Long-term follow-up of 287 meningiomas in neurofibromatosis type 2 patients: clinical, radiological, and molecular features

Abstract

Decision-making criteria for optimal management of meningiomas in neurofibromatosis type 2 (NF2) patients is hampered by lack of robust data, particularly long-term natural history. Seventy-four NF2 patients harboring 287 cranial meningiomas followed up for a mean period of 110.2 months were studied retrospectively. The median number of meningiomas per patient was 3. The mean maximum diameter of meningiomas at diagnosis was 14.3 mm, with a mean annual growth rate of 1.5 mm. Sixty-six percent of tumors showed no or minimal growth. In a subgroup of patients with 3D MRI, 7.3% of meningiomas (28% of patients) had a volumetric growth rate 20% or more per year. Twenty-five de novo meningiomas appeared during the follow-up (8.7%) and demonstrated a higher growth rate than other meningiomas (6.6 mm/year). Fifty-six meningiomas (23%) in 34 NF2 patients (45.9%) were operated on during the follow-up period. Among symptomatic resected meningiomas, grades II and III tumors were found in 29% and 6% of cases, respectively, with a remarkable intratumor histological heterogeneity. Single nucleotide polymorphism array analysis of 22 meningioma samples in 14 NF2 patients showed increasing chromosome instability with increasing grade, the most frequent losses being on 22q, 1p, 18q, and 6p. This study provides clues to improve tailored treatment of meningiomas: de novo and brain edema-associated meningiomas require active treatment. Future clinical trials in NF2 need to focus specifically on meningiomas as the primary endpoint and should include patients with meningiomas growing 20% or more per year in order to assess new treatments.

Neurofibromatosis type 2 (NF2) is a rare genetic disorder (birth incidence 1/33 000) characterized by the development of multiple benign tumors of the nervous system.¹ The hallmark of NF2 is the development of bilateral vestibular schwannomas. Meningiomas are the second most frequent tumor type in NF2 and are found intracranially in 45% to 58% of patients.² The cumulative incidence of meningiomas was shown to be close to 80% by 70 years of age in a cohort of 411 patients with proven NF2 mutation.³ Progressive enlargement of the tumor leads to focal or generalized seizure disorders or neurological deficits caused by compression of adjacent neural tissue. NF2-associated meningiomas are often multiple.^4–7 The presence of meningiomas is a marker of disease severity: The relative risk of mortality is 2.5-fold greater in people with meningiomas compared with those without such lesions.⁸ Current literature provides a clear delineation of the natural history, therapeutic modalities, and timing of NF2-associated vestibular schwannomas.^9–12 In contrast, little is known about the natural history of NF2-associated meningiomas.¹³

This study describes the natural history and pathological and molecular features of meningiomas in a large series of NF2 patients with long-term follow-up. These data will potentially serve to establish a tailored individual treatment strategy.

Patients and Methods

Between 1997 and 2009, 119 patients with diagnosed NF2 were routinely followed up at a tertiary NF2 referral center, with annual clinical examination including audiometry and cranial MRI. Diagnosis of NF2 was based on the Manchester criteria⁴ or the identification of an NF2 mutation via blood test in 1 patient with multiple meningiomas. Data were obtained retrospectively from clinical records. The local institutional review board approved the retrospective chart review and tumor-specimen collections. All living patients provided written informed consent. Seventy-four patients (62%) harbored cranial meningiomas on MRI and were included in this study.

Meningiomas were ascertained by contrast-enhanced T1-weighted MRI performed in several institutions, following local protocols. MRI was reviewed expressly for this study and tumor sizes were assessed by the largest diameter on axial images. Dural tail, defined as a region of enhancement extending from the central tumor mass along the dura, was not included in the linear measurements. Only tumors greater than 0.4 cm in diameter were considered in the analysis, to allow clear distinction from cortical veins. When available, 3D T1 gadolinium-enhanced MRI sequences (slice thickness, 0.7 mm, no skip) were used to calculate tumor volumes using manual segmentation with Osirix 3.9.3 software (Pixmeo). When a hypersignal of the brain parenchyma surrounding a meningioma was visible on T2-weighted or fluid-attenuated inversion recovery (FLAIR) MRI sequences, this meningioma was considered edema associated, whatever the size of this edema. De novo meningiomas were defined as tumors that were undetectable on the previous imaging.

Detection of NF2 gene mutation was performed by direct sequencing (BigDye Terminator chemistry on ABI3110XL, Applied Biosystems) and gene dosage (Multiplex Ligation-dependent Probe Amplification, MRC Holland) of the 17 NF2 coding exons. D.H. and S.G. conducted histological review of resected meningiomas. Tumors were graded according to the WHO 2007 grading scheme. Progesterone hormone receptors were quantified by immunohistochemistry on a Ventana BenchMark XT immunostainer, using an NCL-PGR-312 antibody (Novocastra) at 1/200 dilution, following standard procedures.

DNA was extracted from frozen samples when available as described elsewhere.¹⁴ Affymetrix genome-wide human single nucleotide polymorphism (SNP) arrays were performed following the manufacturer's instructions to assess loss of heterozygosity and copy number variations in DNA extracted from meningioma samples. Matched blood DNA from 10 patients was used for normalization. Analyses were conducted using dChip software as described elsewhere.¹⁴ Statistical analyses were performed in GraphPad Prism 4.0 software; Mann–Whitney U-tests or χ² tests were used to compare groups. Values were expressed as means (95% CI). Statistical significance was presumed P <.05 (2-tailed). Correlation between linear and volumetric variations was searched by third-order nonlinear regression analysis in GraphPad Prism 4.0 software, with a Kolmogorov–Smirnov normality test.

Results

Patients

At least 1 cranial meningioma was found in 74 (62%) of the 119 NF2 patients followed at our institution. Main characteristics of the patients are given in Table 1. Nineteen patients (24.3%) were <18 years old at diagnosis. The mean duration of follow-up after diagnosis was 110.2 months (92.5–128), totaling 680 patient/years of follow-up. Of 74 patients, 14 (19%) had a family history positive for NF2. Sixty-five patients were alive at the end of 2009. Nine patients (12%) died during the follow-up period. The causes of death were as follows (n patients): sudden death (3), postoperative (3), suicide (1), progressive disease (1), unknown (1). At last follow-up, 9 patients (12%) were free of any treatment for NF2-related tumor. Fifty-one patients (69%) underwent at least 1 surgery for vestibular schwannoma, and 17 patients (23%) for spinal tumors.

Table 1.

NF2 population characteristics

	With Cranial Meningioma	Without Cranial Meningioma
N patients	74 (62.1%)	45 (37.2%)
Female	48 (64.9%)	24 (53.3%)
Mean age, y, at diagnosis (range)	29.1 (4–82)	31.3 (8–88)
Mean age, y, at first symptoms (range)	24.6 (2–80)	28 (8–88)
Siblings with NF2	14 (19%)	10 (22%)
NF2 mutation identified	35 (53.8% of 65)	14 (40% of 34)
Truncating	18 (51.4%)	4 (29%)
Nontruncating	17 (48.6%)	10 (71%)
Bilateral vestibular schwannoma	61 (82.4%)	40 (88.9%)
Other cranial nerve tumors	41 (55.4%)***	7 (19.4% of 36)
Spinal tumors	57 (86.4% of 66)**	16 (55.2% of 29)
Spinal ependymoma	33 (50% of 66)*	8 (27.6% of 29)

Statistical difference between patients with/without cranial meningioma. *P < .01, **P < .001, ***P < .0001.

Of the 74 patients with meningiomas, 65 underwent mutation testing, and 35 had a pathogenic NF2 mutation identified (53.8%; Table 1). No mutation was identified in exon 5 or exons 14–17. Somatic NF2 mutations were identified in tumor samples from 7 of 8 patients whose blood screening was negative and were associated with a deletion of the second allele in each case, suggesting NF2 mosaicism. In 1 case, the mutation (c.169C > T) was indeed present in blood DNA at a lower level, indicative of an NF2 somatic mosaicism. No second tumor was available for the 6 other patients to ascertain the diagnosis of a somatic mosaicism. Apart from the mosaic case, 4 unrelated patients had the same c.169C > T mutation in exon 2. All these patients had a severe phenotype with bilateral vestibular schwannomas, multiple meningiomas, and spinal tumors. Interestingly, the mosaic patient had the same tumor spectrum but a later onset of symptoms, ie, at 31 years of age vs 12, 15, 16, and 16 years for germline mutation carriers. This observation was consistent with a milder phenotype in mosaic patients.

Image-based Natural History of Meningiomas

Seventy-four NF2 patients harbored 287 cranial meningiomas. Fifty-two percent of patients had 3 or fewer meningiomas, and 28% had 7 or more. The number of meningiomas per patient did not differ according to gender (5.3 meningiomas [3.6–6.9] in females vs 4.6 (3.1–6.1) in males, P = .77]. The locations of the 287 meningiomas are depicted in Fig. 1. These tumors were most frequently located at the convexity and along the falx (206 meningiomas, 71.8%). The second most frequent location was the base of the skull, with 71 tumors (24.7%). Eight meningiomas were located within the lateral ventricles (2.8%). In 13 patients (17.6%), vast areas of dural enhancement were observed, especially along the falx and/or convexity, and occasionally formed micronodules.

Fig. 1.

Location of 287 cranial meningiomas in 74 NF2 patients. Fifty meningiomas were resected during the 110.2 months follow-up period (triangles). Two hundred thirty meningiomas were observed (dots). Dots and triangles are distributed according to region of occurrence. There is a marked predominance for falx and convexity meningiomas (71.8%).

The mean maximum diameter of meningiomas at diagnosis was 14.3 mm (13.2–15.4) and at the end of follow-up 21.3 mm (19.6–22.9) (P < .0001). The mean annual growth rate during the follow-up period was 1.5 mm (1.1–1.8). Meningiomas demonstrated no or minimal growth (less than 1 mm/y) on serial MRI in 182/277 cases (65.7%). Only 27/277 (9.7%) grew more than4 mm/year. Pattern of growth is depicted in Fig. 2A. Meningiomas growing 1 or more mm/year were more often in males and people 30 years old or younger and were more often associated with brain edema and with an initial tumor diameter greater than 25 mm (Table 2).

Fig. 2.

Patterns of meningioma growth. (A) Mean annual increase of the largest diameter of 287 meningiomas. Fifty-seven meningiomas were resected during the 110.2 months follow-up period (black columns), and 230 were observed (gray columns); (B) 68 meningiomas in 18 patients were evaluated with 3D 0.7 mm thick MRI sequences. Their pattern of growth was similar to the whole series (panel A). (C) There was a strong correlation between largest diameter and volume measurements in the series of 68 meningiomas.

Table 2.

Univariate analysis of prognostic factors associated with tumor growth in 277 NF2 meningiomas

Variable	Growing Meningiomas (>1 mm/y), n = 93	Nongrowing Meningiomas (≤1 mm/y), n = 184	P Value	Statistical Test
Mean age, y, at diagnosis of meningioma (95% CI)	24 (22–26)	30 (28–32)	.0004	Mann–Whitney
Peritumoral edema	20 (21%)	6 (3%)	<.001	χ2
Male sex	41 (44%)	48 (26%)	<.01	χ2
Initial tumor diameter >25 mm	15 (16%)	13 (7%)	<.02	χ2
Mean meningioma burden, y (95% CI)	5.5 (5–5.9)	5.2 (4.9–5.5)	ns	Mann–Whitney
Location at skull base	22 (24%)	47 (26%)	ns	χ2
Truncating NF2 mutation	29 (31%)	51 (28%)	ns	χ2

We decided to evaluate changes in volumes and linear measurements to find a correlation in a subgroup of patients with raw volumetric data. In 18 NF2 patients, at least 2 3D T1 gadolinium-enhanced sequences were available and enabled the calculation of exact changes in volumes of 68 individual meningiomas. Main meningioma characteristics were similar between this subgroup of 68 meningiomas and the 287 meningiomas in the larger series (Fig. 2B). The mean volume of meningiomas at diagnosis was 2.4 cm³ (1.2–3.5) and at the end of follow-up 3.3 cm³ (2.0–4.5) (P = .008). The mean annual growth rate during a mean follow-up period of 35.6 months (30.9–40.3) was 0.34 cm³ (0.16–0.52). In this group, 5/68 meningiomas (7.3%) in 5/18 patients (28%) had an annual growth rate 20% or more (Fig. 2C). There was a strong correlation between linear and volumetric variations (polynomial third order R²= 0.94, Kolmogorov–Smirnov P < .0001), and an increase in volume of 20% or more corresponded to a linear increase of 3% or more.

Meningioma Surgery in NF2 Patients

Fifty-six meningiomas (23%) in 34 NF2 patients (45.9%) were operated on during the follow-up period. Resection was decided owing to symptoms in 36/56 meningiomas (64.3%) (epilepsy in 5, intracranial hypertension in 15, neurological deficit in 16) or to significant growth on serial imaging (13/56, 23.2%). In addition to these cases, 12.5% (7/56) were removed incidentally during operations for which resection of a different lesion was the primary operative goal (“en route”). Four patients (7%) experienced postoperative worsening of their neurological status. No recurrence was detected in patients with total resection of their meningioma on MRI (mean postoperative delay: 51.7 months [95% CI, 33–71], n = 43). Thirteen meningiomas were partially removed. Remnants grew in 6 cases (2 orbitosphenoidal, 2 parasagittal, 1 petrous apex, 1 falx tentorium angle) and required additional treatment in 3 patients (repeated surgery in 2, radiotherapy in 1).

De Novo Meningiomas Are Rare but Display More Aggressive Behavior

At the end of follow-up, 25 de novo meningiomas (8.7%) appeared in 12 NF2 patients (16.2%). When multiple meningiomas appeared, they were asynchronous. The mean age when the tumor was detected was 26.5 years (range, 11.1–62.8). At diagnosis, the mean size of de novo meningiomas was similar to that of other meningiomas (14.2 mm [10.4–17.9] vs 14.3 mm [13.1–15.5], nonsignificant). De novo meningiomas demonstrated a higher annual growth rate than other meningiomas (6.6 mm [3.4–9.9] vs 0.97 mm [0.7–1.2], P < .0001). Accordingly, the mean size at the end of follow-up was also higher than other meningiomas (28.3 mm [22.3–34.2] vs 20.6 mm [18.9–22.3], P < .002), suggesting a higher proliferation potential of de novo meningiomas. De novo meningiomas were more often associated with brain edema (24% vs 9.5%, P < .05). Hence, de novo meningiomas required surgery more often (36% vs 18.3%, P < .05). In this group, pathological examination detected 5 grade I tumors, 2 grade II tumors, and 1 grade III tumor.

Brain Edema Highlights Aggressive Meningiomas

Brain edema on MRI was observed in 31/287 meningiomas (10.8%). Meningiomas associated with brain edema displayed more aggressive behavior than meningiomas without edema: They were larger at diagnosis (mean maximum diameter 24.8 mm [19–30.6] vs 13.3 mm [12.3–14.4], P < .001) and at the end of follow-up (mean maximum diameter 43.3 mm [36.9–49.6] vs 19.2 mm [17.7–20.7], P < .00001), with a higher annual growth rate (5.3 mm [2.3–8.2] vs 1.1 mm [0.8–1.3], P < .00001). They more often required surgery (26/31 [84%] vs 31/256 [12%], P < .0001), and on pathological examination, a higher proportion of meningiomas associated with brain edema were graded II or III (13/19 grade II or III [68%] vs 4/22 [18%], P < .01). Postoperatively, brain edema decreased in all patients.

Neuropathological Features of NF2-Associated Meningiomas

Among the 57 resected meningiomas, 41 samples were available for histological review in 27 NF2 patients. There were 29 grade I tumors, 10 grade II tumors, and 2 grade III tumors. Seven meningiomas, all grade I, were resected en route. Thirty-five percent of meningiomas that were symptomatic or growing on serial MRI when resected were grade II or III. The mean number of mitotic figures was 0.6 in grade I, 3.2 in grade II, and 16 in grade III (P < .0001). It is noteworthy that no grade II/III meningioma arose at the skull base. Among grade I meningiomas, predominant histological subtypes were transitional in 16 (55%), fibroblastic in 10 (35%), and meningothelial in 3 (10%). One of the 2 grade III meningiomas was papillary. Remarkably, meningioma samples demonstrated a major heterogeneity of histological subtypes. Each specimen demonstrated a predominant histological subtype, but it was often associated with 1 to 6 additional and clearly identifiable subtypes (mean, 2.6). To illustrate this observation, Fig. 3 summarizes different histological aspects noticed in a single meningioma sample. Twelve patients were operated on for multiple meningiomas. In 9/12 cases (75%), individual meningiomas were of different subtypes or grades. One collision tumor was observed in a patient operated on for a right cerebellopontine angle tumor, which displayed intricate meningioma and schwannoma on pathological examination. By immunohistochemistry, 28% of grade I meningiomas and 25% of grade II/III meningiomas stained positively in more than 50% of nuclei for progesterone receptor. These progesterone receptor immunoreactivity scores were lower than commonly reported in sporadic meningiomas, and we did not observe the usual decrease with increasing histological grade.¹⁵

Fig. 3.

Association of multiple histological subtypes within a right frontal convexity meningioma specimen in a 58-year-old NF2 patient. The main histological subtype is anaplastic (A), with numerous mitotic figures (arrow heads). The specimen also shows definite meningothelial (B), papillary (C), psammomatous (D), fibroblastic (E), and transitional (F) regions. Additional regions display microcystic (G), angiomatous (H), and cartilaginous metaplasia (I) components.

Genomic Characteristics of NF2-Associated Meningiomas

We performed high-density SNP analysis of 22 meningioma samples in 14 NF2 patients (Fig. 4). There were 14 grade I tumors, 6 grade II tumors, and 2 grade III tumors. Ten paired blood samples were used as references. We observed 24 chromosome gains, 81 losses, and 7 homozygous deletions. Only 1 grade I specimen had no chromosome alteration detected. Among chromosome gains, 85% involved regions smaller than 5 Mb. Conversely, 60% of losses involved whole chromosome arms. Consistent with previous descriptions, the frequency of chromosome arm losses increased with increasing histological grade: The mean number of chromosome arm losses was 2.2 (range, 0–9), 5.7 (range, 1–10), and 4.5 (7 and 9) in grades I, II, and III, respectively (P = .03). As expected, the most frequent arm loss was on chromosome 22q (21/22 samples, 95%), which consisted of terminal deletions encompassing the NF2 gene region. Other frequently observed chromosome losses were on 1p in 27%, 10q and 19p in 18% each, and 6q and 10q in 14% each. This observation is in agreement with previous reports using lower-resolution techniques.¹⁶ Taking into account the histological grade, the only recurring losses in grade I were on 10p and 19p (2 cases each). Grades II and III tumors displayed higher chromosome instability, including frequent losses on 1p (5/8 vs 1/14 grade I, P < .01), 18q (2/8 vs 0/14 grade I, P < .05), and 6p (2/8 vs 0/14 grade I, P < .05). De novo meningiomas did not show higher chromosome instability: they harbored a mean 11.2 imbalanced chromosome segments (6.9–15.5) (sum of gains, losses, and loss of heterozygosity), similar to other meningiomas (12.7 [4–21.3], P = .72).

Fig. 4.

6.0 SNP array analyses of 22 NF2-associated meningiomas. Patient identification numbers are depicted on the top. *De novo meningiomas. When available, control blood samples (B) were hybridized along with grade I (1), II (2), or III (3) meningioma samples. The gray scale displays copy number changes. The most frequent arm loss was on chromosome 22q. Grades II/III meningiomas are associated with a higher chromosome instability than grade I meningiomas (P = .03).

For 6 NF2 patients, multiple meningiomas (2 or 3) were analyzed. Chromosome losses and/or gains varied from 1 sample to another in the same patient. In particular, the breakpoint on 22q, the second stochastic NF2 inactivating event leading to tumor development according to Knudson's double hit hypothesis, was different between individual tumors in the same patient.

Discussion

Grades II/III meningiomas accounted for 35% of symptomatic or growing meningiomas resected in NF2 patients. This finding reflects a selection bias rather than a true aggressive phenotype. Indeed, only symptomatic or growing meningiomas come to be considered for treatment. All meningiomas resected en route were actually grade I, supporting the hypothesis that stable meningiomas in NF2 patients are probably grade I. Taken with the total number of meningiomas in NF2 patients, the real proportion of grades II/III meningiomas would be 12/287 (4.2%). Thus, contrary to other published series, meningiomas in NF2 patients are not histologically more aggressive than sporadic ones.^17,18 Besides, de novo meningiomas are rare (8.7%) but are more aggressive, requiring a closer follow-up and more aggressive treatment. First-line treatment in this cohort was surgery. Fractionated radiotherapy was used only once at recurrence. Besides, as stated in the last consensus recommendations for current treatments in NF2, most meningiomas occur in surgically accessible locations, and hence surgery is generally considered first-line therapy if treatment is needed.¹⁹ In contrast, there are no definitive data for or against the use of radiosurgery as a primary treatment modality for NF2 meningiomas. Only 2 small retrospective single-center series have reported successful radiosurgical treatment of NF2 meningiomas.^20,21

The most striking finding on pathological review of these meningiomas was the frequent association of multiple meningioma subtypes within the same specimen. One hypothesis to explain this finding is that the genetic alteration responsible for the development of meningiomas, namely NF2 inactivation, occurs early in development and thus affects progenitor meningeal cells that are not yet committed to a specific lineage.²² Remarkably, in our series, meningiomas occurred less frequently at the base of the skull (24.7%) than in sporadic cases (48% in a recent meta-analysis).²³ This finding is in line with published data suggesting that skull-base meningiomas are more often of meningothelial subtype and arise through NF2-independent pathways.^24,25 Similarly, the embryonic origin of the meninges seems to be different at the convexity and around the brainstem (neural crest and mesoderm derived, respectively), suggesting that the cells of origin of these meningiomas are different.²⁶

This study clearly shows that a significant proportion of meningiomas in NF2 patients demonstrate no or minimal growth on serial imaging. Thus, in our view, proactive treatments are not to be considered in meningiomas of NF2 patients. Tumor growth potential has to be evaluated before considering any treatment, the potential long-term side effects of which must be taken into account. In contrast to vestibular schwannomas, where size matters in terms of surgical results, meningiomas have a wider therapeutic window, allowing long-term evaluation without compromising patients’ outcomes.

Current trials focus on vestibular schwannomas as the primary endpoint, evaluating either volumetric changes by MRI or improvement in hearing.¹⁹ Based on this series, we suspect that these trials will not be able to evaluate a potential effect on meningiomas. Actually, only 7.3% of meningiomas (in 28% of patients) had a growth rate 20% or more per year. About two-thirds of NF2 patients harbor meningiomas. This suggests that only 17% of unselected NF2 patients would be relevant for evaluating an effect on meningioma growth. It is unlikely that the power of the study will be enough to show efficacy, thus specific clinical trials targeting meningiomas as the primary endpoint are mandatory. Best efficiency will be achieved by selecting patients with growing meningiomas, eg, 20% in volume on 3D MRI.²⁷

Funding

This work was supported by the Brain Science Foundation and Association Neurofibromatoses et Recklinghausen.