A 52-year-old man with an Eastern Cooperative Oncology Group (ECOG) performance status of 1 initially presented with an isolated episode of collapse in October 2013. Prior to this presentation, he and his family had noticed behavioral changes, worsening short-term memory, and headaches not relieved by simple analgesia. A brain MRI scan depicted a large bifrontal mass lesion ( Figure 1A ) which straddled the midline on either side of the interhemispheric fissure in the anterior cranial fossa, measuring 7cm in width and almost 6cm in the anteroposterior dimension, consistent with a meningioma. Due to his escalating symptoms, he underwent a resection. The resection was performed via a bi-coronal flap. The elevation of one flap intraoperatively was noted to be difficult due to adherence to the meningioma. The meningioma was enucleated and a frozen sample at the time of surgery confirmed a meningioma. The base of the tumor arose from the underside of the sagittal sinus. The anterior aspect adhered to the falx, and the ipsilateral layer of the falx and surrounding tissues underwent diathermy. A Simpson 2 resection was achieved. This resected specimen confirmed a WHO grade 2 meningioma with focal rhabdoid changes. The specimen was moderately cellular, with large, ovoid nuclei and centrally located nucleoli with frequent mitosis and copious amounts of eosinophilic cytoplasm. Some cells showed intracytoplasmic ill-defined hyaline structures consistent with rhabdoid change. The proliferation index as assessed by MIB-1 was elevated at approximately 10%-15%. Adjuvant radiation of 60Gy/30 fractions was completed following recovery from surgical resection and a discussion at a neurosurgical multidisciplinary team meeting. Following this, the patient began a surveillance program with MRI scans performed every 3 months initially, which demonstrated postoperative encephalomalacia and gliosis within both the anterior and parasagittal frontal lobes. The interval increased to 6 months until the end of 2016 when a brain MRI scan demonstrated a local recurrence ( Figure 1A ). This MRI brain scan with contrast that showed the recurrence of his disease depicted two areas of lobulated enhancing tissue within the resection cavity. However, it was also noted that there was surrounding T2/FLAIR hyperintensity within the frontal lobes, consistent with post-radiation change. During this period, the patient remained asymptomatic with an ECOG performance status of 1 and he continued to work. However, due to further progression over two subsequent MRI scans, which showed increasing areas of enhancement extending from the prior resection cavity to the right frontal horn resulting in a midline shift, and discussion with his neurosurgical team, he opted to proceed with re-resection in September 2018. A histological examination demonstrated a recurrence of the WHO grade 2 meningioma. The patient completed a further course of radiation therapy (54Gy/30fractions), and recovered well, returning to work following recovery from radiation. Microsatellite instability (MSI) testing of the specimen resected in 2018 confirmed a microsatellite stable (MSS) tumor. Further surveillance MRI was performed at 3-month intervals until May of 2020 when the patient presented with an episode of collapse associated with a 2-month history of grade 1 dyspnoea and dry cough. Diagnostic imaging with a CT scan of his thorax, abdomen, and pelvis with contrast performed during this admission demonstrated multiple bilateral pulmonary masses ( Figure 1A ). Bronchoscopy and biopsy of the pulmonary lesions confirmed metastatic meningioma.

The standard of care adjuvant therapy in high-grade or atypical meningiomas is radiation, however, in this case, referral for consideration of systemic treatment was made due to the recurrence after radiation of new and progressive systemic disease. In July 2020, the patient was referred to medical oncology where he was commenced on sunitinib based on data from other case reports (15). These case reports showed that, in a small exploratory cohort of 13 patients, sunitinib is active in patients with recurrent/atypical meningioma with a progression-free survival of 5.2 months (15). He remained on sunitinib for 14 months and tolerated it well, with regular disease monitoring including both CT and MRI. A repeat CT scan of the thorax, abdomen, and pelvis and a brain MRI scan performed after 14 months of treatment, identified progression of the disease in the thorax, but stable disease in the brain. At this point, the patient was switched to bevacizumab based on a systemic review carried out by Franke et al. (16) but his disease progressed further after 4 months of this treatment, with progression again isolated to the disease in the lungs.

Following a discussion at a medical oncology multi-disciplinary team meeting, whole genome sequencing (WGS) was requested in November 2021 and discussed at a neurosurgical molecular tumor board (MTB) in June 2022. Detailed descriptions of the WGS, mutation, and copy number variant analyses are supplied in the Supplementary Material . Unfortunately, due to significant clinical deterioration and presentation with multiple episodes of seizure over the months preceding the MTB, the patient was not administered any further lines of systemic anti-cancer treatment.

WGS was performed on samples taken from the primary tumor resection (M1-T), the re-excised recurrent tumor (M2-T) after adjuvant radiation therapy, the lung metastases (M3-T) before treatment with sunitinib, and one paired blood sample for tumor-normal analysis ( Figure 1B ; Supplementary Figure S1 ). WGS data was used for genomic characterization of somatic single nucleotide variants (SNVs), insertions and deletions (InDels), and copy number alterations (SCNAs) across all tumor samples. We used the Bi et al.’s genomic sequencing of high-grade meningioma tumor samples (7) and Nassiri et al.’s molecular subgroups (11) as a reference for molecular classification and characterization of the meningioma tumors profiled here ( Supplementary Figure S2 ).

The genome-wide somatic copy number profiles were remarkably similar across all tumor sample timepoints ( Figure 2A ). Chromosome 22q loss encompassing the NF2 gene was present in all tumors along with SCNAs previously reported in Grade 2 and Grade 3 meningioma primary tumors including chr1p, chr6q, chr10, chr14q, chr18p, and q copy number loss. In the NF2 gene, a C>T nonsense (stop gained) mutation was detected in the primary tumor (M1-T) and lung metastases (M3-T) at variant allele frequency (VAF) estimates of 32% and 100% ( Figures 2B, C ). This suggests that in combination with chromosome 22q loss, biallelic NF2 inactivation occurred in these tumors. Structural variant calling identified a large number of intrachromosomal inversions in both the primary tumor (n=114) and lung metastases (n=398) ( Figures 2D, E ) with few interchromosomal translocations (TRAs) identified.

Overall, the SCNA profiles here are characteristic of higher-grade meningioma tumors and of the MG4 proliferative and hypermitotic tumor subtypes reported by Nassiri et al. and Choundary et al. respectively (11, 12). Next, the tumor mutational burden (TMB) was calculated as the total number of nonsynonymous somatic mutations divided by the exome region length in megabases (Mb) ( Figure 2F ). This analysis identified a twofold increase in TMB from 0.450 (log TMB/Mb) in the primary meningioma tumor to 0.935 (log TMB/Mb) in the recurrence and 1.1467 (log TMB/Mb) in the lung metastases. The primary and recurrence TMBs were in the same range as previously reported nonsynonymous TMB estimates for higher-grade, NF2-altered recurrent meningioma tumors ( Supplementary Figure S2 ). Next, in order to investigate which mutagenic process may have led to a stepwise increase in TMB and chromosomal instability with disease progression, we performed a mutational signature analysis of somatic single nucleotide variants ( Figure 2F ; Supplementary Figure S3 ). Mutational signatures extracted from tumor samples were fitted to Catalogue Of Somatic Mutations In Cancer (COSMIC) reference signatures (Signatures 1-30). In both the primary meningioma and lung metastases we detected Signature 3 [associated with homologous repair deficiency (HRD)], Signature 18 (C>A; etiology unknown; found in neuroblastoma) and Signature 29 (Rare; C>A mutation in tobacco chewing). Signature 4, a C>A mutation pattern associated with exposure to tobacco smoking [this case is an ex-smoker (~25 pack years)] and APOBEC-associated Signature 13 were detected only in the lung metastases, with Signature 8 (C>A; found in medulloblastoma) found only in the primary meningioma. None of these signatures were detected in the recurrence, which was overall dominated by a C>T mutation context pattern associated with three mutational signatures: Signature 1 (ageing), Signature 11 (alkylating agents in glioblastoma and melanoma), and Signature 30 (BER deficiency). In the recurrence, we did detect a predicted loss of function driver mutation in SETD2, a reported DNA MMR regulator (17), however, the MSS classification given by MSI testing, performed by the molecular pathology team in 2018, was microsatellite stable.

Next, in order to better understand which, if any, driver mutations besides NF2 may be associated with cancer initiation, progression, and metastases, we used Cancer Genome Interpreter (CGI) to assign known or predicted novel driver status to nonsynonymous mutations. Following this, for clonal evolution analysis, PyCloneVI and ClonEvol were used to infer clonal population structure and clonal ordering from copy number and purity-adjusted VAFs from somatic mutations across all tumor sample timepoints ( Figure 3A ; Supplementary Figure S4 ). Five tumor clones (labeled Cluster/Clone #1 to #5) were detected across the primary meningioma, recurrence, and lung metastases sample timepoints ( Figure 3C ). The primary tumor was composed of two clones: Clone #3 and Clone #5, with Clone #3 present at a clonal prevalence of 0.915. This clone was stably maintained in both the recurrence and lung metastases at clonal prevalence values of 0.676 and 0.968 respectively. Nonsynonymous predicted driver mutations in Clone #3 included NF2 (p.Q389*), as previously described above, BRCA2 (p.E51K), and SEC23B (p.G285V) ( Figure 3B ). This suggests Clone #3 led to the initiation and development of the primary meningioma tumor with NF2 the main driver event. Our analysis also suggests that the presence of BRCA2 may represent a secondary or potentially modifying genetic event; however, this interpretation warrants cautious consideration, particularly given the established role of NF2 mutations in meningioma. Clone #5 was detected only in the primary tumor and contained mutations in genes previously reported to be frequently mutated (> 2 tumors) in high-grade meningioma including LRP1B (7). In addition to Clone #3, the recurrence was composed of two other clones, Clone #1 and Clone #4. Clone #4 contained the driver mutation, previously detailed above, in SETD2 (p.S1885N) and it appears to have not seeded the lung metastases, however, Clone #1 did. The lung metastases were composed of Clone #3, Clone #1, and Clone #2 ( Figures 3D, E ). Although SMARCB1 mutations have previously been shown to be co-mutated in atypical meningiomas (13), we did not identify any SMARCB1 mutations in our study.

Next, having characterized the clonal composition, evolution, and dynamics of the meningioma lung metastases, we used this information along with CGI annotation to identify which mutations were drug response biomarkers ( Table 1 ). The only predicted drug-responsive biomarkers detected were NF2 (p.Q389* mutation/copy number loss or deletion) and BRCA2 (p.E51K mutation) in Clone #3, present in all three tumors and SETD2 (p.S1885N mutation) present only in Clone #4 in the recurrence ( Figure 3F ).