Outline enhanced diagnostics and potential treatment strategies - - PowerPoint PPT Presentation

5/25/2018 1

Pathognomonic mutations in rare gynecologic cancers: from discovery to enhanced diagnostics and potential treatment strategies

David G. Huntsman

BC Cancer Agency Vancouver General Hospital University of British Columbia

Outline

• To understand rare cancers as histo-

molecular entities

• Pathognomonic mutations: where

diagnostic is prognostic and predictive

• FOXL2, DICER, SMARCA4 and other

examples

Almost every type of ovarian cancer is a rare disease

Research into rare cancers joys and challenges

• Direct benefit to those

who have them

• Forme Fruste models

to understand common cancers

• Satisfaction of rapid

knowledge translation

• Limited shots on goal

need for better pre- clinical research

5/25/2018 2 ETV3-NTRK3 fusions : the poster child for generalizability of discoveries from rare cancers

• Fusion discovered in extremely rare sarcoma

(1998)

• Fusion found in vanishingly rare breast cancer

subtype

• Fusion found in rare kidney cancer
• Fusion found rarely in AML and other common

cancers

Congenital fibrosarcoma Celluar mesoblastic nephroma Secretory carcinoma Acute myeloid leukaemia

Drilon et al NEJM 2018

Histomolecular entity versus molecular subtypes

• f a common cancer: the mountain analogy
• Histomolecular entity
• Molecular subtypes of

common cancer

Single pathogenetic pathway and dominant treatment Multiple pathogenetic pathways and treatment

• ptions

What is the value of diagnosis in a post genomic era

• A useful diagnosis

must be

– Clinically relevant – Biologically plausible – Reproducible

• A histologic or

genomic feature alone does not make a diagnosis

Diagnostic gravity Need for other biomarkers

5/25/2018 3

What about diagnostic biomarkers

US National Research Council, Nov 2nd 2011

5/25/2018 4 FOXL2 mutation 4 granulosa cell tumors

• f the ovary

Shah et al (2009) NEJM

V O A 4 9 V O A 1 5 9 V O A 1 1 9 V O A 2 9 2

5/25/2018 5 FOXL2 IHC and mutational analysis as a standard diagnostic (Kommoss et al Mod Path2013)

Confirmation of FOXL2 aGCT specific c.402C>G mutation by Sanger Sequencing

TaqMan based digital mutation assay for FOXL2 aGCT specific c.402C>G mutation

18

Missense Mutations Truncating Mutations *No mutations at C134W

Cross-cancer mutation summary for FOXL2 (147 studies ) None of the cancers included in c-bioportal had the C134W mutation as none were GCT Question : what does a cancer specific mutation mean to patients What does getting the correct diagnosis mean to patients Anniina Färkkilä Mikko Anttonen Markku Heikinheimo Leila Unkila-Kallio Hugo Horlings Hannah van Meurs Maaike Bleeker Melissa McConechy Winnie Yang Blake Gilks Aline Talhouk Stefan Kommoss Sarah Brucker

3 European Centers (n =336) Finland, The Netherlands & Germany

McConechy, Färkkilä, Horlings et al JNCI 2016

5/25/2018 6

• Patient’s disease

relapsed after 2.6 years and died of disease 3 years from Diagnosis

• She had been

misdiagnosed as GCT – personal

• pportunity cost

Overall survival of women with molecularly define AGCT is not distinct from population based controls

5/25/2018 7 Why is the recurrent FOXL2 mutation so specific for adult type GCT

In adults expression is restricted to specialized stroma of

• vary, uterus and fallopian tube

Image from protein atlas

FOXL2 in Ovarian and Mullerian Stroma

Ovary Endometrium Fallopian tube Endocervix

The FOXL2 mutation is diagnostic for aGCT and its oncogenicity linked to granulosa cell pathophysiology

Granulosa cell Theca cell Oocyte

Corpus luteum Follicle

Luteinized granulosa cell FOXL2-mutant granulosa cell

Ovulation

AGCT

Other targets What else is going on in GCT genomes?

• TERT promoter mutations in

1/3

• Smattering of other cancer

mutations including targetable PIK3CA pathway mutations

• LOH of chromosome 22

5/25/2018 8

Molecularly defined AGCT The power of a correct diagnosis

• Prior studies overestimated death from disease due to contaminating

misdiagnosed cases.

• Missed diagnosed cases dominate early relapses and likely past trials

studied in clinical

• Usually an indolent disease or managed well by primary and secondary

surgeries

• Treatment needed for inoperable cases of molecularly defined AGCT
• Medium time to relapse >7 years therefore current follow-up strategy

likely useless – needs testing

• Cell free DNA could be used as an adjunct to monitor patients in trials
• If no means of targeting FOXL2 derived then

analysis for other targetable mutations through inclusion in a basket type trial may be feasible

SCLT Background

• Rare histologicially diverse mixed sex

cord-stromal tumour

• Mostly sporadic
• Occasionally in setting of DICER1

Syndrome

Well differentiated (5) Moderately diff. (34) Poorly diff. (7) Heterologous diff (8*) Retiform (2*) FOXL2 (100% positive)

Gyne Onc 2011

Recurrent DICER1 somatic mutations in

• varian sex cord-stromal tumors

Heravi-Moussavi A. NEJM. 2012

• mutations at four

metal binding sites in RnaseIII b domain.

• High frequency in

Sertoli-Leydig cell tumor (SLCT).

• Often present in

tumors related to DICER1 syndrome (such as PPB, SLCT), a condition caused by germline DICER1 haploinsufficiency.

5/25/2018 9

DICER1 mutations

• In presence of

germline truncating mutations the hotspot mutations are a specific somatic second hit

Moussavi, Anglesio and Cheng et al NEJM 2011

Mutations impact critical aspartic acids in the RNase IIIb domain

Moussavi, Anglesio and Cheng NEJM 2012

Mutations at DICER1 Hotspots lead to global loss of 5p miRNAs

Wang et al. Neoplasia 2015

SLCT miRNA-seq

Anlglesio et al J Path 2012

mES cell model-Nanostring miRNA profiling Mutations at DICER1 Hotspots lead to global loss of 5p miRNAs

Anglesio MS. J Path. 2012

5/25/2018 10

Analyzing miRNA biogenesis defects in

• SLCTs (FFPE) by miRNA sequencing

SLCT (no DICER1 hotspot mutation): 8 SLCT (with DICER1 hotspot mutation): 4

DICER1 status 5p 3p total reads 5p% YW_1 no hotspot 15689 17151 32840 47.775 YW_2 no hotspot 3001 5114 8114 36.979 YW_3 no hotspot 11556 6804 18360 62.940 YW_4 no hotspot 12762 8186 20948 60.922 YW_5 no hotspot 34054 18029 52083 65.385 YW_6 no hotspot 9971 3415 13386 74.487 YW_7 no hotspot 12391 6056 18447 67.173 YW_8 D1709N 2327 7155 9483 24.544 YW_9 D1810H 1400 5750 7150 19.582 YW_10 D1810Y 2339 5734 8073 28.968 YW_11 D1810Y 915 4942 5858 15.623 YW_12 no hotspot 79437 23398 102835 77.247

P < 0.001 Wang et al Neoplasia 2015

DICER1 hotspot mutations in SLCT

• Kato et al Human Path 2017 -6 of 10 –association

with androgenic effects

• De Kock – AJSP 2017 all of 30 mod or high grade

cases had hotspot mutations

• Conlon – Mod Path 2015 20/32 cases had hotspot

mutation no associations

• Larger studies with consistent mutation detection

approaches will be required to determine the frequency of these mutations in SLCT and other tumors and whether there is diagnostic utility

Mutation Distribution by Patient Age

• 45 cases reviewd by

Gilks, Kommoss and Clement

• Tested for Dicer1 &

Foxl2 hotspot mutations and FOXL2 by IHC

A Karnezis et al, submitted

10 20 30 40 50 60 70 80 90 100 DICER1 FOXL2 WT/WT Age P < 0.0001 P = 0.005 P = 0.0003

41% 22% 37%

Do SLCT represent three diseases

• DICER1 mutant (younger, more androgenic

symptoms, moderately/poorly differentiated, retiform or heterologous elements),

• FOXL2 mutant (postmenopausal, abnormal

bleeding, moderately/poorly differentiated, no retiform or heterologous elements)

• DICER1/FOXL2 wild type (intermediate age, no

retiform or heterologous elements, including all well differentiated tumors).

A Karnezis et al, submitted

5/25/2018 11

How do miRNA processing defects lead to cancer in other contexts

A proposed model for DICER1 mutation in

• varian SLCT development: way too complex –

animal models will be required to sort this out

• Described by Scully

1979

• A rare disease in

young women, mean age 22

• Sometimes familial
• Undifferentiated

small cells

• Elevated serum

calcium level (60%)

• Very lethal 35% 2yr

survival

Small cell carcinoma of the ovary, hypercalcemic type (SCCOHT)

SMARCA4 inactivating mutation- the only recurrent somatic mutation in SCCOHT

Missense mutation 1 1,647 a.a. Germline Somatic Splice site mutation Nonsense mutation QLQ Bromo HELICc DEXDc SNF2_N BRK HSA SMARCA4 Kupryjanczyk et al. Polish J Pathol 2013. Ramos, Karnezis et al. Nature Genetics 2014. Witkowski et al. Nature Genetics 2014. Jelinic et al. Nature Genetics 2014 Ramos et al. Rare diseases 2014 20 40 60 80 100 24 48 72 96 120 144 168 % cell confluence Time (hours)

pLDpuro-GFP pLDpuro-SMARCA4

5/25/2018 12 SCCOHT can be familial: associated with germline SMARCA4 mutations

Witkowski, Nature Genetics 2014

What is the significance?

• Diagnostic value?
• Understand the biology, such as cell of origin?
• Therapeutic indication?

BRG1 LOSS

• All cases:83/91 tumours

(91%)

• 15/2324 other ovarian

cancers including 15/360 CCC (4%)

• Karnezis and Ramos et

al J Path 2015

SWI/SNF ATP-Dependent Chromatin Remodeling Complex

Figure modified from Riccio, Nat Neurosci 2010

BAF47/INI1 ARID1A

SCCO-012 SCCO-002 SCCO-014 SCCO-015 PDX-040 PDX-065 BIN67 SCCOHT1 ACTB BCL11B DPF3 SMARCA2 PHF10 SMARCD2 DPF2 SMARCA4 BCL11A BRD9 BRD7 ACTL6A ARID1A SMARCB1 ARID1B PBRM1 SMARCC1 BCL7B DPF1 SMARCD1 SMARCE1 ACTL6B SS18 SMARCC2 SS18L1 ARID2 SMARCD3

3 2 1

• 3
• 2
• 1

Log2 ra o

5/25/2018 13

SMARCA4/SMARCA2 dual deficiency in SCCOHT

1 2 3 4 Relative mRNA level

SMARCA2 SMARCA4

SMARCA4 SMARCA2 Vinculin BIN67 SCCOHT1 SVOG3e KGN

B

Karnezis, Wang, Ramos et al. J Path 2015 Jelinic et al, Mod Pathol. 2016 Jan; 2991):60-66

SWI/SNF-deficiency in rare cancers

site mutation Age of diagnosis SMARCA2 expression

AT/RT CNS, multiple location SMARCB1 SMARCA4 (rarely) <3y loss? Extra-cranial MRT Kidney, soft tissue, liver SMARCB1 SMARCA4 (rarely) Average ~15 months (80%<2y) loss in 70% tumors and 10/11 cell lines SCCOHT

• vary

SMARCA4 SMARCB1 (rarely) mid-20s (1-43y) loss Renal medullary carcinomas (RMC) Kidney SMARCB1 Epithelioid sarcoma (ES) Soft tissue SMARCB1 Young adults Synovial sarcoma (SS) Soft tissue SS18-SSX fusion SMARCA4- deficient thoracic sarcoma (SMARCA4-DTS) Mediastinum, pleura lung SMARCA4 p53 Median 35 (30- 70y); 59 (44-76) loss

Le Loarer et al. Nat Gen 2015

Is SCCOHT a rhabdoid tumor of the ovary

Fahminiya et al. Oncotarget 2016

H&E

Rhabdoid-like cells in SCCOHT

• Shared clinicopathological features
• Common genetic features (mutation in SWI/SNF

complex)

Similarity between SCCOHT and ATRT, MRT

5/25/2018 14 Small cells Large cells – rhabdoid Useful to Conceptualize SCCOHT as a Malignant Rhabdoid Tumour of the Ovary (MRTO)? Yes

• Similar morphology

– Tumours with variable numbers of small, large and rhabdoid cells

• Similar genetics

– Mutations & silencing of core SWI/SNF members

• Similar treatments?

– TBD

Should We Rename SCCOHT as MRTO? No

• SCCOHT is a much better name than MRTO based on

clinicopathologic criteria

– Half of tumors composed exclusively of small cells – Large cells are minority of cells, if present – Rhabdoid cells are usually minority of large cells, if present – Most patients have hypercalcemia

• Neither name suggests cell of origin
• Similar but distinct genetics
• Little evidence that SCCOHT and AT/RT or other

rhabdoid tumours respond to similar therapies

• A more useful/effective nomenclature is needed for

this group of tumours that encapsulates cell context, genetics, histology and treatment similarities

Small cell hypercalcemic ovarian cancer -goals

• 1- to reverse engineer the cell of origin
• 2- to provide the clinical trials community

with a well tested and biologically rational treatment option to test in future patients

5/25/2018 15

Polycomb repressive complex 2 (PRC2)

SUZ12 EZH2 Ciferri et al. Elife 2012 RbAp48 AEBP2

EED

• EZH2: catalytic subunit
• EED: a scaffolding protein

supporting a catalytically competent conformation of EZH2 and stimulate the catalytic methyltransferase activity of PRC2

• Dependency of SWI/SNF deficient-

driven tumors on PRC2 for

• ncogenic transformation

SCCOHT cells are sensitive to EZH2 suppression A C

Strong staining of EZH2

2 4 6 8 10 12 EZH2 EZH1 EED SUZ12

Relative gene expression Normal ovary SCCOHT

*

B

NOY1 SVOG3e OVTOKO JHOC5 G401 BIN67 SCCOHT1 COV434 OVCAR8 ES-2 OVISE EZH2 H3K27me3 19/24 (79%)

5/24 (21%)

Actin

D

SMARCA4 - + - + - BIN67 SCCOHT1 H3K27Me3 Vinculi n EZH2 SMARCA 4 Total H3 COV434

E

0.2 0.4 0.6 0.8 1 1.2 1.4

Relative EZH2 mRNA level GFP SMARCA 4

* ** ** Variable staining of EZH2

Wang et al J Path 2017 Similar results in Chan-Penebre E et al Mol Can Ther 2017

SCCOHT cells are sensitive to EZH2 catalytic inhibition

SCCOHT lines Other ovarian lines SCCOHT lines Other ovarian lines

*** **

EPZ-6438

IC50 (μM)

9-day drug survival assay

20 40 60 80 100 120 140

0.01 0.1 1 10 % Survival EPZ-6438 (μM)

BIN67 SCCOHT1 COV434 JHOC5 OVISE OVTOKO RMG1 20 40 60 80 100 120 140

0.01 0.1 1 10 % Survival GSK-126 (μM)

Wang et al J Path 2017

200 400 600 800 1000

10 15 20 25 30 35 40 Tumor volume (mm3) Days post inoculation BIN67 xenograft model Vehicle 150 mg/kg GSK126

Treatment started (QD)

** *** ** *** *** *** ***

GSK126 0.1 0.2 0.3 0.4 0.5 Vehicle GSK126 final tumor weight (g)

***

BIN67 xenograft model

A B

Model for SCCOHT

EZH2/EED/HDAC inhibitors

Progenitor cells Neuronal differentiation

HDAC

X X

Stalled differentiation and

• ncogenic transformation

X

HDAC

Smarca2 negative

5/25/2018 16

Myc in SCCOHT

1 2 3 4 5 6 7 8 9 10 Negative Scant (<1%) and faint 1-50%, definite >50% Cases

Myc expression in SCCOHT primary tumor

• Depletion of Myc associated with decrease the viability of SCCOHT cells.
• Whether Myc status affects the efficacy of PRC2 inhibitors remains to be

determined

• Our SCCOHT cell lines express high levels of Myc but SCCOHT don’t !

Myc

Myc BIN67 SCCOHT-1 EPZ-6438 0 3 7 10 0 3 7 10 0 4 7 days COV434 Actin

Proteomic analysis identified myc suppression by EZH2 inhibitor

• Four anecdotal response to anti-PD1

immune modulation therapy

• 8/11 cases had PD-L1 expression and

robust T-cell infiltration

• Immunogenic microenvironment

resembles that of tumours that respond to PD-1/PD-L1 blockade

JNCI -2018

Rare ovarian cancers: The Good, the Bad and the Ugly

• Good – new diagnostics

and better understanding of SCCOHT, SLCT and AGCT

• Dearth of model systems to

develop best possible new treatments – but PDX’s for SCCOHT

• Clinical trials difficult to

develop, fund and implement (limited shots on goal thus as a community we must put forward the best options )

UGLY

Cell context and mutation: The evil twins of cancer

Partners in crime from the origin of cancers through to the development of acquired resistance to targeted therapy

5/25/2018 17

More stories of cell context: Synthetic lethality in ARID1A- deficient tumors

ARID1A- deficient OCCC YES1 (src tyrosine kinase) inhibition PI3K/Akt inhibition EZH2 inhibition ARID1B inhibition PARP inhibition ATR inhibition (Bitler et al. Nat Med. 2015) (Helming et al. Nat Med; Kelso et al. Elife 2017) (Samartzis et al. Oncotarget 2014) (Miller et al. MCR2016)

(Williamson et al. Nat

• Com. 2016)

(Shen et al. Cancer Discovery 2015) HDAC6 inhibition (Bitler et al. NCB 2017) Epigenetic remodeling signaling Cell survival pathways DNA damage response pathways

Example: Dedifferentiated endometrial carcinomas

Undifferentiated Differentiated

Case number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 Sequencing results SMARCA4 fs x 2 fs + R3 97 fs fs x 2 fs + fs fs + fs x 2 fs x 2 fs x 2 fs fs x 2 fs fs x 2 SMARCB1 * R4 0X fs SMARCA2 ARID1A fs fs x 2 W 14 fs x 2 S2 64 fs Q1 82 fs fs x 2 Q1 17 fs x 2 fs + fs + Q5 43 fs + fs x 2 fs x 3 fs Q5 43 fs fs x 2 * Q1 09 * E1 71 R1 98 R7 50 fs ARID1B fs x 2 Q1 34 fs + fs + fs fs fs fs + fs + * * R7 50 ARID2 PBRM1 E9 13 SMARCC1 SMARCC2 fs PHF10 Protein (IHC) ARID1A Endometrioid +

• +

+

• +
• +
• +

+ + + + +

• +

+

• +

+ +

• +

+ + + + Undifferentiated +

• +

+

• +
• +
• +

+ + + +

• +

+ +

• +

+ + + + ARID1B Endometrioid + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Undifferentiated + + + + + + + + + + + + + + + + + + +

• +

+ + + + + + + + + + +

BRG1-deficient ARID1A/ARID1B co-inactivated INI1- deficient BRG1/INI1-intact

70% of DDEC shows SWI/SNF inactivation

• 18 of 40 (45%) BRG1/INI1 inactivation
• 10 of 40 (25%) ARID1A/ARID1B inactivation

Two-thirds of SWI/SNF-def DDEC are MMR- deficient/MSI-H

Karnezis AN et al, Mod Pathol. 2016 Mar;29(3):302-14 Coathamn et al, Mod Pathol. 2016 Dec;29(12):1586-1593

ARID1A

ARID1A

5/25/2018 18

ARID1B

Conclusions

• Whilst the targetability of some mutations

(BRAF, ARID1a) are highly cell context dependent and other mutations (TRK fusions) are not

• DICER1 mutations mark a distinct subtype
• f SLCT that occur in younger patients
• A correct diagnosis of cancer type is still

important –molecular features can help (FOXL2, BRG1 etc)

Thanks

• My lab: Dawn Cochrane, Niki Boyd, Michelle Woo, Yemin Wang, Winnie

Yang, Clara Salamanca, Tony Karnezis, Basile Cloutier-Tessier, Yang Xia, Jenn Ji, Shary Chen, Jessica Pilsworth, Kendall Greening, Kate Dixon, Vivian Lac, Germain Ho, Katelyn Onderwater, Evan Gibbard, Amy Lum, Genny Trigo-Gonzalez, Julie Ho, Jamie Lim, Mehrane Nazeran, Forouh Kalantari and Janine Senz

• Sohrab Shah -- Bioinformatics: Jairhu Ding, Yikan Wang, Ali Bashashati,

Gavin Ha, Andrew McPherson

• GSC: Marco Marra, Martin Hirst, Gregg Morin
• Collaborators: Stefan Kommoss, Martin Kobel, James Brenton, Anne-Marie

Mes-Masson, David Bowtell, Barbara Vanderhyden, Aikou Okamoto, Buddy Weissman, Jeff Trent, and Sam Aparicio

• OvCaRe BC: Blake Gilks, Dianne Miller, Paul Hoskins, Nelly Auersperg,

Brad Nelson, Cal Roskelly, Anna Tinker, Mike Anglesio, Paul Yong, Jessica McAlpine, Aline Talhouk, Gillian Hanley, Mark Carey