OLOGICAL SCIENCES O DEPT. OF CLINICAL & BIO UNIVERSITY UNIVERSTY OF TORINO Prognostic and predictive biomarkers in early stage NSCLC Giorgio V. Scagliotti University of Torino Department of Clinical & Biological Sciences giorgio.scagliotti@unito.it Prognostic vs. Predictive Biomarkers Prognostic marker Indicates survival lbenefit/detriment t i tregardless of therapy Stage, tumor size, sex Lymphadenectomy Outcome improved with increasing number of recovered lymph nodes; plateau reached at 11 5 yr OS: 58% with lymphadenectomy vs 42% without (P <.0001) Predictive marker Predicts for differential benefit from a particular therapy Varlotto JM, et al. Cancer. 2009;115:851-858. 1
LACE : A Pooled Analysis of 5 Cisplatin based Adjuvant CT Trials Overall survival : absolute benefit of 54% 5.4% Survival according to type of death The absolute effect of chemotherapy at 5 years was a decrease of 6.9% for lung cancer death and an increase of 1.4% for non lung cancer death. Pignon JP et al. JCO 2008;26:3552 O Adjuvant CT ±post op RT, in operable NSCLC: two meta analyses of individual patient data 13 trials, 2660 patients HR 0.88 (95 CI : 0.81 0.97) P<0.009 34 trials, 8447 patients HR 0.86 (95 CI : 0.81 0.92) P<0.00010001 4% benefit 4% benefit NSCLC Meta analyses Collaborative Group Lancet 2010; 375:1267 2
O Long term Results of Pathological Stage I NSCLC 321 pathological stage I Mean FU time 61+/ 46 months Significant PFs at MV analysis: # nodes removed (< 15) HR 1.66 (1.12 2.38) Tumor size HR 1.10 (1.06 1.16) 16) Smoking history (< 20 packs/year) HR 1.61 (1.22 2.12) Wu YC et al. Eur. J. Cardiovasc. Surg. 2003;24:994 O Never Smokers with Stage I II Lung Cancer 1/99-12/02 AJCC stage I-II = 164 cases HR of death for smokers 2.22 (1.20-4.15) P=0.012 Toh C K et al. JCO 2006; 24:2245 3
OS: All OS: size 4cm CALGB 9633: Carbo/Tax as Adjuvant Therapy in Stage IB NSCLC DFS: All OS: size<4cm Strauss et al. J Clin Oncol 26: 5043 5051, 2008 T Descriptors in the TNM7 Staging System 15,234 patients with sufficient pt descriptor information (M0) When comparing overall survival between groups of patients defined by tumor size, we found that survival differences were optimized at size cutpoints of 2, 3, 5, and 7 cm. These tumor size cutpoints were chosen on the basis of pathologic measurements from completely resected cases in the learning set and were then tested in the remaining pathologic and clinical data. Rami Porta T et al J Thorac Oncol 2007;2: 593 602 4
Prognostic & Predictive Molecular Tools Human Genetics SNPs Haplotypes Sequencing Expression Profiling Specific transcript levels Total RNA profiling Proteomics Specific biochemical markers Protein profiling Importance of Molecular Staging Early-Stage NSCLC Treatment Protocol Stage IA+B: Observation Stage II-IIIA: Adjuvant therapy Diagnosis Tumor is resected and measured Who to treat? 27% of stage IA and 42% of stage IB patients recur and die Q: How to identify individuals at higher risk? 41% of patients with stage II NSCLC are cured with surgery alone and do not need adjuvant treatment Q: How to identify patients that can be cured by surgery alone? Q: How can patient selection among those given adjuvant therapy improve HR and cure rate? 5
Strategies for the Development of a Prognostic Signature Sotiriou C. NEJM 2009;360:790 Why Prognostic Signatures in Early Stage NSCLC? More effective than standard prognostic factors (tumor size, differentation, vascular invasion, tumor margins) in identifying high risk completely resected Stage I patients who might benefit from adjuvant chemotherapy. Able to identify stage II IIIpatients III who have a low risk of recurrence in the absence of adjuvant chemotherapy. 6
Lung Metagene Model Potti A. et al. NEJM 2006:335:570 O CALGB 30506 Schema (Stage IA/IB) Resection T (1.75 to 4.0) N0 Patients + Array N=1296 LM Score <0.55; 850 LM Score > 0.55; 446 Observation N=425 Randomize Adjuvant Chemotherapy N=425 Randomize Observation N=223 LM Scores Blinded to Investigators Adjuvant Chemotherapy N=223 7
Gene Expression Survival Prediction in Lung Adenocarcinoma : Validation Study All stages Stage I only All stages with covariates it Stage I only with ihcovariates Training-testing multi-institution validation study (UM,HLM,CAN/DF,MSK), 442 adenocarcinoma Eight Classifiers Shedden K. et al. Nature Med. 2008; 14::822 O Gene overlap between NSCLC prognostic signatures Overlap in genes of recent NSCLC survival signatures is limited to 5 of a total of 327 genes used Roepman P. et al. Clin. Cancer Res. 2009;15:284 8
Use of Microarrays in NSCLC Need for complicated methods. Large number of genes used in gene profilings. In most of the studies need of fresh tissue. Lack of both reproducibility and independent validation of the results. Genes varied considerably and only few genes have been consistently included. Gene expression profiles can vary according to the microarray platform and the analytic strategy used. Critical Assessment of Gene Expression Signature in Lung Cancer 16 studies published between Jan. 2002 and Feb 2009. Little evidence for any of the gene expression signatures to be ready for clinical application. Serious problems in the design and analysis of many of the studies. Guidelines needed. dd Focused study planning to address medically relevant questions Use of unbiased analysis methods Subramanian J & Simon R. JNCI 2010; 102: 464 9
Extratumoral Vascular Invasion is a Significant Prognostic Factor in Resected NSCLC Vascular invasion in Victoria blue van Gieson staining (B, intratumoral and D extratumoral) Shimada Y et al. J.Thor. Oncol. 2010; 5:970 O Genomic & Clinical Variables in a Breast Cancer Prediction Model Sotiriou C. et al. NEJM 2009; 360:790 10
Relevant Predictive Biomarkers and NSCLC Class Agent Biomarkers Robustness Cytotoxic drugs Cisplatin ERCC1 RRM1 BRCA1 ++ + + 21 Gemcitabine RRM1 + Pemetrexed Paclitaxel FPGS TS MAPtau Beta tubulin III Targeted therapies Erlotinib EGFR mutation FISH EGFR K Ras wt RASSF1A / 9pLOH Bevacizumab circulating VEGF + + + +++ + + + PF 02341066 EML4 ALK ++ LACE BIO: Biomarkers in NSCLC Results Prognostic and predictive values of p27, p16, and cyclin E for adjuvanttherapy therapy inresected NSCLCnot confirmed OS DFS Biomarker Prognostic, HR Predictive, P Value Prognostic, HR Predictive, P Value p27 0.97.83 NR NR p16 0.97.95 NR.79 Cyclin E 1.09.20 NR.21 Results emphasize importance of robust validation of potential biomarkers prior to clinical use Pirker R, et al. Chicago Multidisciplinary Symposium in Thoracic Oncology 2010. Abstract 6. 11
P53 and K ras in JBR 10 253 patients, 132 (52%) +ve for p53 overexpression P53 +ve tumors more frequent in males, squamous carcinoma and wild type ras P53 mutation and k-ras mutation neither prognostic or predictive Tsao MS et al. J. Clin. Oncol. 2007; 25:5240 BIO ALPI : p53, K ras, Ki67 Univariate Analysis Variable HR 95% CI P Value p53 (n=387) 1.03 0.98 1.08.19 K ras (n=108) 1.70 0.99 2.93.06 Ki67 (n=395) 1.05 0.99 1.12.14 Multivariate Analysis (adjustment by stage) Variable HR 95% CI P Value p53 (n=387) 1.03 0.98 1.08.30 K ras (n=108) 1.42 0.82 2.47.20 Ki67 (n=395) 1.02 0.95 1.08.63 Scagliotti GV et al. JNCI 2003; 95:1453 12
Gene Expression Modulation in 64 Cases of Resected NSCLC DNA Repair Genes Saviozzi S. et al. Cancer Res. 2009 O DNA Synthesis & Repair Genes RRM1 and ERCC1 in Lung Cancer (N=187) Automated quantitative determination of RRM1, ERCC1 & PTEN RRM1 & ERCC1 determinants of survival after surgical treatment in early stage NSCLC Zheng Z. Et al. NEJM 2007, 356:800 13
RRM1: A Prognostic and Predictive Biomarker for DFS RRM1 as a prognostic ost marker in surgically resected NSCLC [1] RRM1 as a predictive marker for chemotherapy survival in advanced NSCLC [2] 1. Bepler G, et al. J Clin Oncol. 2004;22:1878-1885. 2. Rosell R, et al. Oncogene. 2003;22:3548-3553. Probability Survival (%) 100 75 50 25 0 1.0 0.8 0.6 0.4 0.2 0 Resected NSCLC High RRM1 Low RRM1 Stage IV NSCLC gem/cis P =.002 0 20 40 60 80 100 120 Mos Low RRM1 High RRM1 P =.003 0 20 40 60 80 100 120 Mos ERCC1 in Early Stage NSCLC & Adjuvant CT N= 761; ERCC1 negative= 56%, benefit of adjuvant chemotherapy only in ERCC1 negative tumors (HR for death 0.65, CI95% 0.50 0.86). In surgery alone ERCC1 positive tumors survived longer NO DATA ACCORDING TO HISTOLOGY AND ERCC1 EXPRESSION Olaussen KA et al, NEJM 2006; 355:983 14
ERCC1 ITACA Adjuvant Trial Pharmacogenomics: Yes or No? High Low TS TS High Profile 4 Low High Control = investigators choice of cisplatin-based doublet Primary endpoint = OS; Sample size = 700 patients Profile 3 Profile 2 Low Profile 1 Taxanes Control Pem Control Cis/Gem Control Cis/Pem Control Disease segmentation based on oncogenic events From an organ based disease to a molecular classifications of rare diseases «Druggable» genomic alterations 30 15
IPASS shows prognostic & predictive value of EGFR mutations Modified by Wolf J, PeerView Press 2010 O % 100 80 60 40 20 0 2nd/3rd line Crizotinib in ALK + patients compared with Crizotinib naive controls WT/WT Control (n=125) ALK Positive (n=23) Median Survival 11 months 6 months 1-yr Survival 47% 44% 2-yr Survival 32% 44% 12% ALK Crizotinib (n=30) not reached 70% 55% from 2 nd /3rd line crizotinib HR=0.49, p=0.02 0 1 2 3 4 Years Shaw A et al. Presented at 2011 ASCO; Abstract 7507 16
Correlative studies Adjuvant Trials Great opportunity to understand the genomic landscape of mutated tumors through deep sequencing Clonal evolution biopsies at the time of relapse Novel technology development Consenting issues for correlative studies Heterogeneity of tumor cell dependency as the basis for resistance to therapeutics targeting oncogene addiction Sharma S V, Settleman J Genes Dev. 2007;21:3214 3231 17
Personalized Adjuvant Therapy (PAT) EGFR An Alliance Trial EGFR TKI Node Negative Randomization Powered for efficacy Resected I, II and III NSCLC (except T 1b N0) Mutations in EGFR TK Placebo EGFR TKI Primary Endpoint: Overall Survival Node positive Up to 4 cycles of platinum based chemo Randomization Placebo Arm A Arm B Arm C Arm D?? Grade Nodal status Stage Future Linking Genomic and Clinical Features Histology Tumor size Clinical Space Oncogenic Pathways EGFR Prognosis signatures Drug Sensitivity Genomic Space Cancer Biology Pti Patient toutcome Patient Outcome Stage EGFR Oncogenic Pathways Tumor Size Prognosis Hist Drug Sensitivity Cancer Biology Clinico Genomic Space Nodal Status REFINED PROGNOSIS And PREDICTION 36 18