Médecine de précision en oncologie
Antoine Hollebecque, MDDépartement d’Innovation Thérapeutique et Essais précoces (DITEP)
Gustave Roussy
Académie Nationale de PharmacieMercredi 3 octobre 2018
Precision medicine: definition from ESMO
Yates, Ann Oncol, 2018
Which rationale for this concept ?
Genomic characterization of cancers have shown that frequent cancers include large numberof genomic segments
Impact of targeting mechanisms of cancer progression
0 20 40 60 80
VHL mut poor outcome
VHL pro inter
VHL good prono
average
ER expression
ERBB2 amp
EGFR
ALK
AR expression
B-RAF
KIT
Estimated survival of metastatic cancers according to molecular alteration (months)
Identification and targeting molecular mechanisms involved in cancer progression improves outcomeSimilar concept applies with immunotherapeutics (PDL1)
B-RAF inhibitor in NSCLC
(V600E BRAF mutation)ESMO 2014
ALK inhibitor in NSCLC
(ALK rearrangment)NEJM 2010
EGFR inhibitor in NSCLC(EGFR mutation) Lancet Oncol 2012
ROS1 inhibitor in NSCLC(ROS1 rearrangment) NEJM 2014
*
**
*
*
*
Nonsmoker
Smoker, ≤
40 pack
years
Smoker, >
40 pack
years
*
*
*
~4% NSCLC
~10% NSCLC
SDPDNE
PR
Best Confirmed Response
38036034010080
60
40
20
0
-20-40-60-80-
100
Max
imu
m P
erc
en
t R
ed
uct
ion
fro
m B
ase
line
Me
asu
rem
ent
~2% NSCLC
~1% NSCLCCourtesy of B. Besse
Increase efficacy
Crizotinib
4 years
Speed-up drug development
Molecular profiling
Targeted therapy accordingto the molecular profile
Tumor Specimen
Can molecular profiling improve patient outcome ?
Patients with lethal Cancer
Intelling reports
Gustave Roussy molecular screening programMOSCATO
MOLECULAR SCREENING
CGH Array & NGS & WES & RNAseq
CLINICAL DECISION
Max 21 calendar days
FRESH TUMOR
BIOPSY PATHOLOGICAL CONTROL
TREATMENT
• High through-put analysis in a high volume phase I center
• Monocentric
• Target accrual => 1000 patients
Antoine Hollebecque et al., ASCO 2013; Charles Ferte et al, AACR 2014
Design of MOSCATO
• Primary Objective: To show that broad molecular screening improves outcome Stastistical hypothesis: > 25% of patients treated according to their genomic alteration
will experience a clinical benefit defined by a PFS ratio > 1.3
• Secondary Objectives To assess the feasability of this approach
To improve tumor response
To assess the percentage of patients treated with a selected therapy
To assess the frequency of genomic alterations
To speed-up drug development through enrichment of trials in biomarker-defined patients (stratified medicine)
PFS 1 PFS 2
PFS1
PFS 2
> 1.3Relevant Molecular Targeted Agent(MOSCATO)
Standard Therapy
TumorProgression
TumorProgression
Objectives of MOSCATO
FGFR1 amplification
Ion Torrent PGM – Life Technologies(Ampliseq CHP2 + custom
n=75 genes, Dec 2013)
CGH array Agilent(180K, Whole genome coverage)
+
Torrent_Suitev2.2Confim
ed bySangerSequencing
DNAextrac on10–50ng
Mul plexPCR1450ampliconsIonTorrent/PGM(>500Xcoverage)
StandardizedReport
Analysis Improvement: 30 genes then 50 then 75 genes screened
then moved partially in may 2014 to WES + RNASeq
MOLECULAR SCREENING
CGH & NGS & RNAseq & WES
CLINICAL DECISION
TREATMENTFRESH TUMOR
BIOPSY PATHOLOGICAL CONTROL
Methods of MOSCATO
• Clinical Decision– Weekly Tumor Board discussion
– Clinicians / Biologists / Bioinformaticians
• Treatment– Phase I/II trial compounds
• > 60 phase I trials @ Gustave Roussy
• > 20 Actionable Targets
– Off label use of EMA approved MTA
Hanahan D, Weinberg R.A. Cell 2011.
CLINICAL DECISON
TREATMENTMOLECULAR SCREENING
CGH & NGS & RNAseq & WES
FRESH TUMOR
BIOPSY PATHOLOGICAL CONTROL
Methods of MOSCATO
adults1000
RESULTS of MOSCATO
Biopsy and decision in MOSCATO 01
are performed in clinically relevant timing
Median 25 daysIC95 12-56 days
Median 24 daysIC95 0-59 days
No Actionable Target based on NGS or CGH: n=434
Molecular portrait (NGS or CGH) n=844
NGS+CGH: n=740 / NGS alone: n=98 / CGH alone: n=6
Actionable Target, n= 411(IHC alone n=1)
Received Matched Treatmentn=199
Patients includedn= 1110
Pediatrics N=74
Screen failure N=87No possible biopsy n=28SAE: n=25Consent withdrawal: n=11 Clinical deterioration or death: n=5 Other: 11Missing: n=7
No NGS nor CGH: n=103
Adults included n= 1036
rapid clinical deterioration: n=62
waiting for treatment: n=37
exclusion criteria n=18
trial not open n=12
absence of progressive disease n=6
patient refusal n=3
Other n=64, unknown n=6:
Primary endpoint completedn= 193
PFS1 missing: n=5PFS2<1.3*PFS1 and not yet progressed n=1
Molecular tumour board (MTB) n=949
Summary of MOSCATO results
49 %
19%
Massard et al. Cancer Discov 2018
Results
PFS ratio > 1.3 = 33%
IC95 PFS ratio = 26%-39%
H0 rejected with a p-value < 0.001
Molecular Targeted Agent(MOSCATO)
Previous Therapy
TumorProgression
TumorProgression
Results for primary endpoint
Massard et al. Cancer Discov 2018
CR + PR = 11% (22/193)
CR + PR + SD = 63% (122/193)
Best percent changefrom baseline
(RECIST1.1)
+ 20%
- 30%
Results for secondary endpoints
Massard et al. Cancer Discov 2018
PatientsN=43 (4%)
BiopsiesN=48
Fit for analysisN=38 (79%)
No biopsy = 3Cellularity < 10% = 6
Not realized = 1
Druggable molecular aberration(s)N=25 (66%)
TreatedN=19 (76%)
MOSCATO-01N=1168
50%
Rebiopsy procedures =1 patient x 3
3 patients x 2
No alteration foundN=5
Winner tumor typee.g. cholangiocarcinoma
Verlingue et al. Eur J Cancer 2017
Median PFS ratio = 1.52
IC95 [0,08-7,1]
PFS ratio > 1.3 = 44%
Response rates Progression free survival
6-month PFS rate = 42%PR+CR = 32%
Winner tumor typee.g. cholangiocarcinoma
ERBB3 alterationsN=33 (3.7%)
NPVN=13
Received therapiesN=28
Other treatmentsN=17
VUPN=13
HotspotsN=6
AmpN=2
HER3 partners inhibitorsN=11
NPV : non pathogenic variantVUP : variant of unknown pathogenicity
Amp: amplification
1 patient with 2 alterations in ERBB3
Winner Pathwaye.g. ERBB3
HER3 partners’ inhibitors:- Trastuzumab and/or lapatinib- Afatinib
Overall survival
Winner Pathwaye.g. ERBB3
PFS < 100 days
PFS > 200 days
HER2 directed monoclonal antibodies
HER2/pan HER directed small molecules
Ongoing treatment
Extracellular domain
Kinase domain
III
III
IV
Docking domain
S846I
E928G
A159TV140L
G661S
Q865H
Transmembrane domain
K329E
ERBB3
Current questions ?
• Should we use large panel of genes ?
• How to develop drugs in rare genomic segments ?
• Prediction of sensitivity ?
• Detecting subclones and monitoring clonal evolution ?
Current questions ?
• Should we use large panel of genes ?
• How to develop drugs in rare genomic segments ?
• Prediction of sensitivity ?
• Detecting subclones and monitoring clonal evolution ?
• Because it allows testing the few relevant genes in a single assay
• Because it allows finding alterations in other genes for which relevance to cancer is shown but there is no drug approval
• Because it allows having « global picture » of cancer genome and mutationalprocesses
• Because it allows modeling disease biology in each patient
Should we use large panel of genes ?
EGFR
ALK
RET
BRAF
MET
ERBB2
ROS1
others
“FDA approved Thermo Fisher Scientific’s Oncomine™ Dx Target Test as the first next-generation sequencing (NGS)-based companion diagnostic that screens tumor samples against panels of biomarkers”
Should we use large panel of genes ?
Femme 30 ans Cholangiocarcinome intra-hépatique (Avril 2011) En progression après:
Mai 2011: Radioembolisation (Yttrium90) Mai-Sep 2011: Folfox Fev-Juin 2012: Folfox Juin-Dec 2012: Gemcitabine-Cisplatine Jan-Avr 2013: Sunitinib Avr-Mai 2013: Paclitaxel Juin 2013: MOSCATO
No mutation
tNGS
Translocation FGFR2-CCAR1
WES et RNAseq
Should we use large panel of genes ?
26/11/2014 10/02/2015
Fusion FGFR2-CCAR1 → FGFR inhibitor
Should we use large panel of genes ?
Current questions ?
• Should we use large panel of genes ?
• How to develop drugs in rare genomic segments ?
• Prediction of sensitivity ?
• Detecting subclones and monitoring clonal evolution ?
Drug development in rare genomic entities
Drug development in rare genomic entities
Drug development in rare genomic entities
Larotrectinib
Drug development in rare genomic entities
Andre F, NEJM, 2018
Basket protocol
1 molécule → 1 ou plusieurs anomalies moléculaires → plusieurs types tumoraux
Drug development in rare genomic entities
Drug development in rare genomic entities
Hyman DM et al. NEJM 2015
Efficacy according to histology ?e.g. BRAF
Hyman DM et al. NEJM 2015
Drug development in rare genomic entities
Mutation BRAF V600E
Mélanome
Mutation BRAF V600E
Cancer du colon
Kopetz et al. J Clin Oncol 2015Chapman et al. N Engl J Med 2011
ORR = 5%
ORR ≈ 50%
Level of evidence for actionability
Current questions ?
• Should we use large panel of genes ?
• How to develop drugs in rare genomic segments ?
• Prediction of sensitivity ?
• Detecting subclones and monitoring clonal evolution ?
Genomic predictors of sensitivity
Anti-PD-1 in NSCLC
Snyder et al NEJM 2014 Rizzi et al Science 2015
Anti-CTLA-4 in Melanoma
Number of non-synonymous
mutations
Generation of neoantigens
Mutational burden
Sensitivity to anti-PD1
Pembrolizumab et MSI-High
Genomic predictors of sensitivity
Le et al. Science 2017
Co-existing mutations could be associated with resistanceto single agent targeted therapies
Each color represent one drug/target couple
driver alone Driver +other mutation
Wild type(no driver)
Sen
siti
vity
to t
arge
ted
age
nts
Lefebvre & Yu, Personal data
What is the clinical impact of multiple genomic alterations ?
Targeted Therapies
Current questions ?
• Should we use large panel of genes ?
• How to develop drugs in rare genomic segments ?
• Prediction of sensitivity ?
• Detecting subclones and monitoring clonal evolution ?
EGFR MET
11 Jul 2012 30 Aug 2012 23 Oct 2012 17 Dec 2012
Erlotinib
PR (-36%) PD (New Lesions)SD (-28%)Baseline
12-Jul-2012C1J1
17-Dec-2012Fin de Traitement
CA
19
.9 é
volu
tio
nC
T-s
can
évo
luti
on
CG
H é
volu
tio
n
EGFR
Detecting subclones and monitoring clonal evolution ?
Perspectives
• Big data
• CfDNA
PerspectivesNew generation of software for target prediction & database to train
New generation of software
Database to train software
Trials testing the clinical utilityof large panel of genes
(Watson …)
Big Data« no amount of algorithmic finesse can squeeze out information that is not present » (NEJM, 2017)
Integration of multicomponent predictors: example of IO
Blank, Science, 2016
PerspectivesMultigene panels to detect genomic alterations in plasma
Jovelet, Clin Cancer Res, 2015
Performance of NGS on plasma DNA to detect actionable mutation
Approach to detect tumor-specific proteins and RNA in blood ?
Top Related