Meningioma transcriptomic landscape demonstrates novel subtypes with regional associated biology and patient outcome - PubMed (original) (raw)

. 2024 Jun 12;4(6):100566.

doi: 10.1016/j.xgen.2024.100566. Epub 2024 May 23.

Damian Almiron-Bonnin 2, Nicholas Nuechterlein 3, Sonali Arora 1, Matt Jensen 4, Carolina A Parada 5, Chengxiang Qiu 6, Frank Szulzewsky 1, Collin W English 7, William C Chen 8, Philipp Sievers 9, Farshad Nassiri 10, Justin Z Wang 10, Tiemo J Klisch 7, Kenneth D Aldape 11, Akash J Patel 7, Patrick J Cimino 3, Gelareh Zadeh 10, Felix Sahm 9, David R Raleigh 8, Jay Shendure 6, Manuel Ferreira 5, Eric C Holland 12

Affiliations

Meningioma transcriptomic landscape demonstrates novel subtypes with regional associated biology and patient outcome

H Nayanga Thirimanne et al. Cell Genom. 2024.

Abstract

Meningiomas, although mostly benign, can be recurrent and fatal. World Health Organization (WHO) grading of the tumor does not always identify high-risk meningioma, and better characterizations of their aggressive biology are needed. To approach this problem, we combined 13 bulk RNA sequencing (RNA-seq) datasets to create a dimension-reduced reference landscape of 1,298 meningiomas. The clinical and genomic metadata effectively correlated with landscape regions, which led to the identification of meningioma subtypes with specific biological signatures. The time to recurrence also correlated with the map location. Further, we developed an algorithm that maps new patients onto this landscape, where the nearest neighbors predict outcome. This study highlights the utility of combining bulk transcriptomic datasets to visualize the complexity of tumor populations. Further, we provide an interactive tool for understanding the disease and predicting patient outcomes. This resource is accessible via the online tool Oncoscape, where the scientific community can explore the meningioma landscape.

Keywords: Oncoscape; UMAP; brain tumor; bulk RNA-seq; meningioma; meningioma subtypes; patient prognosis prediction; recurrent.

Copyright © 2024 The Authors. Published by Elsevier Inc. All rights reserved.

PubMed Disclaimer

Conflict of interest statement

Declaration of interests Although the majority of Oncoscape is open source, a subset of the technology and computational algorithms presented in this paper are covered by serial no. 63/595,717, and N.N., S.A., M.J., and E.C.H. are listed as inventors.

Figures

None

Graphical abstract

Figure 1

Figure 1

Generating the meningioma reference UMAP and coloring by clinical and genomic metadata (A) Method overview. (B) UMAP colored by datasets. (C) Tumors with (blue) and without (yellow) loss of chromosome 22. (D) Tumors with (blue) and without (yellow) NF2 mutations. (E) Tumors with (blue) and without (yellow) NF2 gene fusions. (F) NF2 expression. (G) YAP1 gene fusions. (H) Tumors with (blue) and without (yellow) mutations in TRAF7/KLF4/AKT1/SMO. (I) WHO grade: grade 1 (yellow), 2 (green), and 3 (red). (J) Recurred (blue) and primary (yellow) tumors. (K) Patients’ gender (female pink, male blue). Region 1 marked by the red dashed line. (L) Age at sample acquisition. Region 2 marked by red dashed line. na, not available. Figure 1 in Oncoscape. See also Figure S1 and Table S1.

Figure 2

Figure 2

Various grading systems show regional patterns across the UMAP (A) UMAP colored by the Baylor RNA classification: A/NF2wt_ben = NF2 wild-type benign (blue), B/NF2loss_int = NF2 lost intermediate (green), C/NF2loss_mal = NF2 lost malignant (red). (B) UMAP colored in by UCSF DNA methylation-based classification of a subset of tumors: hypermitotic, red; immune-enriched, green; Merlin-intact, blue. (C) UMAP colored in by the Toronto methylation profile of a subset of tumors: MG1/immunogenic, orange; MG2/benign_NF2wt, (benign NF2 wild type) blue; MG3/hypermetabolic, green; MG4/proliferative, red. (D and E) UMAP colored by GSVA scores calculated using UCSF gene set (D) upregulated and (E) downregulated in most aggressive meningioma. 1 suggests upregulation, −1 suggests downregulation of the respective gene set. (F) Ratio of GSVA scores from upregulated and downregulated gene sets. Black arrows indicate the regions with the most aggressive tumors marked by the ratio. Figure 2 in Oncoscape. See also Figure S2.

Figure 3

Figure 3

Meningioma subtypes with distinct time to recurrence (A) Seven regions identified by DBSCAN denoting meningioma subtypes A–G. Unclustered samples (n = 37) are shown in gray. (B) Kaplan-Meier plots for the seven regions based on time to recurrence (AvsB and AvsC p < 0.0001; CvsD and BvsC p < 0.02); n(A) = 317, n(B) = 172, n(C) = 131, n(D) = 101. (C) Subclusters of cluster A (A1 and A2). (D) Kaplan-Meier plots showing recurrence-free rates of cluster A subclusters (A1 vs. A2: p < 0.0001); n(A1) = 230, n(A2) = 36. (E) Subclusters of cluster C (C1, C2, C3, and C4). (F) Kaplan-Meier plots showing recurrence-free rates of cluster C subclusters (C1 vs. C2, p < 0.0001; C3 vs. C4, p = 0.26; C1 vs. C3, p = 0.01); n(C1) = 26, n(C2) = 5, n(C3) = 13, n(C4) = 100. Figure 3 in Oncoscape. See also Figure S3.

Figure 4

Figure 4

Regionally enriched gene fusions and copy-number alterations (A) Fusion burden in each tumor derived from high-confidence gene fusions called using RNA-seq. (B and C) Examples for regionalized fusions. (D) Burden of CNA in each tumor (loss of chromosome arms). (E–H) Loss (−1), gain (1), or intact (0) status of (E) chromosome 1p, (F) chromosome 6q, (G) chromosome 14q, and (H) chromosome 10q in each tumor. (I) Manhattan plots showing losses (blue) and gains (red) of each chromosome arm in clusters A, B and C. (J) Kaplan-Meier plot showing the recurrence-free rate of patients in cluster A with intact and deleted chr 1p; p < 0.0001, n(del) = 194, n(intact) = 86. (K) chr 6q, p < 0.0001, n(del) = 96, n(intact) = 184. Figure 4 in Oncoscape. See also Figures S4–S7 and Tables S2 and S3.

Figure 5

Figure 5

Biological significance of meningioma subtypes (A) Visualization of GSVA scores across the UMAP for selected Gene Ontology Biological Processes (GO BP) terms. 1 suggests upregulation, −1 suggests downregulation of the respective gene set. (B) Top 15 GO BP terms enriched in clusters A and B. (C) Summary of biological significance of each cluster. (D) Mouse embryonic cell types enriched in each cluster (top hits). Cell type similarities are as listed. Welch’s two sample t test; p < 2.2e−6. (E) Expression profiles for genes known to be involved in embryonic limb development. (F) Kaplan-Meier plots showing correlation between recurrence-free rate and HOXD13 levels; p = 0.0022, n(high) = 66, n(low) = 69. Figure 5 in Oncoscape. See also Figure S8–S10 and Tables S4, S5, and S6.

Figure 6

Figure 6

Evolution of multiple tumors from the same patient (A) Primary and recurred tumors from the same patient. Arrows show the direction from the first tumor to the second tumor of a specific patient. Tumors from a single patient are distinguished by the colors. pt, patient. (B) Multiple individual tumors occurred within the same patient. Each patient distinguished by different colors (e.g., pt1.1 = patient 1 tumor 1, pt1.2 = patient 1 tumor 2). (C) Primary and progressed tumors (e.g., pt1.1 = patient 1 tumor 1, pt1.2 patient 1 tumor 2 [progressed]). See also Table S7.

Figure 7

Figure 7

Overlaying new patients on to the reference UMAP (A) Two of 100 UMAP embeddings produced by 100 pre-trained UMAP models trained with different random states. (B) New patient VST data are mapped onto all 100 UMAP embeddings using the pre-trained UMAP models. (C) For each UMAP embedding, the nearest 100 neighbors are chosen subject to a radius R determined by cross-validation. (D) Example plot of the reference UMAP with samples colored by the frequency of each reference sample in our reference dataset being a nearest neighbor of a new patient. (E) Illustration of the placement of a new patient at the centroid of the nearest neighbors weighted by the frequency vector in (D) after outlier exclusion. (F) The ground-truth location of a reference sample during cross-validation. (G) The placement of a reference sample using our placement method during cross-validation. (H) Comparison of the ground-truth placement and the centroid it is mapped to during cross-validation. (I) The distribution of the distances between the ground-truth placement of a reference sample and its centroid placement for all reference samples during cross-validation. (J) Kaplan-Meier curves for location grade predictions within WHO grade 1, 2, and 3 meningiomas. See also Figures S11 and S12.

References

    1. Sahm F., Schrimpf D., Stichel D., Jones D.T.W., Hielscher T., Schefzyk S., Okonechnikov K., Koelsche C., Reuss D.E., Capper D., et al. DNA methylation-based classification and grading system for meningioma: a multicentre, retrospective analysis. Lancet Oncol. 2017;18:682–694. doi: 10.1016/S1470-2045(17)30155-9. -DOI -PubMed
    1. Louis D.N., Perry A., Wesseling P., Brat D.J., Cree I.A., Figarella-Branger D., Hawkins C., Ng H.K., Pfister S.M., Reifenberger G., et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23:1231–1251. doi: 10.1093/neuonc/noab106. -DOI -PMC -PubMed
    1. Nassiri F., Liu J., Patil V., Mamatjan Y., Wang J.Z., Hugh-White R., Macklin A.M., Khan S., Singh O., Karimi S., et al. A clinically applicable integrative molecular classification of meningiomas. Nature. 2021;597:119–125. doi: 10.1038/s41586-021-03850-3. -DOI -PMC -PubMed
    1. Choudhury A., Magill S.T., Eaton C.D., Prager B.C., Chen W.C., Cady M.A., Seo K., Lucas C.H.G., Casey-Clyde T.J., Vasudevan H.N., et al. Meningioma DNA methylation groups identify biological drivers and therapeutic vulnerabilities. Nat. Genet. 2022;54:649–659. doi: 10.1038/s41588-022-01061-8. -DOI -PMC -PubMed
    1. Wang E.J., Haddad A.F., Young J.S., Morshed R.A., Wu J.P.H., Salha D.M., Butowski N., Aghi M.K. Recent advances in the molecular prognostication of meningiomas. Front. Oncol. 2022;12 doi: 10.3389/fonc.2022.910199. -DOI -PMC -PubMed

MeSH terms

LinkOut - more resources