“At The Leading Edge of Precision Medicine-Sorting
Through a Complicated Landscape”
Phil Stephens; Foundation Medicine Inc
©2014 Foundation Medicine, Inc. | Confidential 2
Molecularly Targeted Therapy Is Rapidly Evolving
©2014 Foundation Medicine, Inc. | Confidential 3
Genomics is Identifying Novel Cancer Genes at an Unprecedented Rate
2003 2016
Increasing Options for NSCLC Molecular Management
Modified and updated from Pao, W, Hutchinson KE. Chipping away at the lung cancer genome. Nat Med 2012 Mar 6;18(3):349-51.
©2014 Foundation Medicine, Inc. | Confidential 4
From Test (2/2012)… … To Treatment (3/2013)
Oncogenic NSCLC RET Fusions: Discovery to Clinic In ~1 Year
Identification of new ALK and RET gene fusions from colorectal and lung cancer biopsies
Lipson D, Capelletti M, et al
RET, ROS1 and ALK fusions in lung cancer
Takeuchi K, Soda M,, et al
KIF5B-RET fusions in lung adenocarcinoma
Kohno T, Ichikawa H, et al
©2014 Foundation Medicine, Inc. | Confidential 5
FoundationOne® For Solid Tumors
• One comprehensive genomic profile for all solid tumors
• Interrogates entire coding sequence of 315 cancer-related genes
• Requires small amounts of tissue routine FFPE samples
including needle biopsies and FNAs
• Validated high accuracy
• 11 day median turn around time from receipt of block
• Medicare approved for NSCLC; Colorectal ca, Melanoma and
AML imminent
• FDA approval Q4 this year with broad Medicare approval
©2014 Foundation Medicine, Inc. | Confidential 6
Analytic Validation Demonstration of high accuracy and reproducibility required
for clinical use
Base Substitutions (MAF 5-100%)
Sensitivity: >99.9% PPV: >99%
Insertions/Deletions (1-40bp, MAF 10-100%)
Sensitivity: 98% PPV: >99%
Copy Number Alterations (zero or ≥8 copies)
Sensitivity: >95% PPV: >99%
Gene Fusions
Sensitivity: >95% (>99% for ALK fusion1) PPV: >99%
Frampton et al, Nature Biotechnology 2013
1. Yelensky et al, Presented at AACR 2014.
©2014 Foundation Medicine, Inc. | Confidential 7
¼ EGFR mutations, 35% of ALK & ROS1 fusions missed by single gene testing
CGP detects clinically relevant alterations in 65% of
patients with pan-negative NSCLC tested at MSKCC
MSKCC testing: EGFR, ERBB2, KRAS, NRAS, BRAF, MAP2K1, PIK3CA, AKT1, ALK,
ROS1 & RET
● 65% of patients harbored a clinically relevant alteration by FoundationOne
26% of patients: targeted therapy in NCCN guidelines
39% of patients: Approved therapy or active MSKCC clinical trial
Clinical Cancer Research Dec 2014: Broad, hybrid capture-based next-generation sequencing
identifies actionable genomic alterations in "driver-negative" lung adenocarcinomas. 158/1078-
0432.CCR-14-2683
©2014 Foundation Medicine, Inc. | Confidential 8
• KIF5B-RET expression in Ba/F3 cells
led to oncogenic transformation
• KIF5B-RET transformed cells are
sensitive to RET inhibitors
Data suggest RET kinase inhibitors should be tested in prospective
trials in NSCLC patients with RET fusions
KIF5B (exons 1-15) RET (exons 12-20)
Kinesin Coiled coil Tyrosine kinase
KIF5B-RET
Collaboration with Pasi Janne, Dana
Farber
• Sequencing cohort of ~600 NSCLC
patients revealed frequency of ~1% in
Caucasians and 6.3% in Asian without
known driver mutations
Translation
Novel KIF5B-RET Fusion Identified In First Clinical Case
©2014 Foundation Medicine, Inc. | Confidential 9
Case Presentation 1: NSCLC Patient
• Middle aged female never smoker
• Stage IV-B lung adenocarcinoma
• Sequenom negative
• ROS1 (FISH) & KIF5B-RET (PCR) testing negative
• Pemetrexed (Alimta), cisplatin and bevacizumab (Avastin) started x 6
cycles then pemetrexed and bevacizumab maintenance every 3 weeks
• 18 months after diagnosis progression of disease with new bony
metastasis
©2014 Foundation Medicine, Inc. | Confidential 10
Repeat CT scan after 28 days of anti-RET therapy:
disappearance of paramediastinal and near
complete resolution of pleural disease.
Baseline CT scan showing paramediastinal and
pleural-based nodularities in left upper lobe.
Case Presentation 1: NSCLC Patient
Multiple patients with RET fusions are known to have responded to a RET inhibitor
©2014 Foundation Medicine, Inc. | Confidential 11
0
5
10
15
20
25
30
35
% t
um
ors
wit
hin
su
bty
pe
HER2/ERBB2 Amplifications are standard of care
©2014 Foundation Medicine, Inc. | Confidential 12
0
5
10
15
20
25
30
35
% t
um
ors
wit
hin
su
bty
pe
HER2/ERBB2 Alterations Across 27 Tumor Types
©2014 Foundation Medicine, Inc. | Confidential 13
0
5
10
15
20
25
30
35
% t
um
ors
wit
hin
su
bty
pe
rearrangements insertions/deletions
base substitutions amplifications
Four times the number of patients harbor druggable
HER2 alterations than typically tested for
Multiple case reports of durable patient responses
HER2/ERBB2 Alterations Across 27 Tumor Types
©2014 Foundation Medicine, Inc. | Confidential 14
EGFR alterations were identified across 13 different tumor
types
6.8% total cases
©2014 Foundation Medicine, Inc. | Confidential 15
53 year old female diagnosed with metastatic Inflammatory Breast Cancer
involving liver and bone in June, 2010
EGFR Exon 21 L858R Point Mutation identified
Present in 10% of lung adenocarcinomas
Associated with unprecedented sensitivity to EGFR-TKIs such as gefitinib
(Iressa) and erlotinib (Tarceva)
Diagnosis confirmed as Breast Cancer (not a lung metastasis)
Case Presentation 2: Identification of EGFR
mutation in an inflammatory breast cancer patient
©2014 Foundation Medicine, Inc. | Confidential 16
• EGFR mutation identified would never have been tested for in a patient with
breast cancer
• Genomic profile led to therapy with FDA-approved agent (off label)
• Erlotinib commenced resulting in symptomatic and radiographic
improvement: response ongoing greater than 18 months
September, 2012 November, 2012
Case Presentation 2: Identification of EGFR
mutation in an inflammatory breast cancer patient
©2013 Foundation Medicine, Inc. | Confidential 17
Cancer Immunotherapies (CIT) are Transformative For Some Patients
Therapy Target Approved
indications
Ipilimumab
(BMS)
CTLA-4 Melanoma
Nivolumab
(BMS)
PD-1 Melanoma,
NSCLC,
RCC, HL
Pembrolizumab
(Merck)
PD-1 Melanoma,
NSCLC,
Head and
Neck
Atezolizumab
(Roche/GNE)
PD-L1 Bladder ca,
NSCLC
Avelumab
(Pfizer)
PD-L1 Merkel cell
Additional checkpoint inhibitors at various points across the cycle are in development
Picture adapted from Mellman et al. Nature, 2011
Anti-CTLA4
Anti-PD1/PD-L1 Pembrolizumab
(Merck)
PD-1 MSI
UNSTABLE
©2013 Foundation Medicine, Inc. | Confidential 18
Despite Growing Success, Many Patients Will Not Respond to CIT
Response rates for the 5 currently approved checkpoint inhibitors typically range
from ~10-30% for solid tumors
Biomarkers that potentially modulate response to cancer immunotherapy
PD1/PD-L1 expression
Tumor infiltrating lymphocytes
Tumor Mutation Burden (TMB) / neoantigen load
Tetrapeptide neoepitopes
Loss of function JAK1/2 alterations (interferon unresponsiveness)
Loss of function B2M alterations (MCH/I presentation)
©2013 Foundation Medicine, Inc. | Confidential 19
TMB Correlates with CIT Response in Several Tumor Types
Snyder et al., NEJM, 2014
High TMB Melanoma
MSI-High Colorectal Cancer
Le et al., NEJM, 2015
Van Allen et al., Science, 2015
Rizvi et al., Science, 2015
High TMB NSCLC
High TMB Melanoma
20
TMB Correlates with Whole Exome Sequencing (WES) vs. 1.1 Mb assay
High correlation between mutations per Mb in WES vs. FOne (1.1 Mb assay)
~1
Mb
as
say:
Mu
tati
on
s/M
b
WES Mutations/Mb
R2=0.976
R² = 0.881
1.0
10.0
100.0
1000.0
0.1 1.0 10.0 100.0 1000.0
F1 T
MB
(m
ut/
Mb
) WES TMB (mut/Mb)
High
Intermediate
Low
Status concordance: >90%
Correlation at low TMB breaks down but can distinguish low TMB from intermediate and high
©2013 Foundation Medicine, Inc. | Confidential 21
Melanoma (n = 65) NSCLC (n = 464) Urothelial carcinoma (n = 94)
M Kowanetz et al, Annals of Oncology, 2017 Balar AV et al. Lancet, 2017
CGP derived TMB Predicts CIT Response in three Indications
DB Johnson et al., Cancer Immunol Research, 2017
PFS
OS
Time (days)
©2013 Foundation Medicine, Inc. | Confidential 22
High TMB may Predict CIT Response Across Most Tumor Types
Razelle Kurzrock’s group: 63 patients from 19 tumor types (excluding NSCLC and melanoma) demonstrated
that HIGH TMB was independently associated with better outcome to CIT (multivariable analysis).
RR for patients with high TMB 58% vs. low to intermediate TMB 20%; (P = 0.0001)
Goodman A et al. Manuscript submitted
• Distribution box plots ranks indications according to mutations/Mb
• Red stars indicate approved indications, blue stars indicate likely approvals
Mu
tati
on
s/M
b
Distribution of Mutational Burden Across All Indications at FMI (n = >100,000)
©2013 Foundation Medicine, Inc. | Confidential 23
Why is TMB not as Predictive of CIT Response in NSCLC?
M Kowanetz et al, Annals of Oncology, 2017
Progression Free Survival Overall Survival
1619 NSCLCs with genomic profiling and PD-L1 IHC
Correlation between TMB and PD-L1 IHC expression low
Are there NSCLC specific genomic modulators of response to CIT?
©2013 Foundation Medicine, Inc. | Confidential 24
STK11 Alterations may be Immunosuppressive in NSCLC
STK11 alterations enriched in
TMB HIGH, PD-L1 LOW tumors
STK11 mutant NSCLCs may do
worse on immunotherapies
HR = 2.59; P = 0.0314 Bonferroni P = 3.23*10-12
While provocative, this observation requires validation in additional cohorts
©2013 Foundation Medicine, Inc. | Confidential 25
BRAF and c-MET Alterations may be Immunogenic in NSCLC
BRAF alterations enriched in TMB LOW,
PD-L1 HIGH tumors c-MET alterations enriched in TMB LOW,
PD-L1 HIGH tumors
These observations require validation in additional cohorts
P = 1.43 x 10-4
P = 4.47 x 10-4
©2013 Foundation Medicine, Inc. | Confidential 26
MDM2 Amplification is Associated with CIT Hyper-Progression
6/13/2016
8/18/2016
Atezolizumab 8/18/2016
Kato et al. Hyper-progressors after Immunotherapy. Analysis of Genomic Alterations Associated with Accelerated Growth Rate. Clin Cancer Research. 2017. Mar 29.
©2013 Foundation Medicine, Inc. | Confidential 27
The majority of BRCAwt patient tumors exhibit LOH
signature (“BRCA-like”)
Tumor BRCA/LOH Status as Determined by HRD Test (N=121, ARIEL2 Pt1)
Source: Swisher, et al. SGO 2015 presentation of clinical efficacy and safety data
BRCAmut
25%
BRCA-like 42%
Biomarker negative
33%
• 17 germline BRCAmut
• 12 somatic BRCAmut
• 1 indeterminate
• High genomic LOH
©2013 Foundation Medicine, Inc. | Confidential 29
ARIEL2 part 1 designed to assess rucaparib efficacy in
three prospectively defined HRD subgroups
©2013 Foundation Medicine, Inc. | Confidential 30
Duration of response in BRCAmut, BRCA-like, and
Biomarker Negative patients
Negative patients
©2013 Foundation Medicine, Inc. | Confidential 31
• Challenges of ctDNA based genomic profiling
©2013 Foundation Medicine, Inc. | Confidential 32
Blood Based Tumor Detection is Extremely Challenging
• Clinical requirement for a highly sensitive and
specific assay
Tumor content in blood may be less than 1% of
total cell free DNA in plasma
TUMOR
TISSUE
Circulating tumor DNA
released from tumor tissue
HEALTHY
TISSUE
Cell free circulating
DNA released from
healthy tissue
©2013 Foundation Medicine, Inc. | Confidential 33
ctDNA Assays Are Not Appropriate for Many Cancer Patients
Up to 40% of patients will be denied standard of care treatment using
ctDNA assays alone as they shed no detectable ctDNA into bloodstream
% Patients with at least one cancer-related
alteration
0%
20%
40%
60%
80%
100%
Kurzrock, et al (n=171) Schwaederle, et al(n=439)
FoundationCOREDatabase (n=21,155)
ctDNA
58%
FOne
98%
FOne
96%
©2013 Foundation Medicine, Inc. | Confidential 34
Recent study: Comparison ctDNA and tissue for ER+ breast
cancer
Temporally matched ctDNA-tissue samples
• 94% of short variant mutations found in tissue were also found in
ctDNA
• 2 mutations in ESR1, were observed in ctDNA only
• High blood vs tissue in NSCLC, colon and melanoma
©2013 Foundation Medicine, Inc. | Confidential 35
Recent study: blood vs. tissue in pancreatic cancer
• Concordance between tissue vs. cfDNA very low
• ctDNA-based assay missed KRAS variants in 14/20 tissue +ve patients (70%)
• ctDNA-based assay missed TP53 variants in 15/18 tissue +ve patients (83%)
• Similar results in brain cancers and sarcomas
Pishvaian MJ et al, Oncotarget, 2016
©2013 Foundation Medicine, Inc. | Confidential 36
Foundation Medicine’s ctDNA Assay Optimized to Address
the Extreme Challenges of ctDNA Diagnostics
Genes interrogated
ABL1 CDKN2A FOXL2 MDM2 PIK3CA
AKT1 CRKL GNA11 MET PTEN
ALK** CTNNB1 GNAQ MPL PTPN11
ARAF DDR2 GNAS MTOR RAF1
BRAF EGFR** HRAS MYC RET**
BRCA1 ERBB2 IDH1 MYD88 ROS1**
BRCA2 ERRFI1 IDH2 NF1 SMO
BTK ESR1 JAK2 NMYC TERT
CCND1 EZH2 JAK3 NPM1 TP53
CD274 FGFR1 KIT NRAS VEGFA
CDH1 FGFR2 KRAS PDCD1LG2
CDK4 FGFR3** MAP2K1 PDGFRA**
CDK6 FLT3 MAP2K2 PDGFRB
• ctDNA assay
optimized for both
sensitivity and
specificity for all
alteration classes
• Base substitutions, short
INDELs, copy number
changes and fusions
• >5,000x coverage
• ~150,000 bps targeted
** Assayed for gene fusions
©2013 Foundation Medicine, Inc. | Confidential 37
ctDNA summary
• Pros of ctDNA
• Can provide options for patients with no tissue biopsy
• Blood draw easy; no TAT to find block/pathology issues
• If exists, finds heterogeneity and resistance mutations (EGFR; T790M)
• Appropriately priced/accurate monitoring will transform care
• Cons of ctDNA
• ~40% patients have insufficient ctDNA to broadly profile
• Less genes interrogated as higher coverage required
• No mutation burden; amplifications missed in most patients
• Most physicians do not understand the nuances of ctDNA and the
potential for patient is high with poorly validated assay
©2013 Foundation Medicine, Inc. | Confidential 38
Thank you!