Introduction

Traditionally, the clinical course of intracranial meningiomas has been a matter of debate and histological morphology and biological behavior of the tumour have been determinants of utmost importance [3, 8]. Nevertheless, considering basal meningiomas, anatomical relationships gain much more concern, regarding the cranial nerves and brain vasculature [15, 25]. The presence of multiple vital neurovascular structures and their anatomical variations make tumour removal much more difficult, and this has always been a dilemma in neurological surgery with variable levels of surgical risk [1, 16, 18, 27]. Considering these factors as a whole, one will recognize that surgery of these tumours requires the highest degree of state-of-the-art. Whatever the philosophy, expertise and experience of the surgeon, prognostication remains an unanswered question regarding surgical indication, selection of treatment modalities including radiosurgery, total or partial resection, and the surgical corridor of choice [13, 17, 20, 29]. Various factors have been proposed by several authors to predict surgical resectability such as peri-tumoural edema, presence of nourishing pial vessels, age of the patient, location of the tumour, duration of symptoms, preoperative neurological status, vessel encasement, tumour size, presence and severity of the associated hydrocephalus, extent of resection, and co-morbid conditions such as hypertension and diabetes mellitus [10, 12, 19, 26].

Accordingly, Levine et al (1999) have proposed a grading system to predict the extent of resection and outcome for these complex tumours [19]. This system has yielded a high degree of correlation between the extent of resection and the tumour grade. However, optic nerve, facial-auditory and caudal cranial nerve involvement have not been addressed. In this study, for the sake of simplicity and communicability, we have postulated a modified grading system based on a series of 124 patients trying to possibly achieve a more significant prognostic correlation.

Patients and methods

Between April 1996 and February 2003, 141 patients suffering from skull base meningiomas, who were candidates for elective surgery, have been referred to this center. Patients with a pathologically proven diagnosis were included, while those with an equivocal histology, co-morbid conditions necessitating non-surgical management and/or inadequate data were excluded from the study. Subjects of the investigation included 90 females (73%) and 34 males (27%) with a mean age of 48.2 years (F: M ratio = 2.6:1, Fig. 1). The topographic locations of the tumours as well as their relative frequency have been shown in Fig. 2. Patients were graded according to both the Levine-Sekhar (LS) and our modified grading systems. The LS grading system uses six variables for determination of the grade in which every positive criterion equals one score and the sum of these options makes the patient’s total score, (Table 1) while our modified system employs six criteria consisting of five based on regional surgical anatomy and pre-operative data as well as a former option employed in LS grading system, namely, history of radiosurgery (Fig. 3). The corresponding scores and grades of both systems have been tabulated in Table 2. Illustrative cases have been shown in Fig. 4 (A to C).

Fig. 1
figure 1

Number of cases in different decades of life according to sex. Note the female preponderance between 4th and 6th decades of life in which the female sex hormonal burst is at its zenith. This supports the former viewpoint of etiological role of female sex hormones in meningiomas

Full size image
Fig. 2
figure 2

Topographic prevalence of tumours, with the most common presenting symptom in parentheses

Full size image
Table 1 Employed variables of Levine-Sekhar and our modified grading systems
Full size table
Fig. 3
figure 3

Schematic diagram depicting the topographic location of cranial nerves and vasculature of the skull base and the corresponding score attributed to their involvement. Note that each score correlates with specific anatomical region within the skull base, the sum of which yields the total score

Full size image
Table 2 Corresponding grades and scores in Levine-Sekhar and our modified grading systems
Full size table
Fig. 4
figure 4

Illustrative cases showing skull base meningiomas with different grades. A, Sagittal paramedian T1-weighted MR image with Gadolinium enhancement of a clinoidal meningioma with grade II (anterior and middle fossa involvement, visual disturbance and 6th nerve palsy) in a 51-year-old male. The tumour was totally resected through a pterional approach. B, Axial T2-weighted MR image of a cerebellopontine angle meningioma with grade II (5th cranial nerve involvement, facial-auditory nerve complex involvement and disturbance of caudal cranial nerves) in a 50-year-old female. The tumour was subtotally resected through a retrosigmoid approach. C, T1-Weighted mid-sagittal MR image of a huge petroclival meningioma with grade III (papilledema, 6th nerve palsy, facial-auditory nerve complex involvement, hoarseness and multiple fossa involvement). The tumour was partially resected through a retrosigmoid approach. The patient doveloped massive cerebellar infarction post-operatively

Full size image

Statistical analysis included student’s t-test to compare independent samples and chi-square exam to assess statistical significance of non-parametric data. We also used linear regression, calculated the Pearson correlation coefficient for evaluation of the grading systems, and computed the z-score for comparison of the correlation coefficients of the original and modified grading systems.

Results

Mean hospital stay of the 124 eligible cases was 16 days. This was 12 days for the total resection group and 22.2 days for the partial resection group; and the difference was statistically significant (P<0.001). The mean duration of preoperative symptoms was 27±4 months, and this time did not show any correlation with tumour resectability (P = 0.68, â = 80%). We did not have a single case of cranial base meningioma recurring after previous radiosurgery; therefore this factor was not applicable to our patients. The variables utilized in the modified system for grading of cranial base meningiomas were (1) presence of optic apparatus involvement, i.e., visual loss, optic atrophy and/or papilledema, (2) signs of cavernous sinus neural involvement, i.e., involvement of cranial nerves III, V and/or VI, (3) 7th and/or 8th nerve complex involvement, (4) dysfunction of at least one of the caudal cranial nerves, (5) vessel encasement and/or multiple fossa involvement by the tumour and (6) recurrence after radiosurgical treatment. The presence of these abnormalities per se correlated significantly with sub-total resection of the tumour (P<0.05).

Of the total 124 patients, 66 (52%) had undergone gross total resection of the tumour and the tumour had been partially resected in the remainder 58(48%). Twenty two tumours fulfilled the necessary criteria for grade zero according to our modified grading system while six of them were not resected totally. Linear regression analysis was performed for totally resected tumour proportion (TRP), versus "score" which yielded the following regression equations:

$$TRP = - 0.128 \times {\left( {LS\,Score} \right)} + 0.7613\quad {\text{and }}TRP = - 0.1487 \times {\left( {Modified\,Score} \right)} + 0.8955$$

for the two grading systems. The Pearson correlation coefficients for scores of the LS and modified grading systems were found to be 0.9585 and 0.9976, respectively, where Table 2 their difference was statistically significant (P<0.01, Fig. 5).

Fig. 5
figure 5

Linear regression analysis for scores of totally resected tumours according to LS grading system and our modified grading system. Numbers on top of columns indicate the total number of patients having undergone gross total resection in each group (LS: Levine-Sekhar, TRP: Totally Resected Percent)

Full size image

Collapsing the score data of the two systems yielded new data (grades) being depicted in Fig. 6. The overall mortality and complication rates were 11.3% and 26% respectively. The most prevalent complications included (1) cranial nerve involvement (27%), (2) hemiparesis due to infarction (11.2%), (3) meningitis (4.8%), (4) CSF leakage (4%), (5) hydrocephalus and hematoma (2.5%), (6) cerebellar dysfunction (9.5%), and (7) non-neurological complications (7.5%). Fourteen patients passed away in the immediate post-operative period. Two deaths were not directly related to surgery (renal failure and pulmonary embolism), and cerebellopontine angle and sphenoid wing meningiomas had the highest mortality rate. However, more patients should be studied to conclude any statistical inference between grade and mortality/morbidity rates.

Fig. 6
figure 6

Linear regression analysis for grades of totally resected tumours according to LS grading system and our modified grading system. Numbers on top of columns indicate the total number of patients having undergone gross total resection in each group (LS: Levine-Sekhar, TRP: Totally Resected Percent)

Full size image

Discussion

Meningiomas of the cranial base have been under special focus prior to the evolution of highly sophisticated approaches and instrumental armamentarium for skull base surgery, and it is not surprising that these tumours still remain among the most challenging problems that mandate a special treatment tailored for each patient [2, 4, 9, 11, 19, 21, 22, 30]. Their management requires extensive knowledge, expertise and experience in the realms of histology, molecular biology, hormonal biochemistry and functional/surgical anatomy of the skull base [9, 23, 24], as well as a modern interdisciplinary facility. Nevertheless, what seems crucial to every neurosurgeon at the time of confrontation with such a tumour, is the answer to this question: “Can It Be Resected Totally?” There has been a great deal of industrious effort to find a logical and satisfactory answer to this question since Harvey Cushing described these challenging tumours for the first time, [6]. However, there have been, of course, many solutions proposed to confine the complexity of the matter.

Follow-up studies on natural course of petroclival meningiomas have shown that the extent of resection is an important factor in predicting the clinical outcome of the malady. Therefore prediction of the degree of resection accomplished pre-operatively may prevent a fait accompli at the time of surgery [12, 28].

Also, Cusimano et al, have performed a major retrospective study on the tumours of the cavernous sinus [7]. Their results have shown that meningiomas of the cavernous sinus are the most difficult tumours to resect, have the worst outcome among all tumours of the region and cause the most devastating complications after surgery [7]. Their investigation emphasizes the importance of prognostication for meningiomas of the cavernous sinus region, and presumably their results may be applicable to the whole cranial base.

Goel et al, have performed a great study on 70 tuberculum sellae meningioma cases. They proposed a justified grading system based on clinical and radiological criteria. Their clinical criteria included (1) duration of symptoms, (2) visual status of the patient before surgery, (3) other neurological deficits, (4) diabetes mellitus and hypertension regarded as co-morbid factors, and (5) age of the patient. Their grading system also uses such radiological criteria as the presence of peri-tumoural edema, internal carotid and/or anterior cerebral artery encasement, tumour nature and severity of the associated hydrocephalus if present [10]. Although it casts special scrutiny on tuberculum sellae meningiomas, the major drawbacks are lack of simplicity and applicability to the whole cranial base. The present modified system has both simplicity and bedside applicability based solely on pre-operative findings.

Sindou has proposed “peri-tumoural edema” and “presence of nourishing pial vessels to the tumour”, as factors influencing surgical prognosis in meningiomas [26], yet not suggested any classification for various locations of meningiomas remembering that basal meningiomas seem to be unique in many aspects. Besides, the “presence of nourishing pial vessels to the tumour” could not be assessed prior to surgery. Our modified grading system puts the cranial base meningiomas at the center of focus, and furthermore, is easily applicable prior to surgery.

The first investigation to construct a grading system for prediction of outcome and extent of resection in skull base meningiomas has been performed by Levine et al yielding a high degree of correlation between the grade of the tumour and degree of resection [19]. However, only three of the cranial nerves (namely III, V and VI) are addressed, and other neural structures such as the optic, facial, auditory and lower cranial nerves, whose significant contribution and correlation to tumour resectability have been proven in the current study, do not appear in the LS grading system (P<0.05).

One of the important determinants in resectability is the involvement of cranial nerves. There is a common belief among neurological surgeons that resection of a tumour with potential harm to the cranial nerves is much more difficult than that of a mass that has not produced nerve dysfunction [17]. It is prudent to assert that tumour resectability is not affected by the involvement of different cranial nerves in the same way.

We have found a significant correlation between visual disturbance, optic atrophy with subtotal resection of the tumour (P<0.05). Although discrete involvement of cranial nerves III, V, and VI well correlated with subtotal tumour resection (P<0.05), these variables have been included solely as a binary criterion, such that its value switches between 0 (no III, V and VI neuropathy) and 1 (at least one of the nerves is dysfunctional) (P<0.01). Trochlear nerve dysfunction has been omitted because of high probability of misdiagnosis of its dysfunction [19].

Paresis of the facial nerve was also found to be an ominous sign for tumour resectability, since it could not be easily preserved during surgery, and this is a reason to attempt subtotal resection. Besides, most of facial palsies caused by tumoural compression of the nerve are associated with some degree of auditory nerve involvement. This has been the basis for inclusion of these variables (CN VII and/or VIII dysfunction) as a single criterion. The presence of facial and/or auditory nerve dysfunction well correlated with subtotal resection of the tumour (P<0.05). This may be indicative of their dysfunction as a key factor in neurosurgical decision-making.

The lower cranial nerves IX, X, XI are usually involved in petroclival, cerebellopontine angle, and foramen magnum meningiomas. Dysfunction of any of these nerves has been assumed a single score, yielding a significant correlation with the extent of resection of the tumour (P<0.05).

Levine et al did not address the very nature of cranial nerve palsies in the subgroup having undergone radiosurgery so that the net effect of this factor, i.e., its influence on resectability may be either due to inherent difficulty of surgery, perceived at the first surgical attempt which led to radiosurgical treatment or secondary changes in tumour consistency induced by radiosurgery. On the other hand, cranial nerve dysfunctions directly related to radiosurgical complications may be erroneously attributed to the invasive behavior of the tumour and factitiously escalate the corresponding score and grade. Although none of our patients had a previous radiosurgery session for treatment of a cranial base meningioma, we did not exclude this variable considering its high predicting value in Levine’s survey [19]. Expectant studies may further evaluate this feature.

Levine et al have reported a series of 232 patients having undergone surgery for basal meningiomas and proposed a grading system based on preoperative data [19]. They have mentioned that involvement of the optic nerve and lower cranial nerves were found to be of little predictive value, ergo being omitted in their study possibly because of paucity of posterior fossa cases in Levine’s survey. Therefore larger number of patients might be necessary to yield statistical significance regarding these points [14]. On the other hand this sub-category comprised an important bulk in our patients (35.5%). Considering Fig. 1, one recognizes that the involvement of lower cranial nerves has been observed commonly in tumours involving lower regions in the posterior cranial base [5, 28, 31].This means that many tumours of this region might attain a low grade in the original system proposed by Levine et al while they could not be resected totally because of intricate neurovascular relationships, hence diminishing the predictive value of the LS grading system in large series of patients suffering from posterior fossa meningiomas. Although the difference between the Pearson correlation coefficient obtained from the LS grading system and that of our modified grading system seems to be negligible (0.9683 versus 0.990), analysis of the difference (z-score) unmasks a significant difference between them (P<0.01).

Grade zero patients did well and all except six underwent a total resection. The cause for subtotal resection has been limited access to the petroclival location in two patients and recurrent tumour in three. The last one was a jugular foramen meningioma in a 35-year-old lady without any preoperative cranial nerve symptoms.

Conclusion

Numerous studies have coined predictive factors for resection and outcome of different types of cranial base meningiomas. However few of them have considered cranial base meningiomas on the whole, obviously caused by complexity of the matter. These studies have focused on various aspects of these tumours and with the introduction of newer surgical approaches and the novel therapeutic technique of radiosurgery, new horizons have appeared in promotion of the outcome for these patients. Our study has tried to revise the LS system, possibly enhancing the simplicity and communicability as well as correlation significance.