USCAP 2013: Clinical Implementation of Molecular Testing for Targeted Therapy of Breast Cancer Antonio C. Wolff, MD, FACP, FASCO Professor of Oncology, Johns Hopkins Kimmel Cancer Center Adjuvant therapy changed the natural of early stage breast cancer. Starting in the 1970s, randomized trials have shown a survival benefit from adjuvant chemotherapy in operable, node-pos disease. Since 1985, the Oxford Overview led the rapid adoption of adjuvant chemotherapy (for ER-neg disease) and tamoxifen (for ER-pos disease), and this trend accelerated after the 1988 NCI Clinical Alert on adjuvant chemotherapy in node-neg disease. However, this was not without controversy. In 1990, an NIH Consensus Development Conference concluded that some degree of improvement (from adjuvant therapy) may be so small that they are outweighed by the disadvantages of therapy. Soon, adjuvant chemotherapy also became a standard in ER-pos, node-neg disease. Starting in the mid 1990s, data began to suggest that not all patients with ER-pos disease benefitted from chemotherapy. Decision algorithms based on routine clinicopathologic factors (eg, TNM, grade, ER, and HER2) proved quite useful for decision-making for the average patient. But by the early 2000s, new molecular tools to stratify recurrence risk (prognosis) and likelihood of benefit from chemotherapy (prediction) became available in clinical practice to further individualize clinical decisions. Prospective-retrospective studies tested the prognostic/predictive utility of Oncotype DX. Soon, it became clear that it and routine clinicopathologic parameters offered independent prognostic utility in ER-pos disease. The TailoRx (node-neg) and RxPONDER (node-pos) trials are now prospectively testing Oncotype DX to guide the decision to add chemotherapy to endocrine therapy in patients with a lower/intermediate recurrence score. Mammaprint became the first FDAcleared IVDMIA assay, and the MINDACT study directly addresses this predictive utility question by having randomized patients with a discordant result (compared to Adjuvant! Online) also receive chemotherapy or not. Patients with HER2-pos disease are offered HER2-directed therapy plus chemotherapy. Consequently, a clinical decision gap exists in triple-neg disease (TNBC), which affects 50,000 women just in the US each year. Decision algorithms for them are exclusively based on clinicopathological factors like TNM, grade, and HER2, as molecular assays like Oncotype DX and MammaPrint have no role in this setting. New molecular assays based on DNA methylation, immune markers, or other gene expression signatures are under development. In the meantime, adjuvant chemotherapy is routinely offered to all TNBC patients with at least T1b tumors, despite the fact that most patients with ER-neg, node-neg disease remain disease-free long-term when treated with just locoregional therapy. Various studies examined clinical decisions based on standard clinicopathologic parameters alone or with knowledge of molecular assay results. However, data beyond costs of care are needed to assess their independent and complementary role to improve the most important clinical outcome to patients (i.e., survival). Until then, access to accurate routine clinicopathologic markers for all patients worldwide remains critical to ensure the most optimal outcome to all patients in high and low income countries.
Clinical Implementation of Molecular Testing for Targeted Therapy of Breast Cancer Antonio C. Wolff, MD, FACP, FASCO Professor of Oncology Breast Cancer Program awolff@jhmi.edu @awolff 102 nd USCAP Annual Meeting March 2013 This presentation is the intellectual property of the presenter; please contact him for permission to reprint and/or distribute
How Do Clinicians Make Decisions About Adjuvant Therapy? 1980-2005 Before Overview/Since Overview Chemo decisions based on anatomy: Nodal status Tumor size (add endocrine Rx if ER+) Since 2005 Post Overview? Greater focus on biologic subtype: ER+/HER2-, HER2+, or Triple negative
Breast Cancer Phenotypes Alone Not Enough ER+ 65% 75% All Breast Cancer HER2+ 15% 20% Triple Neg 15% Accurate predictive biomarker testing is critical as current phenotypes determine what type of therapy to give (strong NPV) However, they do not tell us who will benefit (modest PPV) Nor do they alone tell us in absolute terms whether to give it (modest prognostic utility)
Oxford Overview 2010 Main Findings per the Authors In all meta-analyses involving taxane or anthracycline regimens, proportional reductions in early recurrence, any recurrence, and breast cancer mortality were largely independent of age, nodal status, size, differentiation (high or intermediate; relatively few low grade), or ER status (ER-poor or ER+) Even in strongly ER+, chemo did at least somewhat affect outcome, although not necessarily to exactly the same extent as in less strongly ER+ In other words, Overview principles do still apply EBCTCG, Lancet 379:432, 2012
A Simplified View of Breast Cancer Taxonomy Nottingham Grading System TNM for tubule formation, nuclear pleomorphism, and mitotic count Add tumor size and node status Nottingham Prognostic Index Current Practice Add HER2 status, perhaps add LVI Add ER status Adjuvant! Online NCCN and St. Gallen Guidelines: As many as 60% of early stage breast cancer patients considered for adjuvant chemo, even though only a small number might benefit from it
Prognostic Molecular Classification RFS and OS Pratt and Perou, 2011
Semi-Unsupervised Gene Expression Array Analysis of a Cohort of Breast Cancers Identifies Several Intrinsic Subtypes normal-like claudin-low luminal A luminal B HER2-enriched basal-like Can we reconcile with traditional phenotypes? a) ER+: proliferation may help differentiate luminal A vs B adapted from L. Carey, Clin Breast Cancer 10:188,2010
Intrinsic Molecular Classification of Breast Cancer Intrinsic Subtypes Basal-like IHC markers Mostly triplenegative (not always!) Proliferation & Grade High Ki67, high grade Luminal A Mostly ER+ Luminal B Mostly ER+ Low Ki67, low grade Often high Ki67, high grade HER2+ HER2 over expressed High Ki67 Perou et al, Nature 2000;406:747-52
Intrinsic Molecular Classification of Breast Cancer Intrinsic Subtypes IHC markers Proliferation & Grade Rx Implications Basal-like Mostly triplenegative (not always!) High Ki67, high grade Worse natural history, quite responsive to chemo (eg, preop chemo & pcr) Luminal A Mostly ER+ Luminal B Mostly ER+ HER2+ HER2 over expressed Low Ki67, low grade Often high Ki67, high grade High Ki67 Indolent, responsive to endocrine Rx Less/unresponsive to endocrine Rx, more responsive to chemo Worse natural history, quite sensitive to anti- HER2 Rx Perou et al, Nature 2000;406:747-52
Treatment Modality Indication Comments Endocrine therapy Anti-HER2 therapy Chemotherapy Threshold for Rx Modalities St Gallen 2009 Any ER staining b ASCO/CAP HER2-positive (>30% intense and complete staining (IHC) or FISH >2.2+) b ER-negative, PgR-positive is probably artefactual 73 May use clinical trial definitions A. In HER2-pos disease (with anti HER2 therapy) Trial evidence for trastuzumab is limited to use with or following chemotherapy b B. In triple neg disease Most patients b, c Combined endocrine therapy + anti-her2 therapy without chemotherapy in strongly ER-positive, HER2-positive is logical but unproven No proven alternative. Most at elevated risk C. In ER-pos, HER2-neg disease (with endocrine Rx) Variable according to risk b See next table *Most factors are continuous but a binary decision needs to be made at some level Goldhirsch et al, Ann Oncol 2009 10
Chemoendocrine Rx if ER-pos/HER2-neg St Gallen 2009 ER, PgR Clinicopathological Features Relative Indications for Chemoendocrine Rx Lower ER and PgR level Factors Not Useful for Decision Relative Indications for Endocrine Rx Alone Higher ER and PgR level Histological Grade Grade 3 Grade 2 Grade 1 Proliferation a High Intermediate Low Nodes Peritumoral Vasc. Invasion (PVI) Node positive (4 or more involved nodes) Presence of extensive PVI Node positive (1-3 involved nodes) Node negative Absence of extensive PVI pt-size > 5cm 2.1 5 cm 2cm Patient Preference Use all available treatments Multi-Gene Tests Avoid chemotherapyrelated side effects Gene Signature b High score Intermediate score Low score Goldhirsch et al, Ann Oncol 2009 11
Chemoendocrine Rx if ER-pos/HER2-neg St Gallen 2009 Footnotes: a) Conventional measures of proliferation include assessment of Ki67 labelling index (e.g. Low = 15%; Intermediate = 16-30%; High = > 30%) and pathological description of the frequency of mitoses. The reliability of these measures will vary in different geographic settings. First generation genetic signatures contain genes sampling the ER, HER2 and proliferative pathways. Metaanalysis suggests that much of the prognostic information in these signatures resides in their sampling of proliferative genes, but their respective total scores may be the only form in which information is provided at present, and could be used in this component of assessment of relative indications for chemotherapy. b) The Panel agreed that validated multi-gene tests, if readily available, could assist in deciding whether to add chemotherapy in cases where its use was uncertain after consideration of conventional markers 2011 modification: if Ki67 low (<14%, cut-point based on comparison with PAM50) or an alternative measure of proliferation = Luminal A
International Ki67 in Breast Cancer Working Group A BIG/NABCG Collaboration Comprehensive recommendations on preanalytical and analytical assessment, and interpretation and scoring of Ki67 were formulated and geared toward achieving a harmonized methodology, create greater between-laboratory and between-study comparability, and allow earlier valid applications of this marker in clinical practice. Please attend oral presentation S4-6 An international Ki67 reproducibility study by Nielsen et al on Thursday Dec 6 @ 4:30 pm Dowsett et al, JNCI 103:1, 2011
Commercially Available Prognostic Multigene Signatures in Breast Cancer Reis-Filho and Pusztai, Lancet, 378:1812, 2011
Adoption of Gene Expression Profiling and Chemo Use in NCCN 17 sites (n = 7,375) 2006 2008 p value GEP 14.7% 27.5% < 0.01 Use of Chemo 53.9% 47% < 0.01 Lower odds of testing: Black vs White race (OR 0.70; 95% CI 0.54-0.92) High school or less education vs more (OR 0.63; 95% CI 0.52-0.76) Testing was associated with lower odds of chemo use OR 0.70; 95% CI, 0.62 to 0.80 In small LN-neg cancers, testing associated with higher odds of chemo use OR 11.13; 95% CI 5.39-22.99 In large LN-neg or any LN+ cancers, testing was associated with lower odds of chemo use OR 0.11; 95% CI 0.07-0.17 Is the test being ordered to confirm the clinician s bias? Hassett, JCO 30:2218, 2012
Adoption of Gene Expression Profiling and Chemo Use in NCCN Hassett, JCO 30:2218, 2012
Oncotype DX and Standard Path Measures E2197 Goldstein et al, JCO 26:4063, 2008 ATAC Dowsett et al, JCO 28:1829, 2010 Appear to offer complementary and independent prognostic info
Completed Trials TailoRx # MINDACT* Patients LN-/ER+ (10,500) LN-/ER+ or (6,000) Assay Oncotype DX MammaPrint Tissue FFPE Fresh Frozen Randomized Group Randomization Observation Group RS 11-25 (40%) Endocrine ± chemo RS < 11: endocrine RS > 25: both Discordant w/ Adjuvant! Online - high/low or low/high (32%) Rx choice based on clinical vs genomic risk Both low risk: endocrine or nothing Both high risk: chemo ± endocrine # Trial Assigning IndividuaLized Options for Treatment (Rx) *Microarray for Node-Negative Disease may Avoid Chemotherapy
Time-Dependent Effects of Adjuvant Rx SEER 1992 2007 Data on long term follow-up may not have been long enough for ER+ disease Jatoi I et al, J Clin Oncol 2011;29:1
NSABP B-13 Local Rx ± Adjuvant Chemo in ER-neg After 14y of observation (no chemo) of patients with stage 1 ER-neg breast cancer: ~ 49% disease-free ~ 69% alive Fisher, JNCI Monograph 30:62, 2001
Semi-Unsupervised Gene Expression Array Analysis of a Cohort of Breast Cancers Identifies Several Intrinsic Subtypes normal-like claudin-low luminal A luminal B HER2-enriched basal-like Can we reconcile with traditional phenotypes? a) ER+: proliferation may help differentiate luminal A vs B b) HER2-enriched:? c) Basal-like:? Luminal and HER2 clusters are low, usually ER/PgR/HER2 neg by IHC, and commonly described as TNBC
Microarray vs Clinical Phenotype Triple-negative, but not basal ~ 25% other subtypes IHC triple-negative vs basal-like ~ 75% concordance gene array Basal, but not triple-negative ~ 25% ER+, PgR+, or HER2+ on clinical assays When we talk about TNBC, we are mostly (but not entirely) talking about basal-like molecular subtype. However, ER/PgR-neg tumors have clinical heterogeneity suggesting different underlying molecular subtypes. Adapted from T. Traina
What is Triple Negative Disease? Metzger-Filho et al, JCO 30:1879, 2012
TCGA Molecular Portraits Phenotype, (Epi)Genotype, and Proteomic Features Subtype Luminal A Luminal B Basal-like HER2 ER+/HER2-87% 82% 10% 20% HER2+ 7% 15% 2% 68% TNBC 2% 1% 80% 9% TP53 pathway TP53 mut 12%; MDM2 gain 14% TP53 mut 32%; MDM2 gain 31% TP53 mut 84%; MDM2 gain 14% TP53 mut 75%; MDM2 gain 30% PIK3CA/PTEN pathway PIK3CA mut 49%; PTEN mut/loss 13%; INPP4B loss 9% PIK3CA mut 32% PTEN mut/loss 24% INPP4B loss 16% PIK3CA mut 7%; PTEN mut/loss 35%; INPP4B loss 30% PIK3CA mut 42%; PTEN mut/loss 19%; INPP4B loss 30% RB1 pathway Cyclin D1 amp 29%; CDK4 gain 14%; low expression of CDKN2C; high expression of RB1 Cyclin D1 amp 58%; CDK4 gain 25% RB1 mut/loss 20%; cyclin E1 amp 9%; high expression of CDKN2A; low expression of RB1 RB1 mut/loss 20%; cyclin E1 amp 9%; high expression of CDKN2A; low expression of RB1 mrna expression High ER cluster; low proliferation Lower ER cluster; high proliferation Basal signature; high proliferation HER2 amplicon signature; high proliferation Copy number Most diploid; many with quiet genomes; 1q, 8q, 8p11 gain; 8p, 16q loss; 11q13.3 amp (24%) Most aneuploid; many w/ focal amp; 1q, 8q, 8p11 gain; 8p, 16q loss; 11q13.3 amp 51%; 8p11.23 amp 28% Most aneuploid; high genomic instability; 1q, 10p gain; 8p, 5q loss; MYC focal gain 40% Most aneuploid; high genomic instability; 1q, 8q gain; 8p loss; 17q12 focal DNA mutations PIK3CA 49%; TP53 12%; GATA3 14%; MAP3K1 14% TP53 32%; PIK3CA 32%; MAP3K1 5% TP53 84%; PIK3CA 7% TP53 75%; PIK3CA 42%; PIK3R1 8% DNA methylation - Hypermethylated phenotype for subset Hypomethylated - Protein Expression High oestrogen signalling; high MYB; RPPA reactive subtypes Less oestrogen signalling; high FOXM1 and MYC; RPPA reactive ssubtypes High expression of DNA repair proteins, PTEN and INPP4B loss signature (pakt) High protein and phosphoprotein expression of EGFR and HER2 The Cancer Genome Atlas, Nature, 490:61, 2012
A Model for Breast Cancer Development over Molecular Time Chromosome instability and cancer-specific signatures Molecular Time International Cancer Genome Consortium, Cell 149:994, 2012
Intratumoral Mutational Heterogeneity Gerlinger et al, NEJM 366:883, 2012
Non-Small Cell Lung Cancer From Histology to Driver Mutations Pao and Girard, 2011
ASCO Tumor Marker Guidelines Breast Cancer ER/PgR to decide if to offer endocrine therapy HER2 to decide if to offer anti-her2 therapy 21-gene score and upa/pai-1 for prognosis Colorectal cancer KRAS mutations to select patients for EGFR antibody Rx Lung cancer ALK translocations to select patients for crizotinib EGFR mutations to select patients for TKI therapies
HER2 Measures by non-ihc/non-ish Methods 1. More difficult to distinguish between HER2+ and HER2- tumors in ER+ tumors (ie, those that would be sent for Oncotype DX) 2. Prospective trials showed benefit from trastuzumab in patients originally identified as HER2+ by IHC and FISH 3. Correlations between HER2 mrna and benefit from trastuzumab await testing ER-positive ER-negative Pinhel et al, BCRT 14:R46, 2012
What to Use Beyond Routine Path Markers for ER-pos Disease? PAM50: Evaluation of prognostic utility ongoing Mammaprint: FDA-cleared as a prognostic marker Predictive utility awaits MINDACT BIG3-04 trial Oncotype DX: Prognostic utility in various settings Predictive utility in small node-neg NSABP B20 and smaller node-pos SWOG 8814 prospective-retrospective studies Prospective data from TailoRx and RxPONDER trials
What Should Doctors Do For Now? Greater attention to preanalytical (cold ischemia) and analytical issues (assay validation/proficiency testing) This is a BIG deal! Standard and molecular assays may be complementary Clinical context matters Guideline-based use is appropriate First generation of molecular assays (proliferation) may add little to patients in the grey zone Functional assays of pathway activation? -Omics? Mutational analyses? Molecular assays must not yet override results with standard assays that were used in clinical trials for correlation with outcome Would anyone recommend against adjuvant tamoxifen if the tumor is ER-pos by IHC but ER-neg by mrna?
CAP Laboratory Accreditation Program No. of Labs 1400 1200 1000 800 600 400 200 Predictive Marker Proficiency Testing Publication of ASCO/CAP ER guideline Publication of ASCO/ CAP HER2 guideline 0 2004 2005 2006 2007 2008 2009 2010 2011 HER2 IHC 125 156 188 659 907 1000 1050 1150 HER2 FISH 174 191 210 263 295 320 315 317 ER IHC 97 139 168 233 283 370 550 1276 Wolff, Hammond, and Hayes, JNCI 104:957, 2012
Inflection Point Predictive Quantitative Prediction What is/are the risk(s)? What will happen? What is it? What has happened? Diagnostic Descriptive Diagnostics Qualitative Quantitative Adapted from Matt Taylor, U Colorado
Toward Precision Medicine IOM 2011
A New World Order? Do not worry about your difficulties in Mathematics. I can assure you mine are still greater. - Albert Einstein (1879-1955) In Mathematics you don't understand things. You just get used to them. - Johann von Neumann (1903-1957) Perhaps New, But Old World Rules Still Apply
Shall We Speak the Same Language? Analytic Validity The analytic accuracy, reliability, and reproducibility related to the marker assay or test in hand Clinical Validity The test has a suitably strong association with a clinical outcome of interest Clinical Utility If use of the marker test to direct patient care has been shown to result in a favorable balance of benefits to harm, leading to improved outcomes compared with non use of the marker test Improvement in outcome may relate to overall survival, disease-free survival, quality of life, or cost of care Marchionni, Wilson, Wolff et al, Ann Intern Med 148:358, 2008 Teutsch, Bradley, Palomaki et al (EGAPP), Genet Med 11:3, 2009 McShane and Hayes, JCO 30:4223, 2012
Bioinformaticians Pathologists The New Breast Cancer Black Box Oncologists Molecular Biologists Evolution of Translational Omics: Lessons Learned and Path Forward IOM Report, March 2012 BRISQ Criteria REMARK Criteria
Clinical Implementation of Molecular Testing for Targeted Therapy of Breast Cancer Antonio C. Wolff, MD, FACP, FASCO Professor of Oncology Breast Cancer Program awolff@jhmi.edu @awolff 102 nd USCAP Annual Meeting March 2013 This presentation is the intellectual property of the presenter; please contact him for permission to reprint and/or distribute
EXTRA SLIDES
Shall We Speak the Same Language? Evidence-Based Medicine The process of synthesizing available research evidence to draw conclusions about how to address specific clinical problems Comparative Effectiveness Research Encompasses broader considerations like cost and feasibility, and places a higher emphasis on realworld settings and recognizing the impact of health care decisions at both the individual and population levels Teutsch et al, Genet Med 11:3, 2009 McShane and Hayes, JCO 30:4223, 2012
EBCTCG 2010 Data in 2005-2010 for all RCTs started between 1973-2003 of: 1. Taxane vs non-taxane (33 trials, 1994-2003) 2. Any A vs std/near std CMF (20 trials, 1978-1997) 3. Higher vs lower dose A (6 trials, 1985-1994) 4. Polychemo vs no chemo (64 trials, 1973-1996) Lancet 379:432, 2012
EBCTCG 2010 Data in 2005-2010 for all RCTs started between 1973-2003 of: 1. Taxane vs non-taxane (33 trials, 1994-2003) 2. Any A vs std/near std CMF (20 trials, 1978-1997) 3. Higher vs lower dose A (6 trials, 1985-1994) 4. Polychemo vs no chemo (64 trials, 1973-1996) Lancet 379:432, 2012
EBCTCG 2010 Data in 2005-2010 for all RCTs started between 1973-2003 of: 1. Taxane vs non-taxane (33 trials, 1994-2003) 2. Any A vs std/near std CMF (20 trials, 1978-1997) 3. Higher vs lower dose A (6 trials, 1985-1994) 4. Polychemo vs no chemo (64 trials, 1973-1996) Lancet 379:432, 2012
EBCTCG 2010 Take Home Message While awaiting the results of these new trials (MINDACT, TAILORx, and RxPONDER), it appears that ER status, differentiation, and the other tumour characteristics available for the present meta-analyses had little effect on the proportional risk reductions with taxane-based or anthracycline-based regimens... Although nodal status and tumour diameter and differentiation are of little relevance to the proportional risk reductions produced by such chemotherapy (and by tamoxifen therapy), they can help in treatment decisions as they are strongly predictive of the absolute risk without chemotherapy, and hence of the absolute benefit that would be obtained by a one-third reduction in that risk.
Therapeutic Targeting of the Hallmarks of Cancer Hanahan and Weinberg, Cell 144:668, 2011