Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Gene signatures: What’s new to predict outcome and drug sensitivity? Christos Sotiriou, MD, PhD Jules Bordet Institute Université Libre de Bruxelles (ULB) Brussels, Belgium What’s new (1)Molecular taxonomy and prognosis (2)Drug sensitivity (chemo, targeted agents) (3)Future directions (RNA seq) Gene prognostic signatures: Genomic Grade Index H/I + MGI Mammaprint Oncotype DX Ma et al. Cancer Cell. 2004 Sotiriou et al. J Natl Cancer Inst. 2006 Paik et al. NEJM, 2004 Van’t Veer et al. Nature,2002 Population: Untreated Tissue: Fresh/Frozen Population: Tamoxifen-treated Tissue: FFPE May add additional information to current clinico-pathological parameters for treatment decision making for some patients Recurrence score (TransATAC) TAM Postmenopausal ER+, No chemo R Anastrazole Anastrazole+ TAM TAM (N=609) Anastrazole (N=622) Dowsett M et al, JCO 2010 (1) Proliferation = driving force Proliferation Genes Non-Proliferation Genes Sotiriou and Pusztai, N Engl J Med 2009 Desmedt et al. Clin Cancer Res, 2008 Wirapati et al. Breast Cancer Res, 2008 Sotiriou and Piccart Nat Rev Cancer, 2007 Most published signatures are not significantly better outcome predictors than random signatures of identical size… Venet et al. PLoS Comput Biol. 2011 Proliferation score (2) Informative for ER+/HER2- BC High proliferative high risk tumors (luminal B) low proliferative low risk tumors (luminal A) ER/HER2- HER2+ ER+/HER2Sotiriou and Pusztai, N Engl J Med 2009 Desmedt et al. Clin Cancer Res, 2008 Wirapati et al. Breast Cancer Res, 2008 Sotiriou and Piccart Nat Rev Cancer, 2007 (3) Currently defined molecular subtypes are mainly driven by ER, HER2 and proliferation Basal-like (ER-/HER2) HER2 + Luminal (ER+) Normal-like Perou et al., Nature 2000 Intrinsic ER/HER2/ classification proliferation Comparison in 5715 breast cancer samples (36 datasets) Haibe-Kains et al., JNCI 2012 Clinical relevance – prognosis PAM 50 Intrinsic classification Single genes (mRNA) ER, HER2, Aurka ER/HER2/proliferation Modules Haibe-Kains et al., JNCI 2012 Clinical utility? Molecular classification (PAM50) vs 1st generation prognostic signatures vs IHC? Attend the next guidelines session… Basallike HER2 + Luminal B Luminal A Next Steps… Identify molecular drivers Patients selection for clinical trials Methods Discovery set -Primary BC (N=997) -Matched normal (N=258) Validation set -Primary BC (N=995) METABRIC consortium Illumina HT-12 (gene expression) Affymetrix SNP 6.0 (CAN, CNV, SNPS) Integration of genomic and transcriptomic analysis of breast cancer (1) Germline and somatic variants influence breast tumor expression architecture (39% 11,198/28609 probes ) • Cis = a variant at a locus has an impact on its own expression • Trans = a variant at a locus is associated with genes at other sites in the genome (2) Patterns of cis outlying expression refine putative breast cancer drivers (3) Integrative clustering reveals 10 novel IntClust molecular subgroups beyond the intrinsic subtypes Clinical outcome * * * * * * * = good outcome * = poor outcome What is the clinical utility of the Intclust classification today? • Limited (prognosis) • Potential to identify targetable drivers in the future • NGS will provide additional information What’s new (1)Molecular taxonomy and prognosis (2)Drug sensitivity (chemo, targeted agents) (3)Future directions (RNA seq) Pooled analysis of gene expression studies to predict neoadjuvant (taxanes and/or anthracyclines) chemotherapy response (pCR) Several molecular Processes and molecular pathways ? Response to chemotherapy M Ignatiadis JCO 2012 All N= 845 pts M Ignatiadis JCO 2012 ER-/HER2- Different processes/pathways are associated with pCR in different BC subtypes HER2+ ER+/HER2- Main message Chemotherapy sensitivity = tumor microenvironment matters! What about targeted agents? • High prevalence of PIK3CA mutation in ER+ BC (irrespective of molecular subtypes LumA vs LumB) • PIK3CA mutation = better clinical outcome (low mTORC1 signaling) Does PIK3CA GS predicts for response to PI3K pathway inhibitors ? Baselga et al. Neoadjuvant study Letrozole (N=27) vs Letrozole + Everolimus (N=31) Sabine et al. Pre-surgical window study Everolimus 15 days (N=23) Total 81 patients used for this analysis Collaboration with Novartis and John Bartlett The PIK3CA-GS is associated with anti-proliferative response to Everolimus (Baselga et al. dataset) Interaction p test= 0.02 Letrozole + Everolimus Interaction p test= 0.02 Letrozole Unpublished data The PIK3CA-GS is associated with anti-proliferative response to Everolimus by PIK3CA genotype PIK3CA wild type PIK3CA mutation Interaction p test= 0.05 Interaction p test= 0.07 Letrozole + Everolimus Letrozole Unpublished data What’s new (1)Molecular taxonomy and prognosis (2)Drug sensitivity (chemo) (3)Future directions (RNA seq) Rational for RNA-Seq 1. Study of the entire transcriptome 2. It can identify small non-synonymous mutations that alter protein-coding sequencing (1-2% of whole genome) 3. Splice variants, gene fusions 4. It can identify RNA editing events Belgian National Initiative N = 60 BC + normal breast Exome seq RNA seq (+Affymetrix) Methylation Infinium 450K (1) Excellent correlation between Affymetrix and RNA-seq 2.0 Correlation for all genes AFFY vs ILLUMINA RNA−seq 1.0 0.5 0.0 Density 1.5 75% genes correlate at 0.82 −1.0 −0.5 0.0 0.5 Spearman correlation Quantiles5%: −0.025; 25%: 0.41; 50%: 0.69; 75%: 0.82; 95%: 0.92 1.0 (2) Identification of the 4 main molecular subtypes (3) Lack of correlation between Allred score and ER mRNA transcripts… RNA seq = Higher dynamic range HER2 Lum B Basal Lum A (4) Good correlation between HER2 copy number and mRNA levels Basal Lum A HER2 Lum B (5) Some expression basedsignatures could be reproduced GGI using RNA-seq data GGI AFFYMETRIX vs ILLUMINA RNA−seq ● ● Low−risk High−risk Discordance ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● 1 ● ● 0 ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ●● ●● −1 GGI scores (ILLUMINA RNA−seq) 2 ● ● ● ● ● Spearman correlation 0.98 ●● ●● ● ● ● ●● ● −1 0 1 GGI scores (AFFY) Spearman correlation: 0.98 2 Mammaprint using RNA-seq data MAMMAPRINT AFFYMETRIX vs ILLUMINA RNA−seq ● ● ● ● ● ● ● 0.0 ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● −0.2 ● ●● ● ● −0.4 ● ● ●● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● −0.6 MAMMAPRINT scores (ILLUMINA RNA−seq) 0.2 ● Low−risk High−risk Discordance ● ● ● Spearman correlation 0.97 ● ● ● ●● −0.4 −0.2 0.0 MAMMAPRINT scores (AFFY) Spearman correlation: 0.97 0.2 0.4 Many other things… 1. Study of the entire transcriptome 2. It can identify small non-synonymous mutations that alter protein-coding sequencing (1-2% of whole genome) 3. Splice variants, gene fusions 4. It can identify RNA editing events Messages 1. Novel molecular classification - Clinical utility? - Potential to identify targetable drivers in the future… 2. Tumor microenvironment matters for chemotherapy response! 3. Predict response to targeted agents may be easier (i.e. PIK3CA signature). 3. RNA-seq: lot of promises… Sotiriou’s Translational lab Christine Desmedt Michail Ignatiadis Sherene Loi Françoise Rothé Marion Maetens Debora Fumagalli Hatem Azim Stefan Michiels David Brown Sandeep Singhal Vinu Jose Laurence Buisseret Samira Majjaj Naïma Kheddoumi Ghizlane Rouas PierreYves Adnet Delphine Vincent Laurence Simon Dominique Roels Martine Piccart (IJB) Major Collaborators IRIBHM: Vincent Detours David Gacquer Marc Abramowicz ULB-Epigenetics: François Fuks Sarah Dedeurwaerder Montreal University: Benjamin Haibe-Kains Immunology lab-IJB: Karen Willard-Gallo MDACC: Lajos Pusztai Fraser Symmans Les Amis de l'Institut Bordet MEDIC Foundation Sanger: Peter Campbell Mike Stratton Sancha Martin UCL UK: Charles Swanton Institut Gustave Roussy: Fabrice Andre ER+ = 8 pts; HER2+ = 8 pts TNBC = 8 pts 131 fusion transcripts Cancer Res. 2012 Apr 15;72(8):1921-8. 86 fusion transcripts PRIVATE 45 fusion transcripts REDUNDANT