Two molecular subtypes of skull base meningiomas with extracranial extension

Meningiomas are primarily benign intracranial tumors but in rare instances, they may extend extracranially or even originate outside the cranium [9, 11]. While researchers have made significant progress in elucidating the genetic characteristics of intracranial meningiomas [2, 3, 7], the molecular profiling of meningiomas with extracranial involvement remains insufficiently characterized. In this study, the authors aimed to investigate the genetic features of skull base meningiomas with extracranial involvement based on their locations using whole-exome sequencing (WES).

We included consecutive meningioma cases with extracranial involvement that underwent surgical resection at our institution between 2000 and 2024. For tumors with secondary progression from intracranial to extracranial space, the first specimen exhibiting extracranial involvement was selected for the analysis. Patients associated with schwannomatosis were excluded. WES was performed to assess both somatic mutations and copy number alterations (CNAs), including TERT promoter mutations, and homozygous deletions of CDKN2A/B, both of which are associated with 2021 World Health Organization classification of meningiomas [6]. Details of genomic analysis are provided in the supplementary resources and previous reports [4, 5]. Recent advances in multi-omics analyses have enabled refined prognostic stratification of meningiomas through molecular classification. However, the implementation of such approaches in routine clinical practice remains limited due to their technical complexity and cost constraints. To address this, we sought to classify meningiomas using only data obtained from WES, focusing specifically on somatic mutations and CNAs. Meningiomas were initially classified into three molecular groups [10], and subsequent studies have proposed overlapping classification systems that refine or expand upon this original framework [2, 8, 12], including approaches that integrated transcriptomic, epigenomic, and genomic data to capture the biological features of meningiomas with greater resolution. In this cohort, tumors were classified into three groups based on driver mutations and CNAs: Group A comprised NF2-wildtype tumors, including those harboring driver mutations such as TRAF7, KLF4, AKT1, SMO, or POLR2A, as well as tumors without detectable driver mutations or high-risk CNAs. Group B included tumors with NF2 mutation/22q loss but lacking high-risk CNAs. Group C consisted of tumors harboring high-risk CNAs, either with known driver mutations or without other known drivers (Fig. 1A). Speficially, 1p loss, 6p/q loss, 10p/q loss, 14q loss, and 18p/q loss were defined as high-risk CNAs in accordance with previous publications [5, 8, 12]. An 80% threshold of chromosomal arm involvement was used to define arm-level events. Tumor location was classified into four categories: (1) anterior cranial fossa, (2) middle cranial fossa (including the cavernous sinus), (3) posterior cranial fossa, and (4) intraorbital (Fig. 1B).

Fig. 1

(A) Overview of molecular classification of meningiomas with extracranial extension based on whole-exome sequencing data. (B) Representative magnetic resonance imaging. Tumor location was classified into four categories: (1) anterior cranial fossa, (2) middle cranial fossa (including the cavernous sinus), (3) posterior cranial fossa, and (4) intraorbital. (C) Schematic diagram illustrating the association between anatomical tumor location and driver mutations/copy number alterations. Each tumor is plotted according to its location, with colors indicating the corresponding genetic alterations. (D) Driver mutations and copy number alterations identified in the cohort (n = 34). Shown are the World Health Organization grade, histological subtype, age, sex, and representative driver mutations and copy number alterations for each case

Of the 766 consecutive meningioma cases treated during that period, 34 exhibited extracranial involvement and were included in the study. Among them, 17 cases were newly diagnosed meningiomas, whereas the other 17 represented recurrent tumors with newly developed extracranial extension. A total of nine patients had undergone radiotherapy before surgical resection. Most tumors (20 cases, 59%) were located in the middle cranial fossa, followed by posterior cranial fossa (6), intraorbital (5), and anterior cranial fossa (3). The association between driver mutations/CNAs and anatomical location is presented in a figure (Fig. 1C). Among middle cranial fossa meningiomas, 85% (17/20) were classified as Group C, and with one harboring a TRAF7 mutation and another carried a POLR2A mutation. Of the 14 non-middle cranial fossa meningiomas, most were Group A (12/14, 86%). Regarding specific locations, anterior cranial fossa tumors included one AKT1/TRAF7 and one SMO mutation; posterior cranial fossa cases had three TRAF7 and three POLR2A mutations; and intraorbital meningiomas mostly lacked detectable driver mutations, except one with POLR2A mutation. A comprehensive oncoplot summarizing the clinical information, driver mutations, and CNAs of all tumors is shown in a figure (Fig. 1D). Detailed information for each case, including patient background, driver mutations, tumor status (primary or recurrent), prior treatments such as radiotherapy, and clinical course, is provided in a supplementary table (Supplementary Table 1). Notably, Group B tumors were not observed in this cohort. Among 15 tumors of Group A, the most frequently identified driver mutations were POLR2A (5 tumors) and TRAF7 (4), which predominated within this group. Co-occurring mutations were observed in TRAF7-mutated tumors, including KLF4 (2 tumors), AKT1 (1), and PIK3CA (1). Five tumors in Group A lacked detectable driver mutations , three of these were intraorbital meningiomas. Moreover, except for a single tumor, none of the tumors in Group A harbored high-risk CNAs. On the other hand, most Group C tumors were predominantly found in the middle cranial fossa (17/19, 89%). Histopathological classification identified 7 meningothelial, 5 atypical, 5 transitional, 1 fibrous and 1 anaplastic meningiomas. Bone invasion was observed in 17 of 19 tumors (89%). Among NF2-mutant tumors, TRAF7 co-mutations were detected in two. No TERTp mutations were identified, whereas homozygous loss of CDKN2A/B was detected in one case. The most frequent aneuploidy in Group C was 1p loss (14/19, 74%), followed by 6q loss (8/19, 42%), 14q loss (6/19, 32%), 10q loss (5/19, 26%) and 18q loss (4/19, 21%). A comparative analysis of the clinical characteristics between Group A and Group C tumors was conducted using the t-test for continuous variables and Fisher’s exact test for categorical variables. All data were complete with no missing values. This analysis demonstrated no significant differences in age or sex. By contrast, Group C tumors were significantly more prevalent in the middle cranial fossa (89% vs. 20%, P = 0.0001) and exhibited a higher frequency of bone invasion (89% vs. 40%, P = 0.004) (Table 1).

Table 1 Comparison between clinical and radiological characteristics between 2 groups of the cohort

Our study indicates that meningiomas with extracranial involvement can be broadly classified into two distinct molecular subtypes: (1) NF2 wild-type tumors primarily driven by TRAF7 or POLR2A mutations (corresponding to the merlin-intact subtype in methylation-based classifications) and (2) aggressive tumors characterized by NF2 mutation/22q loss accompanied by high-risk CNAs (corresponding to the hypermitotic subtype). Notably, no NF2 mutation/22q loss tumors lacking high-risk CNAs (corresponding to immune-enriched subtype) were identified.

NF2 mutation/22q loss represents the most common genetic alteration in meningiomas. Recent integrative multi-omics analyses have demonstrated that NF2-mutated meningiomas can be classified into distinct prognostic subtypes, namely the immune-enriched and hypermitotic type [2, 8, 10, 12]. In our cohort, detecting both NF2 mutations and copy number alterations (CNAs) allowed a simplified classification of NF2 mutation/22q loss meningiomas. Notably, NF2 mutation/22q loss meningiomas with extracranial extension uniformly exhibited high-risk CNAs, predominantly 1p loss, and were located in the middle cranial fossa with associated bone invasion. The immune-enriched and merlin-intact subtypes were rarely observed, suggesting an association between high-risk CNAs and aggressive tumor behavior.

This study has several limitations. First, it lacked a control group of middle cranial fossa meningiomas without extracranial extension, making it unclear whether the high-risk CNAs observed represent a general feature of middle cranial fossa meningiomas or are specifically associated with their capacity for extracranial spread. Second, although classification was performed using driver mutations and CNAs derived from WES data, transcriptomic or DNA methylation profiling was not conducted due to limited tissue availability. Such profiling could further resolve tumor subtypes, particularly in unclassified or ambiguous cases. Nonetheless, previous studies have demonstrated substantial concordance between genomic and integrated molecular classifications [1]. Finally, our cohort was restricted to consecutively collected cases with extracranial extension, which may bias the findings toward more aggressive or anatomically predisposed tumors.

In conclusion, meningiomas with extracranial involvement and NF2 mutation/22q loss corresponding to the immune-enriched type were not observed in this cohort. Notably, middle cranial fossa meningiomas with extracranial extension frequently harbored high-risk CNAs and exhibited bone invasion. Although these features may reflect a more aggressive clinical phenotype, the absence of a control group of middle cranial fossa meningiomas without extracranial extension limits definitive conclusions. Future comparative studies are warranted to clarify whether high-risk CNAs are truly associated with the capacity for extracranial spread and resistance to stereotactic radiosurgery.