According to the WHO classification, ependymal tumors are classified as subependymomas, myxopapillary ependymomas, classic ependymomas, anaplastic ependymomas, and RELA-fusion-positive ependymomas (RELA-EPN). Among classic ependymomas, the WHO defines rare histological variants, i.e., the clear cell, papillary, and tanycytic ependymoma. In parallel, global DNA methylation patterns distinguish nine molecular groups, some of which tightly overlap with histopathological subgroups. However, the match of the aforementioned histological variants to DNA methylation classes remains unclear. We analyzed histomorphology, clinical parameters, and global DNA methylation of tumors with the initial histological diagnoses of tanycytic (n = 12), clear cell (n = 14), or papillary ependymoma (n = 19). Forty percent of these tumors did not match to the epigenetic profile of ependymomas, using a previously published DNA methylation-based classifier for brain tumors. Instead, they were classified as low-grade glioma (n = 3), plexus tumor (n = 2), CNS high-grade neuroepithelial tumor with MN1 alteration (n = 2), papillary tumor of the pineal region (n = 2), neurocytoma (n = 1), or did not match to any known brain tumor methylation class (n = 8). Overall, integrated diagnosis had to be changed in 35.6% of cases as compared to the initial diagnosis. Among the tumors molecularly classified as ependymoma (27/45 cases), tanycytic ependymomas were mostly located in the spine (5/7 cases) and matched to spinal or myxopapillary ependymoma. 6/8 clear cell ependymomas were found supratentorially and fell into the methylation class of RELA-EPN. Papillary ependymomas with a positive ependymoma match (12/19 cases) showed either a “papillary” (n = 5), a “trabecular” (n = 1), or a “pseudo-papillary” (n = 6) growth pattern. The papillary growth pattern was strongly associated with the methylation class B of posterior fossa ependymoma (PFB, 5/5 cases) and tumors displayed DNA methylation sites that were significantly different when compared to PFB ependymomas without papillary growth. Tumors with pseudo-papillary histology matched to the methylation class of myxopapillary ependymoma (4/6 cases), whereas the trabecular case was anatomically and molecularly a spinal ependymoma. Our results show that the diagnosis of histological ependymoma variants is challenging and epigenetic profiles may improve diagnostic accuracy of these cases. Whereas clear cell and papillary ependymomas display correlations between localization, histology, and methylation, tanycytic ependymoma does not represent a molecularly distinct subgroup.
This is a preview of subscription content, log in to check access.
We thank Anne Reichstein for her dedication and excellent technical support concerning the DNA-methylation analyses. We also thank Ulrike Rumpf, Hannelore Junker-Polzin, Carolina Janko, Tasja Lempertz, and Celina Liza Soltwedel for excellent technical support with tissue sections and stainings. J.N. was supported by the Else-Kröner Fresenius Stiftung and the UKE Nachwuchsförderung. This project was generously funded by the Gert and Susanna Mayer Stiftung. U.S. was supported by the Fördergemeinschaft Kinderkrebszentrum Hamburg.
Supplementary Fig. 1: t-SNE plot including the 45 cases analyzed in this study that were plotted with brain tumor samples of 82 published CNS tumor methylation classes and with cases of nine control tissue methylation classes (n = 2801 ). Supplementary Fig. 2: Histomorphology of cases diagnosed as tanycytic ependymoma that did not match to any molecular ependymoma group based on DNA methylation. (a–d) H&E stainings. Scale bar in a is 50 μm for a–d. (a) Case #9 showed astrocytic tumor cells arranged around vessels. No clear pseudorosettes were detected. (b) Case #10 was suggestive of a low-grade glioma with abundant and evenly distributed small round calcifications (see arrows). (c) Case #11 impressed as an ependymoma with tanycytic morphology. (d) Case #12 showed a fibrillary tanycytic pattern with hyalinized vessels (see arrows). Supplementary Fig. 3: Histomorphology of cases diagnosed as clear cell ependymoma that did not match to any molecular ependymoma group based on DNA methylation. (a–f) H&E stainings. Scale bar in a is 50 μm for a–f. (a) Case #21 showed slightly pleomorphic tumor cells with clear cell morphology (left panel) and areas with ependymal pseudorosettes (right panel), being compatible with a clear cell ependymoma. (b) Case 22 showed thin capillaries, small oligodendroglia-like cells and some larger cells suggestive of “floating neurons” (see arrow). (c,d) Cases #23 and #24 impressed as uniformly clear cell tumors without papillary structures. (e) Clear cells with accentuated cell boundaries and nuclei free vascular spaces. (f) Uniform round cells and calcifications (see arrow), well compatible with a neurocytoma. Supplementary Fig. 4: Molecular characteristics of RELA-fusion-positive ependymoma comparing cases with (n = 6) or without clear cell morphology (n = 65). (a) Unsupervised hierarchical clustering of RELA-fusion-positive ependymoma cases with or without clear cell morphology based on 3 significantly differentially methylated CpG sites (FDR < 0.1, delta beta value ≥ 0.2) did not reveal distinct clusters. Details on all 4 significantly differentially methylated CpG sites can be found in supplementary Table 1. (b) t-SNE plot of RELA-fusion-positive ependymomas. Cases with clear cell morphology (n = 6) located scattered among cases without clear cell morphology (n = 65) (c) Stacked integrated copy number plots of cases with or without clear cell morphology showed similar chromosomal alterations. RELA, NCCM = RELA-fusion-positive ependymoma, clear cell morphology. RELA, NCCM = RELA-fusion-positive ependymoma, no clear cell morphology. FDR = false discovery rate. Supplementary Fig. 5: Histomorphology of cases diagnosed as papillary ependymoma that did not match to any molecular ependymoma group based on DNA methylation. (a–g) H&E stainings. Scale bar in a is 50 μm for a–g. (a,d,e) Cases #39, 42 and 43 showed a pseudo-papillary picture with hyalizined vessels being present in cases #42 and 43. (b) Papillary growing tumor, slightly suggestive of plexus tissue, well compatible with a plexus tumor. (c) Case #41 was a highly vascularized tumor with nuclei free areas and circular arrangement of tumor cells around vessels. (f) Case #44 showed abundant trabecular structures resembling ependymal canals. (g) Case #45 showed tumor cell nests within a myxoid matrix. Supplementary Fig. 6: Molecular characteristics of posterior fossa group B ependymoma, comparing papillary morphology (n = 5) with non-papillary morphology (n = 15). (a) t-SNE plot showing cases with papillary morphology or non-papillary morphology including a data set of published posterior fossa group B ependymoma . Histomorphology was not associated with one of the published 5 subtypes of posterior fossa group B ependymoma . (b) Stacked integrated copy number plots of cases with papillary morphology or non-papillary morphology showed common chromosomal alterations but papillary ependymomas lacked gain of chromosome 11. PFB1-5 = posterior fossa group B ependymoma, subtypes 1-5. (PDF 50941 kb)
Grajkowska W, Matyja E, Pronicki M, Daszkiewicz P, Roszkowski M, Perek D et al (2009) Papillary ependymoma with unique superficial cortical location: immunohistochemical and ultrastructural studies. A case report. Folia Neuropathol 47:354–361PubMedGoogle Scholar
Hovestadt VZM (2017) conumee: Enhanced copy-number variation analysis using Illumina DNA methylation arrays. R package version 1.9.0Google Scholar
1.Institute of NeuropathologyUniversity Medical Center Hamburg-EppendorfHamburgGermany
2.II. Medizinische Klinik und PoliklinikUniversity Medical Center Hamburg-EppendorfHamburgGermany
3.Department of Pediatric Hematology and OncologyUniversity Medical Center Hamburg-EppendorfHamburgGermany
4.Institute of NeuropathologyUniversity Hospital MünsterMünsterGermany
5.Center for NeuropathologyLudwig-Maximilians-UniversityMunichGermany
6.German Center for Neurodegenerative DiseasesMunichGermany
7.Department of Neuropathology, Institute of PathologyUniversity Hospital, Heidelberg and Clinical Cooperation Unit Neuropathology, German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ)HeidelbergGermany
8.Clinical Cooperation Unit NeuropathologyGerman Cancer Consortium (DKTK), German Cancer Research Center (DKFZ)HeidelbergGermany
9.Hopp Children’s Cancer Center Heidelberg (KiTZ)HeidelbergGermany
10.Institute of PathologyUniversity of BaselBaselSwitzerland
11.Department of Neuropathology, Institute of PathologyUniversity of WürzburgWürzburgGermany
12.Department of NeuropathologyCharité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of HealthBerlinGermany
13.Division of Pediatric Neuro-OncologyGerman Cancer Research Center (DKFZ), German Cancer Consortium (DKTK)HeidelbergGermany
14.Department of Pediatric Oncology and HematologyUniversity Hospital HeidelbergHeidelbergGermany
15.Research Institute Children’s Cancer Center HamburgHamburgGermany