A Clinically Applicable Integrative Molecular Classification of Meningiomas

Authors

Nassiri F

Liu J

Patil V

Mamatjan Y

Wang JZ

Hugh-White R

Macklin A

Khan S

Singh O

Karimi S

Corona RI

Liu LY

Chen CY

Chakravarthy A

Wei Q

Mehani B

Suppiah S

Gao A

Workewych AM

Tabatabai G

Boutros PC

Bader GD

de Carvalho DD

Kislinger T

Aldape K

Zadeh G

Doi

10.1038/s41586-021-03850-3

PMID: 34433969 · DOI: 10.1038/s41586-021-03850-3 · Journal: Nature (2021)

TL;DR

Nassiri et al. assembled a TCGA-style multi-omic resource of 121 meningiomas (discovery) plus 80 (validation) enriched for higher-grade tumors. By integrating somatic copy-number, DNA methylation, mRNA, and proteomics using the Cluster of Cluster Algorithm (COCA), they defined four consensus molecular groups (MG1–MG4) — immunogenic, benign NF2-wildtype, hypermetabolic, and proliferative — that outperformed WHO grade and prior methylation-based classifications in predicting recurrence-free survival. Single-cell RNA sequencing confirmed the biological distinctiveness of these groups, and the HDAC inhibitor vorinostat selectively reduced viability of MG4 (proliferative) patient-derived cell lines in vitro and in xenograft models.

Cohort & data

  • Discovery cohort: 121 fresh-frozen meningiomas from the University Health Network Brain Tumor BioBank (Toronto), enriched for WHO grade 2 and 3 tumors, plus 5 healthy meninges controls (PMID:34433969).
  • Validation cohort: 80 independent meningiomas assessed using mRNA signatures (PMID:34433969).
  • Dataset: mng_utoronto_2021.
  • Cancer type: MNG (meningioma).
  • Assays: whole-exome sequencing (median 191X), Illumina EPIC methylation array, mRNA-seq, whole-cell proteomics (6,568 proteins quantified in 96 tumors), and single-nuclear RNA-seq (54,393 nuclei from 8 tumors + 2 healthy meninges).

Key findings

  • COCA integration of six CNA clusters, six DNA methylation clusters, and six mRNA clusters converged on four stable molecular groups (MG1–MG4) (PMID:34433969).
  • MG classification was independently associated with recurrence-free survival after adjusting for WHO grade, extent of resection, and adjuvant radiotherapy (multivariable Cox regression; log-rank P = 5 x 10^-15) (PMID:34433969).
  • MG classification was superior to WHO grade, methylation-based classification, and individual-datatype clustering for predicting time to recurrence (PMID:34433969).
  • NF2 was mutated in 88% of MG1 vs. 9% of MG2 tumors (Fisher exact P = 5.9 x 10^-8). Non-NF2 driver mutations (TRAF7 25%, AKT1 13%, KLF4 13%, POLR2A 6%) were exclusive to MG2 (Fisher exact P = 1.20 x 10^-8) (PMID:34433969).
  • Novel recurrent driver mutations in chromatin-remodeling genes KDM6A and CHD2, and tumor suppressor PTEN, were identified at 3–5% frequency, collectively enriched in MG3/MG4 (Fisher exact P = 0.002) (PMID:34433969).
  • Subthreshold recurrent inactivating mutations were found in CREBBP (q = 0.126), FBXW7 (q = 0.226), and RB1 (q = 0.250) (PMID:34433969).
  • MG4 tumors had significantly greater mutational burden than MG1–3 (Kruskal-Wallis P = 1.6 x 10^-3) (PMID:34433969).
  • Percent genome altered was higher in MG3 (median 16.9%) and MG4 (median 19.5%) vs. MG1 (3.5%) and MG2 (9.6%) (Kruskal-Wallis P = 5.2 x 10^-6) (PMID:34433969).
  • Proteogenomic analysis identified group-specific protein markers: S100B for MG1, SCGN for MG2, ACADL for MG3, and MCM2 for MG4, validated by immunohistochemistry (PMID:34433969).
  • Vorinostat selectively decreased viability of MG4-derived cell lines and attenuated tumor growth and improved survival in MG4 intracranial xenografts (PMID:34433969).

Genes & alterations

  • NF2 – Point mutations + chromosome 22q loss causing biallelic inactivation; near-universal in MG1 (88%), rare in MG2 (9%); some NF2-wildtype MG3/MG4 showed expression silencing without methylation changes.
  • TRAF7 – Point mutations exclusive to MG2 (25%).
  • AKT1 – Point mutations exclusive to MG2 (13%).
  • KLF4 – Point mutations exclusive to MG2 (13%).
  • SMO – Point mutations in a copy-number-neutral subset of MG2.
  • POLR2A – Point mutations exclusive to MG2 (6%).
  • KDM6A – Novel recurrent driver mutations (chromatin remodeling); enriched in MG3/MG4.
  • CHD2 – Novel recurrent driver mutations (chromatin remodeling); enriched in MG3/MG4.
  • PTEN – Novel recurrent driver mutations (tumor suppressor); enriched in MG3/MG4.
  • CREBBP – Subthreshold recurrent inactivating mutations (q = 0.126).
  • FBXW7 – Subthreshold recurrent inactivating mutations (q = 0.226).
  • RB1 – Subthreshold recurrent inactivating mutations (q = 0.250).
  • SMARCB1, DMD – Known meningioma driver genes assessed in the mutational analysis.
  • FOXM1, MYC – Key proliferation-associated transcription factor networks enriched in MG4.
  • S100B, SCGN, ACADL, MCM2 – IHC-validated protein markers for MG1, MG2, MG3, and MG4, respectively.

Clinical implications

  • The four integrative molecular groups provide a classification that outperforms WHO grading for predicting meningioma recurrence, with potential to inform future WHO classification revisions (PMID:34433969).
  • Proteomic markers (S100B, SCGN, ACADL, MCM2) could enable molecular group assignment in conventional neuropathology laboratories via immunohistochemistry, pending independent validation (PMID:34433969).
  • Vorinostat (HDAC inhibitor) showed selective efficacy against MG4 (proliferative) meningiomas in vitro and in vivo, warranting further clinical investigation for a cancer type with no effective medical therapies (PMID:34433969).
  • MG2 tumors (benign NF2-wildtype) harboring TRAF7/AKT1/KLF4/SMO mutations may be amenable to targeted therapies directed at these specific pathways (PMID:34433969).

Limitations & open questions

  • The discovery cohort (n = 121) was enriched for higher-grade tumors, which may affect the prevalence estimates of molecular groups in unselected populations.
  • The validation cohort (n = 80) used mRNA-only signatures; full multi-omic validation in independent cohorts is needed.
  • Vorinostat efficacy was demonstrated only in preclinical models (cell lines and xenografts); in-human trials are required.
  • The study used the 2016 WHO classification; the 2021 WHO update may affect grade-MG correlations.
  • Single-cell data was limited to 8 tumors and 2 healthy meninges, restricting the power of intratumoral heterogeneity analyses.
  • The relationship between molecular groups and response to radiotherapy was not specifically addressed.

Citations from this paper used in the wiki

  • “NF2 was predictably the most commonly point mutated gene, its prevalence differed significantly across MGs… 88% vs 9%, Fisher’s Exact test, P=5.9x10^-8” (p. 4).
  • “COCA combining six copy number clusters with six DNA methylation and six mRNA abundance clusters converged to reveal four novel stable molecular groups (MG 1-4) of meningioma” (p. 3).
  • “MG classification was independently associated with recurrence-free survival even after accounting for known prognostic clinical factors including WHO grade, extent of surgical resection, and receipt of adjuvant radiotherapy” (p. 4).
  • “Treatment with Vorinostat selectively decreased the viability of only MG4-tumor-patient-derived cell lines in comparison to patient-derived cell lines of other MGs” (p. 6).
  • “S100B for MG1, SCGN for MG2, ACADL for MG3 and MCM2 for MG4” (p. 6).

This page was processed by crosslinker on 2026-05-04.