Transoral decompression for craniovertebral osseous anomalies : perioperative management dilemmas.
The surgical outcome of 74 patients, who underwent transoral decompression (TOD) for ventral irreducible craniovertebral junction anomalies between January 1989 to September 1997, was studied to evaluate the perioperative complications and problems encountered. The indications for TOD included irreducible atlantoaxial dislocation (n=24), basilar invagination (n=16), and a combination of both (n=35). Following TOD, occipitocervical stabilization using Jain's technique was carried out in 50 (67.5%) and atlantoaxial fusion using Brooks' construct in 18 (24.3%) patients. The pre- and postoperative radiology was compared to assess the adequacy of decompression and stability. The major morbidity included pharyngeal wound sepsis leading to dehiscence (20.3%) and haemorrhage (4%), valopharyngeal insufficiency (8.1%), CSF leak (6.7%) and inadequate decompression (6.7%). Neurological deterioration occurred transiently in 17 (22.9%) and was sustained in 7 (9.4%) patients. The mortality in six cases was due to operative trauma, exanguination from pharyngeal wound (one each), postoperative instability and inability to be weaned off from the ventilator (two each). Of the 47 (63.5%) patients available at follow up ranging from 3 months to 2 years, 26 (55.3%) showed improvement from their preoperative status while 14 (29.8%) demonstrated stabilization of their neurological deficits. Seven (14.9%) of them deteriorated. Though TOD is logical and effective in relieving ventral compression due to craniovertebral junction anomalies, it carries the formidable risks of instability, incomplete decompression, neurological deterioration, CSF leak, infection and palatopharyngeal dysfunction.
Congenital osseous craniovertebral junction (CVJ) malformations are the most common compressive lesions at CVJ.1 There is high incidence of CVJ anomalies in India,,,,,,,, with a different clinical pattern as compared to the west., Transoral surgery effectively deals with the ventral irreducible anomalies at the CVJ., It ensures direct exposure through familiar anatomy, and avascular median raphe and allows foramen magnum decompression in neck extended position. However, the technique carries the risk of neurological deterioration, due to surgery at the depth of a narrow field with vital cervicomedullary centres underneath. Passage through the potentially infected oral cavity may produce infection with the risk of CSF leak. Presence of myriad bony configurations in the region like angulation or rotation of atlas and asymmetry of facet joints may lead to incomplete decompression. Wide excision of the osteoligamentous components may produce instability., This study examines the perioperative problems encountered while performing transoral decompression (TOD) for congenital CVJ anomalies during an 8-year period and reviews the pertinent literature.
The surgical results and outcome of 74 consecutive patients who underwent TOD for congenital osseous CVJ anomalies between January 1989 and September 1997 at Sanjay Gandhi Postgraduate Institute of Medical Sciences, are presented. The series included 45 ( 60.8%) male and 29 (39.2%) female patients, whose age at presentation ranged from 4 to 78 years (mean 30 years). 18 patients (24.3%) were in the paediatric age group (less than 16 years). A detailed history and neurological examination was recorded.
Radiology : Pre- and postoperative lateral (in flexion and extension) as well anteroposterior (open mouth view) radiographs and tomograms of the CVJ were examined to assess the presence of atlantoaxial dislocation (AAD) and basilar invagination (BI).,, In lateral view, the spinal canal diameter at the foramen magnum,, the spinolamellar line and the clivus-dens angle (the obtuse angle between lines drawn along with clivus and projected into the cervical canal and along the posterior aspect of dens and body of axis) were also measured. Preoperative magnetic resonance imaging (MRI) determined the cervicomedullary compression and coexisting soft tissue anomalies. Pre- and postoperative comparison of adequacy of decompression and fusion was performed using intrathecal contrast CT scan to circumvent the artifacts caused by the metallic wires after posterior fusion on MRI. The radiological findings [Table I] were confirmed by direct observations at surgery.
Management protocol : In patients with AAD, in whom reduction was achieved either spontaneously by neck extension or following application of Crutchfield's cervical traction, posterior stabilization was performed without transoral surgery. TOD was undertaken in a irreducible AAD even after two weeks of cervical traction (the weights being progressively increased from 7 to 8% of body weight to a maximum of 7 kg, (n=24, 32.4%) ; b) BI (n=16, 21.6%) ; and, c) a combination of both (n=34, 45.9%). Adequate osteoligamentous decompression was ensured by visualizing a lax dura. The palate was usually split except in 5 (6.7%) patients with fixed AAD, where it was retracted into the nasopharynx. Tracheostomy was undertaken in 10 (13.5%) of our patients, the indication being prolonged ventilatory compromise due to tongue swelling and pooling of secretions.
After ventral decompression, posterior stabilization was performed in all the patients. Occipitocervical fusion was carried out in 50 (67.5%) patients using Jain's technique, and atiantoaxial fusion in 18 (24.3%) patients using Brooks' construct. Posterior fusion could not be done in six patients who expired following TOD. Between the two procedures, the patients were maintained on cervical traction. However, in 18 (24.3%) patients, TOD and posterior stablization were undertaken under the same anaesthetic., The patients were mobilized early following stabilization and a hard cervical collar was worn for atleast three months.
The duration of symptoms at presentation ranged from less than a month to 10 years (mean: 8 months). Pyramidal and posterior column derangements were the predominant clinical manifestations in the series. A history of quadriparesis following minor trauma was present in 12 (16.2%) patients. Restricted neck movements and specific stigmata (short or webbed neck, dysplastic face, high arched palate or asymmetry of the palpabral fissures) were significant clinical pointers towards the existence of CVJ anomalies [Table I].
A clinical comparison between the pre- and postoperative status of patients was made using the grading system used by Di Lorenzo. Improvement on deterioration was considered when there was a change of atleast one grade from the preoperative status as assessed at follow up ranging from three months to two years. A significant number of patients (n=50, 67.6%) were preoperatively dependent on others for their daily needs and with overt respiratory compromise (n=24, 33.4%) [Table II] [Table III].
Morbidity and mortality : The complications occurring after TOD are listed in Table IV. The perioperative mortality in the series was 6 ( 11.1 %). One patient sustained operative trauma leading to quadriplegia with respiratory arrest. Two patients could not be weaned off from the ventilator following TOD. All these patients had spastic quadriparesis with poor respiratory reserve and were in functional grade IV at admission. Two patients, who were being maintained on traction following ventral decompression and were awaiting posterior stablization, became quadriplegic with respiratory arrest while being turned, probably due to postoperative instability. One patient developed exanguinating secondary haemorrhage from the pharyngeal wound.
Outcome : 47 (63.5%) patients were available at follow up ranging from 3 months to 2 years. Out of these, 26 patients showed improvement from their preoperative status, while 14 showed stabilization of their neurological deficits and 7 deteriorated [Figure 1].
Reoperations : In 4 (5.4%) patients in the series, who did not show adequate neurological improvement at follow up, residual ventral compression was detected on postoperative CT studies. In 2 patients, the lateral remnants of axis were preventing the cord from moving forwards, while in the other two, the vertical decompression was inadequate [Figure 2]. A review of the preoperative radiology of these 4 patients revealed that two of them had a high BI with the body of axis invaginating into the foramen magnum, while two had an obliquely aligned odontoid with asymmetrical facet joints. In one (1.4%) adult patient, there was regrowth of bone in the region of the excised axis causing recurrence of symptoms after one and a half years. Following repeat TOD, three of these five patients showed marked clinical improvement.
Instability : TOD involves resection of the anterior longitudinal ligament, anterior arch of atlas, odontoid and part of the axis body, apical and alar ligaments and tectorial membrane. This diffuse osteoligamentous excision at the most mobile region of the spine predisposes to instability and requires internal stabilization.,, Dickman et al demonstrated significantly enhanced translational movements of the atlas in the anteroposterior axis and angular movements during flexion/extension and lateral bending. Keeping this in mind, after TOD, posterior fusion using wire and autologous rib graft was performed in all the patients in this series. Two patients, awaiting posterior stabilization following TOD, developed quadriplegia due to instability, while being turned to prevent bed sores. Since then , we have preferred to perform ventral decompression and posterior stabilization under the same anaesthetic, as recommended by Crockard et al.,
However, the additional posterior procedure can add significant morbidity., Dickman et al have observed that congenital AAD are more stable than rheumatoid or traumatic dislocations and require internal stabilization in only 45% of cases. We retrospectively identified 8 patients in this series in whom no mobility or distraction was detected between the atlas and axis after traction and following TOD. The immobility could have been due to articular and periarticular adhesions and fibrosis causing lateral atlantoaxial joint osteoarthritis.3 Perhaps a stabilization procedure was not warranted in these patients. However, in case posterior fusion had not been undertaken after TOD, a long term followup with a high index of clinical suspicion (especially for the advent of severe neck or suboccipital pain or neurological deterioration) as well as meticulous radiological monitoring, would have been necessary to detect incipient instability.
Radiological assessment of instability is extremely difficult after ventral decompression as the bony landmarks (anterior arch of atlas and dens) have been excised. On dynamic lateral flexion and extension views of the CVJ, an excessive displacement of the spinal-cranial axis, change of the clivus dens angle or spinolamellar line; and on AP view, an offset at the lateral occipitoatlantoaxial joints, may be used as indicators of instability. In difficult circumstances, a three dimensional CT scan or MRI may be required.
A ventral stabilization procedure would obviate the need for additional surgery for posterior fusion., However, it was not undertaken due to the propensity of devitalized bone in a potentially contaminated area to become infected. Moreover, extensive excision of the stabilizing structures at the CVJ during TOD prevent the occurrence of a stable anterior bony union.
Inadequate decompression : The transverse exposure from lower clivus to C 2-3 interspace is limited to 3-4 cm due to the emergence of hypoglossal nerves, vertebral artery and the eustachian tube laterally. Rostral exposure of the clivus during TOD is facilitated by soft and/or hard palatal split or open-door maxillotomy, and caudal exposure (especially with mouth opening less than 2.5 cm) by median labiomandibular glossotomy. The end point of anterior decompression in transoral surgery is heralded by the dural bulge following excision of the tectorial membrane. The four patients in the series, who had residual compression after TOD had either a high BI or an oblique odontoid due to C 1-2 facet joint asymmetry associated with a malrotated atlas The existence of these anomalies made intraoperative assessment of adequacy of decompression difficult. Consequently, after TOD, the residual remnants of axis prevented the cord from moving forwards. The sublaminar posterior fusion further compromised the canal diameter.
These initial setbacks prompted certain steps in order to ensure adequate foramen magnum decompression. A preoperative cervical traction was used to distract the odontoid from the foramen magnum. The identification of midline is crucial as an asymmetrical approach or a disregard of the atlantal rotation may lead to incomplete decompression or vertebral artery injury. Therefore, a wide exposure of the bony elements was carried out upto the facet joints on either side. Lateral decompression was ensured by drilling the odontoid and axis beyond the lateral dural margins [Figure 3]and vertical decompression by removal of the odontoid and body of axis, rostral to their intersection with Weckenheim's clival canal line projected into the cervical canal [Figure 4]. In patients with rotation of atlas on axis, after rendering the atlantoaxial complex mobile by ventral decompression and opening of the lateral joints, the centre of posterior arch of atlas was approximated to the spine of axis during posterior fusion to facilitate a three dimensional reduction of the deformity., Cerebrospinal fluid leak : Five patients in the series had CSF leak. A high BI in three and AAD causing severe dorsal angulation of the dens in two, were the predisposing factors. Perhaps, dural infolding over the tip of the odontoid precipitated the dural breach. To prevent this complication while drilling the odontoid, Pasztor's recommendations were followed. In the deeper part of the dens, the steel drill was changed for a diamond drill which prevented damage to the posterior longitudinal ligament or the dura when the drill penetrated the posterior cortical layer of the dens. The dens was resected in its whole width at its base to separate it totally from the body of the axis. Drilling this part was easier because the spongy bone at the base of dens has lower biomechanical strength due to a bone mass reduction of 55% and a poor trabecular interconnection as compared to the rest of the odontoid. Then, the posterior cortex of the dens was pulled up from the posterior longitudinal ligament and dura followed by removal of its apex. A Valsalva maneuver performed by the anaesthesiologist helped in localizing the site of CSF leak. A small leak in the dura usually healed with lumbar drainage. Larger leaks required a fascia-fat patch, CSF drainage and oral acetazolamide administration. Thrombin-fibrin glue has also been found to be effective for dural patch graft of autologous fascia or oxidized cellulose., Occasionally, epithelization of the tract may occur down to dura, causing persistent meningitis requiring re-exploration and separation. A vascularized sternocleidomastoid flap or transposition rotation flap based at the nasal septum may be utilized to fill the defect in the posterior pharynx.
Neurological deterioration : Fifty patients in the series were preoperatively in grade III or IV, being partially or totally dependent on others for their daily needs. The repetitive trauma due to craniovertebral anomalies leads to anterior horn cell destruction, gliosis of gracile and cuneate nuclei and demyelination of the corticospinal tracts and posterior columns., Stagnant hypoxia secondary to venous stasis or occlusion of the vertebral or spinal arteries,,, and preexisting microscopic intracranial abnormalities also contribute to the neural damage. Thus, 21 of the 47 patients seen at follow up did not show significant neurological recovery. Twenty four patients had an overt respiratory compromise. Many others probably had the syndrome of afferent respiratory dysfunction in which, following injury to the reticulospinal tracts in the ventrolateral quadrants of the cervical cord segments one to three, normal facilitatory and modulatory feedback pathways are no longer available to make fine adjustments in the respiratory servo mechanism. Thus, gross responses are handled satisfactorily so that routine blood gases and vital capacity are deceptively normal and the condition remains undetected preoperatively. However, even minor trauma on an already compromised cord may cause respiratory deterioration. Sleep normally depresses all vital functions exacerbating the respiratory depression.,, The common features in patients who deteriorated and those who could not be weaned off from the ventilator following transoral surgery were the presence of advanced spastic quadriparesis and respiratory compromise. On MRI, they showed evidence of marked spinal cord compression with thinning of the cord and hyperintense cord signals. We agree with the observation of Tuite et al that the higher rate of neurological morbidity may be related to greater severity of preoperative neurological deficits. One of our patients had transient haemodynamic instability and bradycardia during TOD. The developed quadriplegia with respiratory arrest following reversal from anaesthesia. This clinical syndrome of complete cervicomedullary transaction could have been due to accentuation of cord damage by the recurrent posterior displacement of the odontoid while drilling. However, Dastur et al have observed respiratory instability and quadriplegia following posterior decompression of foramen magnum and have implicated intramedullary haemorrhage from the sulcal branches of anterior spinal artery due to the sudden release of pressure. The possibility of this `release haematomyelia' causing complete cervicomedullary transaction in this patients cannot be ruled out. Spinal cord injury during the sublaminar passage of wires or instability of CVJ during repositioning for posterior stabilization may also add to neurological injury. We also feel that drilling bones with normal cortical margins may be technically more demanding than removal of partially destroyed bones in rheumatoid arthritis and neoplastic conditions and may in part be responsible for our higher rate of neurological deterioration when compared to the previously reported series.,,,,, To prevent this, while drilling the odontoid, a lateral rather than a downward pressure was applied and a thin posterior cortical surface of odontoid was left which was then removed after elevating it from the posterior longitudinal ligament. Neurophysiological monitoring with evoked somatosensory or brainstem auditory potential helps in predicting potential brain-stem or cord injury.
Pharyngeal Sepsis : Transoral surgery is based on the principle that local resistance of the oral tissue to its own bacterial flora facilitates healing. However, a high incidence of wound sepsis and dehiscence was noted in the series despite pre- and postoperative oropharyngeal toilet with antiseptic mouth washes, systemic antibiotic administration based on culture and separate muscle and mucosal layer closure with vicryl. Four patients had delayed retropharyngeal bleeding, which in one was sufficient to cause exanguination and consequent cord ischaemia leading to quadriplegia. This bleeding could have been from the vertebral venous plexus or from a false aneurysm of the vertebral artery secondary to sepsis., The approach through the potentially infected oral cavity, retropharyngeal haematoma due to straying from the midline and dividing the constrictor muscles rather than the midline avascular raphe, excessive usage of cautery to mobilize longus coli from the vertebral body and superinfection due to anaerobic and candida organisms can contribute to sepsis. The propensity to pharyngeal wound dehiscence increased in nutritionally compromised patients. Decreased oral intake further contributed to delayed wound healing. Attempts to combat this complication included a strict midline approach, avoiding excessive cautery and meticulous pharyngeal wound closure. A large retropharyngeal cavity was obliterated with fat and the last suture at the caudal end of the pharyngeal wound was left open in order to permit drainage and prevent haematoma formation. Early ambulation of the patients prevented pooling of saliva at the apex of the incision which is the weakest point. In case of dehiscence, healing was allowed to proceed by secondary intention as extensive mucosal oedema and infection preclude any attempts at secondary suturing.
The oral cavity may be avoided by the use of transcervical, transclival, extrapharyngeal,, anterolateral or transcondylar approaches to the CVJ. However, the obliquity of the approaches distorting the perspective, extensive soft tissue dissection especially the facial and hypoglossal nerves and vascular structures of the neck, interposition of the angle of jaw and the inability to achieve sufficient decompression especially on the contralateral side prevented us from adopting these procedures. Moreover, the transcondylar approach further destabilizes the CVJ by resection of the occipital condyles.
Rhinolalia and nasal regurgitation : The soft palate had to be split in order to resect the lower clivus to gain access to a highly placed odontoid. However, in 5 patients, the palate was retracted into the nasopharynx using a silk ligature brought out through the nose, and not split. Rhinolalia and palatal wound dehiscence was corrected by secondary suturing of the palate. Persistent nasal intonation and difficulty in swallowing after TOD may be due to abnormal palatal and pharyngeal closure resulting from a scarred pharynx, a large dead space in the posterior pharyngeal wall or lower cranial nerve deficits. Corrective measures include palatal prosthesis or pharyngoplasty.
Transoral surgery achieves anterior decompression in ventral irreducible osseous anomalies at the CVJ by the most direct approach. The resultant instability requires posterior stabilization. Failure to recognise the spectrum of bony configurations at the CVJ occasionally leads to remnant osseous compression after TOD, preventing forward movement of the dura and cord. Neurological deterioration and respiratory compromise may occur in the presence of marked spinal cord compression. CSF leak and palatopharyngeal dehiscence secondary to sepsis add to the morbidity. However, a good chance of neurological improvement after effective anterior decompression makes this formidable procedure a worthwhile one.