Genomic tests to personalize therapy of metastatic breast cancers Fabrice ANDRE Gustave Roussy Villejuif, France
Future application of genomics: Understand the biology at the individual scale Patients with lethal BC Tumor Specimen Molecular profiling Targeted therapy according to the molecular profile The deliverable: A tool to decipher the molecular mechanisms involved in cancer progression in each patient How can genomics help the oncologist understanding the biology in each patient in order to better treat them?
Outline Genomic tests to decipher cancer biology at the individual level: Oncogenic drivers Lethal subclones & intratumor heterogeneity Mutagenesis processes & DNA repair defects Dialogue between cancer cells and immune system Development & implementation of genomics in metastatic breast cancer patients Which model for clinical research? Model of implementation Conclusion
Genomic tests to identify oncogenic drivers at the individual scale Oncogenic drivers: Malignant transformation Cancer progression Targeting oncogenic drivers: oncogene de-addiction = tumor shrinkage Normal cell Cancer progression Residual disease Can we identify oncogenic drivers in individuals, in order to shrink the tumor?
Oncogenic drivers What are the genomic alterations at the DNA level? Targeting pathway Multiple alterations Private alterations
% of primary tumors Candidate «actionable» genomic alterations (with drugs under development) 35 30 25 1. 10-20 candidate actionable drivers, 2. Most of them occuring in <10% of patients 20 15 10 TNBC TNBC TNBC 5 0 Which drivers are associated with response in breast cancer? Stephens et al, Nature, 2012, TCGA, Nature, 2012, Gewinner, Cancer Cell, 2009, Andre, Clin Cancer Res, 2009, Turner, Oncogene, 2010
Best % change from baseline Targeting the driver: BYL activity in a population enriched in PIK3CA genomic alterations 100 80 60 n=72* 40 20 0 20 40 PD PD PD PD PD PD SD PD PD PD UNK UNK SD SD PD PD PD SD SD SD SD PD PD PD SD SD SD SD SD PD SD SD SD PD SD SD SD SD SD SD SD SD UNK SD SD SD PD SD SD SD SD SD SD SD SD PD UNK SD SD SD SD SD SD SD PR PR SD SD PR SD SD 60 PR 80 100 Breast Colorectal Head and neck Other Ovarian PIK3CA mutations could be a relevant target
Targeting FGFR using multikinase inhibitors Dovitinib Lucitanib Soria, Annals Oncol, 2014 Andre, Clin Cancer Res, 2013
Target screening & selection: SAFIR01 trial 403 Metastatic breast cancer patients Metastases accessible to biopsies 18 investigation centers Biopsy of metastases % of cancer cells >50% DNA extraction Whole genome CGH array (gene copy numbers) (n=287) Sanger sequencing hot spots PIK3CA/AKT1 5 genomic centers from academic hospitals Identification of a targetable genomic alteration Treatment proposition during a team webex conference 1 bioinformatician Targeted therapy according to the genomic profile (n=52) 18 investigation centers Andre F, Lancet Oncol, 2014
Target screening & selection: SAFIR01 trial Antitumor activity was observed in patients with AKT1 mutation or EGFR amplification Andre F, Lancet Oncol, 2014
% of primary tumors Which driver alterations have been associated with objective responses in mbc? 35 30 25 20 multikinase 15 10 5 0 Stephens et al, Nature, 2012, TCGA, Nature, 2012, Gewinner, Cancer Cell, 2009, Andre, Clin Cancer Res, 2009, Turner, Oncogene, 2010
Take home messages Few driver alterations associated with objective responses when targeted with specific therapy Three of them are very rare : how to develop drug in rare segments? Several candidate drivers have not yet been investigated: FGFR2, MAP3K1, MAP2K4, PTENmut
Oncogenic drivers What are the genomic alterations at the DNA level? Targeting pathway dependancy Multiple alterations Private alterations
mtor activation assessed by IHC Andre F, Lancet Oncol, 2014
Oncogenic drivers What are the genomic alterations at the DNA level? Targeting the pathways Multiple alterations Private alterations
Sensitivity to targeted agents What are the implications of co-existing mutations? p< 2.2e-16 Each color represent one drug/target couple driver alone Driver + other mutation Wild type (no driver) Co-existing mutations could be associated with resistance Lefebvre & Yu, Personal data
Drug combination to optimally target multiple drivers Combining Her2-inh and PI3K inh to treat patients with Her2+++ / PIK3CA mutated cancers Saura C et al. Clin Cancer Res 2014;20:1935-1945
Oncogenic drivers What are the genomic alterations at the DNA level? Targeting pathway dependancy Multiple alterations Private alterations
% of primary tumors Private alterations: is it worth detecting them? Recurrent alterations (>1%): 10-20 genes private alterations (0 to 1%): >200 genes 35 30 25 20 Is it worth detecting them? 15 10 5 0 private alterations (STK11, IGFR, ckit, mtor )
Treating private mutations: example with ERBB3 mutations before after Lapatinib / trastuzumab Bidard, Annals Oncol, in press
Beyond the drivers: How the use of genomic tests could help avoiding resistance? Oncogenic drivers Oncogene de-addiction Normal cell Cancer progression Residual disease Resistant disease
Outline Genomic tests to decipher cancer biology at the individual level: Oncogenic drivers Lethal subclones & intratumor heterogeneity Mutagenesis processes & DNA repair defects Dialogue between cancer cells and immune system Development & implementation of genomics in metastatic breast cancer patients Which model for clinical research? Model of implementation Conclusion
Lethal subclones in metastatic breast cancers
ESR1 mutations and resistance to endocrine therapy Mutated in metastases Not in primary (using low coverage) Mediates resistance to therapy Could be targeted by ER degradors Toy, Nature Genetics
Overall Survival ESR1 mutations and prognosis 1.00 Multivariate analysis: HR=4.60 (2.04;10.35), p<0.0001) 0.80 0.60 Median OS: 11 months 0.40 0.20 ESR1 WT (n=75, 83%) ESR1 mutations (n=16, 17%) 0.00 0 3 6 9 12 15 18 21 24 27 Months ESR1 mutations are associated with OS < 1 year and could represent a genomic segment where fast track approvals based on phase II data could be considered Arnedos, suppl Annals of Oncol, 2014
Identification of lethal clones: whole exome sequencing of metastases n=62 n=20 n=11 TSC1/2 identified as new driver in metastatic samples Outlier responder after treat with everolimus Arnedos, ESMO, 2014
Identification of resistant subclones: Match-R trial Patients with + biomarker tumor exposed to a targeted therapy and an initial response Sensitive Resistant In vitro Cell lines Mouse avatar Tumor biopy or effusion Trial started jan 2015
Outline Genomic tests to decipher cancer biology at the individual level: Oncogenic drivers Lethal subclones & intratumor heterogeneity Mutagenesis processes & DNA repair defects Dialogue between cancer cells and immune system Development & implementation of genomics in metastatic breast cancer patients Which model for clinical research? Model of implementation Conclusion
Identify mutational processes and DNA repair defects in each single patient Oncogenic drivers Oncogene de-addiction Normal cell Cancer progression Residual disease Resistant disease Mutational processes & DNA repair defects: Generate accumulation of DNA alterations
Genomic tests to identify defects in DNA repair pathways: illustration with BRCA Patients with BRCA1 and 2 mutations! (basal-like or luminal, but all with genetic instability)
Whole exome sequencing to decipher the mutational processes and DNA repair defects in each patient Pattern of mutations by sequencing (without focus on specific genes) BRCA1/2 deficiency MMR deficiency POLE mutation APOBEC APOBEC (+ NER, BER..) Alexandrov, Nature, 2013 Synthetic lethality with DNA repair defect (PARPinh and BRCA defects) Targeting mutational process?
Mutational processes and metastatic breast cancers: a role for APOBEC? Mutation rate Mutational signature Hormonal receptor status PIK3CA mutation ESR1 mutation Apobec germline deletion TpC>G/T C>T Cluster of patients present with ER+, PIK3CA mutations, high mutation rate, TpC>G/T mutations
Outline Genomic tests to decipher cancer biology at the individual level: Oncogenic drivers Lethal subclones & intratumor heterogeneity Mutagenesis processes & DNA repair defects Dialogue between cancer cells and immune system Development & implementation of genomics in metastatic breast cancer patients Which model for clinical research? Model of implementation Conclusion
Potential applications of genomics to identify immune defects in individuals with mbc Dendritic cell Is the host capable to create adaptive immune response following immunogenic cell death? What is the panel of tumor antigens? Cancer cell Can the tumor present antigens? (TAP1, HLA-A mutations ) What is the immunosuppressive network induced by cancer cells? Cytotoxic T lymphocytes What is the TcR repertoire?
Mutational load and efficacy of anti-ctla4 Ab
Neoantigen detection by WES and efficacy of anti-ctla4 Ab
Outline Genomic tests to decipher cancer biology at the individual level: Oncogenic drivers Lethal subclones & intratumor heterogeneity Mutagenesis processes & DNA repair defects Dialogue between cancer cells and immune system Development & implementation of genomics in metastatic breast cancer patients Which model for clinical research? Implementation Conclusion
How to develop genomic medicine? Two different models Drugs are evaluated in genomic segments: Stratified Medicine The method for treatment allocation is evaluated in the overall population Large scale screening using high throughput genomics (SAFIR01, AURORA ) «all comers» randomize d PIK3CA mut FGFR1 ampl others Standard of care +/-PI3K inh Standard of care +/-FGFR inh Standard of care +/-other inh Treatment driven by high throughput genomics Standard arm High throughput genomics is used to SCREEN patients to generate cohorts where drugs are tested. Each genomic alteration defines a new rare entity The benefit of using high throughput genomics is evaluated in all comers
Stratified Medicine: Testing a therapy in a population defined by a genomic alteration Drugs are evaluated in genomic segments: Stratified Medicine Large scale screening using high throughput genomics (SAFIR01, AURORA ) PIK3CA mut FGFR1 ampl others Standard of care +/-PI3K inh Standard of care +/-FGFR inh Standard of care +/-other inh High throughput genomics is used to SCREEN patients to generate cohorts where drugs are tested. Each genomic alteration defines a new rare entity
% of primary tumors Challenges in stratified medicine for mbc 35 30 25 20 15 10 AKT1 mutations : 4% of BC Minimal number of patients needed for a randomized trial: 200 Number of patients to be screened for the mutation: 5 000!!!!! 5 0 Accrual is the challenge of stratified medicine in mbc: How to screen genomic alterations in 5-10 000 metastatic breast cancer patients? Stephens et al, Nature, 2012, TCGA, Nature, 2012, Gewinner, Cancer Cell, 2009, Andre, Clin Cancer Res, 2009, Turner, Oncogene, 2010
How to overcome the accrual challenges of stratified medicine? PIK3CA FGFR1 AKT ERBB2 ESR1 PTEN BRCA1 rare genomic segments: need to screen large number of patients with mbc to perform therapeutic trials BRCA2 Cluster several genomic alterations into single pathway Scale-up capacities of molecular screening (use of circulating DNA, nationwide screening, International groups ) Approve drugs based on phase II in genomic segments associated with very poor outcome Perform part of the development in the preoperative setting in early BC Test the efficacy of the concept, not of each drug (strategy trials)
Personalized Medicine: testing whether the method to decipher biology and allocate therapy improves outcome The method for treatment allocation is evaluated in the overall population «all comers» randomize d Treatment driven by high throughput genomics Standard arm The benefit of using high throughput genomics is evaluated in all comers Controversies: what is the appropriate standard arm? How to make sure that difference between experimental and standard arms is homogenously distributed between drugs?
Illustration of personalized medicine trial: SAFIR02 NGS Array CGH: 51 genomic alterations Therapeutic phase Targeted therapy according to genomic alteration (8 different therapies) Genomic alteration R mbc Her2-negative Standard of care No alteration: follow-up
MOSCATO trial Monocentric Target Accrual = 900 patients BIOPSY FRESH TUMOR PATHOLOGICAL CONTROL MOLECULAR analysis CGH Array & NGS CLINICAL DECISION TREATMENT Hypothesis: PFSmed perso > PFS previous line of therapy Presented by: Antoine Hollebecque et al., ASCO 2013
MOSCATO: study flow Ferte C et al AACR 2014 52
53 Ferte C et al AACR 2014
Outline Genomic tests to decipher cancer biology at the individual level: Oncogenic drivers Lethal subclones & intratumor heterogeneity Mutagenesis processes & DNA repair defects Dialogue between cancer cells and immune system Development & implementation of genomics in metastatic breast cancer patients Which model for clinical research? Implementation Conclusion
Challenges in implementing genomic medicine: Offering free and equal access to genomic tests 28 public molecular laboratories in France for genomic testing. Andre F et al. Clin Cancer Res 2012
Conclusion Application of genomics Optimal technology Targets Level of evidence associated with the target* Drivers (DNA) Drivers (RNA/proteins) Lethal subclone Next generation sequencing if multiple genes validated How many genes???? Gene expression Phosphoprotein assays Ultradeep sequencing Circulating DNA ERBB2 amplification PIK3CA mutations AKT1 mutations ERBB2 mutations ER expression mtor activation CDK4/6 activation ESR1 mutations PTEN mutations I II III III I ND ND III IV DNA repair Targeted sequencing Whole exome sequencing SNP arrays BRCA1/2 mutations I / II Immune system Whole exome sequencing RNA-seq PDL1 overexpression Neoantigens Mutation load ND
Acknowledgements M Jimenez S. Delaloge M. Arnedos JC Soria C Swanton Fundings