Classifications génotypiques dans les cancers du sein
publicité
Classifications génotypiques dans les cancers du sein DES Onco Lille 1er avril 2016 Christine Desmedt PhD Translational Research Unit Université Libre de Bruxelles Institut Jules Bordet Brussels, Belgium (R)Evolution of molecular testing in breast cancer 1980 ER status (protein) 1990 ER/PR/HER2 status (protein) IHC HER2 status (DNA) 2000 Gene expression (RNA) 2011-… Next generation Sequencing (RNA, DNA) Copy number aberrations (DNA) Biochemistry FISH 1 marker 3 markers Gene expression signatures RNA seq, Exome, whole genome 1. Mutations Types of genome alterations Meyerson Nat Rev Genet 2010 Types of genome alterations Stratton et al. 2009 Substitutions • Synonymous = mutation which results in the same (or a sufficiently similar) amino acid • Missense = point mutation where a single nucleotide is changed to cause substitution of a different amino acid • Nonsense = point mutation in a sequence of DNA that results in a premature stop codon and in a truncated, incomplete, and usually nonfunctional protein product • Splice site = mutation which may alter splicing of the mRNA • Stop codon readthrough = allows protein translation to continue through a stop codon Examples of substitutions Synonymous Missense Nonsense Insertions/Deletions (indels) • Insertion = add one or more extra nucleotides into the DNA • Deletion = remove one or more extra nucleotides into the DNA - Can be in-frame or a frameshift - The earlier in the sequence the deletion or insertion occurs, the more altered the protein produced is Examples of cancer genes Driver/Passenger mutations • Driver mutations= mutations which confer selective advantage for the cancer cells, and are causally implicated in oncogenesis • Passenger mutations = the remainder. Represent the majority of the mutations in a tumor. 2. DNA sequencing Library construction Library= double stranded DNA which is characterized by a length and some specific sequences at 5’ and 3’ end. The construction of a library is the first step of most technologies. www.illumina.com Sample multiplexing Sample multiplexing allows different samples to be simultaneously sequenced during a single experiment. Samples are differentiated by their unique ‘barcode’. www.illumina.com Coverage Sequence coverage represents the number of sequenced reads that cover the site; this affects the ability to detect point mutations. Physical coverage measures the number of fragments that span the site; this affects the ability to detect the rearrangement, based on paired reads that map to different chromosomes Meyerson Nat Rev Genet 2010 Coverage Some considerations: Cancer is heterogeneous: 1. Mix of tumour and non-tumour cells 2. Mix of clonal and subclonal mutations Example: Detection of a heterozygous subclonal mutation which is only present in 20% tumour cells in a sample which contains 50% tumor cells, average coverage of 100X: 100*0.5*0.5*0.2= 5 mutated reads DNA sequencing Whole genome sequencing Targeted sequencing = 3GB! *whole exome (~30Mb) *large panel of genes *small panel of genes Targeted sequencing (1)pull-down approach Targeted sequencing (2)amplicon sequencing single-tube per fragment assay multiplex PCR assay RainDance micro droplet PCR Single-end vs paired-end You sequence using the adapter for one end, and then once you’re done you start over sequencing using the adapter for the other end. This means your two reads are the reverse complement of the 100 3′-most bases of the Watson strand and the Crick strand; these reads are assumed to be identical to the 100 5′-most bases of the Crick strand and Watson strand respectively. Sequencing data FASTA BAM VCF •Raw data •Aligned data • Binary Alignment Map •Called data • Variant Call Format 3. Publically available information COSMIC cBioPortal Important international cancer initiatives • TCGA • ICGC To conclude! • Sequencing technologies are continuously evolving • The choice of the sequencing technology and sequencing kits depends on different parameters • Standardization of data analysis is important 4. NGS applied to breast cancer Outline 1. 2. 3. 4. 5. 6. INTER-tumor heterogeneity Mutational signatures INTRA-tumor heterogeneity Primary/metastatic heterogeneity Treatment-induced changes Liquid biospies 1. INTER-tumor heterogeneity With the help of gene expression arrays… At least 4 biologically distinct groups changed the perception of the disease Perou et al, Nature 2000; Sorlie et al, PNAS 2001; Hu et al, BMC Genomics 2006; Parker et al. JCO 2009 With the help of next generation sequencing… Mutational landscape BEFORE next generation sequencing… Growth factor receptors: Chromatin modification: EGFR, ERBB2, FGFR1 PI3K pathway: AKT1, PIK3CA, PTEN DNA damage control: Transcriptional regulation: BRCA1, BRCA2, TP53 JNK pathway: Cell cycle: MEK pathway: CCND1 Mutational landscape AFTER next generation sequencing… Chromatin modification: ARID1A, ARID1B, ARID2, KMT2C, KMT2D, KDM5C, HIST1H3B, NCOR1, SETD2 Growth factor receptors: EGFR, ERBB2, ERBB3, FGF3, FGFR1, FGFR2 PI3K pathway: AKT1, AKT2, AKT3 INPP4B , PIK3CA, PIK3R1, PTEN DNA damage control: Transcriptional regulation: ATM, ATR, BRCA1, BRCA2, FANCD2, MDM2, MLH1, MSH6, TP53 CFCB, CTCF, CHEK2, ESR1, FOXA1, GATA3, SF3B1, TBX3, RUNX1 JNK pathway: MAP2K4, MAP3K1 MEK pathway: BRAF, KRAS, NF1 Cell cycle: CCND1, CDK4, CDKN1B, CDKN2A, RB1 Long tail of mutations present at low frequency Stephens et al. Nature 2012 Mutational landscape can differ according to: 1. Molecular subgroups TCGA Nature 2012 Mutational landscape can differ according to: 2. Histological subtype Enrichment of specific alterations in lobular vs ductal breast tumors Desmedt et al. submitted; Ciriello et al. Cell 2015 Invasive lobular breast cancer (ILBC) • • • • • • ILBC ~5-15% of breast cancers Small, discohesive epithelial cells Mostly ER+ Characterized by the loss of CDH1 More late relapses compared to IDBC Peculiar metastatic pattern NOTE: the terms ‘ductal’ and ‘lobular’ carcinoma have no implication with regard to the site of origin within the mammary ductal-lobular system Comparison of recurrence patterns with IDBC Pestalozzi et al. JCO 2008 ER- ER+ Differential benefit from aromatase inhibitors BIG 1-98 trial (SABCS 2012) ABCSG 8 trial (SABCS 2014) Three Large Efforts To Better Characterize Lobular Tumors Total Nr ILBC Ciriello et al. Cell 2015 TCGA Desmedt et al. J Clin Oncol In Press IJB Michaut et al. Scient Rep 2016 RATHER 127 630 144 FFPE samples Frozen samples ̌ ̌ Source of tumor DNA Frozen samples Availability of gDNA Exome Sequencing ̌ ̌ (all) Targeted Sequencing Copy Number RNA-Seq ̌ ̌ (Affy SNP 6.0) ̌ Expression arrays miRNA Sequencing DNA methylation RPPA ̌ ̌ ̌ ̌ ̌ (47.5%) (38%) (from seq) ̌ ̌ (Affy 133 Plus2) ̌ (360 cancer genes) (kinases + cancer genes) (Affy SNP 6.0) ̌ Mutational Landscape of Lobular Tumors Mutational landscape of ILBC Desmedt et al. n=414 Ciriello et al. n=127 Michaut et al. n=138 Comparison ILBC vs IDBC Desmedt et al. JCO In Press Comparison ILBC vs IDBC Ciriello et al. Cell 2015 E-Cadherin (CDH1): the most frequently altered gene in ILBC In terms of substitutions and indels (~65%) - Spread accross all the gene - Most are truncating mutations Ciriello et al. Cell 2015 At the copy number level (~90%) Desmedt et al. JCO In Press Association with mRNA and protein expression Ciriello et al. Cell 2015 Association with mRNA and protein expression Absence of expression by IHC in 85% of cases, but more frequent when bi-allelic loss Michaut et al. Sci Rep 2016; Desmedt et al. JCO In Press CDH1 & clinicopathologic characteristics • Enrichment of CDH1 mutations in multifocal tumors • No prognostic value PI3K pathway: the most frequently altered pathway in ILBC PI3K pathway (>50%) Mutual exclusivity! Michaut et al. Sci Rep 2016 Alterations and PI3K/AKT protein activation • pAKT in ~40% of samples • AKT1 and PTEN alterations associated with activation, but not PIK3CA mutations Ciriello et al. Cell 2015 PI3K alterations & clinicopathologic characteristics • PIK3CA mutations enriched in low proliferative, low grade and HER2-negative tumors • AKT1 mutations associated with increased risk of early relapse Increased Rate of HER2 and HER3 mutations HER2 and HER3 mutations (9%) Desmedt et al. JCO In Press Structural visualization HER2 HER3 HER3 E928G mutation Enhanced dimerization affinity Enhanced EGFR kinase activity Littlefield et al. Sci Sign 2014 HER2/3 mutations & clinicopathologic characteristics • HER2 mutations enriched in ER and PgR negative tumors • HER3 mutations enriched in HER2-amplified tumors • HER2 mutations associated with increased risk of early relapse Alterations involving the Estrogen Receptor or critical regulators ESR1 copy number gains (~25%) Desmedt et al. JCO In Press FOXA1 & GATA3 as critical regulators of the ER-dependent transcriptional program Liu et al. Cell 2014 FOXA1 mutations (9%) FOXA1 & GATA3 mutations & clinicopathologic characteristics • No association Some Alterations are Enriched in Specific ILBC Histological Subtypes Survival Analyses Survival analyses Conclusions 1. Genomic characterization of ILBC by 3 consortia; 2. Identification of differentially mutated genes compared to IDBC; 3. ILBC subtypes do represent distinct disease entities since they are associated with specific clinical features and genomic alterations; 4. Patients with HER2, HER3 and AKT1 mutated tumors might benefit from targeted treatment; 5. High frequency of FOXA1 mutations and ESR1 gains deserves further investigation for potential differential endocrine treatment strategy Targetable alterations Arnedos, M. et al. (2015) Precision medicine for metastatic breast cancer—limitations and solutions Nat. Rev. Clin. Oncol. doi:10.1038/nrclinonc.2015.123 Targetable alterations • Genome-driven therapeutical studies mainly in metastatic setting; • Best to be carried out in large international initiatives to garantee adequate characterisation of the alterations and access to the drugs Aiming to Understand the Molecular Aberrations in Metastatic Breast Cancer The AURORA Program AURORA 1. To improve the understanding of metastatic breast cancer; 2. To discover biomarkers of response and/or resistance to systemic therapy using genomic and transcriptomic data of “exceptional responders” and “rapid progressors”. 3. To assess the feasibility of implementing a global molecular screening platform of metastatic breast cancer. 4. To build new therapeutic hypotheses based on findings generated by TGS. 5. To evaluate the prognostic relevance of genomic alterations detected in tumour metastatic biopsies and archived primary tissue. 6. To correlate molecular alterations in patients with the efficacy endpoints; 7. To identify patients with candidate driver alterations in their tumours that can be matched to biomarker-driven clinical trials. Screening Samples Samples Collection Patient enrollment • Serum • Plasma • Metastatic lesion tissue, archived (< 6 months) or newly collected: 1 FFPE biopsy core (or sections) + 1 Frozen biopsy core • Primary tissue, archived : 1FFPE block (or sections) • Whole blood: 1x10mL Store at site Central Pathology Laboratory Frozen tissue Blood ER/PgR HER2 Ki67 Storage AURORA Platform FFPE tissues DNA Results reporting TGS RNA RNA sequencing Messages • No two tumors are genomically identical; • Targetable alterations described in breast cancer; • Stay cautious…: – Assessment of the alteration – Availability of germline DNA – Theoretical expectation does not always match clinical results – Cancer is a dynamic disease relying on several hallmarks 2. Mutational signatures Desmedt et al. Can Met Rev in press Somatic mutations occur in all cells of the body throughout life Adapted from S Nik-Zainal Adapted from S Nik-Zainal 3083 tumors 27 ≠ tumor types 7042 tumors 30 ≠ tumor types Exomes=2957 Whole genomes=126 Exomes=6535 Whole genomes=507 1) Mutation frequency varied markedly across cancer types 1000-fold! Lawrence M et al. Nature 2013 2) Mutation frequency varied markedly across patients within a cancer type 1000-fold! Lawrence M et al. Nature 2013 3) Identification of 21 mutational signatures (substitution class + sequence context) Alexandrov L et al. Nature 2013 4) The presence of mutational signatures vary across human cancers Alexandrov L et al. Nature 2013 5) Most individual cancer genomes exhibit more than one mutational signature Alexandrov L et al. Nature 2013 6) Associating cancer etiology and mutational signatures Alexandrov L et al. Nature 2013 Messages • Better understanding of the etiology • Most frequent signatures in breast cancer= – Age-associated signature – HRD-associated – APOBEC • Still many mutations of cryptic origin • Mutational processes can vary over the course of the disease 2. INTRA-tumor heterogeneity Intra-tumor heterogeneity Already evident for pathologists since many decades: - Different cell populations - Difference regarding proliferation/grade - Differences regarding hormonal and HER2 receptors - … Example HER2-heterogeneity • Regional heterogeneity present in 18% of the cases with HER2 amplification detected on whole slides; • Associated with worse disease free survival. Seol et al. Mod Pathology 2012 Three strategies 1. Sequencing of single ‘bulk tissue tumour samples; 2. Multi-region sampling and sequencing; 3. Single-cell sequencing Three strategies 1. Sequencing of single ‘bulk tissue tumour samples; 2. Multi-region sampling and sequencing; 3. Single-cell sequencing Driver/Passenger mutations • Driver mutations= mutations which confer selective advantage for the cancer cells, and are causally implicated in oncogenesis • Passenger mutations = the remainder. Represent the majority of the mutations in a tumor. Clonal versus subclonal alterations • Clonal: present in all the tumor cells; arose most probably early in the development of the tumor. • Subclonal: only present in a group of tumor cells (subclones), arose most probably late in the development of the tumor. Subclones present in each tumor Nik-Zainal et al. Cell 2012 Shah et al. Nature 2012 ‘Drivers’ can be present subclonally Three strategies 1. Sequencing of single ‘bulk tissue tumour samples; 2. Multi-region sampling and sequencing; 3. Single-cell sequencing Multi-region sampling (example 1) Yates et al. Nat Med 2015 Multi-region sampling (example 1) • Potentially targetable mutations were subclonal in 26% of the primary tumors; • Spatial heterogeneity Yates et al. Nat Med 2015 Example 2= Multifocal breast cancer • Definition: presence of multiple unilateral synchronous invasive lesions; • Frequency: 21-24% according to the most recent estimates; • More frequent axillary lymph node involvement; • Some studies report that it is associated with worse prognosis. Andea et al. Cancer 2004; Cabioglu et al. J Am Coll Surg 2009; Weissenbacher et al. BCRT 2010; Tot et al. Patholog Res Int. 2010; Ustaalioglu et al. Am J Clin Oncol 2011; Tot Clin BC 2011; Tot et al. Human Path 2011; Moutafoff et al. Gyn Obst & Fert 2011; Lynch et al. Ann Oncol 2012… Characterization of multifocal breast cancer CAP: It’s OK to only characterize the largest lesion if all lesions have similar histology and grade Lester et al. Arch Pathol Lab Med 2009 Discordances regarding ER, PgR, HER2 and grade % discordancy RO RP HER2 Grade 3 - 4.4 11 - 15.9 6- 9.7 5.5 - 18.6 Bethine et al. Am J Clin Pathol 2013, Buggi et al. Ann Oncol 2012; Pekar et al. Cancer 2014 Genomic heterogeneity of multifocal breast cancer STEP1: Targeted sequencing of 360 cancer-related genes N=36 pts (171 samples) All samples from the same pt needed to have same grade, ER, PgR and HER2 status Identification of 3 groups Homogeneous (N=11, 31%) L1+L2 Intermediate (N=13, 36%) L1 L2 Heterogeneous (N=12, 33%) L1 L2 Heterogeneous group Inter-lesion distance matters Lesions closer to one another tend to be genomically more similar to one another Lesions further apart from each other tend to be more different from each other Genomic heterogeneity of multifocal breast cancer STEP2: Low coverage whole genome sequencing for identification of rearrangements and copy number alterations N=8/36 pts All samples from the same pt needed to have same grade, ER, PgR and HER2 status Rearrangements All lesions of a same patient share common somatic rearrangements Messages regarding multifocal breast cancer • It is relatively frequent (≈1/4); • It is important to assess ER and HER2 on more than 1 lesion; • Even if lesions ‘look’ similar at the pathology level, one third can present different mutations; • Despite mutational heterogeneity, all investigated samples shared common ancestor: intra-mammary metastases Three strategies 1. Sequencing of single ‘bulk tissue tumour samples; 2. Multi-region sampling and sequencing; 3. Single-cell sequencing Example of a ER-positive tumor Example of a ER-negative tumor No two single tumour cells are genetically identical, calling into question the strict definition of a clone!!! Messages • Primary tumors are composed of different subclones; • Expansion of dominant clone which could be responsible of the ‘diagnosability’ of the disease; • Actionable alterations can be present subclonally/ in different lesions from MFBC; • Clinical relevance to be further investigated. 4. Primary/metastatic heterogeneity Questions 1. How do primary tumors progress to metastatic disease? 2. Is the primary/metastatic heterogeneity clinically relevant? 3. What about contralateral breast tumors? Questions 1. How do primary tumors progress to metastatic disease? 2. Is the primary/metastatic heterogeneity clinically relevant? 3. What about contralateral breast tumors? What we know The invasionmetastasis cascade What we don’t! AND/OR? Metastatic cascade Parallel progression Naxerova & Jain, Nat Rev2015 Phylogenetics Tree of life= Evolutionary trajectory of the disease Exome/genome-wide studies • • • N= 1 pt with only one metastatic site Naxerova & Jain, Nat Rev2015 Autopsy-based BC studies Study Nr of pts Main findings Viadana (1973) 647 Comparison of metastases in young and older pts: 1. More extensive disease in younger pts 2. More liver, thyroid and bone mets in younger pts 92 Comparison of metastatic pattern IDC vs ILC: 1. More lung mets in IDC 2. More bone peritoneal, car meningitis mets in ILC Parham (1989) 85 Confirmation of cancer-related death in 75% of the cases with BC history : tendency to over-estimate BC as cause of death. Cummings (2014) 197 1. 2. 3. 4. Juric (2015) 1 1. Comparison of primary and metastatic lesions. 2. Heterogeneity between lesions regarding PTEN alterations, which correlated to response to PI3K inhibition Harris (1984) (<1970) (1972-83) (1973-86) (1960-79) Pts with CNS mets more likely to present with bone mets More liver and gynecological mets in young pts (n=55): ER and PgR downregulation in mets compared to prim. (n=6): CGH analysis: Prim differs from mets, but mets are similar Autopsy patients from our study Eligibility criteria: (1) Patients died from breast cancer; (2) Availability of FFPE tissue blocks from the primary breast tumor, a non-cancerous tissue as germline reference and at least one metastatic sample; (3) Minimum 30% tumor cellularity at central pathological review; (4) >1µg of dsDNA for from at least the primary breast tumor, a non-cancerous tissue as germline reference and at least one metastatic sample N=10 patients Brown et al. submitted Patients and samples Time between death and autopsy Average=2.8 days (range= 1.5 - 4.2) Nr of distant metastatic samples/pt Average=3 (range= 1-4) Nr of patients with multiple primary samples 7 (range= 2-8) Molecular subtype 5 ER-/HER2-, 2 HER2+, 3 ER+/HER2- Age at diagnosis 4 young patients (≤40), 3 between 40-60, 3 older patients >60 Histologic subtype 9 IDC and 1 ILC Treatment 2 treatment naive, 8 with systemic treatment (3/8 with neo-adjuvant treatment) 131 Strategy Substitutions/indels (mutations): 1/ Whole-exome sequencing 2/ Deep re-sequencing Copy number alterations (CNAs): Affymetrix Oncoscan FFPE Express 2.0 assay Brown et al. submitted Disease progression = common metastatic precursor METASTATIC CASCADE MODEL Brown et al. submitted Disease progression PARALLEL PROGRESSION MODEL Brown et al. submitted Questions 1. How do primary tumors progress to metastatic disease? 2. Is the primary/metastatic heterogeneity clinically relevant? 3. What about contralateral breast tumors? What about drivers/targets? Meric-Bernstam et al. Mol Can Ther 2014 Questions 1. How do primary tumors progress to metastatic disease? 2. Is the primary/metastatic heterogeneity clinically relevant? 3. What about contralateral breast tumors? Contralateral breast cancer (CBC) • CBC is more frequent in BRCA1/BRCA2 mutated or lobular tumors, as well in younger patients; • By convention, CBC is treated as a new primary tumor independent of the first cancer; • However, this assumption has until recently not been investigated in detail. Contralateral breast cancer (CBC) • Two recent studies now demonstrated that 1012% of CBC are actually not second primaries but metastases from the first cancer, and that in this case they are associated with worse survival. Alkner et al. BCR 2015, Klevebring et al. BCRT 2015 How does it come that we still do not know enough about metastatic disease? • Key element for these studies= adequate samples • So far: mainly primary tumor samples collected; • Longitudinal sampling and autopsy cohorts still very rare 5. Treatment induced changes Resistant clones might be unmasked by neo-adjuvant chemotherapy In 25% of the patients: identification of new subclones including oncogenic mutations; Most probably subclones which were already present (same evolutionary age). Yates et al. Nat Med 2015 Genomic alterations ESR1 1. Substitutions 2. Gene fusion events 3. Copy number gains Deroo, Korach JCI 2006 ESR1 mutations • Rare in primary tumor & frequent in metastases; • Presence linked to nr and duration of endocrine treatment; • Mutations (Y537/D538)= constitutive activation in absence of estrogen resistance to AIs & more important dose needed of Tamoxifen and Fulvestrant Y537/D538 mutations agonist conformation of ESR1 Toy et al. Nat Genet 2013 Alterations involving PTEN Convergent loss of PTEN leads to clinical resistance to a PI(3)Ka inhibitor Juric et al. Nat 2014 Messages • Breast cancer most frequently progresses via the metastatic cascade model; • Existence of primary/met heterogeneity, but also inter-met heterogeneity! • Treatment can induce emergene or increase of resistant genomic alterations. 6. Liquid biopsies Liquid biospies = détection du cancer dans le sang Crowley et al. Nat Rev 2013 In some tumors, cells disseminate early Klein C et al. Science 2008 CTC Enrichment Biological properties Antibody (EpCAM+) selection Physical properties Beads Microfluidics Size Leucocyte Depletion Density Protein Secretion Invasion Electrical Charges + -+ + -+ CTC detection: poor outcome in metastatic breast cancer OS 1.0 0.8 0.6 Level I evidence that CTC detection is associated with worse prognosis in MBC 0.4 0.2 Cohort CTC <5 CTC >=5 Patients 1033 911 Events 371 558 0.0 0 6 12 18 24 30 Months N= 177 pts, 49% (≥ 5CTCs) HR = 4.26 p<0.0001 Cristofanilli M et al. NEJM 2004 N= 1.944 pts, 47% (≥ 5CTCs) HR = 2.77 p<0.0001 Bidard FC et al. Lancet Oncology 2014 36 Liquid biospy: tracking of the disease In the early setting Acknowledgments BCTL: Christos Sotiriou Michail Ignatiadis Françoise Rothé Marion Maetens Hatem Azim Amir Sonnenblik David Brown Vinu Jose Sylvain Brohé David Venet Samira Majjaj Naïma Kheddoumi Ghizlane Rouas Delphine Vincent Floriane Dupont Bastien Nguyen Roberto Salgado (UZA) Jeanne Letor Dominique Roels Molecular Immunology: Karen Willard-Gallo Soizic Garaud Laurence Buisseret Gert Vanden Eynden (UZA) Pathology: Denis Larsimont Medicine dpt: Martine Piccart All patients & their families University of Milan: Elia Biganzoli Marco Fornili VIB-K.U.Leuven: Dominiek Smeets Thomas Van Brussel Diether Lambrechts Thierry Voet University of Genova: Gabriele Zoppoli Alberto Ballestrero Wellcome Trust Sanger Institute: Gunes Gundem Serena Nik-Zainal IEO: Lucy Yates Giancarlo Pruneri Peter J. Campbell Giuseppe Viale UCL: Christine Galant Marseille (IPC): François Bertucci IRST Elisabetta Pietri Dino Amadori University of Budapest: Borbala Szekely Marcell Szasz Janina Kulka
