Molecular landscape and clinical correlates of olfactory groove meningiomas: a multi-institutional study

In Brief

Researchers investigated the clinical, radiological, and molecular profiles of 123 olfactory groove meningiomas (OGMs) in an international study. SMO, AKT1, and PIK3CA mutations were present in 70% of OGMs, influencing tumor behavior, symptoms, and progression-free survival. SMO-mutant OGMs had an elevated recurrence rate, despite WHO grade I histology, even if completely resected. These findings highlight the importance of molecular profiling and highlight a need for adjuvant or targeted therapies in SMO-mutant OGMs, even after complete resection.

Abstract

OBJECTIVE

The aim of this study was to investigate the relationship between the clinical and radiological characteristics of olfactory groove meningiomas (OGMs) and their molecular profiles.

METHODS

The authors performed targeted next-generation and whole-genome sequencing in 123 OGM samples collected from 4 international institutions, focusing on known meningioma-driver genes. They compared the molecular data with the clinical and radiographic features of the tumors. Patient and tumor data, including age, sex, radiological features, and overall survival, were retrospectively collected and analyzed.

RESULTS

The study cohort comprised 90 females (73%) and 33 males (27%), with a median age at diagnosis of 57 years (range 25–87 years). The majority of tumors (88.6%, n = 109) were classified as WHO grade I meningioma. Known driver mutations were found in 86.2% of patients (n = 106), with the most common mutations found in the SMO^L412F/W535L and AKT1^E17K genes, each present in 36 cases (29.3%), followed by mutations in PIK3CA/PIK3R1 (19 cases, 15.4%; 14 PIK3CA and 5 PIK3R1), TRAF7 alone (7 cases, 5.7%), POLR2A^Q403K (4 cases, 3.3%), and TRAF7/KLF4^K409Q (3 cases, 2.4%), while 17 patients (13.8%) did not harbor known meningioma driver mutations (wildtype group). Within molecular subgroups, patients with AKT1 mutations were the youngest (median age 51 years, range 30–87 years) and patients with TRAF7-only mutations were the oldest (median 66 years, range 28–76 years). The median tumor volume at diagnosis was 18.04 cm³. SMO-mutant tumors were significantly larger (median volume 19.5 cm³) than both AKT1-mutant (median 7.5 cm³, p = 0.021) and TRAF7/KLF4-mutant (median 4.9 cm³, p = 0.002) tumors. Tumor-associated hyperostosis of the sphenoid planum was common (58.5%), led by PIK3CA/PIK3R1, SMO, and wildtype groups (73.7%, 72.2%, and 70.6%, respectively), compared with a notably lower rate in AKT1-mutant tumors (25%) (p < 0.001). Tumor invasion of the ethmoid sinuses occurred most frequently in the TRAF7-only mutant OGMs (42.9%), followed by PIK3CA/PIK3R1-mutant (31.6%) and wildtype (23.5%) OGMs. The mean progression-free survival (PFS) was 144.4 months (95% CI 123.8–165 months). Patients with SMO-mutant OGMs exhibited a significantly shorter mean PFS of 92.0 months (95% CI 70.1–113.9 months) compared with 158.2 months (95% CI 134.9–181.5 months) for SMO-wildtype OGMs (p = 0.004), identifying a tumor type that might benefit from adjuvant treatment after resection.

CONCLUSIONS

This study revealed that 70% of OGMs harbor SMO, AKT1, and PIK3CA mutations, influencing tumor behavior, symptoms, and outcomes, supporting molecular profiling for personalized treatment in OGM management.

Meningiomas are the most common primary intracranial tumors, comprising up to 40% of all primary CNS neoplasms.^1,2 Among these, olfactory groove meningiomas (OGMs) represent about 8%–13% of cases, uniquely characterized by their ability to induce hyperostosis of adjacent bones.^3,4 Notably, OGMs extend into the ethmoid sinuses and nasal cavity in approximately 15% of instances, often presenting as large- or giant-size tumors.^4,5 These growths can be closely associated with marked frontal lobe edema and significant dysfunction of the frontal lobe.

A Simpson grade I removal, which entails complete excision of the tumor along with affected dura and bone, particularly when large or invasive, has been rare due to these factors and is often associated with considerable morbidity.⁶ This surgical complexity contributes to the notable high recurrence rates observed with OGMs, with late recurrences averaging around 23% within 10 years.^4,7 Treatment options in the recurrence setting range from observation for minimal growth to total resection followed by radiotherapy depending on the tumor’s progression.⁸ However, medical therapies have thus far demonstrated limited effectiveness in managing meningiomas, further emphasizing the need for innovative treatment modalities and comprehensive management strategies.⁹

Known prognostic factors in OGM are patient age, tumor size, WHO grade, and the extent of resection.^7,10–12 However, recent breakthroughs in genetic analysis have unveiled specific driver mutations, including AKT1, SMO, and PIK3CA, illuminating the complex molecular underpinnings of meningiomas.^13–19 These discoveries have not only enhanced our understanding of the molecular landscape of these tumors, but also identified mutations with distinct anatomical distributions and associations with key clinical characteristics, including histology, WHO grade, tumor size, and age at diagnosis.^20–22

Considering the limited knowledge regarding the genetic alterations and their clinical implications in OGM, the objective of this study was to investigate the relationship between the clinical and radiological features of OGMs and their molecular profiles.

Methods

Sample and Data Acquisition

One hundred twenty-three patients diagnosed with an OGM who underwent surgical treatment and had available clinical data and tumor tissue for molecular analysis between July 2002 and October 2023 were included in this study. The patients were from 4 independent neurosurgical centers: Huashan Hospital (n = 59); Dresden University Hospital (n = 55); Graduate School of Medical and Dental Sciences, Kagoshima University Hospital (n = 4); and Oxford University Hospitals (n = 5). Informed consent was obtained in accordance with the standards of each facility, and ethics approval for the study was granted by the institutional review boards. All patients provided informed consent. To ensure accurate inclusion of OGM, the exact topography on MRI and radiological features were carefully reviewed. OGMs were defined as tumors arising in the midline over the cribriform plate and frontosphenoidal suture, with the majority occupying the anterior cranial fossa extending from the crista galli to the tuberculum sellae and some also extending to the ethmoid sinuses or into the nasal cavity. Study inclusion required the availability of relevant clinical data (such as treatment details and follow-up outcomes) and tumor characteristics that were collected retrospectively to ensure comprehensive data analysis. Patients with a radiation-induced meningioma or neurofibromatosis type 2 were excluded from the study. Histological features, including subtypes and WHO grades, were independently reviewed by one member in each of the Departments of Pathology at Huashan Hospital, Dresden University Hospital, and Oxford University Hospitals. This review was conducted to validate the histological characteristics of the meningiomas according to the 2016 WHO classification of tumors of the CNS.

Radiological Data Acquisition

Axial and sagittal T1-weighted postcontrast gadolinium-enhanced MR images, as well as CT scans, were collected for all cases. The presence of hyperostosis of the sphenoid planum and invasion of the ethmoid sinus were evaluated using preoperative CT and MRI scans. To ensure accuracy and consistency, the preoperative CT scans and MR images were independently reviewed by 4 experienced neurosurgeons (L.H. at Huashan Hospital, T.A.J. at University Hospital Dresden, S.F. from Kagoshima University Hospital, and S.J. at Oxford University Hospitals). For the cases from Huashan Hospital, tumor volumes were computed by manually segmenting the tumors on contrast-enhanced T1-weighted MR images with OSIRIX software, using other sequences (nonenhanced T1-weighted, FLAIR, and T2-weighted sequences) for further reference. The volume measurement function of the software was then used to record the tumor volume.¹⁷ For the remaining cases, tumor volumes were computed by manually segmenting the tumors on contrast-enhanced T1-weighted MR images with ITK-SNAP software, using other sequences (nonenhanced T1-weighted, FLAIR, and T2-weighted sequences) for further reference. The volume measurement function of the software was then used to record the tumor volume.²³

Tumor DNA Sequencing

Because of the multi-institutional nature of this study, samples underwent targeted sequencing using three different protocols, as partially previously described.^24,25 All known meningioma driver genes (AKT1, SMO, PIK3CA, TRAF7, NF2, POLR2A, and KLF4) were included in the genetic panels across all the institutions.

Next-Generation Sequencing Protocol at Huashan Neurosurgical Center (n = 59)

The tumor DNA was extracted from 10 slides of 15-μm scrolls taken from archived formalin-fixed paraffin-embedded blocks using the GeneRead DNA FFPE kit (QIAGEN), following a standard technique. The extracted DNA was then profiled using a self-designed targeted panel sequencing consisting of 184 genes known to be frequently mutated in CNS tumors, including the common pathological genes relevant to meningiomas, as previously described. These genes include NF2, TRAF7, KLF4, AKT1, SMO, PIK3CA, PIK3R1, PTEN, SMARCE1, BAP1, CDKN2A/B, TERT promoter, ARID1A, SUFU, SMARCB1, POLR2A, DMD, and PBRM1. Sequencing was performed using a custom hybrid capture approach (Agilent Technologies) on a Miniseq instrument (Illumina), with a mean coverage of more than 500-fold. Internal next-generation sequencing (NGS) controls were performed to ensure accurate sample assignment.

NGS Protocol at University Hospital Dresden (n = 59)

The fresh-frozen tumor tissues from the University Hospital Dresden (n = 55) and from Kagoshima University Hospital (n = 4) underwent sequencing using the NGS protocol at University Hospital Dresden. The tumor DNA was purified using the AllPrep DNA Universal Kit for fresh-frozen tissue (QIAGEN), following the manufacturer’s instructions. The regions of interest were amplified using a custom-designed amplicon panel, according to the QIAseq Targeted DNA V3 Panel, May 2017 protocol (QIAGEN), which was custom-designed and manufactured by our group.²⁰ The panel covered mutation hotspots or whole genes where loss of function is a known mechanism of action. The included meningioma-relevant genes were AKT1, CDKN2A, KLF4, NF1, NF2, PIK3CA, PIK3R1, POLR2A, PTEN, SMARCB1, SMO, STAG2, SUFU, TP53, TRAF7, and TERT promotor. During library preparation, unique molecular barcodes and sample-specific indices were incorporated following the protocol. The indexed libraries were then quantified using a Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific) and sequenced in paired-end mode (2 × 150 bp) on the Illumina NextSeq platform. HG19 was used as the reference genome for bioinformatic analyses.

For all 118 samples, further bioinformatics analysis was performed using the Biomedical Workbench from CLC (version 21.0.3, QIAGEN) with a customized analysis algorithm. The analysis included filters such as coverage ≥ 100 and allele frequency ≥ 5%.

NGS Protocol at Oxford University Hospitals (n = 5)

Whole-genome sequencing (WGS) for the Oxford University Hospital samples was carried out by the National Health Service (NHS) Genomic Medicine Service via Genomics England. The sample collection and DNA extraction requirements are described in the Sample Handling Guidance (version 4.0) available at https://files.genomicsengland.co.uk/forms/Sample-Handling-Guidance-v4.0.pdf. The Illumina North Star pipeline (version 2.6.53.23) was used for the primary WGS analysis. The full methods are available at Genomics England.²⁶

All results were validated at a local Genomic Tumor Advisory Board comprising Oxford University Hospitals clinical scientists, clinical geneticists, neuropathologists, neuro-oncologists, and neurosurgeons. To ensure the quality control of sequencing across institutions, strict quality control measures were in place at each institution, and standardized bioinformatics pipelines were used for variant calling and filtering.

Follow-Up

All patients were monitored postoperatively with a median follow-up duration of 55 months (range 6–218 months). MRI was performed annually in the majority of patients during follow-up to detect any recurrence, with the first assessment occurring 6 months after surgery. Progression-free survival (PFS) was defined as the time from the initial surgery to tumor recurrence/progression, with censoring at the last follow-up visit.

Statistical Analysis

Statistical analysis was performed using IBM SPSS Statistics software (version 28). Categorical variables were compared using the chi-square test and Fisher’s exact test. Correlations between clinical variables (age, sex, hyperostosis, and ethmoid sinus infiltration) and tumor mutational status (SMO, TRAF7, AKT1, POLR2A, PIK3CA, PIK3R1, KLF4, and NF2) were analyzed using the Mann-Whitney U-test and Fisher’s exact test. Differences in tumor volumes across molecular groups were assessed using Welch’s t-test to accommodate unequal variances among groups. Clinical continuous variables were compared using the Student t-test; p < 0.05 was considered statistically significant.

Results

Clinical and Tumor Characteristics

In this study, we evaluated 123 patients with OGM who were treated across 4 neurosurgical centers, each with comprehensive molecular data available. The cohort comprised 90 females and 33 males, resulting in a sex ratio of 2.7:1. The median age at the time of diagnosis was 57 years, with a broad age range spanning from 25 to 87 years (Table 1).

TABLE 1.

Clinical and molecular characteristics of OGMs: an overview of 123 cases by submolecular group

	All Patients, n = 123	SMO Mutant, n = 36	AKT1 Mutant, n = 36	PIK3CA/PIK3R1 Mutant, n = 19	TRAF7 Alone, n = 7	TRAF7/KLF4, n = 3	POLR2A, n = 4	WT, n = 17
Median age at Dx, yrs	57 (25–87)	62 (25–75)	51 (30–87)	53 (29–78)	66 (28–76)	49.3 (41–62)	55.5 (51–69)	55 (38–74)
Sex ratio, F/M	2.7:1	3:1	2.6:1	2.6:1	6:1	2:1	3:1	3.2:1
WHO grade	109 grade I; 13 grade II; 1 grade III	33 grade I; 3 grade II	31 grade I; 5 grade II	17 grade I; 2 grade II	4 grade I; 2 grade II; 1 grade III	All grade I	All grade I	15 grade I; 2 grade II
Hyperostosis of sphenoid planum	72 (58.5)	26 (72.2)	9 (25)	14 (73.7)	4 (57.4)	1 (33.3)	0	12 (70.6)
Tumor extension into ethmoidal cells	25 (20.3)	6 (16.7)	5 (13.9)	6 (31.6)	3 (42.9)	0	0	4 (23.5)
Median tumor vol, cm3	18.04	19.5	7.5	18.6	29.5	4.9	6.7	45.5
GTR (Simpson grade I–III)	96%	100%	100%	91.6%	100%	100%	88.2%	93%
Postop RT	8 (6.5)	2 (5.6)	2 (5.6)	0	1 (14.3)	0	1 (25)	2 (11.8)
Recurrences	15 (12.2)	8 (22.2)	1 (2.8)	1 (5.3)	1 (14.3)	0	1 (25)	3 (17.6)

Dx = diagnosis; WT = wildtype.

Values are presented as the number of patients (%) unless specified otherwise.

The median tumor volume at initial diagnosis was 18.04 cm³, with a wide variation in size (range 0.6–291.06 cm³). Preoperative CT scans indicated that tumor-associated hyperostosis of the sphenoid planum was present in 72 patients (58.5%), while preoperative MR images showed that 25 patients (20.3%) presented with tumor extension or invasion of the ethmoid sinus.

Regarding tumor classification, the majority of the meningiomas were determined to be WHO grade I (109 cases, 88.6%), with the meningothelial subtype being the most prevalent (87 cases, 70.7%). Following this were transitional meningiomas (14 cases, 11.4%); atypical (WHO grade II; 13 cases, 10.5%); fibrous (5 cases, 4%); less common subtypes including psammomatous, secretory, angiomatous, each represented by 1 case (0.8%); and 1 case of meningioma WHO grade III (0.8%). This distribution underscores the predominance of lower-grade meningiomas within our cohort, with a significant majority presenting as the meningothelial subtype.

Molecular Profile of OGMs

Our genomic analysis uncovered distinct genetic mutation patterns in OGMs, highlighting the molecular diversity of these tumors. Clinically significant driver mutations were identified in 106 of the patients (86.2%). The most common mutations were found in the SMO^L412F/W535L and AKT1^E17K genes, each present in 36 cases (29.3%), followed by mutations in PIK3CA/PIK3R1 (19 cases, 15.4%; 14 PIK3CA and 5 PIK3R1), TRAF7 alone (7 cases, 5.7%), POLR2A^Q403K (4 cases, 3.3%), and TRAF7/KLF4^K409Q (3 cases, 2.4%). In total, 35 cases (28.4%) harbored a TRAF7 mutation, occurring either alone or as co-mutations with AKT1, PIK3CA, or KLF4. Notably, only one NF2 mutation was identified across the cohort. Additionally, in 17 patients (13.8% of the cohort, consisting of 13 females and 4 males), no known meningioma driver mutations were detectable. Taken together 85 cases (69.1%) harbored one of the three clinically actionable mutations: AKT1^E17K, SMO^L412F/W535L, and PIK3CA.

Within our cohort, meningiomas with an AKT1 mutation (n = 36) often exhibited additional mutations in TRAF7 (n = 19, 53%), while 16 meningiomas harbored an AKT1 mutation only. Co-occurring mutations in SMO-mutant meningiomas included NF1 (n = 2) and SUFU (n = 1) mutations. Importantly, KLF4^K409Q mutations were consistently observed alongside TRAF7 mutations. Moreover, a mutually exclusive relationship was observed between SMO, AKT1, and POLR2A mutations (p < 0.001) (Fig. 1). Notably, none of the meningiomas in our study harbored a TERT promoter mutation or a CDKN2A/B deletion.

FIG. 1.

Oncoprint illustrating the mutational landscape of OGMs. Each column represents an individual tumor, and key genes commonly altered in meningiomas are shown in the upper rows. The most frequently mutated genes in this cohort include AKT1, SMO, PIK3CA, TRAF7, and KLF4. Mutually exclusive mutation patterns are observed among several driver genes, supporting the existence of distinct molecular subgroups within OGMs. The Oncoprint highlights the predominance of NF2-wildtype (WT) tumors and the enrichment of AKT1 and SMO mutations. Tumor characteristics, such as location and histological subtype, are shown in the lower rows. Figure is available in color online only.

Interestingly, we did not observe any significant interinstitutional molecular differences in our cohort. Key driver genes (AKT1, SMO, PIK3CA, TRAF7, POLR2A, and KLF4) were consistently represented across institutions.

Correlations Between Genetic Findings and Patient Demographics

Within the cohort (median age 57 years, range 25–87 years), individuals with AKT1-mutant OGMs constituted the youngest subgroup, with a median age of 51 years (range 30–87 years). In contrast, those with TRAF7 mutations and SMO mutations were the oldest, with median ages of 66 (range 28–76 years) and 62 years, respectively. However, these age differences across molecular groups did not reach statistical significance (Supplementary Fig. 1). Notably, there was a consistent female predominance in all molecular groups, with the TRAF7-only mutant group showing the most pronounced female-to-male ratio of 6:1 (Table 1).

Correlations Between Genetic Findings and Tumor Features

The overall median tumor volume at initial diagnosis was 18.04 cm³ across all tumors. Among tumors with known driver mutations, TRAF7-only mutant OGMs had the largest median volume, measuring 29.5 cm³, followed by SMO-mutant OGMs with a median volume of 19.5 cm³. In contrast, AKT1-mutant and TRAF7/KLF4-mutant tumors were associated with the smallest volumes, at 7.5 cm³ and 4.9 cm³, respectively.

Statistical analysis using Welch’s t-test identified significant differences in tumor volumes across molecular groups. Specifically, SMO-mutant tumors had significantly larger volumes than both AKT1-mutant (p = 0.021) and TRAF7/KLF4-mutant (p = 0.002) tumors (Fig. 2). Notably, despite their smaller size, 41.6% of the AKT1-mutant tumors presented with seizures at diagnosis, while patients with SMO-mutant tumors, despite the larger tumor size, commonly exhibited nonspecific symptoms such as headache or hyposmia, or were even asymptomatic (Supplementary Table 3). This variation in clinical presentation may help explain the observed disparity in tumor size between AKT1- and SMO-mutant tumors. Additionally, TRAF7-only mutant OGMs showed a significant difference in volume when compared with TRAF7/KLF4-mutant tumors (p = 0.021).

FIG. 2.

Left: Sagittal gadolinium-enhanced T1-weighted MR image obtained in a representative case of a small AKT1-mutant meningothelial WHO grade I OGM, without infiltration of the ethmoid sinus and with a sphenoid planum hyperostosis. Right: Sagittal gadolinium-enhanced T1-weighted MR image obtained in a representative case of a large SMO-mutant meningothelial WHO grade I OGM, with infiltration of the ethmoid sinus and hyperostosis of the sphenoid planum.

A significant finding was the occurrence of tumor-associated hyperostosis of the sphenoid planum in over half of the cases (58.5%), which was notable in the PIK3CA/PIK3R1, SMO, and wildtype groups (73.6%, 72.2%, and 70.5%, respectively), compared with a markedly lower incidence in AKT1-mutant tumors (25%) (p < 0.001) (Table 1, Fig. 3), which highlights the distinct pathological features associated with specific genetic alterations. Furthermore, tumor invasion of the ethmoid sinuses occurred in 25 cases (20.3%), most frequently in the TRAF7-only mutant OGMs (42.9%), followed by PIK3CA/PIK3R1-mutant (31.6%) and wildtype OGMs (23.5%) (Table 1). The variation in ethmoid sinus infiltration rates suggests the potential for distinct patterns of tumor extension among the mutation groups.

FIG. 3.

Coronal (left) and sagittal (right) CT scans obtained in an OGM patient with a PIK3CA mutation and hyperostosis of the sphenoid planum.

Correlations Between Genetic Findings and Tumor Histology

Our analysis further examined the relationship between genetic mutations and tumor histology. While not reaching statistical significance, a notable proportion of tumors with TRAF7-only mutations appeared to exhibit WHO grade II or III histology, suggesting a trend toward higher tumor grades. In separate comparisons, TRAF7-only mutations showed a statistically significant difference in the prevalence of the meningothelial subtype when compared with AKT1 mutations (p = 0.0071) and SMO mutations (p = 0.0017), underscoring a distinct correlation between these specific genetic alterations and histological characteristics.

Correlations Between Genetic Findings and Patient Outcome

The median follow-up time for the dataset was 55.0 months (range 6.0–218.0 months). A complete tumor resection, defined as Simpson grades I–III, was achieved in 96% of cases, with comparable rates across all molecular groups. Postoperative radiation therapy (RT) was administered in 6.5% of cases, specifically in patients with WHO grade II meningiomas with residual tumor. Importantly, there were no significant differences in RT administration between the molecular subgroups. In terms of survival outcomes, in total, 15 patients (12.2%) developed a tumor recurrence during the follow-up period; 8 recurrences were in patients with SMO-mutant OGMs (53.3% of all recurrences, 22.2% of all SMO-mutant OGMs). This demonstrates a significantly higher recurrence rate in SMO-mutant OGMs compared with the remaining groups together (p = 0.038), despite 91.6% of SMO-mutant OGMs being WHO grade I meningiomas.

The mean PFS was 144.4 months (95% CI 123.8–165 months). Among patients with WHO grade I OGMs, Kaplan-Meier survival analysis demonstrated significant differences in PFS between SMO-mutant and SMO-wildtype OGMs (Fig. 4). Patients with SMO-mutant OGMs exhibited a significantly shorter mean PFS of 92.0 months (95% CI 70.1–113.9 months) compared with 158.2 months (95% CI 134.9–181.5 months) in SMO-wildtype OGMs (p = 0.004). The proportion of patients without disease progression was also lower in the SMO-mutant group (75.0%) compared with patients with SMO-wildtype OGMs (89.3%). These findings reinforce the association between SMO mutations and early tumor progression. More importantly, our multivariate analysis (HR 4.5, 95% CI 1.4–13.6; p = 0.009) demonstrated that SMO mutation was the only significant risk factor for decreased PFS, independent of WHO grade and other clinical factors, such as extent of resection or tumor size, further supporting the prognostic impact of molecular alterations in OGMs (Table 2). Comprehensive patient data supporting these findings are detailed in Supplementary Table 1.

FIG. 4.

Kaplan-Meier estimates of PFS in WHO grade I OGMs based on molecular subgroups. Left: Comparison of PFS between SMO-mutant OGMs (red) and SMO-wildtype OGMs (blue). SMO-mutant OGMs demonstrate a significantly higher recurrence rate (p = 0.004), highlighting the prognostic impact of SMO mutations. Right: Kaplan-Meier curves illustrating PFS across different molecular subgroups, including AKT1-mutant (orange), SMO-mutant (red), PIK3CA-mutant (green), wildtype (blue), and other mutations (black). There is a significant difference in PFS among the groups (p = 0.020), with SMO-mutant OGMs showing the highest recurrence risk. Figure is available in color online only.

TABLE 2.

Multivariate logistic regression analysis of risk factors associated with recurrence in OGMs

	HR	95% CI	p Value
SMO mutant	4.5	1.4–13.6	0.009
Age >65 yrs	0.8	0.1–4.2	0.923
Sex	0.9	0.5–1.6	0.589
Ethmoidal cell infiltration	0.6	0.1–2.5	0.433
Hyperostosis	0.3	0.1–2.1	0.581
Tumor vol >15 cm3	1.3	0.2–6.7	0.524
Complete resection	0.1	0–1.5	0.098
WHO grade I	0.8	0.1–6.1	0.927

Among the evaluated factors, SMO mutation was the only variable significantly associated with recurrence Other variables did not show statistically significant associations.

Discussion

OGMs represent a distinct subset of meningiomas, accounting for 34% of meningiomas in the anterior fossa.^27–29 Our comprehensive analysis offers a detailed examination of the mutational landscape specific to OGMs, revealing the prevalence and implications of key genes such as SMO, AKT1, PIK3CA, TRAF7, POLR2A, and KLF4. Notably, our study highlights the striking predominance of the clinically actionable SMO, AKT1, and PIK3CA mutations, which collectively appear in 69% of OGM cases. The high rate of SMO mutations in our cohort is particularly significant given the rarity of SMO mutations in meningiomas located elsewhere (3%–5%), suggesting a unique pathogenic role in OGMs (Fig. 1, Supplementary Table 1).^13,14 Moreover, our analysis identified a higher incidence of AKT1 mutations in OGMs (29%) than previously reported, where AKT1 mutations were detected in 13%–15% of OGMs.³⁰ This likely reflects our larger sample size and focused examination of OGMs compared with other studies (Supplementary Table 2).^30,31 Additionally, all OGMs except one in our cohort were NF2-wildtype, further emphasizing their distinct molecular profile. These findings underscore the importance of tumor subtype specificity in genetic studies of meningiomas to uncover unique mutational patterns. Taken together, these results highlight the consistency of key molecular alterations in OGMs and support the robustness of our findings across institutions. We did not observe significant interinstitutional molecular differences, with key driver genes (AKT1, SMO, PIK3CA, TRAF7, POLR2A, and KLF4) consistently represented. While racial and ethnic factors can influence tumor biology in some cancers, our study did not reveal such disparities in OGMs.

Beyond genetic alterations, our study is the first to comprehensively examine the correlation between mutational status and distinct tumor characteristics, revealing novel associations that have implications for surgical management. Specifically, we observed a strikingly high prevalence of tumor-associated hyperostosis of the sphenoid planum, particularly pronounced in tumors without AKT1 mutations. Furthermore, OGMs with either TRAF7-only or PIK3CA/PIK3R1 mutations frequently demonstrated ethmoid sinus invasion. These novel findings underline the connection between genetic profiles and specific tumor characteristics, influencing the complexity of surgical interventions, notably the feasibility of achieving Simpson grade I resection that requires the complete excision of the tumor along with affected dura and bone.

In addition, we have confirmed that SMO-mutant OGMs are significantly larger at diagnosis compared with AKT1-mutant OGMs, as previously described by Strickland et al.³² Interestingly, AKT1-mutant tumors, despite their smaller size, were more associated with a higher incidence of seizures, while SMO-mutant tumors, although larger, often present with nonspecific symptoms or remain asymptomatic. This highlights the influence of molecular subtype on clinical presentation and the potential need for tailored management strategies. Furthermore, here we describe that TRAF7-only mutant OGMs are significantly larger than TRAF7/KLF4-mutant tumors, despite sharing a common mutation. This finding suggests a more indolent growth pattern in TRAF7/KLF4-mutant tumors, an observation consistent with our prior findings in foramen magnum meningiomas.³³ This supports the hypothesis that certain genetic alterations influence tumor growth dynamics.

From a prognostic standpoint, recurrences were observed in 12.2% of all cases. Our data reveal that patients with SMO mutations exhibit shorter PFS and a higher recurrence rate compared with those with AKT1 mutations. This observation is consistent with previous studies, including one that screened 79 patients with OGM for SMO and AKT1 mutations.³⁰ Remarkably, in our cohort, 50% of the patients who experienced tumor recurrence had SMO-mutant meningiomas. This is significant considering that 91.7% of SMO-mutant tumors were classified as WHO grade I meningiomas, while a higher proportion of WHO grade II meningiomas was observed in the AKT1- and TRAF7-only mutant OGMs. This trend suggests that the recurrence rate of SMO-mutant meningiomas is not merely a function of tumor grade. Youngblood et al. found a significant correlation between the recurrences of SHH meningiomas and the extent of resection.³⁸ Importantly, the vast majority of our cases underwent gross-total resection (GTR; 96%), with no significant differences in resection extent across molecular subgroups. SMO-mutant OGMs showed the highest rate of recurrence (22.2%), despite 100% of these cases undergoing GTR. This reinforces our finding that SMO mutation is an independent risk factor for decreased PFS, as confirmed in our multivariate analysis, and challenges the notion that incomplete resection could be responsible for the increased recurrence rate within this molecular subgroup. Furthermore, the rare use of RT across all molecular subgroups suggests that this treatment modality does not explain the observed trends.

The therapeutic potential highlighted by our study is substantial. The identification of SMO mutations in OGMs, especially those associated with larger tumor sizes and more aggressive clinical features, suggests that FDA-approved SMO inhibitors such as vismodegib and sonidegib^34–36 could provide new postoperative treatment opportunity. This is particularly relevant for patients with large, incompletely resected, or recurrent SMO-mutant OGMs, who may be at high risk of recurrence. Clinical trials testing SMO inhibitors for these patients could significantly shift the treatment landscape by integrating medical therapies alongside surgical approaches and provide insight into the optimal timing, duration, and efficacy of the targeted therapies.

This insight underlines the necessity for molecular screening of OGMs, particularly because SMO-mutant meningiomas, constituting the group with larger tumor sizes, present challenges in achieving complete resection compared with tumors with AKT1 and TRAF7/KLF4 mutations. Moreover, the SMO-mutant OGMs have the worst prognosis in terms of recurrence, potentially necessitating further surgical interventions or RT. From a biological perspective, it is interesting to note that mutations in AKT1 and the PIK3CA/PIK3R1 genes, which are critical regulators of the PI3K pathway, are seen in OGMs in a notably younger age group, with similar female-to-male ratios, yet distinct tumor characteristics. Tumors with PIK3CA/PIK3R1 mutations are larger and more frequently exhibit bony hyperostosis and tumor infiltration into the ethmoid cells than AKT1-mutant tumors. Of note, the NCT02523014, Alliance A071401 trial, is currently enrolling patients with AKT1- or PIK3CA-mutant meningiomas. Given the high frequency of AKT1 and PIK3CA mutations in OGMs, these tumors represent an ideal group for study in this trial.

Peyre et al. previously reported on a small cohort of patients with aggressive, diffuse midline skull base meningiomas, including 2 cases with only TRAF7 mutations detected.³⁷ In our dataset, 7 patients with OGM exhibited TRAF7-only mutations, including the patient with a WHO grade III meningioma in our study, presenting even a larger tumor size at diagnosis compared with SMO-mutant OGMs. While the small number of cases in our research limits the broad applicability of these findings, it underscores the importance of understanding the TRAF7 gene function for discovering new therapeutic options for this subset of OGMs.

While our study offers valuable insights into the molecular characteristics of OGM, it is essential to acknowledge its limitations. The use of different genomic panels across institutions and the inherent limitations of our sample size could affect the generalizability of our findings. Additionally, the limited sample of nonmeningothelial cases might have influenced our ability to detect significant associations between mutations and specific histological subtypes.

Conclusions

This comprehensive study highlights the unique genetic and clinical profile of OGMs, where mutations in SMO, AKT1, and PIK3CA were notably frequent, present in 70% of cases. SMO-mutant OGMs, which are rare in other meningiomas, were associated with larger tumor size and an elevated recurrence rate—even after complete resection and despite WHO grade I histology—suggesting a more aggressive tumor behavior that could benefit from targeted therapies such as FDA-approved SMO inhibitors. Interestingly, AKT1-mutant tumors, despite their smaller size, presented more frequently with seizures, underscoring the influence of genetic subtype on clinical symptoms. Furthermore, TRAF7-only and PIK3CA/PIK3R1 mutations were linked with specific features, including ethmoid sinus invasion, complicating resection. These findings suggest that molecular profiling may contribute to the diagnosis, treatment, and prognostic assessment of OGMs, highlighting the potential role of genetic analysis in guiding future personalized treatment approaches.

Acknowledgments

This project was funded by the Bundesministerium für Forschung, Technologie und Raumfahrt (Federal Ministry of Research, Technology and Space [BMFTR]) under the BMFTR’s Advanced Clinician Scientist Programs (funding code: 01EO2101, project title: CAMINO).

Disclosures

Dr. Cahill reported personal fees from Servier, Boston Scientific, and Incephalo; and advisory board (equity option) from Pyramid Biosciences outside the submitted work. Dr. Brastianos reported consulting for Advise Connect Inspire, Atavistik Bio, and Axiom HealthCare Strategies; personal fees from CraniUS, Genentech, Eli Lilly, InCephalo, Kazia, and MPM; scientific advisory board of CraniUS and Kazia; grant support to MGH and support for clinical trial (including drug) from Mirati Merck and Eli Lilly; clinical trial support/drug for clinical trial grant support from Bristol Myers Squibb, AstraZeneca, Pfizer, and Kazia; grant support to MGH from Kinnate; nonfinancial support from Genentech-Roche (clinical trial support/drug for clinical trial); and non-financial support from Verastem (drug support for research) outside the submitted work. Dr. T. A. Juratli reported personal fees from CSL Behring outside the submitted work.

Author Contributions

Conception and design: TA Juratli, Alkhatib, Hua, Podlesek, Teo. Acquisition of data: TA Juratli, Alkhatib, Hua, Beyer, Prilop, Podlesek, Jeyaretna, Fujio, Günther, Cicek, Herold. Analysis and interpretation of data: TA Juratli, Alkhatib, Hua, Jeyaretna, Zolal, Günther, Herold, Cahill, Brastianos. Drafting the article: TA Juratli, Alkhatib, Hua, Beyer, Podlesek, Fujio, Zolal, JH Juratli, Cahill, Brastianos. Critically revising the article: TA Juratli, Hua, Beyer, Podlesek, Jeyaretna, Fujio, Zolal, Cicek, JH Juratli, Cahill, Wakimoto, Teo, Gong, Pinzer, Eyüpoglu, Brastianos. Reviewed submitted version of manuscript: TA Juratli, Hua, Alkhatib, Podlesek, Jeyaretna, Fujio, JH Juratli, Cahill, Wakimoto, Teo, Gong, Schackert, Eyüpoglu, Brastianos. Approved the final version of the manuscript on behalf of all authors: TA Juratli. Statistical analysis: Hua, Zolal. Administrative/technical/material support: Fujio, Zeugner, Gong. Study supervision: TA Juratli, Gong, Eyüpoglu.