CLINICAL STUDIES

Surgical Results for Meningiomas of the Craniocervical Junction

Medical School of Hannover, Neurosurgical Clinic, Nordstadt Hospital, Hannover, Germany

OBJECTIVE: We have undertaken a retrospective analysis of 38 patients who were operated on for 40 meningiomas of the craniocervical junction between September 1977 and August 1995 to determine which factors influenced resectability, complications, and postoperative outcomes.
METHODS: Radiological examinations, clinical data, and operation notes were evaluated, and additional follow-up information was obtained from outpatient examinations, telephone calls, and questionnaires.
RESULTS: Four groups could be distinguished according to dural attachment as follows: 1) 15 spinocranial meningiomas originated from the spinal canal and extended intracranially; 25 craniocervical meningiomas of intracranial origin were divided into 2) meningiomas of the lower clivus (10 patients with 11 tumors), 3) lateral meningiomas (11 patients with 12 tumors), and 4) posterior meningiomas (2 patients). Standard midline or lateral suboccipital approaches with opening of the foramen magnum and laminectomy of the involved cervical segments were sufficient for the great majority of tumors. In seven instances only, drilling the posterior third of an occipital condyle was needed. Twelve of 15 spinocranial meningiomas and 13 of 25 craniocervical meningiomas could be removed totally. One patient underwent ventriculoperitoneal shunting only. With a rate of 63% of totally removed and 30% of subtotally removed meningiomas in this region, we observed clinical recurrences for two patients only. Complications were encountered in 30% of patients, predominantly with recurrent and/or infiltrative or en plaque meningiomas. Whereas motor weakness and gait ataxia tended to improve postoperatively, cranial nerve deficits usually remained unaltered.
CONCLUSION: The relationship of the tumor to neighboring structures, i.e., the vertebral artery in particular, determines its resectability. We recommend using extreme caution with recurrent or en plaque meningiomas and tumors associated with extensive arachnoid scarring.
(Neurosurgery 39:1086­1095, 1996)

Key Words: Arachnoid scarring, Craniocervical junction, Meningioma

Meningiomas of the craniocervical junction may originate from the clivus, the cerebellopontine angle, or the occiput and extend into the spinal canal across the foramen magnum (craniocervical meningiomas) or may originate in the spinal canal and grow rostrally into the cranial cavity (spinocranial meningiomas) (3). Surgical removal poses several problems. The tumor may encase the vertebral, basilar, and their perforating arteries and cranial nerves, may invade bony structures of the craniocervical junction, or may be densely adherent to the brain stem. A variety of surgical approaches have been advocated to facilitate resection and limit postoperative morbidity. We present our experience with these challenging tumors over a period of 17 years and analyze which factors determine postoperative outcome, predispose to complications, and limit the resectability of meningiomas in this area.

PATIENTS AND METHODS

Meningiomas of the craniocervical junction were defined according to their tumor matrix, i.e., origin of the tumor or dura insertion. All tumors in this study arose between the level of the lower third of the clivus and C2 (5). According to their origin, four types could be characterized as follows (Table 2): 1) spinocranial (tumor matrix below level of foramen magnum and tumor mass ventral or lateral to brain stem and spinal cord) (Fig. 1), 2) craniospinal (tumor matrix intracranially), clivus (tumor matrix localized at the lower clivus and tumor mass ventral to brain stem and spinal cord) (Fig. 2), 3) craniospinal, lateral (tumor matrix localized in the cerebellopontine angle or the dura opening for the verterbral artery and tumor mass predominantly lateral to brain stem and spinal cord) (Fig. 3), 4) craniospinal, posterior (tumor matrix localized in the region of the cisterna magna and tumor mass dorsally to cerebellum and spinal cord) (Fig. 4).

Whereas the major tumor mass was situated intracranially in craniocervical meningiomas, it was found in the spinal canal of spinocerebral meningiomas (3). Additionally, encapsulated and en plaque growing tumors could be distinguished according to intraoperative findings. En plaque meningiomas were characterized by extensive dura infiltration, absence of a tumor capsule, and violation of tissue planes.

The approach consisted of a laminectomy or hemilaminectomy of the upper cervical spine, depending on the extent of the meningioma, and a craniectomy with opening of the foramen magnum either in the midline (midline approach) or toward the sigmoid sinus as a lateral suboccipital approach (lateral approach) in the semisitting position. The positioning, as well as the remainder of the operation, was performed under monitoring of somatosensory evoked potentials (SEPs). Depending on the localization of the tumor, the posterior third of an occipital condyle or facet joints were resected to gain a more lateral access. After dural opening, the arachnoid was inspected for evidence of scarring. Arachnoid dissection was performed only as required for safe removal of the meningioma. Care was taken to preserve the arachnoidal plane between tumor and surrounding nervous structures. In general, the meningioma was debulked first before dissection along these arachnoidal planes was performed to mobilize the tumor from vessels, cranial nerves, or brain stem. Whenever possible, the tumor matrix was excised with the tumor. If this was considered too difficult or hazardous, the matrix was cauterized with bipolar forceps. In cases with an infiltrative or en plaque growth pattern, no attempt was made to remove the tumor radically unless it was clearly separable from neighboring vessels or cranial nerves.

Total removal was defined as complete resection of the tumor mass with resection or cauterization of the tumor matrix. Subtotal resection was defined as resection of the tumor mass with small tumor remnants left behind at important structures, such as the vertebral artery, perforating arteries, or cranial nerves. Partial resection was defined as incomplete resection of the tumor mass.

Postoperatively, every patient underwent computed tomography and, since the advent of MRI, gadolinium-enhanced MRI after 3 and 12 months. Further scans were obtained if new clinical symptoms evolved or if tumor removal had been incomplete. Clinical examinations were performed before discharge and after 3 months for all patients. Additional follow-up information was obtained by further outpatient examinations, regular telephone calls, or questionnaires.

A recurrence was defined clinically as neurological deterioration after surgical treatment, independent of evidence for recurrent tumor growth on CT or MRI scans. Surgical morbidity was defined as a new postoperative neurological deficit or an aggravation of a preexisting symptom without subsequent recovery.

Means are presented plus or minus the standard deviation. For statistical analyses, Student\'s t tests for paired or unpaired variables were used, provideed the Komolgorov-Smirnov tes indicated normal data distribution. A multiple regression analysis was performed to determine which factors were of significant importance for the resectability of a particular tumor, the probability of recurrence, and postoperative outcome. A difference was considered significant if a P value of 0.05 was reached.

RESULTS

Patient data

Neck pain or headache (31%), gait ataxia (23%), and motor weakness (21%) were generally reported as the first symptom. The remaining patients complained about cranial nerve deficits (13%), dysesthesias (8%), or sensory changes (3%) at the beginning.

At admission, the major concerns were gait ataxia (28%), cranial nerve deficits (23%), and pain (21%). Motor weakness (15%), sensory changes (5%), and dysesthesias (5%) were mentioned less often as the main complaint. Craniocervical meningiomas were more likely to present with cranial nerve deficits than were spinocerebral meningiomas, whereas the reverse relationship was observed for pain, dysesthesias, and sensory deficits. Four patients presented with obstructive hydrocephalus. A more detailed overview on preoperative neurological symptoms is provided in Table 3. We did not observe differences in presentation for patients with encapsulated (n = 29) or infiltrative meningiomas (n = 10). The average Karnofsky score at presentation was 66 ± 19, indicating significant preoperative neurological deficits for the majority of patients.

Localization and extension of meningiomas

Approaches and surgical findings

All except three craniocervical meningiomas were removed using a lateral approach. One patient in bad general health with a recurrent clivus meningioma underwent ventriculoperitoneal shunting for accompanying hydrocephalus only. For exposure of 2 of 12 lateral and 4 of 10 clivus meningiomas, partial resection of an occipital condyle was required (Figs. 2 and 3). In one patient, a clivus meningioma was misdiagnosed, using CT scans before MRI was available, as a tumor of the fourth ventricle, and it was attacked via a midline approach. After partial removal, clinical symptoms caused by tumor growth developed after 8 years and a complete resection was performed via a lateral suboccipital approach (Table 5). Both posterior meningiomas were removed using a midline approach (Table 4).

Displacement of the vertebral artery was observed in seven instances and encasement in 16 (Fig. 3). Displacement and encasement of the basilar artery was found in two patients. Extensive arachnoid scarring around the tumor was found in 15 patients (5 patients at first surgery and all 10 patients with recurrent tumors).

Surgical results

Complications occurred in 30% of patients (Table 6). Postoperative tracheostomy was required in four patients, and prolonged feeding via a gastric tube was required in three patients. One patient required a gastrostomy. The most severe complication was aspiration caused by postoperative lower cranial nerve dysfunction affecting four patients. Two of these developed pneumonia and died, indicating a surgical mortality of 6%. Transient worsening of caudal cranial nerve function was observed in four patients, and another four patients with lateral meningiomas acquired transient facial palsies. Six patients demonstrated a temporary aggravation of their gait ataxia postoperatively. A multiple regression analysis revealed that recurrent tumor, arachnoid scarring, craniocervical type, and no preoperative lower cranial nerve deficit were independent predictors of postoperative aspiration (P < 0.01, respectively).

Two patients acquired a permanent postoperative deficit (surgical morbidity 5%). One 67-year-old patient with a lateral en plaque growing meningioma, which was exposed from a lateral suboccipital approach and found to be very adherent to medulla, cranial nerves, and the vertebral artery, acquired lower cranial nerve deficits that did not recover. An 18-year-old patient with neurofibromatosis Type 2 was operated on for a lateral craniocervical meningioma via a lateral suboccipital approach. No difficulties were encountered during surgery, with complete resection of the encapsulated tumor. Postoperatively, the patient demonstrated an incomplete paraparesis with a spinal level at L1. A spinal tumor or a compressive lesion in the area of operation was ruled out by myelography. The patient made an incomplete recovery.

Patients were followed for an average period of 21 ± 33 months (maximum, 10 yr). We observed two recurrences for a partially removed meningioma after 8 years and a completely removed meningioma after 11 months (Fig. 3). Both patients were operated on again, and a complete and a subtotal removal were achieved, respectively. Currently, both patients are free of recurrence.

In general, patients benefited significantly from surgery. Although the majority of preoperative cranial nerve deficits tended to remain unchanged postoperatively, sensory changes (t test for paired variables, P = 0.042), dysesthesias (t test for paired variables, P = 0.0134), pain (t test for paired variables, P = 0.0061), and gait ataxia (t test for paired variables, P = 0.0111) improved during the 1st postoperative year. The average Karnofsky score increased from 63 ± 17 to 73 ± 12 (t test for paired variables, not significant), indicating that the majority of patients were able to live independent lives postoperatively (Table 7).

DISCUSSION

Surgical approach