cord compression when the condition advances. Radiotherapy combined with chemotherapy is the usual treatment of choice.2,9-11 Recourse to surgical treatment becomes necessary in the presence of neurological complications or spinal instability resulting in severe pain. Nonetheless, unique anatomical features of the craniovertebral junction pose difficult surgical problems. Advances in the treatment of primary tumours have improved the survival rate, which is greater for patients presenting with upper cervical metastases, either as multiple other metastases or as the first manifestation of malignancy.10,12 The development of newer surgical techniques providing stable fixation of the upper cervical spine and craniocervical junction has improved treatment strategies and results, especially in cases with instability.13-21

METHODS

We retrospectively reviewed the hospital records and radiographs of patients presenting with malignant C1 or C2 tumours from July 2000 to June 2004 at the Prince of Wales Hospital, Hong Kong. Records of 12 patients (9 female and 3 male), with a mean age of 56.5 years (range, 35-71 years), were reviewed. At the time of presentation, 7 patients had received treatment for a known primary malignancy of the lung (n=4; patients 2, 7, 10, and 12), colon (n=1; patient 5), nasopharynx (n=1; patient 3), or breast (n=1; patient 8). The atlanto-axial spine was the first presentation of malignant disease in the remaining 5 patients, whose subsequent diagnoses were multiple myeloma (n=1; patient 9), carcinoma of the thyroid (n=1; patient 4), malignant fibrous histiocytoma (n=2; patients 6 and 11), and adenocarinoma of unknown origin (n=1; patient 1) [Table].

Demographic data, duration of symptoms, and severity of neck pain at the time of presentation using the visual analogue scale (VAS),22 neurological deficit using Frankel grade23 attributable to atlantoaxial instability or tumour mass, sites of the primary tumours, and treatment history were charted (Table). Preoperative, postoperative, and follow-up radiographs and computed tomographic (CT) and magnetic resonance imaging (MRI) scans were analysed with regard to the extent of the tumour involvement, other spinal metastases, instability, and evidence of cord compression.

Management consisted of non-surgical and surgical approaches, determined with input from referring physicians and oncologists concerning each patient's overall medical condition, including fitness for surgery, life expectancy, and tumour characteristics. Surgeries were performed for surgically fit patients with life expectancy of more than 3 months.

RESULTS

All patients presented with localised unremitting suboccipital neck pain, while patient 3 complained of pain radiating to the shoulder. Neck motion in all patients was very limited. Only patients 5 and 10 had motor weakness (Frankel grade D) on presentation. The mean interval between the onset of neck pain and diagnosis was 2.8 months (range, 2 weeks-6 months). Of the 7 patients with known primary tumours, the interval between the diagnosis of the primary tumours and the onset of upper cervical metastases ranged from one month in a patient with carcinoma of lung, to 11.7 years in a patient with carcinoma of breast. Nine patients had additional spinal involvement in the subaxial cervical, thoracic, or lumbar spine, whereas patients 4, 6, and 11 had no metastases on additional spinal sites (Table).

Radiology

Plain radiographs, CT scans, and MRI scans of the upper cervical spine were performed on all patients. Tumour involvement in the atlanto-axial spine is summarised in the Table. Lesions in 11 of the 12 patients could be adequately visualised on plain lateral radiographs (Fig. 1a), while in patient 1 the destruction of the bone was visualised through an open-mouth view (Fig. 1b). According to Phillips and Levine,11 tumour involvement is classified as: type A-diffuse C2 body with or without pathological fracture (n=11, patients 2-12; Figs. 2a and 3a), type B-arch of C2 (n=0), or type C-lateral masses of C1 resulting in lateral tilting or subluxation (n=3, patients 1, 3, and 5; Figs. 1b, 1c, and 1d). For type A involvement, CT scans showed that all 11 patients had pathological fractures of the cortices. The atlas was displaced anteriorly due to pathological odontoid fractures in 7 (patients 2-5, 9,10, and 12; Fig. 2a). MRI scans demonstrated spinal cord impingement in patients 5 and 10, but no cord signal change (Fig. 2b). Patients 7, 8, and 11 showed no C1/C2 malalignment despite tumour infiltration of the dens or the body of C2 (Fig. 3a). No flexion or extension radiographs were performed to define instability due to the risk of iatrogenic fracturesubluxation and neurological dysfunction.

Treatment

Patients 5 and 8 were treated nonoperatively because they had multiple metastases and were not fit for surgery. Patient 5 died 3 months later, and patient 8 received a course of radiotherapy and survived further 11 months. Seven women and 3 men underwent surgical stabilisation. The main surgical indication was pain. Patient 10 presented with mild weakness and requested surgery because of severe neck pain. Skull traction with either a halo ring or skull tong was applied pre- and peri-operatively to reduce and maintain atlanto-axial alignment. A halo ring was used for postoperative immobilisation in patients 1 and 11. All patients underwent fibre-optic intubation with minimal manipulation of the head and neck. Patients were turned to prone position on a Jackson spine table (Orthopedic Systems Inc, Union City [CA], US). Fullimage intensification control of the neck position was allowed during the operation. The types of instrumentation used and the levels of fusion achieved are summarised in the Table. All patients received postoperative radiotherapy as determined by the oncologists.

An occipitocervical plate (CerviFix; Synthes, Bettlach, Switzerland) was used in patients 1 and 11 (Fig. 1e), in whom fixation involved the occiput. Patient 11 underwent a second-stage intralesional tumour excision and anterior bone grafting surgery using a lateral retropharyngeal approach.24 A halo-jacket was applied postoperatively for 2 months. Patients 2, 3, 9, and 12 underwent occipitocervical fusion modified from Ransford et al.13 A 5-mm Luque rod was preoperatively bent to match the occipitocervical lordosis (Fig. 2c). The rod was fixed to the skull by wires passing through suboccipital burr holes, and to the cervical spine by segmentai fixation with Wisconsin or sublaminar wires. Three burr holes on each side, approximately 1 cm apart and 2 cm from midline, were drilled in the occipital bone between the inion and the foramen magnum with due attention to the proximity of the transverse sinuses. Flexible wire (gauge 20) was passed in the extradural plane through the 2 burr holes on each side of the midline. Usually 4 wires could be passed to fix the occiput to the rod. Segmentai fixation to the spine was achieved with sublaminar wires for C1, and either sublaminar or spinous process wiring for the C2 and the subaxial vertebrae. Iliac bone grafts were applied generously over the area of fusion. A Philadelphia neck collar was applied for 2 months postoperatively.

Fusion did not involve the occiput in patients 4, 6, 7, and 10. The methods of fixation used were screws and rods (Starlock/Cervifix; Synthes, Bettlach, Switzerland). Screws were inserted through the pedicle of Cl into the lateral mass. For the C2 vertebra, either a transarticular C1/C2 screw was applied (patient 4) or the screw was inserted into the pedicle (patient 6). For the subaxial vertebrae, screws were either inserted through the pedicle into the body (patients 7 and 10; Fig. 3b) or into the lateral masses (patients 4 and 6). A Philadelphia neck collar was applied for protection for postoperative 2 months.

Outcome

All 10 operated patients tolerated the surgery well. There were no neurological complications or wound infections. In patient 1, a small amount of cerebrospinal fluid leaked through the suction drain; the leakage subsided gradually with the discontinuation of suction, and an antibiotic was prescribed to prevent infection. All patients were satisfied with the amount of pain relief. The mean VAS pain score of the 10 operated patients improved from 9.3 to 1.9. The mean survival period of 9 patients after surgery was 13.1 months. At the time of review, patient 11 was still living at postoperative 11 months. In all other patients, the cause of death was progression of systemic cancer. No patient died within one month of surgery. Implant loosening was noted in patient 6, just before she died as a result of the progressive destruction of bone. Bony union was achieved in patients 1, 4, and 9.

DISCUSSION

Atlanto-axial malignancies seldom present with neurologic deficit due to the wide mid-sagittal diameter. Spinal cord compression at this level often results from pathological fracture-subluxation rather than epidural tumour.2,50,10,11,25 Up to 83% of the patients had some degree of C1-C2 subluxation when first diagnosed.5,9 In our series, 9 patients showed major C1-C2 instability (patients 7, 8, and 11 showed no C1C2 instability). The occiput-C1-C2 complex is responsible for 40% of cervical flexion-extension and 60% of cervical rotation. Stability of the upper cervical spine depends very much on the lateral articular mass of both C1 and C2 and the transverse ligament of C1.26 Tumour destruction of the lateral mass and dens, leading to pathological fracture, together with inherent mobility in this region accounts for the high incidence of spinal subluxation. Painful limitation of neck motion, especially rotation, may therefore be the first (if not the only) clinical sign of upper cervical metastases.2,9 Radiation therapy and external immobilisation can provide symptomatic relief in the early stage of the disease.2,60,10,11 Plain tomography or CT are helpful in the diagnosis of patients who have a history of malignant disease and present with neck pain but have negative radiological findings. Regular bone scans can pick up early lesions27 and are recommended in patients suffering from malignancies with relatively favourable prognoses, such as breast or nasopharyngeal carcinoma.

Up to 55% of patients with metastatic bone disease had spinal lesions.28 It was not uncommon for patients with upper cervical metastases to have additional subaxial cervical, thoracic, or lumbar spine involvement. Up to 70% of cancer patients with pathological odontoid fracture had other vertebrae involved; this complicated the decision as to whether to proceed with an aggressive approach to treatment.9,10 Mean survival period after the diagnosis of cervical metastases was 14.7 months.29 For metastases in the C1-C2 region, the mean survival period for patients treated conservatively and surgically were 5 to 11 months and 8 to 26 months, respectively.5,8-11 The goal of treatment is, therefore, to improve quality of life during the short subsequent survival period.5,7,8

Another aim of surgical treatment is to provide rigid fixation so that patients do not require prolonged external support. Although recent studies have advocated aggressive anterior resection and stabilisation for metastatic tumours,7,12,30-32 posterior instrumentation has been shown to be effective for the relief of symptoms and the prevention of paraplegia.5,8-11,25-33 Only in rare occasions when neurologic compromise persists after posterior reconstruction, anterior stabilisation and decompression become necessary.34

In our series, tumour lesions involved the dens, the body of the axis, and the lateral articular mass of the atlas. As it is often difficult to secure the transarticular fixation for atlanto-axial instability, screws can be inserted into the lateral masses of C1 instead.19,21 While the Magerl-Gallie transarticular screw wiring technique is commonly used to treat atlanto-axial instability,35 the screw-rod construct for C1-C2 pedicle screws has been shown to provide the same degree of biomechanical stability as that of the Magerl-Gallie transarticular screw-wiring construct.36 This, together with the great pullout strength of cervical pedicles screws, allows very rigid fixation for the cervical spine.18-37 For a solitary C2 metastasis, the level of fusion may be limited to C1 to C3 if the screws can be inserted into the lateral masses of C1 and the pedicle of C3. Fixation within the cervical region does not overly stress the bone-implant interface during head mobilisation, and thus reduces the chance of loosening. We found that polymethylmethacrylate cement is not necessary for patients treated with wires and bone grafts only, as has been advocated elsewhere.9,11 Preoperative CT scans are therefore imperative in order to delineate the extent of osseous destruction and the degree of misalignment.

If the lateral masses of C1 were eroded, fusion had to be extended to the occiput. The technique used to wire the cervical spine and the external occipital protuberance with corticocancellous bone did not provide the stability necessary for this group of patients38; occipitocervical fusion using a Luque rod and segmental fixation provided the rigid fixation required. Many modifications of this technique use a similar concept.13,14,16,17 Advantages include simplicity, secure fixation to the occiput, and the ability to contour the rod to adjust the occipitocervical angle and to correct any kyphotic deformity. Postoperative halo immobilisation was unnecessary in patients 2, 3, and 12 using this method.

We used occipital screws for occipitocervical fusion in patients 1 and 11. Because the pullout strength of the screws varies with the thickness and bony quality of the occipital cortices, these patients were immobilised with halo-vests for 2 months, which compromised their quality of life.

CONCLUSIONS

Malignant lesions in the upper cervical spine often involve the lateral masses, vertebral bodies, and the dens, and cause pathological fracture and subluxation giving rise to intense suboccipital neck pain. Spinal cord compression, nonetheless, is uncommon in this anatomical region. An aggressive approach utilising posterior stabilisation of the upper cervical region can improve quality of life by relieving pain and preventing neurological complications.

© 2005 Western Pacific Orthopaedic Association Provided by ProQuest LLC. All Rights Reserved.