Introduction

Foramen magnum meningiomas account for 2.5 % of all intracranial meningiomas and 4 % of those found in the posterior fossa; 90 % are located ventrally and ventrolaterally [13, 14, 20, 60–62]. Meningiomas arising in the foramen magnum are classified as a separate subgroup according to the first and still undisputed reference scheme devised by Castellano and Ruggiero in 1953 [18]. With continuing advances in imaging technology in the last decades, tumor involvement of the skull base area can now be more precisely delineated. The foramen magnum extends anteriorly from the junction of the lower and middle third of the clivus to the upper edge of the C2 body, laterally from the jugular tubercle to the upper aspect of the C2 lamina, and posteriorly from the anterior edge of the occipital bone to the C2 spinous process. Since foramen magnum meningiomas involve the clival area, they share several of the surgical difficulties associated with this location [6, 11, 25, 37, 38, 50, 52–54, 61].

Published surgical series addressed diagnostic problems before the 1970s and technical surgical refinements thereafter [5, 16, 17, 19, 21, 31, 39, 59]. Development of the dorsolateral approach and its later refinements provided better tumor exposure and enhanced safety in tumor removal. Nevertheless, controversy surrounds management in individual cases as regards tumor variation and range of surgical options in a multiple-step approach. The focus of this article is to draw attention to uncertainties and difficulties in the surgical management of foramen magnum meningioma.

Case material

The series included 64 patients (48 females and 16 males; average age, 59 years; range, 27–82) operated on for new tumors between January 1990 and September 2010 at the Department of Neurosurgery, Verona University Hospital, and retrospectively analyzed. The series included 29 patients of advanced age (>65 years). The mean duration of disease was 33 months (range, 1 month to 17 years). Symptoms and neurological signs are listed in Tables 1 and 2. The predominant signs at admission were quadriparesis, gait ataxia, sensory disturbance, and 9th-10th cranial nerve deficits, resulting in severe disability as assessed by the Karnofski score: 34 patients were able to carry out activities of daily living (>70), 11 only self-care (70–60), and 19 required total assistance (<60).

Table 1 Presenting symptoms
Full size table
Table 2 Changes in clinical picture
Full size table

Neuroradiological workup included contrast-enhanced magnetic resonance imaging (MRI) in all 64 patients, computed tomography (CT) in 56 (plus bone algorithm in 52), and bilateral vertebral angiography in 18 (Figs. 1, 2, and 3). The tumor position at the foramen magnum was: midline ventral in 24 cases and ventrolateral in 40, according to their extension, whether it was symmetric or not, often beyond the midline in both cases. The mean diameter of the mass measured along the sagittal plane was 35 mm. The extent of foramen magnum occupancy reflects the severity of brain stem compression, which in this series was <50 % in 11 cases, 50–70 % in 20, and >70 % in 33, leaving only a thin slice of brain stem and creating an enormous challenge for the surgeon and a considerable risk for the patient. Twelve patients presented with hydrocephalus accompanied in seven by signs of intracranial hypertension. The meningioma was supplied by an arterial feeder in 13 out of 18 patients. In our practice, angiography was carried out for exclusively diagnostic purposes early in this patient series. In selected cases, when MRI features suggested high vascularity, angiography was scheduled, and endovascular embolization was prepared. Embolization was performed in hypervascularized tumors with feeders from the external carotid artery or from branches of the vertebral artery. Clinical and radiological assessments were performed at different time points during the postoperative follow-up period (mean, 138 and 92 months, respectively) in all patients.

Fig. 1
figure 1

Exemplifying ventrolateral foramen magnum meningiomas with prevalent clival growth. Preoperative axial (a), sagittal (b), and coronal (c) T1-weighted MR images showing a right large enhancing mass exerting a considerable mass effect on the lower brain stem. Postoperative axial (d), sagittal (e), and coronal (f) T1-weighted enhanced MR images showing no residual tumor or brainstem re-expansion

Full size image
Fig. 2
figure 2

Exemplifying ventrolateral foramen magnum meningiomas with prevalent spinal growth. Preoperative axial (a) T1-weighted MR image showing a small left enhancing mass compressing and displacing the brain stem. The tumor leaves a comparably small corridor to the ventral rim of the foramen, requiring maximal bone resection medially to the sigmoid sinus and at the atlo-occipital junction, as demonstrated on the 3D CT scan (b). Intraoperative view at the dural opening showing wide exposure after bone resection (c). Postoperative axial (d) T1-weighted enhanced MR image showing no residual tumor or brain stem re-expansion

Full size image
Fig. 3
figure 3

Angio-MR image showing the relation of the meningioma to two critical vascular structures: the jugular bulb, laterally, and the vertebral artery, medially

Full size image

Results

Surgical approach

The patients were placed in a semi-sitting position. The patient's head was fixed in a Mayfield headrest, then cautiously flexed and tilted to the opposite side. The incision was made in a hockey-stick fashion. After downward reflection of the suboccipital muscles, the posterior arch of the atlas was resected from the midline to the vertebral groove, and the course of the artery between C0 and C1 was identified. Low suboccipital craniectomy was performed. The medial rim of the sigmoid sinus was unroofed up to the jugular bulb, and the foramen magnum was opened. The next steps were performed under an operating microscope. While protecting the underlying vertebral artery with a spatula, the posteromedial half of the occipital condyle and a lesser portion of the superior lateral mass and facet of C1 were drilled in just over half of the cases (n = 35). Then, 1–2 cm of the dura anterior to the entrance of the vertebral artery was exposed. Drilling was continued intracranially and rostrally until the jugular tubercle was flattened (n = 21). The hypoglossal nerve canal running below it was identified extradurally.

The dura was opened posterior to the sigmoid sinus, and the incision was continued downward and posterior to the entrance of the vertebral artery in a curvilinear fashion, thus allowing lateral mobilization of the vertebral artery when the dura was reflected and tented to the margins of the craniectomy. This maneuver also permitted retraction of the sigmoid sinus anteriorly. Transection of the dentate ligament may further enhance the working space by relaxing the brain stem, whose exposure is not required, leaving the dura covering it.

Operative findings

Presentation of the tumor and its relation to neural and vascular structures differed from case to case depending on the site of dural attachment and rostrocaudal extension. Several recurrent patterns could be identified: midline ventrally located tumors pushed the 7th through the 12th cranial nerves backwards; ventrolateral tumors pushed the ninth–tenth and possibly the 7th-8th cranial nerves upwards; the 11th was pushed backwards; and the 12th was often encased by the tumor. The vertebral artery was identified at its dural entry in most cases, but in 31 (48 %), it was encased by the tumor in this area, where it was ventrolateral in 23 cases and ventral in 8. In the remaining 33 cases, it was displaced posteriorly. In the cases of large tumors, the posterior inferior cerebral artery (PICA) was usually displaced medially and the anterior inferior cerebral artery (AICA) and basilar artery displaced superiorly. Tumor encasement of these structures created surgical difficulties (Table 3), compounded by excessive bleeding (n = 2, 3 %), fibrous and dense tumor consistency (n = 28, 44 %), and heavy calcification (n = 7, 11 %). The tumor showed aggressive behavior with invasion of the dura in 5 (8 %) cases, and absence of the arachnoid plane was noted in 21 (33 %) cases, involving the vertebral artery in nine, the basilar artery in five, and the brain stem in seven.

Table 3 Cranial nerve and arterial encasement
Full size table

Surgical results

Radical removal was achieved in 52 (81 %) patients; in the remaining 12, remnants were left attached to the perforators originating from the bottom of the basilar artery (n = 3) or to the vertebral artery (n = 10); in four of these ten patients, a heavy calcification closely adhering to the brain stem was left untreated. Only one elderly patient developed nonsurgery-related complications (bronchopneumonia). Operative mortality was nil. To minimize surgical risk, total removal was not attempted in some elderly patients. Long-tract signs (quadriparesis, hemiparesis, sensory deficit) and cerebellar deficits improved remarkably during the postoperative course in 74 and 77 % of cases, respectively; pre-existing cranial nerve deficits improved in 27 % of cases during the perioperative period (Table 2). New cranial nerve dysfunction (n = 23) developed, attributable to the 9th through the 12th cranial nerves in particular. All patients with dysphagia were treated with conservative management, and none underwent permanent ancillary surgery or vocal cord augmentation. Clinical oropharyngeal assessment and endoscopic evaluation of laryngeal function were performed as soon as the patient was able to cooperate, before and immediately after extubation. The aims are prevention of aspiration, adequate feeding, and rehabilitation. The first issue is addressed by keeping the patient in the supine position, preferably contralateral to the impaired side, encouraging self-aspiration and maximizing respiratory physiotherapy. Intensive and meticulous rehabilitation included orofacial exercise and swallowing therapy. The majority (71 %) were able to swallow a sufficient oral diet within 1 month after surgery, meanwhile requiring an endonasal feeding tube, and the remaining patients were independent within 3 months after surgery and required temporary gastrostomy.

Table 2 reports the pre- and post-treatment clinical characteristics. Functional recovery (Karnofski score >70) was achieved in all but the six patients who had presented with severe disability before surgery. All but one patient with total tumor removal showed no residual disease. This patient had an aggressive meningioma and was reoperated on with subtotal removal 8 years later; this is the only reoperation in this series. Three of the 13 patients with subtotal removal received gamma knife radiosurgery soon after the operation. All are presently under observation (mean, 88 months), and none has shown evidence of tumor regrowth.

Discussion

The prognosis of foramen magnum meningiomas has improved with continuing advances in surgical technique and approach. An initial turning point in the treatment of foramen magnum meningioma came with the advent of the surgical microscope and CT in the mid-1970s, followed by the development of lateral approaches and the advent of MRI in the mid-1980s [1, 29, 34, 47, 57, 58]. In 1978, Seeger was the first to briefly describe an extreme-lateral approach for craniospinal lesions, including suboccipital craniectomy and drilling of the occipital condyle and of the jugular tubercle [51]. In 1987, Gilsbach et al. reported their results in 12 patients operated on at the same institution [26]. The dorsolateral approach introduced by Koos et al. in 1985 and later described by Perneczski in 1986, a member of the same school [35, 44], resembles the lateral suboccipital approach first described by Heros in 1985 [30]. Both approaches improved the exposure of the lower clivus and the upper cervical canal for removal of midline lesions, with minimal retraction of the cerebellum and complete sparing of the brain stem. Since then, posterolateral approaches, however named (dorsolateral transcondylar or far lateral), have become a flexible corridor encompassing different degrees of craniocaudal and mediolateral exposure [7, 12, 22, 23, 46, 55]. Mastoidectomy, condylectomy, extent of cervical laminectomy, and vertebral artery mobilization may all be performed and tailored to the individual case [10, 56]. In their study, Salas et al. distinguished six possible variations of the dorsolateral approach: transfacetal, retrocondylar, partial transcondylar, complete transcondylar, extreme-lateral transjugular, and transtubercular, each specific to different pathologies and locations, reserving the partial transcondylar (n = 14), the retrocondylar (n = 5), and the extreme-lateral transjugular (n = 4) for meningioma [48]. The most controversial step in the technique involves mobilization of the vertebral artery, occasionally applied to increase anterior exposure in intradural tumors [2, 4, 7, 8].

The choice of patient position (supine, park bench, semi-sitting) will usually depend on the surgeon's preference and may influence the extent of bone resection. In our experience, the semi-sitting position remains the single most important step in the procedure. The semi-sitting position offers a view from below, thus reducing the need for lateral exposure while permitting tumor debulking from below and delivered downward, away from the lower cranial nerves [40]. With the patient so positioned, anteriorly located tumors require tilting of the head in order to lower and shift the contralateral anterior rim of the foramen magnum forward to bring it within the surgeon's visual field. We fully agree with other operators that overuse of a novel surgical approach may result in unnecessary additional risks for the patient.

Results of tumor resection

The main difficulties encountered in the surgical treatment of foramen magnum meningioma have been investigated in detail by operators highly experienced in treating skull base tumors. Table 4 compares three variables in surgical tumor removal (case series size, tumor resection, and mortality rates) gleaned from studies published between 1978 and 2010. The discrepancies closely reflect differences in population characteristics, since tumor recurrence, extradural growth, vascular encasement, and arachnoid scarring are all widely accepted as factors arguing against radical excision [4, 24, 49, 54]. In our series, for example, tumor fragments were left attached to the arteries and the brain stem because of the absence of an arachnoidal plane in some cases.

Table 4 Review of the literature on extent of removal and mortality
Full size table

Difficulties may be related to involvement of vascular components or the cranial nerves. Vascular dissection may be problematic in hard tumors encasing vessels and when arachnoid scarring occurs. In surgical planning, MRI imaging is useful for predicting vascular encasement, although it can slightly overestimate this feature and yields no information on cleavage. The vertebral artery carries the highest risk of injury, being involved during different stages of the procedure: exposure, drilling the condyle, mobilization, and debulking [9, 41, 60]. We did not mobilize the vertebral artery, leaving the peripheral venous plexus intact, in agreement with most operators who consider it a useless maneuver [13, 33, 40, 60]. A few surgeons may still use it in selected cases presenting with complications [4, 53]. Encasement of the PICA and the basilar artery also carries a higher risk of neurological sequelae from dividing the perforators. In meningiomas with vertebrobasilar artery encasement, extradural encasement and reoperation are associated with an elevated risk of vessel rupture and incomplete removal which occur in 41 and 51 % of cases, respectively [53, 60]. In our series, there were no cases of vascular injury.

Surgical technique is key to the management of tumor removal from neural structures. According to Sekhar et al., the brain stem, together with the cavernous sinus, was a recurrent site of remnants in meningiomas involving the clivus, with incomplete removal in 22 % of cases [52, 53]. The expectation of finding the lower cranial nerves in a certain position according to tumor size and position gives the surgeon the mental comfort indispensable for carrying out the operation. This is particularly true in cases where neural encasement is expected, as in large tumors. In ventrolateral tumors, the 11th cranial nerve is pushed posteriorly, the 9th–10th nerves elevated and stretched, and the 12th nerve encased or pushed downwards. Although cranial nerve position is fairly predictable in ventral tumors, these tumors still pose particular challenges for the surgeon. Only George et al. gave comparable importance to the position of cranial nerves classified according to anteroposterior tumor location [14, 24].

We found 65 cranial nerves embedded in tumors which were then dissected and freed along the arachnoidal plane, without causing intraoperative rupture. In their study on a larger series of clival meningiomas, Sekhar et al. reported similar results [52].

Clinical outcome

Perioperative mortality in recent surgical series was reportedly low or absent (Table 4). During the postoperative course, neurological impairment was mainly attributable to cranial nerve dysfunction. Dysphagia remains the most serious complication and is additionally burdened by risks linked to aspiration. Samii et al. reported general complications in 30 % of cases, with aspiration pneumonia the most serious complication (10 %). According to multiple regression analysis, tumor recurrence, arachnoid scarring, prevalent cranial extension, and absence of preoperative lower cranial nerve dysfunction were all found to be significantly associated with aspiration pneumonia [49]. Some studies reported 9th-10th cranial nerve impairment as the most common complication occurring in 43 to 60 % of cases [4, 33, 60], while two others found 12th cranial nerve deficit to be the most frequent complication (6 and 28 %, respectively) [40, 48]. In addition to lower cranial nerve dysfunction, Samii et al. mentioned transient 7th cranial nerve deficits (10 %), gait ataxia (15 %), and persistent paraparesis after anterior spinal artery injury in one case [49].

Overall neurological morbidity can be classified as transient or permanent. The ratio between the two is considered superior to global morbidity as an indicator of outcome quality (range, 27–62 %) [32, 45]. In two relatively recent studies, the prevalent permanent morbidity was 25 and 18 %, and the transient morbidity was 6 and 9 %, respectively [32, 45], while five other studies reported higher transient than permanent morbidity: 12 and 8, 35 and 22, 38 and 19, 50 and 5, and 61 and 2 %, respectively [9, 33, 40, 49, 60]. As a CSF fistula is often mentioned in the various series, it should be considered the most threatening local complication [3, 4, 9, 13, 32, 40, 42, 60] (Table 5).

Table 5 Review of the literature on complications and neurological morbidity
Full size table

Our findings confirm previous observations that postoperative impairment is due to cranial nerve dysfunction, especially of the 9th–10th and 12th cranial nerves (44 and 33 %, respectively), leading to life-threatening disturbance of respiration or impaired protection of the tracheobronchial tree, particularly during the early postoperative period. The difficulty in predicting this complication, either by preoperative compensation or clinical postoperative assessment, warrants careful management [60]. We prefer physiotherapist care rather than ancillary surgery if disability is not severe [32, 60]. With this strategy, mild aspiration occurred in only two cases with otherwise very successful results in our series. During the early postoperative course, 74 and 77 % of neurological signs associated with long-tract dysfunction or cerebellar deficits improved, whereas only 27 % of cranial nerve deficits did. Taken together, these data summarize the peculiarity of foramen magnum meningiomas: they may behave as either clival tumors or spinal tumors depending on the type and outcome of neurological dysfunction [33]. This observation is in agreement with other studies which classified patient outcome by tumor location and extension [13, 33, 49, 60]. This, together with a favorable transient/permanent neurological morbidity ratio, explains why the final outcome compared favorably between our and other series, despite severe clinical presentation and advanced age: 91 % of patients resumed a normal productive life within a few months after surgery, and only one patient was reoperated on during the 7-year follow-up period [9, 33]. We also concur with other operators on conservative management in the removal of fibrous tumors intimately adherent to cranial nerves or vessels or both, especially if the surgeon is less experienced, where gamma knife radiosurgery may be a valid ancillary treatment [32, 42, 45].

Conclusions

In this series, observed in the MRI era, mortality was nil, morbidity was very low, and outcome after surgery for ventral and ventrolateral foramen magnum meningioma was good. All patients underwent surgery in the semi-sitting position via the dorsolateral approach. A surgical approach optimized by tailoring the procedure to the single case and knowledge of the position of neurovascular structures can aid in the attempt to reach the gold standard: radical removal at the first operation. Nevertheless, incomplete resection may lead to a long steady state which warrants careful assessment of intraoperative risks. In most cases, functional outcome after long-tract and cerebellar decompression was very good. Transient morbidity was mainly attributable to lower cranial nerve deficits, which need to be adequately managed to prevent major complications.