sign in sign up
<<
Home , Esthesioneuroblastoma

Molecular Pathology of CNS Tumors with Limited Resources

H.K. Ng Chinese University of Hong Kong

Full ppt at http://www.acp.cuhk.edu.hk/hkng/ – the with the worse survival among all common now Median survival is 14 months with standard-of-care treatment Should the diagnosis of glioblastoma just be HISTOLOGICAL ?

Currently, no single Biomarkers for glioblastoma

Endothelial proliferation necrosis Histologic criteria for glioblastoma were quite old

Kernohan grading system 1949 Necrosis as a diagnostic marker of glioblastoma - 1985 But IDH genotype is a powerful prognosticator

IDH genotyping is more accurate than what pathologists seen down the microscope

Andreas von Deimling

Hartmann C et al. Acta Neuropathologica 2010

WHO 2021 – led by IARC

Ten neuropathologists - Brat, Dan - Ellison, David - Figarella-Branger, Dominique - Hawkins, Cynthia - Louis, David - Perry, Arie - Ng, H K - Von Deimling, Andreas - Reifenberger, Guido - Wesseling, Pieter

Two clinicians - Pfister, Stefan - Soffietti, Riccardo

New blue book out in late Spring 2021 WHO classification Of CNS Tumors 2021

Unfortunately, even common tumors are quite a handful In contrast to other tumors Disclaimer : Just don’t shoot the messenger Molecular diagnostics of brain tumors is an obligation

To have or not to have ?

Is this a question anymore ? Aims of Molecular Pathology

• More precise classification • Prognosis • Prediction to treatment / drugs • May suggest need for genetic counselling

• Special situation of recurrent tumors BRAF inhibitors Currently used for cancers

• BRAFV600E inhibitors • Selumetinib (AZD6244) • FDA approved in adults – MEK inhibitor in trial • Vemurafenib – Phase II for low-grade • Dabrafenib • Sorafenib (PBTC) • MEK inhibitors • FDA approved in adults • Trametinib • Cobimetinib Local cases of response to BRAF inhibitors

Woo, Poon , Ng. Oncotarget 2019 Patient from Sickkids, Toronto Info from internet Potential Predictive biomarkers for reporting

• 1p19q • MGMT • PDL-1 • BRAF • EGFR amplification • IDH • EGFRvIII • PTCH • Mismatch repair proteins • NTRK • AKT1 • ALK • ………………… • FGFR-TACC • MET • ROS1 • PDFRA • Her2 • Hereditary tumor syndrome is the only known aetiology outside brain tumors other than irradiation • Common hereditary syndromes NF1 – gliomas, OPG NF2 – , bilateral acoustics, spinal Li Fraumeni – , young , FAP – WNT MB Lynch – young glioblastomas MMR deficiency – young glioblastomas Nevoid basal cell carcinoma – nodular Cowden’s syndrome – gangliocytomas Dicer1 – ……… WHO 2021 20 % of MB Have a genetic predisposition Esp SHH MB Stefan Pfister 2018

My algorithm

• Molecular IHC • Single gene : Sanger sequencing and FISH and MGMT • Multiple genes by single gene tests • Genomics : Target sequencing, RNAseq

• For diagnostic clarification : methylation profiling • Special case : medulloblastoma • Commercial or “outside” sources To do genomics or not Where else where can I get the tests done ?

Molecular tests for brain tumors are available In China too

http://www.acp.cuhk.edu.hk/hkng/

Please see my website For details of genomics

Not a pay / commercial service

Partially funded by Children Cancer Foundation and S K Yee You can do all the IHC and single and multiple single gene studies So what I am doing, you can mostly do too

世上无难事

Molecular diagnostics of brain tumors is an obligation

To have or not to have ?

Is this a question anymore ? Recurrent brain tumors

• patients die of recurrence and not of sepsis or met • Glioma patients die of infiltration esp brain stem and probably not of RICP • not surgically resected in the past • Usually meant close to the end in the past but slowly changing due to re-operation • Limited amount of tissue based molecular studies – few studies showed conservation of driver mutations • Great potential for research for molecular evolution Recurrent brain tumors – how to report ?

• Distinguish between the three if possible : radionecrosis, residual and recurrence • IDH genotype maintained • Usually had RT already, so drug targets become important • Glioma - Hypermutator phenotype induced by Temozolamide • Glioma – may lose their G-cimp high status Glioma – may assume more mesenchymal type or stem cell phenotype • MB – maintain their molecular groups but mutation profiles change So let’s start the algorithm and have fun Immunohistochemistry The simplest thing Molecular IHC : put it in another line

IDH-1 Molecular Immunohistochemistry of CNS tumors

• IDH1 – adult gliomas, but also chondrosarcomas and cholangiocarcinoma - Much rarer than IDHwt (90%) for glioblastomas - Negative can exclude - Not IDH2 mutation - Sequencing when patient is young and astrocytic - Classifier (WHO2016) and prognosticator for both low and high grade Isocitrate dehydrogenase (IDH)1

Median survival : 31 months for mutated GBM (IDH1 or 2), 15 months for wild-type

Yan et al. NEJM 2009 Gupta et al. 2011 IDH – oncogene or tumor suppressor gene?

NADP+ NADPH NADPH NADP+ Isocitrate α-ketoglutarate 2-hydroxyglutarate (2-HG) Wild-type IDH Mutant IDH

Role in tumorigenesis….? INI1 loss

• Lost in ATRT • But also in poorly differentiated chordomas • CRINET ATRT – INI-1 BRAF V600E

• Potential response to inhibitors • PXA • Some pilocytic astrocytomas (5-10%), especially those with poorer prognosis • • Some AYA glioblastomas • Epitheloid glioblastoma • Papillary craniophrayngiomas • Erdheim-Chester • PLNTY

Clear, oligodendroglia-like cells in pLGG Polymorphous low grade neurepithelial tumor of the young From WHO 2021

About 50% BRAF mutated or have other MAPK aberrations

CD34 BRAF

？ 8/M A form of PLNTY or pediatric low grade Glioma with MAPK activation Yang, Ng. Brain Pathology， in press Papillary

BRAF

From WHO 2021 Epitheloid glioblastoma

WHO 2021 BRAF ATRX - loss

• Pitfalls : normal cells • Diagnosis of not oligo • Better prognosis in astrocytomas and glioblastoma • DIPG • G34 pediatric high grade gliomas • Gliomas associated with NF1 • with piloid features (+CDKN2A loss) ATRX H3K27M • Midline H3K27M mutant gliomas (DIPG) • Pitfalls : may occur outside midline; usually poor prognosis DIPG – H3K27M mutant mid-line glioma

H3K27me3 - loss

• Posterior fossa (PFA) : poorer prognosis • Poor prognosis in meningiomas, usually higher grade • Often : DIPG • Oligodendroglioma • Pitfall : normal cells Ependymal tumors location, genetics and prognosis

Supratentorial Ependymomas (30%) -RELA-fusion, poor prognosis (infant to adult) -YAP1-fusion, good prognosis (infants, children) -, good prognosis (adults)

Posterior Fossa Ependymomas (60%) -PF-EPN-A, poor prognosis (infants, children) -PF-EPN-B, good prognosis (children, adults) -subependymoma, good prognosis (adults)

Spinal Ependymomas (10%) (children to adults) -NF2 mutation -good prognosis

Myxopapillary Ependymoma (adults) (100%) in Conus Medullaris -good prognosis

Modified Pajtler et al., Cancer Cell , 2016 • Supratentorial: - two distinct types: • YAP 1 group – good survival

• C11orf95-Rela fusion sub-type: • pathologic activation of the NF-kappa B signaling pathway • 48 - 70% of supratentorial primaries

Posterior Fossa: - two distinct types: • PF Group B – older children/adults • PF Group A - younger children/infants • CIMP+ (CpG island methylator phenotype) • 78% of posterior fossa primaries • Very poor prognosis IHC for H3K27me3

Venneti & Judkins Brain Pathology 2013 Loss of H3K27me3 poor prognosticators in meningiomas G34 – mutant pediatric high grade glioma H3K27me3 Beta-catenin

• Pitfall : nuclear staining; background dirty • Wnt medulloblastoma • Adamantinomatous craniopharyngioma Wnt MB - Beta-catenin Adamaninomatous craniopharngioma

Beta-catenin G34 V/R pediatric high grade gliomas

IHC efficacy not yet fully known Fontebasso & Jabado Brain Pathology 2013

MSH2, MSH6, PMS2, MLH1

• Pitfall : do the normal cells stain ? • Must do in pediatric high grade gliomas • One allele lost – Lynch syndrome; biallelic loss : BMMR deficiency • Implications for genetic counselling • Mutations usually denotes hypermutators – may be resistant to temozolamide SHH medulloblastomas

• Yap1 more sensitive • Gab1 more specific • P53 poor prognosis • P53 mutated SHH – genetic counseling WHO 2016 Filamin Filamin

SHH medulloblastoma

Yap1 p53

• Interpretations is difficult • Unfortunately, we still have to use IHC • Astrocytoma not oligodendroglioma • Giant cell glioblastoma • Worsened prognosis in NF1 patients • SHH MB subdivision Giant cell glioblastoma ROS1 / ALK ?

• Brain met • Infantile high grade gliomas • Better off doing FISH or RNAseq Infantile gliomas are a separate group in WHO 2021

2019

Infant high grade gliomas comprise multiple subgroups Characterised by novel target fusions and better survivals Clarke M, Mackay A…….Ng HK……Jones C Cancer Discovery 2020 Others

• Lin28a – ETMR • L1CAM – RELA ependymoma • BAP1 loss in high grade meningiomas • NKX2-1 FOXR2 embryonal tumors (PNET) • BCOR BCOR altered embryonal tumor (PNET) • NUTM1 Some CIC-altered CNS • EZHIP – H3wt DIPG, PFA

BAP1 loss in Rhabdoid meningiomas

Shankar & Santagata Neuro-oncology 2017 Loss of BAP1 poor prognosticator In meningiomas

Shankar, Santagata Neuro-oncology 2017 Single gene studies DNA repair by MGMT

CH3 G C

MGMT

MGMT CH3

G C DNA repair by MGMT

CH3 G C

MGMT

MGMT CH3 Transversion Strand break A T G Repair

G C enhanced chemosensitivity resulting from epigenetic inactivation of the DNA repair gene MGMT

hypermethylation of CpG island in the promoter region of MGMT

transcriptional silencing

O 6 guanine . MGMT ↓ DNA cross linking by alkylating agents

MGMT = O6-methylguanine-DNA methyltransferase MGMT

• Predictor to RT-Chemo in adult glioblastomas • Most common method : methylation specific PCR • Does not help towards morphology • Good prognosis in adult glioblastoma Methylation-specific PCR for MGMT

GBM

1 2 3 4 met+ met- H2O M U M U M U M U M U M U M U

M = methylation U = unmethylation met+ = methylation control met- = unmethylation control MGMT gene silencing and benefit from RT and temozolomide in GBM

Roger Stupp

46% vs 14% respond o TMZ/RT

M. Hegi et al, NEngJMed 352:997-1003, 2005 Need for single gene results, including molecular ICH

• Genomics take a long time • Many single gene studies are critical for the diagnosis and therapeutics (IDH, 1p19q, BRAF fusion etc) Procedure of Sanger Sequencing DNA F& R Primer Polymerase DNTP MgCl2

1)Tissue from tumour area Cell 2)Proteinase K lysate

Prepare cell PCR

lysate amplification ◦ Denaturation: 95◦c,20sec Annealing: 60◦c,20sec Extension: 72◦c,30sec

DNA Perform gel Sequencing electrophoresis

Sequencing reaction IDH1 Chromatogram

Wild type R132H R132L

A G G T C A T A G G T C T T

Mutation Mutation

Arginine->Histidine Arginine->Leucine (>90%) (rare) Useful single gene Sanger sequencing

• IDH1 and 2 • BRAF V600E • H3.3 and H3.1 • Beta catenin – exon 3 • PDGFRA K385 L/I : myxoid glioneuronal tumor BRAF V600E lowers survival in pediatric low grade glioma

Lassaletta….Ng…. Tarbori Journal of Clinical Oncology 2017 Histone mutations in pediatric high grade gliomas 6 years old female

Thalamic GBM

K27M-H3.3 mutations

→ → (AAG ATG, lysine methionine) TERT promoter mutations TERT mutation

• Two hotspots • Not covered by the usual target sequencing or WES • Technically a little more demanding • Diagnostic uses : - Molecular glioblastomas (70-80%) even when histologically low grade astrocytoma - Poor prognosticator in glioblastomas - Poor prognosticator in meningiomas - Pitfall : found also in Life history of a glioblastoma

TERT is a common end point

From Roel Verhaak

From Wesseling and Verhaak Two hotspots in TERT TERT wild type mutations

C250T C228T

C C

Mutation Mutation

Single gene FISH

DNA Fluorophore-conjugated + nucleotides

Labeling

Protease Fluorophore-tagged digestion probes

Denaturation Hybridization

Tissue Double-stranded Single-stranded section DNA DNA Hybrids

Post-hybridization wash Mounting with counterstain

Signal scoring

Imaging

Fluorescence microscope Green/red = Target and reference probes Blue = DAPI-stained nucleus/chromosomes Desirable FISH tests for molecular brain tumor diagnostics

Ng. Glioma 2020 FISH procedure

DNA Fluorophore-conjugated + nucleotides

Labeling

Protease Fluorophore-tagged digestion probes

Denaturation Hybridization

Tissue Double-stranded Single-stranded section DNA DNA Hybrids

Post-hybridization wash Mounting with counterstain

Signal scoring

Imaging

Fluorescence microscope Green/red = Target and reference probes Blue = DAPI-stained nucleus/chromosomes FISH for 1p19q

Red = 1p Red = 19q Green = 1q Green = 19p

EGFR – traditional biomarker for glioblastoma but only 20-25% Also a molecular target for therapy Pediatric glioma is quite a separate group from adults

BRAF FISH is probably the most important test you Need to have if you have to diagnose pediatric glioma

pilocytic CCF website BRAF Gene Rearrangement (Fusion)

Two normal signals (orange) plus a smaller third signal near one of the large signals (arrows)

Rela

M/3 L frontal lobe tumor

LICAM

Or p65/RELA Li, Chan, Ng. NOA 2019

CDKN2A/B (FISH or methylation) : major prognosticator in IDH mutant astrocytomas Criteria for molecular Astrocytoma Grade IV

CIMPACT-NOW Brat et al. 2020

Also, risk for recurrence for meningiomas Other uses of Single gene FISH (Ng. Glioma 2020)

• 1p loss – poor prognosis in meningiomas, diffuse leptomeningeal glioneuronal tumor • Monosomy 6 : Wnt medulloblastoma • 9q loss – SHH medulloblastoma • 10q loss – traditional molecular marker for glioblastoma • CIC fusion – CIC-rearranged sarcoma • EWSR1 fusions – Ewing sarcoma and intracranial myxoid mesenchymal tumor • MN1 fusion – • C-myc amplification – group 3 or poor prognosis in medulloblastoma • MYCN amplification – poor prognositic group in pediatric HGG, and some medulloblastomas, MYCN-DIPG • PRKCA fusion – papillary glioneuronal tumor • CDKN2A deletion – poor prognosis in RELA-ependymoma, poor prognosis in meningiomas, diagnosis of molecular glioblastoma in IDH mutant gliomas Diffuse leptomeningeal gliomatosis Ng. Pathology 1999

Multiple single gene studies 289 grades II and III diffuse gliomas Methylation 1087 diffuse gliomas Expression (Mayo, UCSF, TCGA) >600 grades II and III aCGH miRNA

Eckel-Passow JE & Jenkins RB et al., 2015 TCGA, 2015 Lower grade gliomas – TCGA -2015

• Group 1 – IDH mut, 1p19q codeleted - > 15 yr survival • Group 2 – IDH mut, 1p19q non-deleted – 8-12 year survival • Group 3 – IDH wt, 1p19q non-deleted – survival 2-3 yrs Single gene stratification of triple negative adult low grade gliomas

Chan, Mao, Ng. New England Journal of Medicine 2016 From single gene to multiple genes

Overall survival Progression free survival – risk stratification risk stratification of pediatric low grade gliomas Highrisk; n=21 High risk; n=21 High Time(months) Time(months) Lowrisk; n=86 Lowrisk; n=86 p<0.0001 p<0.0001 p - Intermediate risk; Intermediate =0.0023 p - =0.0002 n=111 Intermediate risk; Intermediate n=111 p<0.0001 p<0.0001 (BRAFfusion or MYB amplification Lowrisk without H3F3A/TERT/BRAF/MYBwithout alterations) (BRAFV600Eor risk Intermediate (H3F3A or TERTmutation) promoter Highrisk Yang, Ng. Shi, Zhang, At GuangzhouChildren’s Hospital Now neurosurgeona RyanYang Acta Neuropathologica 2018 Molecular grading of low grade diffuse astrocytoma Adult IDHwt low grade astrocytomas can be molecularly graded by a panel of genes

Neuro-oncology 2017

Molecularly lower-grade (n=68) Survival % Survival

Mijiti Aibaidula Molecularly high grade (n=51) Neurosurgeon p <0.001 Shanghai

OS (years) 45/M Frontal lobe Diffuse astrocytoma

IDH wt Molecular low grade

Survived 4.6 years

Diagnosis of a molecular glioblastoma

Capper D, et al. Acta Neuropathologica 2018 Genomics

• All the tests mentioned before you can do in your own histo labs • Genomics – most people send away to institutional facilities or companies Ng, Chan, Kan, Li. Glioma 2020 Genome-wide DNA methylation

Available commercially easily and Also at HKU – paediatrics for Brain tumors

Cecarrelli et al. Cell 2016 G-CIMP (Glioma CpG Island Methylator Phenotype)

Ceccarelli, Noushmehr, Verhaak. Cell 2016 G-CIMP can further grade IDH mutant glioblastoma

A B

Li, Nousmehr, Ng. Neuro-oncology Advances 2019 NGS steps Nucleic acid (wet lab) Fragmentation Purpose of adaptor: Library preparation enable sequence to become End repair and adaptor ligation bound to a complementary counterpart Fragment library

Clonal amplification Template preparation (generation clusters of identical DNA molecules for adequate detection) Local amplification of single molecules, Bead are compartmentalized called “cluster” into water-oil emulsion droplet. Solid surface, flow cell-based Emulsion PCR Bridge amplification Sequencing Sequencing the clonal amplicons

Sequence by synthesis Ion semiconductor sequencing

Generation of fluorescent images Data analysis Conversion to sequence Data analysis in NGS (Dry lab) Target sequencing-example

-classic medulloblastoma -M, 5Y -cerebellum -group 3 tumor

IGV plot showing DDX3X mutation

p.R326C -likely pathogenic Oncoprint developed out of target sequencing

Cohort of adult medulloblastoma Wong, Ng. Acta Neuropathologica Communication, in press RNAseq – commonly used for fusion genes

From Al Seesi et al. 2016 RNAseq of an IDH mutant glioblastoma showing fusion genes

Wong, Ng. Modern Pathology, in revision Commercial NGS panels

• Not specific to brain. But overseas brain institutions may do it for you • NGS results have to be with reference to histology, the clinical scenario. ?? Microdissected for tumor concentration ? Quality of tissue • NGS panels vary with respect to their capacity, eg fusion, CNV, genes • No best or standard method or algorithm • NGS panels vary with respect to their reporting Problems with NGS

• Takes about two months per case. Some key information, eg BRAF fusion, 1p19q are critical towards an immediate diagnosis and management – need to know early • Consultation : only a limited number of slides • How much do you include in a report ? • No clinical trial uses NGS yet Factors to consider in NGS

• Amount and purity of samples • Sensitivity and specificity of the method used • Is the variant pathogenic ? • Is the variant monoallelic or biallelic ? • Is the variant a drug target ? • NGS itself does not make a diagnosis • Disclaimer : not an expert in NGS reporting The methylomes A new kind of “histology”

Without the microscope Principle behind methylation array

Binary IDAT files (1 for red, 1 for green) Methylation array

Infinium MethylationEPIC BeadChip (850K)

Each site is assayed using 1 of the 2 assays below (a) Infinium I assay • Use two bead types for each target CpG: methylated (M) and unmethylated (U) • Methylation status is determined by comparing the intensities from the two different probes in the same color

(b) Infinium II assay • Use one bead type for each target CpG • Methylation status is determined by comparing the two colors Two analyses in one experiment

Methylation status ◦ M = signal intensity of the methylated ◦ U = signal intensity of the unmethylated ◦ % of methylation (β) = M / (M + U)

Copy-number variation (CNV) ◦ Compare M + U in a sample of interest with the (M + U)’s in healthy reference samples ◦ High ratios correspond to areas with a gain of chromosomal material, whereas low ratios represent lost DNA. Procedure in methylation-based classification of CNS tumors in our experience

FFPE samples

Send to BGI or other companies for Illumina Infinium 450K BeadChip

Generation of two unprocessed IDAT files

Upload to MolecularNeuropathology.org

Receive a report methylation profiling, copy number plot, and an estimation of MGMT promoter methylation status. Name Methylation class (anaplastic) pleomorphic xanthoastrocytoma Methylation class CNS Ewing sarcoma family tumor with CIC alteration Methylation class CNS high grade neuroepithelial tumor with BCOR alteration Methylation class CNS high grade neuroepithelial tumor with MN1 alteration Methylation class CNS with FOXR2 activation Methylation class Ewing sarcoma Methylation class IDH glioma, subclass 1p/19q codeleted oligodendroglioma Methylation class IDH glioma, subclass astrocytoma Methylation class IDH glioma, subclass high grade astrocytoma Methylation class anaplastic Methylation class atypical teratoid/rhabdoid tumor, subclass MYC Methylation class atypical teratoid/rhabdoid tumor, subclass SHH Methylation class atypical teratoid/rhabdoid tumor, subclass TYR Methylation class central Methylation class cerebellar liponeurocytoma Methylation class chordoid glioma of the third ventricle Methylation class chordoma Methylation class control tissue, cerebellar hemisphere Methylation class control tissue, hemispheric cortex Methylation class control tissue, hypothalamus Methylation class control tissue, inflammatory tumor microenvironment Methylation class control tissue, pineal gland Methylation class control tissue, pituitary gland anterior lobe Methylation class control tissue, pons Methylation class control tissue, reactive tumor microenvironment Methylation class control tissue, white matter Methylation class craniopharyngioma, adamantinomatous Methylation class craniopharyngioma, papillary Methylation class diffuse leptomeningeal glioneuronal tumor Methylation class diffuse midline glioma H3 K27M mutant Methylation class embryonal tumor with multilayered rosettes Methylation class ependymoma, RELA fusion Methylation class ependymoma, YAP fusion Methylation class ependymoma, myxopapillary Methylation class ependymoma, posterior fossa group A Methylation class ependymoma, posterior fossa group B Methylation class ependymoma, spinal Methylation class esthesioneuroblastoma, subclass A Methylation class esthesioneuroblastoma, subclass B Methylation class glioblastoma, IDH wildtype, H3.3 G34 mutant Methylation class glioblastoma, IDH wildtype, subclass MYCN Methylation class glioblastoma, IDH wildtype, subclass RTK I List of tumor classes assigned by genomeMethylation class-wide glioblastoma, IDHDNA wildtype, subclass methylation RTK II profiles Methylation class glioblastoma, IDH wildtype, subclass RTK III Methylation class glioblastoma, IDH wildtype, subclass mesenchymal Name Methylation class glioblastoma, IDH wildtype, subclass midline Methylation class (anaplastic) pleomorphic xanthoastrocytoma Methylation class Methylation class CNS Ewing sarcoma family tumor with CIC alteration Methylation class infantile hemispheric glioma Methylation class CNS high grade neuroepithelial tumor with BCOR alteration Methylation class low grade glioma, MYB/MYBL1 Methylation class CNS high grade neuroepithelial tumor with MN1 alteration Methylation class low grade glioma, desmoplastic infantile astrocytoma / Methylation class CNS neuroblastoma with FOXR2 activation Methylation class low grade glioma, dysembryoplastic neuroepithelial tumor Methylation class Ewing sarcoma Methylation class low grade glioma, ganglioglioma 9 classes of LGG Methylation class IDH glioma, subclass 1p/19q codeleted oligodendroglioma Methylation class low grade glioma, rosette forming glioneuronal tumor Methylation class IDH glioma, subclass astrocytoma Methylation class low grade glioma, subclass hemispheric pilocytic astrocytoma and ganglioglioma Methylation class IDH glioma, subclass high grade astrocytoma Methylation class low grade glioma, subclass midline pilocytic astrocytoma Methylation class anaplastic pilocytic astrocytoma Methylation class low grade glioma, subclass posterior fossa pilocytic astrocytoma Methylation class atypical teratoid/rhabdoid tumor, subclass MYC Methylation class low grade glioma, subependymal giant cell astrocytoma Methylation class atypical teratoid/rhabdoid tumor, subclass SHH Methylation class Methylation class atypical teratoid/rhabdoid tumor, subclass TYR Methylation class medulloblastoma, WNT Methylation class Methylation class medulloblastoma, subclass SHH A (children and adult) Methylation class cerebellar liponeurocytoma Methylation class medulloblastoma, subclass SHH B (infant) 5 classes of MB Methylation class chordoid glioma of the third ventricle Methylation class medulloblastoma, subclass group 3 Methylation class chordoma Methylation class medulloblastoma, subclass group 4 Methylation class control tissue, cerebellar hemisphere Methylation class melanocytoma Methylation class control tissue, hemispheric cortex Methylation class Methylation class control tissue, hypothalamus Methylation class melanotic Methylation class control tissue, inflammatory tumor microenvironment Methylation class Methylation class control tissue, pineal gland Methylation class papillary tumor of the pineal region group A Methylation class control tissue, pituitary gland anterior lobe Methylation class papillary tumor of the pineal region group B Methylation class control tissue, pons Methylation class , spinal non-CIMP Methylation class control tissue, reactive tumor microenvironment Methylation class pineal parenchymal tumor Methylation class control tissue, white matter Methylation class pineoblastoma group A / intracranial Methylation class craniopharyngioma, adamantinomatous Methylation class pineoblastoma group B Methylation class craniopharyngioma, papillary Methylation class / granular cell tumor / spindle cell oncocytoma Methylation class diffuse leptomeningeal glioneuronal tumor Methylation class pituitary adenoma, ACTH Methylation class diffuse midline glioma H3 K27M mutant Methylation class pituitary adenoma, FSH/LH Methylation class embryonal tumor with multilayered rosettes Methylation class pituitary adenoma, STH densely granulated, group A Methylation class ependymoma, RELA fusion Methylation class pituitary adenoma, STH densely granulated, group B Methylation class ependymoma, YAP fusion Methylation class pituitary adenoma, STH sparsely granulated Methylation class ependymoma, myxopapillary Methylation class pituitary adenoma, TSH Methylation class ependymoma, posterior fossa group A 6 classes of ependymoma Methylation class pituitary adenoma, prolactin Methylation class ependymoma, posterior fossa group B Methylation class plasmacytoma Methylation class ependymoma, spinal Methylation class plexus tumor, subclass adult Methylation class esthesioneuroblastoma, subclass A Methylation class plexus tumor, subclass pediatric A Methylation class esthesioneuroblastoma, subclass B Methylation class plexus tumor, subclass pediatric B Methylation class glioblastoma, IDH wildtype, H3.3 G34 mutant Methylation class retinoblastoma Methylation class glioblastoma, IDH wildtype, subclass MYCN Methylation class schwannoma Methylation class glioblastoma, IDH wildtype, subclass RTK I 7 classes of GBM Methylation class / Methylation class glioblastoma, IDH wildtype, subclass RTK II Methylation class subependymoma, posterior fossa Methylation class glioblastoma, IDH wildtype, subclass RTK III Methylation class subependymoma, spinal Methylation class glioblastoma, IDH wildtype, subclass mesenchymal Methylation class subependymoma, supratentorial Methylation class glioblastoma, IDH wildtype, subclass midline Methylation class hemangioblastoma Methylation class infantile hemispheric glioma Methylation class low grade glioma, MYB/MYBL1 Methylation class low grade glioma, desmoplastic infantile astrocytoma / ganglioglioma Methylation class low grade glioma, dysembryoplastic neuroepithelial tumor Methylation class low grade glioma, ganglioglioma Methylation class low grade glioma, rosette forming glioneuronal tumor Methylation class low grade glioma, subclass hemispheric pilocytic astrocytoma and ganglioglioma Methylation class low grade glioma, subclass midline pilocytic astrocytoma Methylation class low grade glioma, subclass posterior fossa pilocytic astrocytoma Methylation class low grade glioma, subependymal giant cell astrocytoma Methylation class lymphoma Methylation class medulloblastoma, WNT Methylation class medulloblastoma, subclass SHH A (children and adult) Methylation class medulloblastoma, subclass SHH B (infant) Methylation class medulloblastoma, subclass group 3 Methylation class medulloblastoma, subclass group 4 Methylation class melanocytoma Methylation class melanoma Methylation class melanotic schwannoma Methylation class meningioma Methylation class papillary tumor of the pineal region group A Methylation class papillary tumor of the pineal region group B Methylation class paraganglioma, spinal non-CIMP Methylation class pineal parenchymal tumor Methylation class pineoblastoma group A / intracranial retinoblastoma Methylation class pineoblastoma group B Methylation class pituicytoma / granular cell tumor / spindle cell oncocytoma Methylation class pituitary adenoma, ACTH Methylation class pituitary adenoma, FSH/LH Methylation class pituitary adenoma, STH densely granulated, group A Methylation class pituitary adenoma, STH densely granulated, group B Methylation class pituitary adenoma, STH sparsely granulated Methylation class pituitary adenoma, TSH Methylation class pituitary adenoma, prolactin Methylation class plasmacytoma Methylation class plexus tumor, subclass adult Methylation class plexus tumor, subclass pediatric A Methylation class plexus tumor, subclass pediatric B Methylation class retinoblastoma Methylation class schwannoma Methylation class solitary fibrous tumor / hemangiopericytoma Methylation class subependymoma, posterior fossa Methylation class subependymoma, spinal Methylation class subependymoma, supratentorial You have not seen the histology yet, btw Typical report from DKFZ Molecular Classifier Capper D et al. Nature 2018 The methylomes A new kind of “histology”

Wong, Ng Modern Pathology, in revision

IDH glioma, subclass 1p/19q codeleted oligodendroglioma GBM_RTK I A_IDH_HG IDH glioma, subclass astrocytoma GBM_MES

A_IDH IDH glioma, subclass high grade astrocytoma glioblastoma, IDH wildtype, subclass RTK I O_IDH glioblastoma, IDH wildtype, subclass RTK II glioblastoma, IDH wildtype, subclass mesenchymal

IDH mutant glioblastomas (our cohort, n=86) GBM_RTK II Figure 1. Unsupervised clustering of reference cohort samples and 85 IDH mutant glioblastomas using t-SNE dimensionality reduction. The reference cohort of the DKFZ CNS tumor classifier includes 82 tumour and 9 non-tumour classes and they are shown as circles of different colors. The 85 primary IDH mutant glioblastomas of our cohort clustered mainly to the (I) IDH mutant high grade astrocytomas; (2) glioblastoma, IDH wildtype, subclass RTK II and (3) subclass mesenchymal (green triangles). Mutations of IDH in our samples were tested and confirmed by independent PCR and sanger sequencing. Diagnosis of a molecular glioblastoma

Capper D, et al. Acta Neuropathologica 2018 Diagnosis is only a chip away Copy number variation of the genome of cohort of primary IDH-mutant glioblastoma made from methylomes

Wong, Ng. Modern Pathology, in revision Other information obtainable from methylomes

• MGMT • G-CIMP status • 1p19q status DKFZ Methylation Classifier

• Desirable criteria in most entities in WHO 2021 • ? Replace histology • 16,945 tumors as of last year • 10-15% undiagnosable • Best for clarifying diagnoses • Good at assigning tumor type. Lesser use for grading. • Does not usually give druggable targets • Does not tell fusions My problem with a Just Chips approach

The single gene tests, eg 1p19a, Braf, are still very important and fast and essential for clinical trials 173 groups on DKFZ ! Medulloblastoma typing – special considerations • Molecular subtypes originally developed by transcriptomes, not NGS • No single method used

Commonly used methods • Nanostring • Methylomes • Integrated NGS panels • IHC + FISH WHO 2016 - medulloblastoma

Molecular groups were developed out of algorithms using expression arrays primarily MB – implications for treatment

• Pediatric Wnt – ? Scaling down of therapy • SHH p53wt – SHH antagonist • Group 3 non-met – Prophylaxis for met, craniospinal RT • Group 3 met +/- c-myc – prioritze for novel therapy; intrathecal CT for met • Group 4 – chrom 11 loss. ?? Scaling down of treatment • Group 4 met – treatment for met component MB

• Adult Wnt : fewer cases of monosomy 6 • Adult Wnt : do worse • N-myc : may mark cases esp in SHH as unfavorable; in groups 3 and 4 less clear • P53 in SHH – poor prognosis and genetic counseling • Grouping does not differ too much geographically • NGS. Sufu – Gorlin; Ptch – SMO inhibitors • NGS. Grouping does not change in recurrence but mutation pattern may change; mutation pattern also differ geographically – implications for NGS • Recurrent tumors – emergence of p53 and c-myc Nanostring method

• Non-PCR transcriptome-based • Not very good with archival materials • Medulloblastoma grouping – can be difficult with MB mimics • Can also be used for detection fusion genes nanoString nCounter System

-a non-enzymatic multiplexed assay that uses sequence-specific probes to digitally measure targeted mRNA transcript abundance -FFPE tissues, a routine diagnostic material (3-4 4um-thickness slides; 100ng RNA). -Up to 800 genes can be detected simultaneously. Two major steps: -Hybridization of mRNA transcript with molecular barcodes. -Immobilize target-probe complex on the surface and qualified barcodes.

Target-probe complex Probes for qualification of gene expression

-Two types of probes are utilized for generation of target-probe complex

Hybridization region Hybridization region Molecular barcode 50 bp, 50 bp, (6 fluorescent spots) gene specific gene specific

Biotin

Reporter probe

Capture probe Each spot can be one of the four colors.

Barcodes Genes Each gene is assigned with unique DKK2 combination of these barcodes NPR3 Count individual barcode which is equal to count individual mRNA molecule. KCNA1 Procedures of NanoString Technologies

Step 1. Hybridization Step 2. Incubation to form target-probe complex

Nucleic acid Reporter probe

Capture probe Probes, RNA, and hybridization buffer Up to 12 samples can be examined in each run.

Step 4: Hybridized probes bind Step 5: Probe alignment and Step 3. Two round, magnetic bead to streptavidin-coated slide. imaging based purification to remove excess unbounded probes

current, fixative, and imagining. Biotin helps to bind to the slide Application of NanoString nCounter System-MB subgroup affiliation

A list of probes in nanoString-based assay for MB subgroup affiliation - 5–6 signature genes are included for each subgroup.. Genes Subgroup DKK2 - 3 housekeeping are included for normalization. EMX2 GAD1 WNT TNC WIF1 ATOH1 EYA1 HHIP SHH PDLIM3 SFRP1 EGFL11 GABRA5 IMPG2 GROUP 3 MAB21L2 NPR3 NRL EOMES KCNA1 KHDRBS2 GROUP 4 OAS1 RBM24 UNC5D ACTB GAPDH Housekeeping LDHA Provided by Taylor MD Example of raw data generated by the nanoString nCounter Technologies

nCounter RCC Collector Worksheet © 2009 NanoString Technologies v1.6.0 -Contains positive & negative controls, 22 subgroup- Import RCC Files... Delete Data specific genes, and 3 housekeeping genes. File Attributes File name 20150305_Betty23_3552_0120150305_Betty23_3555_0220150305_Betty23_3556_0320150305_Betty23_3559_0420150305_Betty23_3560_0520150305_Betty23_3561_06 ID G2314 G2319 G2471 G2587 G2616 G2644 Owner Date // // // // // // File Version 1.6 1.6 1.6 1.6 1.6 1.6 GeneRLF Taylor_1_C2364Taylor_1_C2364Taylor_1_C2364Taylor_1_C2364Taylor_1_C2364Taylor_1_C2364 Comments

Lane Attributes Lane ID 1 2 3 4 5 6 FOV Count 600 600 600 600 600 600 FOV Counted 595 600 600 600 600 600 Scanner ID DA32 DA32 DA32 DA32 DA32 DA32 StagePosition 2 2 2 2 2 2 Binding Density 0.28 0.29 0.32 0.24 0.12 0.26 Messages

Reporter Counts Code Class Name Accession Positive POS_A(128) ERCC_00117.1 22492 27279 23970 26658 25758 25758 Positive POS_B(32) ERCC_00112.1 7336 8484 7488 8638 8136 8133 Positive POS_C(8) ERCC_00002.1 2001 2175 1928 2154 1995 2040 Positive POS_D(2) ERCC_00092.1 506 594 517 562 546 568 Positive hybridization controls Positive POS_E(0.5) ERCC_00035.1 112 115 103 108 111 123 Positive POS_F(0.125) ERCC_00034.1 45 39 46 43 55 53 Negative NEG_A(0) ERCC_00096.1 16 17 11 8 11 12 Negative NEG_B(0) ERCC_00041.1 5 2 7 5 3 2 Negative NEG_C(0) ERCC_00019.1 2 5 9 2 5 1 Negative NEG_D(0) ERCC_00076.1 2 5 3 6 2 2 Negative NEG_E(0) ERCC_00098.1 15 11 14 8 5 9 Negative NEG_F(0) ERCC_00126.1 19 17 9 15 12 6 Negative hybridization controls Negative NEG_G(0) ERCC_00144.1 3 2 4 1 0 4 Negative NEG_H(0) ERCC_00154.1 2 5 4 5 1 5 Endogenous ATOH1 NM_005172.1 6 10 20 3 9 70 Endogenous DKK2 NM_014421.2 6673 25 105 22 9 37 Endogenous EGFL11 NM_198283.1 180 1232 484 12523 11 843 Endogenous EMX2 NM_004098.3 5858 40 98 23 6 58 Endogenous EOMES NM_005442.2 12 8805 22002 20 964 27 Endogenous EYA1 NM_172059.2 27 23 30 21 5 854 Endogenous GABRA5 NM_000810.2 119 1041 139 451 26 172 Endogenous GAD1 NM_000817.2 8655 559 235 38 27 273 Endogenous HHIP NM_022475.1 2 20 39 6 2 2622 Endogenous IMPG2 NM_016247.2 6749 932 3332 398 14 35 22 subgroup-specific genes Endogenous KCNA1 NM_000217.2 138 2053 2913 8297 133 25 Endogenous KHDRBS2 NM_152688.2 38 27769 15284 8850 215 7 Endogenous MAB21L2 NM_006439.4 371 26 6514 31 12 56 Endogenous NPR3 NM_000908.2 9 29 44 24 10 91 Endogenous NRL NM_006177.3 81 113 376 115 12 44 Endogenous OAS1 NM_016816.2 41 154 170 299 9 95 Endogenous PDLIM3 NM_014476.3 8 62 27 22 7 45 Endogenous RBM24 NM_153020.2 45 4510 5622 1820 259 97 Endogenous SFRP1 NM_003012.3 45 26 230 293 121 506 Endogenous TNC NM_002160.1 21015 134 82 77 3 131 Endogenous UNC5D NM_080872.2 46 3458 11789 10590 378 14 Endogenous WIF1 NM_007191.2 39758 155 691 17 4 14 Housekeeping ACTB NM_001101.2 18823 36884 44343 24489 2433 28702 Housekeeping GAPDH NM_002046.3 14504 32946 48922 26728 3236 72722 Housekeeping LDHA NM_005566.1 8410 18581 16919 5380 806 7527 3 housekeeping genes Example of output for subgroup prediction and normalized data created by R script

Supplementary Table 4: Examples of molecular subgroup prediction Samples G2314 G2319 G2471 G2587 G2616 G2644 Confidence 0.999999949 0.99706154 0.998279143 0.990795936 0.999408376 0.999870941 Generally, the confidence needs to be >0.9 to be trust.

Subgroup WNT Group4 Group4 Group4 Group4 SHH DKK2 13.27 4.106 5.943 4.815 6.564 4.846 EMX2 13.08 4.784 5.843 4.879 5.979 5.494 WNT GAD1 13.64 8.589 7.105 5.604 8.149 7.729 TNC 14.92 6.528 5.586 6.622 4.979 6.67 WIF1 15.84 6.738 8.661 4.443 5.394 3.444 ATOH1 3.146 2.784 3.55 1.941 6.564 5.766 SHH EYA1 5.316 3.986 4.135 4.748 5.716 9.375 HHIP 1.561 3.784 4.514 2.941 4.394 10.99 PDLIM3 3.561 5.416 3.983 4.815 6.202 5.128 SFRP1 6.053 4.162 7.074 8.55 10.31 8.619 EGFL11 8.053 9.729 8.147 13.97 6.854 9.356 GABRA5 7.456 9.486 6.347 9.173 8.095 7.063

Group 3 IMPG2 13.28 9.326 10.93 8.992 7.202 4.766 MAB21L2 9.097 4.162 11.9 5.31 6.979 5.444 NPR3 3.731 4.32 4.688 4.941 6.716 6.144 NRL 6.901 6.282 7.783 7.201 6.979 5.096 EOMES 4.146 12.57 13.65 4.678 13.31 4.391 Group 4 KCNA1 7.67 10.47 10.74 13.37 10.45 4.28 R-script KHDRBS2 5.809 14.22 13.13 13.47 11.14 2.444 Courtesy of Michael Taylor OAS1 5.919 6.729 6.638 8.58 6.564 6.206 RBM24 6.053 11.6 11.69 11.19 11.41 6.236 UNC5D 6.085 11.22 12.75 13.73 11.96 3.444 Advantage and limitation of molecular subgrouping by NanoString

Advantages: -Fast turnaround time: 2 days -Little tissue: 1-3 slides Limitations: -Highly reproducible compared to IHC (intra- and inter- observer variability; antibody lot-to-lot variation). -It can subgroup any sample regardless -Inexpensive diagnosis, but subgroup prediction is reliable when samples are true medulloblastoma. -<10% of samples, we cannot confidently assign -archival cases can be difficult - Older tissues not reliable Application of nanostring to test for fusion genes

2. detection of reported fusion transcripts and duplication events in PLGG 9 KIAA1549-BRAF fusion variants

KIAA1549-BRAF fusion transcripts Others fusion transcripts KIAA1549(exon 13)-BRAF(exon 9) BRAF(exon 7)-MACF1(exon 19) KIAA1549(exon 15)-BRAF(exon 10) CLCN6(exon 2)-BRAF(exon 11) KIAA1549(exon 15)-BRAF(exon 11) ETV6(exon 1)-NTRK3(exon 18) KIAA1549(exon 15)-BRAF(exon 9) FAM131B(exon 1)-BRAF(exon 10) KIAA1549(exon 16)-BRAF(exon 10del74) FAM131B(exon 2)-BRAF(exon 9) KIAA1549(exon 16)-BRAF(exon 11) FAM131B(exon 3)-BRAF(exon 9) KIAA1549(exon 16)-BRAF(exon 9) FGFR1(exon 17)-TACC1(exon 7) KIAA1549(exon 18)-BRAF(exon 10) FGFR3(exon 17)-TACC3(exon 4) KIAA1549(exon 19)-BRAF(exon 9) FXR1(exon 13)-BRAF(exon 10) Duplication/deletion transcripts GNAI1(exon 1)-BRAF(exon 10) FGFR1 tandam duplication MKRN1(exon 4)-BRAF(exon 11) MYBL1 deletion/tandem duplication MYB(exon 6)-MAML2(exon 4) Housekeeping transcripts MYB(exon 9)-PCDHGA1(exon 2) ABCF1 (NM_001090.2) NACC2(exon 4)-NTRK2(exon 13) ALAS1 (NM_000688.4) NTRK3(exon 10)-ETV6(exon 5) CLTC (NM_004859.2) QKI(exon 1)-RAF1(exon 14) HPRT1 (NM_000194.1) QKI(exon 2)-MYB(exon 16) QKI(exon 6)-NTRK2(exon 16) RNF130(exon 3)-BRAF(exon 9) SRGAP3(exon 11)-RAF1(exon 8) SRGAP3(exon 12)-RAF1(exon 10) 9 other BRAF fusion variants ST6GAL1(exon 2)-WHSC1(exon 4)

Courtesy Cynthia Hawkins Medulloblastoma at CUHK

• Nanostring for typing using Mike Taylor’s R script • P53 IHC if SHH • C-myc, N-myc FISH Medulloblastoma without molecular typing set up • Do It Yourself – methylation array and upload DFKZ website • IHC : Gab1, Yap1 • c-myc, MYCN • Monosomy 6 or beta catenin exon 6 sequencing • FISH for 9q • Isochrome 17q • Chromosome 11

Less commonly used genomics for brain tumor diagnostics

• Whole Exome Sequencing (WES) • Whole Genome Sequencing (WGS) • SNP arrays / CGH arrays • Liquid biopsies (single gene, methylation, extrachromosomal vesicles) No longer just an ancillary test for precision Medicine – It is part of the definition in some instances F Stephen Vogel Robin O Peter Burger Barnard

S K Yee Foundation