Introduction

Meningiomas account for 70% of all benign tumours arising in the foramen magnum [15, 28], and meningiomas in this region account for 0.2–3.2% of all meningiomas [1]. Recent developments in surgical techniques have improved the surgical outcomes associated with such tumours [3, 8, 13, 21–23, 27]. Several approaches for the resection of meningiomas have been proposed and modified; these include the lateral suboccipital [20], transcondylar [4], condylar fossa [14], extreme lateral [23], and transoral approaches [16]. Meningiomas in the lower clivus cannot be clinically distinguished from those in the foramen magnum because of the anatomical continuity between the two types of tumours. The latter tumours may occasionally extend superiorly into the lower clivus, and the surgical outcomes in such cases have not been discussed in detail. Therefore, we included lower clival meningiomas in the present series study on foramen magnum meningiomas (FMMs). We aimed to clarify the factors influencing the surgical and follow-up outcomes of meningioma patients who underwent subtotal resection.

Patients and methods

Patient population

We retrospectively examined 23 consecutive cases of meningiomas arising in the foramen magnum and lower one-third of the clivus. All the concerned patients (9 men and 15 women) were surgically treated at the Department of Neurosurgery, Keio University Hospital, between September 1991 and July 2008. The mean age of the patients was 55.5 years (range, 26–70 years). The most common presenting symptoms and signs were paraesthesia and hemiparesis, each of which was observed in eight patients. Paresis of the lower cranial nerves was observed in six patients (glossopharyngeal and vagal nerve palsy in three patients, accessory nerve palsy in one, and hypoglossal nerve palsy in two), hearing disturbance in three, and facial nerve palsy in one (Table 1). Three patients had no clinical symptoms before the operation. All patients except one underwent surgery as the initial treatment.

Table 1 Preoperative signs and symptoms
Full size table

Neuroimaging and classification

We retrospectively reviewed the patients’ clinical records and the findings of radiological examination [magnetic resonance (MR) imaging]. All the tumours originated from the region that has been defined as the foramen magnum by George et al. (anteriorly, from the lower one-third of the clivus to the upper edge of the axis; laterally, from the jugular tubercle to the C2 laminae; posteriorly, from the anterior edge of the squamous occipital bone to the C2 spinous process) [8, 9]. Using a modified version of the classification systems described by George et al. and Samii et al., we anatomically classified the meningiomas into the following three types according to their precise location in the foramen magnum and lower clivus: lateral, midline and lower clival (L-, M-, and C-type tumours, respectively; Fig. 1) [6, 8, 9, 20]. L-type meningiomas (14 cases) were attached to the anterolateral rim of the foramen magnum. M-type menigiomas (three cases) were attached to the anterior midline rim of the foramen magnum and caused deviation of the brain stem posteriorly. In C-type meningiomas (six cases), the bulk of the tumour was located on the lower one-third of the clivus. One patient with neurofibromatosis type 2 (case no. 6) had bilateral tumours on the rim of the foramen magnum. The mean size of the tumours was 26.5 mm (range, 12–50 mm). Extradural invasion was noted in three cases—around the dural ring of the vertebral artery (VA) in two cases and around the hypoglossal canal in one case.

Fig. 1
figure 1

Type of the tumour location: L-type meningiomas were attached to the anterolateral rim of the foramen magnum. M-type menigiomas were attached to the anterior midline rim of the foramen magnum and caused deviation of the brain stem posteriorly. In C-type meningiomas, the bulk of the tumour was located on the lower one-third of the clivus

Full size image

Surgical approach

All the patients were positioned laterally. One of three surgical approaches, namely, the lateral suboccipital (condylar fossa) approach [14, 25], the transcondylar approach, or the presigmoid transpetrosal approach, was selected depending on the type and size of the tumour. C1 hemilaminectomy was also performed in the case of L- and C-type tumours. The jugular tubercle was epidurally resected in four cases where the tumours had extended into the midline lower clivus. Resection of the condyle was limited to the posterior quadrant. The VA was not transposed so as to avoid possible injury to the perforating arteries arising from the VA (Fig. 2). A staged operation via the presigmoid and transcondylar approaches was attempted in the case of one patient (case no. 19) with a large C-type tumour.

Fig. 2
figure 2

Case no. 14. a Preoperative MR image of an L-type tumour compressing the cervicomedullary junction. b Graphic representation of the operative view. A small perforating artery arising from the VA supplies blood to the spine (arrow). c Postoperative MR image showing the site after total tumour resection

Full size image

Clinical and radiological follow-up

For all the patients, follow-up clinical and radiological examinations were performed every 6 months to check for tumour regrowth or recurrence. The morbidity at the time of discharge and the Karnofsky performance status (KPS) at the last follow-up examination (mean interval, 42.8 months) were evaluated. The Kruskal-Wallis test was used to analyse the relationship between surgical radicality and the tumour type. Logistic regression analysis was used to examine the influence of variables such as age, sex, tumour type, tumour size, dural invasion, the surgical approach, and the extent of tumour resection on the permanent morbidity rate.

Results

Gross total tumour resection (GTR) was achieved in 15 patients (65.2%), the GTR rate being 71.4% (10/14), 100% (3/3), and 33.3% (2/6) for L-, M-, and C-type tumours, respectively (Table 2). No significant difference in the resection rate was observed between tumour types in Kruskal-Wallis test. Incomplete resection was attributable to extradural invasion or adhesion of the tumour to the lower cranial nerves, the brain stem, or the VA (Figs. 3 and 4). There was no surgical mortality in this series. However, one patient with an M-type tumour (case no. 15), who was operated in 1992, i.e., before the introduction of the skull base approach, developed respiratory failure 3 days after the surgery and was bedridden. The KPS improved in 14 cases (60.8%) and deteriorated in 4 cases (17.4%). None of the patients who had undergone partial condylar resection complained of instability or restriction of neck movement. Long tract signs that were noted preoperatively largely improved after tumour resection. Hemiparesis, tetraparesis, paraesthesia, and hypaesthesia improved in 63% (5/8), 100% (1/1), 88% (7/8), and 100% (3/3) cases, respectively. However, one patient with a large C-type tumour (case 19) developed hemiparesis after the surgery. Transient and permanent cranial nerve deficit occurred in seven (30.4%) and four patients (17.4%), respectively. Dysphagia occurred in seven patients, two of whom developed permanent cranial nerve deficit; hoarseness occurred in three patients, 1 of whom developed permanent cranial nerve deficit; abducens nerve palsy occurred in four patients, two of whom developed permanent cranial nerve deficit. Permanent paresis of the lower cranial nerves and abducens palsy occurred only in the case of C-type tumours (Table 2). Further, of three patients who developed severe neurological deficits, two had C-type tumours. Logistic regression analysis revealed that tumour extension to the clivus (as in C-type tumours) was a statistically significant, independent factor influencing the permanent morbidity rate (p = 0.005), while age, sex, tumour size, dural invasion, surgical approach, and resection rate were not significant factors in this regard.

Table 2 Data disturibution of 23 cases
Full size table
Fig. 3
figure 3

Case no. 4. a Preoperative MR image showing the left VA surrounded by the tumour (arrow). b Preoperative MR image in the sagittal view. c Postoperative computed tomography (CT) scan showing the residual tumour behind the left VA (arrowhead)

Full size image
Fig. 4
figure 4

Case no. 23. a Preoperative MR image showing a large C-type tumour. b and c Postoperative MR images showing the residual tumour dorsal to the brain stem (arrow)

Full size image

At the last follow-up examination, at a mean of 42.8 months after the surgery (range, 2–186 months), the mean KPS score was 89.5 (range, 30–100). All patients except two (91.7%) were caring for themselves. In 50% of the patients with C-type tumours, the KPS score was less than 80.0 (Table 2). Regrowth of the residual tumour was noted in only one of seven patients who underwent subtotal resection (case no. 11; recurrence rate, 4.3%). In this case, a little tumour was intentionally left because of its adhesion to the VA. Within 40 months thereafter, the residual tumour grew back to the size it was at the time of the first operation. The second operation was not successful because of the high degree of adhesion, and conventional radiation therapy was subsequently initiated. The tumour was histopathologically diagnosed as a clear cell meningioma with high proliferative activity (MIB-1 index, 10%; Fig. 5).

Fig. 5
figure 5

Case no. 7. a Preoperative MR image. b Postoperative MR image showing a small residual tumour around the VA (arrow). c Histopathological specimen of a clear cell meningioma. d MR image obtained at the time of tumour recurrence

Full size image

Discussion

Surgical approach

Ventrally located FMMs are confined to a space surrounded by the brain stem, lower cranial and cervical nerves, and vertebral artery; minimal retraction of these structures is warranted for tumour resection. In order to avoid severe surgical complications similar to those observed in case no. 15 in 1992, the transcondylar approach has widely been applied to reduce the damage that retraction causes to the brain stem. Among the early cases encountered in the present series, the transcondylar approach was widely indicated for tumours located beyond the midline of the ventral foramen magnum. Thereafter, this approach was reserved for relatively small M-type tumours that did not compress the brain stem laterally but did so posteriorly. In cases of lateral deviation of the brain stem (L-type tumours) and those of large M-type and C-type tumours, the surgical corridor was wide enough for tumour resection to be performed without condylectomy (Fig. 1). Samii et al. reported that in most cases, the time-consuming process of extensive drilling into the occipital condyle is not required to achieve adequate tumour exposure [20]. Moreover, the traditional suboccipital approach has proven effective for the treatment of FMMs in several previous series [2, 10, 11, 14, 17]. Suhardja et al. calculated and compared the area of the surgical window obtained when the lateral suboccipital approach and the transcondylar approach were used on cadaveric heads; they concluded that the latter approach afforded a significantly larger surgical field [5, 24]. However, the space created by the tumour was not considered in these previous studies. We emphasise that L-type tumours and large M-type and C-type tumours can be treated without condylectomy.

The necessity of VA transposition has been debated. Some surgeons suggested that transposition of the third part of the VA is essential [1, 13, 18, 19], while others noted that it is not necessary, especially in the case of intradural tumours [3, 4, 8]. Samii et al. reported that compromise of the small perforating arteries arising from the VA could be an important risk factor for postoperative cranial nerve palsy [20]. Therefore, the VA was not transposed in the present series in order to avoid injury to these arteries.

Clinical outcomes

In the present analysis, we identified tumour extension to the lower clivus as the sole factor influencing the permanent morbidity rate. When an appropriate surgical approach was selected, the tumour laterality did not influence the morbidity rate or the surgical radicality. George et al. classified FMMs with regard to three parameters: one in the axial plane, one in relation to the VA, and one in relation to the dura mater [6, 8, 9]. They compared the surgical outcomes in the case of tumours located anterior and lateral to the foramen magnum and found that the tumour location was not a statistically significant factor influencing the outcome. They also compared the surgical radicality in the case of anterior and lateral tumours [8] and concluded that this parameter is not influenced by the tumour type, but rather by extradural extension because of the invasive nature. Arnautovic et al. studied 18 FMM cases and identified radical tumour resection and a higher preoperative KPS score as significant factors influencing the clinical outcomes [1]. Further, Cushing et al. classified FMMs into the craniospinal and spinocranial types depending on their site of origin [7]. Thereafter, Samii et al. classified FMMs into four categories: spinocranial, craniospinal-clivus, craniospinal-lateral, and craniospinal-posterior [20]. They identified tumour extension to the clivus as an independent predictor of postoperative aspiration, because tumours exhibiting such invasion involved the lower cranial nerves and perforating arteries arising from the VA. Except with regard to tumours originating from the posterior margin of the foramen magnum (classified as craniospinal-posterior tumours), the classification system used by Samii et al. and their conclusions are very similar to those in the present study. In the case of tumours extending to the lower clivus, preservation of the lower cranial nerves and the arteries arising from the brain stem should be given priority over total tumour resection for improving the patient’s quality of life, because subtotally resected benign meningiomas have a very low recurrence rate. One of the reasons of the favourable outcome in this series might be due to preservation of small spinal arteries from the vertebral artery. We did not routinely perform intraoperative electrophysiological monitoring; however, in cases in which those must be sacrificed, the risk of brain stem ischemia should be evaluated beforehand by the electrophysiological monitorings [12, 26].

Conclusion

The use of an appropriate surgical approach to treat the FMMs resulted in good surgical outcomes with a mean KPS score of 89.5 and a mean recurrence rate of 4.3%. However, tumour extension to the lower clivus significantly increased the morbidity rate. Our results suggest that in the case of C-type meningiomas, which have a low recurrence rate, the degree of tumour resection should be carefully considered so as to avoid worsening the preoperative condition, except in cases of pathological malignancy.