DNA methylation-based classification and grading system for meningioma: a multicentre, retrospective analysis

TL;DR: Compared with WHO grading, classification by individual and combined methylation classes more accurately identifies patients at high risk of disease progression in tumours with WHO grade I histology, and patients at lower risk of recurrence among WHO grade II tumours.

Abstract: Summary Background The WHO classification of brain tumours describes 15 subtypes of meningioma. Nine of these subtypes are allotted to WHO grade I, and three each to grade II and grade III. Grading is based solely on histology, with an absence of molecular markers. Although the existing classification and grading approach is of prognostic value, it harbours shortcomings such as ill-defined parameters for subtypes and grading criteria prone to arbitrary judgment. In this study, we aimed for a comprehensive characterisation of the entire molecular genetic landscape of meningioma to identify biologically and clinically relevant subgroups. Methods In this multicentre, retrospective analysis, we investigated genome-wide DNA methylation patterns of meningiomas from ten European academic neuro-oncology centres to identify distinct methylation classes of meningiomas. The methylation classes were further characterised by DNA copy number analysis, mutational profiling, and RNA sequencing. Methylation classes were analysed for progression-free survival outcomes by the Kaplan-Meier method. The DNA methylation-based and WHO classification schema were compared using the Brier prediction score, analysed in an independent cohort with WHO grading, progression-free survival, and disease-specific survival data available, collected at the Medical University Vienna (Vienna, Austria), assessing methylation patterns with an alternative methylation chip. Findings We retrospectively collected 497 meningiomas along with 309 samples of other extra-axial skull tumours that might histologically mimic meningioma variants. Unsupervised clustering of DNA methylation data clearly segregated all meningiomas from other skull tumours. We generated genome-wide DNA methylation profiles from all 497 meningioma samples. DNA methylation profiling distinguished six distinct clinically relevant methylation classes associated with typical mutational, cytogenetic, and gene expression patterns. Compared with WHO grading, classification by individual and combined methylation classes more accurately identifies patients at high risk of disease progression in tumours with WHO grade I histology, and patients at lower risk of recurrence among WHO grade II tumours (p=0·0096) from the Brier prediction test). We validated this finding in our independent cohort of 140 patients with meningioma. Interpretation DNA methylation-based meningioma classification captures clinically more homogenous groups and has a higher power for predicting tumour recurrence and prognosis than the WHO classification. The approach presented here is potentially very useful for stratifying meningioma patients to observation-only or adjuvant treatment groups. We consider methylation-based tumour classification highly relevant for the future diagnosis and treatment of meningioma. Funding German Cancer Aid, Else Kroner-Fresenius Foundation, and DKFZ/Heidelberg Institute of Personalized Oncology/Precision Oncology Program.

Summary

Introduction

• The meninges exert a protective function for the entire central nervous system (CNS).
• While 80 % of meningiomas show a benign clinical behavior and can be cured by resection alone, about 20 % recur and need additional treatment such as repeated surgery, irradiation, and systemic chemotherapy4,5.
• Histopathological evaluation aims at the identification of cases at risk for recurrence.
• For various other CNS tumors, molecular profiling has identified distinct subtypes with characteristic aberrations.
• Many of these correlate with prognosis or provide targets for treatment, and therefore support clinical decision making, e.g. epigenetic subgroups in medulloblastoma8-10 and ependymoma11, or isocitrate dehydrogenase (IDH) status in diffuse glioma12-14.

Results

• DNA methylation analysis identifies six distinct methylation classes of meningioma.
• The authors generated genome-wide DNA methylation profiles from a discovery cohort of 497 meningiomas (Suppl. Fig. 1) along with 309 samples of other extra-axial skull tumors that may histologically mimic meningioma variants, including solitary fibrous tumor/hemangiopericytoma, schwannoma, malignant peripheral nerve sheath tumors, chordoma, chondrosarcoma, fibrous dysplasia, and hemangioblastoma.
• These six subgroups were designated as “methylation classes” (MCs).
• Based on further molecular and clinical characteristics outlined below, the four MCs of Group A were designated MC benign 1 through 3 (MC ben-1, ben-2, ben-3) and MC intermediate A (MC int-A).
• The two MCs of Group B were designated MC intermediate B (MC int-B), and malignant (MC mal).

MC predict clinical course with higher accuracy than WHO grading

• The wide spectrum of clinical behavior among WHO grade I and II meningiomas points towards the limited prognostic power of the current classification, particularly at the border between grade I and II.
• The authors further combined MCs exhibiting virtually identical benign (MC ben-1, MC ben-2, MC ben-3) or intermediate (MC int-A, MC int-B) outcome into combined MCs (Fig. 2C).
• The authors next sequenced 304 meningiomas with sufficient material available using a custom hybridcapture next-generation sequencing (NGS) panel dedicated to 40 genes previously reported to be mutant in meningioma (Suppl. Table 1), based on their recently established custom NGS approach for routine brain tumor diagnostics21.
• In MC mal, a higher frequency of CDKN2A deletion was apparent (70%).

Discussion

• The 15 subtypes of meningioma included in the current WHO classification have evolved over decades.
• It prompted allotment of distinct WHO grades to specific meningioma subtypes.
• The very different DNA methylation profiles of Groups A and B despite the shared occurrence of NF2 mutations might suggest that meningiomas arise from two different precursor cell populations.
• Moreover, the fact that patients with meningiomas clustering in Group A share a predominantly benign, with a small proportion exhibiting a intermediate clinical course, and that patients with meningiomas of Group B follow an intermediate to malignant clinical course, may further argue towards a distinct cell of origin with different intrinsic propensities for malignant transformation.
• Similarly strong limitations apply to approaches based on copy-number-profiles:.

An integrated diagnosis for meningioma evaluation

• The WHO 2016 revision of the classification for CNS tumors supports the concept of an integrated diagnosis.
• Adopting this WHO approach to the diagnosis of meningioma, the morphological layer corresponds to the current diagnostic standard, i.e. diagnosing the 15 WHO meningioma subtypes and grading according to the morphological scheme.
• Mutational data may enable inferring the MC for a subset within the MC ben-2, e.g. for AKT1 mutant cases, but not in every instance.
• With methylation analysis performed, one of the six MCs can be diagnosed.
• Collectively, the dataset and accompanying classification scheme proposed here advances meningioma diagnostics from histology into an integrated profiling with higher accuracy of risk assessment for individual patients.

Author contributions

• FS and AvD conceived the project, coordinated data generation, and wrote the manuscript with input from all co-authors.
• FS, D Schrimpf and TH analyzed survival data.
• HGW, ASB, PB, HE, K Kurian, AFO, CM, CJ, KD, MSR, RK, M Simon, AB, M Westphal, KL, AK, JS, VPC, SB, M Platten, DH, AU, WP, WW, MM, M Preusser, CHM, and M Weller collected and interpreted clinical data and/or compiled respective tissue collections.

Acknowledgments

• The authors thank Hai Yen Nguyen, Laura Doerner, Jochen Meyer and Julian Baron for excellent technical assistance.
• The authors also thank the Microarray unit of the Genomics and Proteomics Core Facility, German Cancer Research Center (DKFZ), especially Roger Fischer, Nadja Wermke and Anja Schramm-Glück, for providing excellent methylation services.

Samples

• Samples with clinical data were retrospectively collected from the Dept. of Neuropathology Heidelberg, Germany (local and referral cases), Dept. of Neurosurgery Heidelberg and the FORAMEN network, the Dept. of Neurology and Neuropathology, Zürich, Switzerland, and the Neurological Institute (Edinger Institute) Frankfurt/Main, Germany.
• Additional samples without survival annotation were included from the Dept. of Neuropathology and Neurosurgery Berlin, Bonn, Hamburg, Magdeburg, Münster, Tübingen (all Germany), and Bristol (UK).
• The validation cohort was provided by the Medical University of Vienna.
• Copy-number aberrations were inferred from methylation array data using the R/Bioconductor package conumee.
• Cohort-wide copy number analysis in MCs Methylation-class wide relative copy-number assessment was performed based on 450k data by a proprietary algorithm and controlled by manual inspection of the conumee-based copy-numberprofiles (Stichel et al., in preparation).

Panel and RNA sequencing

• Panel sequencing for genes reported to be mutant in meningioma (Suppl. Table 1) was performed applying a custom hybrid-capture approach as described before.
• Statistical analysis of clinical parameters Distribution of survival times was estimated by the method of Kaplan and Meier and compared between groups with the log-rank test.
• Hazard ratios including 95% confidence intervals based on Cox regression models were calculated.
• Hazard ratio for age is given per 10 year increment.
• Prediction error curves based on the Brier score were computed.

Legends

• Figure 1 Unsupervised clustering of methylation data of 479 meningioma samples (A).
• Unsupervised clustering of matched primary and recurrent samples (matched primary/recurrent samples of identical patient identified by arrows) combined with reference samples from group A and B shows that no shift between groups occurs upon recurrence (B).
• Figure 3 Comparison of WHO grading and methylation-based risk prediction: WHO grade I cases allotted to an intermediate methylation class show PFS similar to the average grade II tumors.
• In turn, WHO grade II cases assigned to a benign methylation class have longer PFS than the average WHO grade II cases (A).
• Copy number variations across all samples that underwent 450k analysis (497) the MCs (B).

Full text

ZurichOpenRepositoryand
Archive
UniversityofZurich
UniversityLibrary
Strickhofstrasse39
CH-8057Zurich
www.zora.uzh.ch
Year:2017
DNAmethylation-basedclassicationandgradingsystemformeningioma:a
multicentre,retrospectiveanalysis
Sahm,Felix;Schrimpf,Daniel;Stichel,Damian;Jones,DavidTW;Hielscher,Thomas;Schefzyk,
Sebastian;Okonechnikov,Konstantin;Koelsche,Christian;Reuss,DavidE;Capper,David;Sturm,
Dominik;Wirsching,Hans-Georg;Bergho,AnnaSophie;Baumgarten,Peter;Kratz,Annekathrin;
Huang,Kristin;Wefers,AnnikaK;Hovestadt,Volker;Sill,Martin;Ellis,HayleyP;Kurian,
KathreenaM;Okuducu,AliFuat;Jungk,Christine;Drueschler,Katharina;Schick,Matthias;
Bewerunge-Hudler,Melanie;Mawrin,Christian;Seiz-Rosenhagen,Marcel;Ketter,Ralf;Simon,
Matthias;etal
DOI:https://doi.org/10.1016/S1470-2045(17)30155-9
PostedattheZurichOpenRepositoryandArchive,UniversityofZurich
ZORAURL:https://doi.org/10.5167/uzh-141060
JournalArticle
AcceptedVersion
ThefollowingworkislicensedunderaCreativeCommons:Attribution-NonCommercial-NoDerivatives
4.0International(CCBY-NC-ND4.0)License.
Originallypublishedat:
Sahm,Felix;Schrimpf,Daniel;Stichel,Damian;Jones,DavidTW;Hielscher,Thomas;Schefzyk,Sebas-
tian;Okonechnikov,Konstantin;Koelsche,Christian;Reuss,DavidE;Capper,David;Sturm,Dominik;
Wirsching,Hans-Georg;Bergho,AnnaSophie;Baumgarten,Peter;Kratz,Annekathrin;Huang,Kristin;
Wefers,AnnikaK;Hovestadt,Volker;Sill,Martin;Ellis,HayleyP;Kurian,KathreenaM;Okuducu,Ali
Fuat;Jungk,Christine;Drueschler,Katharina;Schick,Matthias;Bewerunge-Hudler,Melanie;Mawrin,
Christian; Seiz-Rosenhagen,Marcel; Ketter, Ralf; Simon,Matthias; etal(2017).DNAmethylation-
basedclassicationandgradingsystemformeningioma:amulticentre,retrospectiveanalysis.Lancet
Oncology,18(5):682-694.
DOI:https://doi.org/10.1016/S1470-2045(17)30155-9

Molecular analysis of meningioma increases prognostic power: A methylation-based classification
and grading system
Felix Sahm, MD
1,2
, Daniel Schrimpf, PhD
1
, Damian Stichel, PhD
2
, David T.W. Jones, PhD
3
, Thomas
Hielscher, MSc
4
, Sebastian Schefzyk, MSc
1
, Konstantin Okonechnikov, PhD
3
, Christian Koelsche, MD
1,2
,
David Reuss, MD
1,2
, David Capper, MD
1,2
, Dominik Sturm, MD
3,5
, Hans-Georg Wirsching, MD
6
, Anna
Sophie Berghoff, MD
7
, Peter Baumgarten, MD
8
, Annekathrin Kratz, MD
1,2
, Kristin Huang, MD
1,2
,
Annika K Wefers, MD
1,2
, Volker Hovestadt, PhD
9
, Martin Sill, PhD
4
, Hayley P Ellis, BSc
10
, Kathreena M
Kurian, MD
10
, Ali Fuat Okuducu, MD
11
, Christine Jungk, MD
12
, Katharina Drueschler, MD
13
, Matthias
Schick
14
, Melanie Bewerunge-Hudler, PhD
14
, Christian Mawrin, MD
15
, Marcel Seiz-Rosenhagen, MD
16
,
Ralf Ketter, MD
17
, Matthias Simon, MD
18
, Manfred Westphal
19
, MD, Katrin Lamszus, MD
19
,
Albert
Becker, MD
20
, Arend Koch, MD
21
, Jens Schittenhelm, MD
22
, Elisabeth J Rushing
23
, V Peter Collins,
MD
24
, Stefanie Brehmer, MD
16
, Lukas Chavez, PhD
3
, Michael Platten, MD
13,25,26
, Daniel Hänggi, MD
16
,
Andreas Unterberg, MD
12
, Werner Paulus, MD
28
, Wolfgang Wick, MD
13,28
, Stefan M. Pfister, MD
3,5
,
Michel Mittelbronn, MD
8
, Matthias Preusser, MD
29
, Christel Herold-Mende, PhD
12
, Michael Weller,
MD
6
, Andreas von Deimling, MD
1,2
1.
Department of Neuropathology, Institute of Pathology, Ruprecht-Karls-University Heidelberg, Heidelberg, Germany,
2.
Clinical Cooperation Unit Neuropathology, German Consortium for Translational Cancer Research (DKTK), German Cancer
Research Center (DKFZ), Heidelberg, Germany
3.
Division of Pediatric Neurooncology, German Consortium for Translational Cancer Research (DKTK), German Cancer Research
Center (DKFZ), Heidelberg, Germany
4.
Division of Biostatistics (C060), German Cancer Research Center (DKFZ), and German Cancer Consortium (DKTK), Heidelberg,
Germany
5.
Department of Pediatric Oncology, Haematology and Immunology, Heidelberg University Hospital, and National Center for Tumor
Diseases (NCT), Heidelberg, Germany
6.
Department of Neurology, University Hospital and University of Zurich, Zurich, Switzerland
7.
Institute of Neurology, Comprehensive Cancer Center, CNS Tumours Unit (CCC-CNS), Medical University of Vienna, Austria
8.
Neurological Institute (Edinger-Institute), Goethe University, Frankfurt, and German Cancer Consortium (DKTK), Germany
9.
Division of Molecular Genetics, German Cancer Research Center (DKFZ), Heidelberg, Germany; current address: Broad Institute,
Cambridge, MA, USA
10.
Brain Tumour Research Group, Institute of Clinical Neurosciences, Southmead Hospital, University of Bristol, UK
11.
Department of Pathology, University Hospital Nürnberg, Nürnberg, Germany
12.
Department of Neurosurgery, University Hospital Heidelberg, Heidelberg, Germany
13.
Neurology Clinic, Heidelberg University Hospital, Heidelberg, Germany
14.
Genomics and Proteomics Core Facility, Micro-Array Unit, German Cancer Research Center, Heidelberg, Germany
15.
Department of Neuropathology, Otto von Guericke University Magdeburg, Magdeburg, Germany
16.
Department of Neurosurgery, University Hospital Mannheim, Mannheim, Germany
17.
Department of Neurosurgery, Saarland University, Homburg, Germany
18.
Department of Neurosurgery, Ev. Krankenhaus Bielefeld, Bielefeld, Germany
19.
Department of Neurosurgery, University Hospital Hamburg-Eppendorf, Hamburg, Germany
20.
Department of Neuropathology, University of Bonn, Bonn, Germany
21.
Department of Neuropathology, Charité Medical University, Berlin, Germany
22.
Department of Neuropathology, University Hospital Tübingen, Tübingen, Germany

Sahm et al., p. 2
23.
Department of Neuropathology, University Hospital and University of Zurich, Zurich, Switzerland
24.
Department of Molecular Histopathology, University of Cambridge, UK
25.
Clinical Cooperation Unit Neuroimmunology and Brain Tumor Immunology, German Consortium for Translational Cancer Research
(DKTK), German Cancer Research Center (DKFZ), Heidelberg, Germany
26.
Current address: Neurology Clinic, University Hospital Mannheim, Mannheim, Germany
27.
Institute of Neuropathology, University Hospital Münster, Münster, Germany
28.
Clinical Cooperation Unit Neurooncology, German Consortium for Translational Cancer Research (DKTK), German Cancer Research
Center (DKFZ), Heidelberg, Germany
29.
Department of Medicine I, Comprehensive Cancer Center, CNS Tumours Unit (CCC-CNS), Medical University of Vienna, Austria
Corresponding author
Dr. med. Felix Sahm and Prof. Dr. med. Andreas von Deimling
Department of Neuropathology
Institute of Pathology
Ruprecht-Karls-University Heidelberg, Heidelberg
Im Neuenheimer Feld 224
D-69120 Heidelberg
email: andreas.vondeimling@med.uni-heidelberg.de
phone: +49 6221 56 4651

Sahm et al., p. 3
Summary
Background
The World Health Organization (WHO) classification of brain tumors describes 15 subtypes of
meningioma. Nine of these are allotted to WHO grade I, and three each to grade II and grade III,
respectively. Grading is purely based on histology, with molecular markers lacking. While the current
classification and grading approach is of prognostic value, it harbors shortcomings such as ill-defined
parameters for subtypes and grading criteria prone to arbitrary judgment.
Methods
We investigated genome-wide DNA methylation patterns of 479 meningiomas to identify distinct
methylation classes (MC) of meningioma. The MCs were further characterized by DNA copy-number
analysis, mutational profiling and RNA sequencing. We validated our findings in an independent
cohort of 140 tumors.
Findings
DNA methylation profiling distinguished six distinct MCs associated with typical mutational,
cytogenetic, and gene expression patterns. Meningioma MCs exhibit a more homogeneous clinical
course and allow prognostication with significantly higher power than the current morphology-based
WHO classification. Meningioma MCs more accurately identify patients at high risk of recurrence
among tumors with WHO grade I histology, and patients at lower risk of recurrence among WHO
grade II tumors. DNA methylation-based classification and grading reduces the number of
meningioma subtypes from 15, as historically defined by histology, to six clinically relevant MCs, each
with a characteristic molecular and/or clinical profile.
Interpretation
DNA methylation-based meningioma classification captures biologically more homogenous groups
and has a higher power for predicting tumor recurrence than the current WHO classification. The
approach presented here is highly useful for stratifying meningioma patients for observation or
adjuvant treatment groups. We consider methylation-based tumor classification highly relevant for
the future diagnosis and treatment of meningioma.
Funding
This work was supported by the German Cancer Aid (110670, 110983) and the Else Kröner-Fresenius
Foundation (A_60). We thank the DKFZ-Heidelberg for funding by HIPO H033.

Sahm et al., p. 4
Research in context
Evidence before this study
Meningiomas, the most frequent primary intracranial tumors, are diagnosed and graded according to
the WHO classification of brain tumors. The recent update of this classification in 2016 has
implemented molecular markers for several brain tumor entities.
However, there are still no established prognostic molecular markers for meningioma. Meningioma
diagnostics is still based on purely histological criteria which are prone to a high inter-observer and
sampling bias. Thus, the relevance of the current grading system for clinical decision making is
heavily debated.
Previous work by several groups, including ours, has shown that DNA methylation signatures are
specific for tumor entities. Importantly, DNA methylation profiling can identify biologically and
clinically relevant subgroups among histologically indiscernible cases. Here, we employed this
concept for the classification of meningiomas. A search in PubMed on October 21 2016 did not
identify articles which used high-resolution DNA methylation profiling for identification of clinically
relevant subgroups across all subtypes and grades of meningioma.
Added value of this study
We demonstrate that classification of meningiomas based on DNA methylation profiling is more
powerful in predicting the clinical behavior than the current WHO classification and grading system.
Our findings on a discovery series were confirmed on an independent validation series. Most notably,
the novel approach was capable of identifying patients at high risk of rapid recurrence which were
expected to have benign tumors based on WHO grading. Likewise, a considerable fraction of patients
with the histological diagnosis of a higher grade meningioma - fostering the consideration of
adjuvant treatment - but no recurrence could upfront be identified as low risk by DNA methylation
profiling.
Implications of all the available evidence
Our data demonstrate that meningioma patients can be more accurately stratified for tumor
behavior by DNA methylation profiling than by the current WHO classification. This greatly improves
the basis for clinical decision making for or against additional therapy after surgery. We expect
epigenetic profiling to be included into the diagnostic routine and implemented into upcoming
updates of the WHO classification for brain tumors.

Frequently asked questions

Q1. What are the contributions in "Dna methylation-based classification and grading system for meningioma: a multicentre, retrospective analysis" ?

Sahm et al. this paper performed a comprehensive characterization of the entire molecular genetic landscape of meningioma in order to identify biologically and clinically relevant subgroups that refine the current classification scheme. 

Q2. What is the significance of the study?

Interpretation DNA methylation-based meningioma classification captures biologically more homogenous groups and has a higher power for predicting tumor recurrence than the current WHO classification. 

Q3. How many MCs are more accurately able to predict recurrence?

DNA methylation-based classification and grading reduces the number of meningioma subtypes from 15, as historically defined by histology, to six clinically relevant MCs, each with a characteristic molecular and/or clinical profile. 

Q4. What are the shortcomings of the current classification and grading approach?

While the current classification and grading approach is of prognostic value, it harbors shortcomings such as ill-defined parameters for subtypes and grading criteria prone to arbitrary judgment.