1. Biomarkers in CRC: Are We Any Closer to Our Goal of Personalized Medicine? Lee M. Ellis, MD Depts. of Surgical Oncology and Cancer Biology U.T. M.D. Anderson Cancer Center Houston, Texas, USA

2. Characteristics of a Good Predictive Biomarker Consistent across numerous studies, lines of therapy, and numerous drugs within the same class Makes biologic sense Relatively easy assay with rapid turnaround time “Yes” or “no” answer –Objective, not subjective –Not subject to lab-to-lab variation

3. We Need to Start Thinking in Terms of “Pathways”, Not Individual Molecules Systems biology …… focuses on the systematic study of complex interactions in biological systems… (integration instead of reduction) to study them.….. Wikipedia HIF  LDH/Glut1/VEGF/R PI3K  PTEN  Akt

4. Intratumoral mRNA expression of genes involved in angiogenesis and HIF-1 pathway to predict outcome to VEGFR tyrosine kinase inhibitor (TKI) in patients enrolled in CONFIRM1 and CONFIRM2 P. M. Wilson, D. Yang, M. M. Shi, W. Zhang, C. Jacques, J. C. Barrett, K. Daneneberg, T. Trarbach, G. Folprecht, G. Meinhardt, H. J. Lenz

5. There Are No Known Predictive Biomarkers for Anti-VEGF Therapy Proposed BiomarkerOutcome Plasma VEGFNot Predictive Primary Tissue VEGF (ISH, IHC) Not Predictive Upstream Mediators of VEGF (ras, raf, P53) Not Predictive Other Angiogenic Mediators (TSP-2) Not Predictive Jubb et al. J Clin Oncol 24:217-227, 2006 Ince et al. J Natl Cancer Inst. 97(13):981-9, 2005 Holden et al. ASCO. Abstract: 3555, 2005 IFL +/- Bev Trial

6. Progression-Free Survival PTK/ZK (n=585) 9.1 mo Placebo (n=583) 7.7 mo HR 0.89 (95% CI) (0.78, 1.03) P-value 0.108 100 80 60 40 20 % PFS Time since randomization (months) 4812162024283203640 0 25 50 75 100 048121620 Time since randomization, mos 75 FOLFOX4 + PTK/ZK FOLFOX4 + Placebo PTK/ZK (n=426) 5.6 mo Placebo (n=429) 4.1 mo HR 0.83 (95% CI) (0.71, 0.98) P-value 0.026 PFS % CONFIRM-1CONFIRM-2 Analyses in this study done on 42-54 patient tumors/arm.

7. High LDH and Predictive Value of FOLFOX + PTK/ZK Overall Survival High LDH Progression Free Survival High LDH CONFIRM1CONFIRM1 CONFIRM2CONFIRM2 Premise of Study –LDH regulated by HIF, and serum LDH may be a predictive marker –Why not investigate genes in the HIF pathway as predictive markers for PTK/ZK efficacy?

8. Important Findings in This Study Expression levels of genes in a pathway may be associated with potential benefit of the addition of PTK/ZK to FOLFOX –Proof of principle (genomic profiling) However, there are divergent results between first line and second line therapy –Did this fail the validation test? or –Is this a reflection of the effect of chemotherapy on inducible genes? –Small numbers of patients Some groups n<50 (sometimes <10) Less than 10% of overall trial

9. PFS with PTK/ZK >HIF-1α 0.021 >LDHA 0.075 >VEGFR2 0.001 CONFIRM 1 Genep-valueGenep-value CONFIRM 2 <HIF-1α 0.021 <LDHA 0.031 11.3 v 7.6 mo 9.4 v 3.5 mo 7.6 v 2.7 mo 7.6 v 1.7 mo 8 v 4.1 mo

10. “The data is the data…you can’t change the data, change the hypothesis” Josh Fidler Unexpected results, such as the divergent data in CONFIRM-1 AND CONFIRM-2, provide an opportunity for the development of new hypotheses regarding potential biomarkers We must be flexible in our thinking, and be able to alter our hypotheses, and subsequent investigations i.e. does exposure to chemotherapy effect the HIF pathway? (topotecan can decrease HIF-1  : G. Mellilo NCI)

11. HIF-1       Regulation: Measuring mRNA May Not Give You the Whole Story IGFR/EGFR/HER2 PI-3-K Akt Transcriptional Regulation (increased mRNA and protein)  HIF-1  mRNA  HIF -1  Protein Hypoxic Regulation (increased protein) Decreased prolyl hydroxylation and ubiquitination Decreased proteosomal degradation HIF -1  Protein Gene transcription

12. There is Not Much Remaining Interest in PTK/ZK Will findings from this study translate to other drugs in the field Markers that may be predictive for one TKI, may not be predictive for another The larger the red circle, the more effective the drug is for the target Human Kinome Tree The Non-Specificity of VEGFR Kinase Inhibitors

13. Characteristics of a Good Predictive Biomarker Ras/EGFRHIF Pathway/ VEGF/R Consistent across numerous studies and numerous drugs within the same class Yes, and consistent upon lines of therapy TBD, not consistent upon lines of therapy Relatively easy assay with rapid turnaround time YesKind of…. Makes biologic senseYes? “Yes” or “no” answer Objective, not subjective Not subject to lab-to-lab variation YesNo

14. Is Gene Expression Profiling IN THIS STUDY Any Better Than Measuring Serum LDH?

15. Is This Enough to Proceed With a Phase III Trial Selecting for Patients By Evaluating HIF Pathway? NO –Divergent results in CONFIRM-1 and CONFIRM-2 require explanation –There are better drugs with better PK that are already in late phase clinical trials –However, tissue should be banked on all Phase III trials for study of potential markers of efficacy of anti-VEGF therapy This is “hypothesis generating” data, not validated data We must continue to search for predictive markers of anti-VEGF therapy

16. ASCO 2005 Plenary Session Here we Here we are again….3 years later and the results for CONFIRM-1 have not yet been submitted for publication. We when enroll patients in clinical trials, they put their trust in us to use their experience to advance our knowledge. We owe it to our patients to publish data from clinical trials in a timely fashion regardless of whether the trial was negative or positive!!! Disclaimer: This is not a reflection on most of the current authors on this study Published? Yes No Yes No Yes CONFIRM-1

17. Evaluation of PTEN expression in colorectal cancer (CRC) metastases (mets) and in primary tumors as predictors of activity of cetuximab plus irinotecan treatment F. Loupakis, L. Pollina, I. Stasi, G. Masi, N. Funel, M. Scartozzi, Petrini I, Santini D, Cascinu S, Falcone A

18. Activation of Downstream Pathways and EGFR MoAB Resistance 18BMS-ASCO-CRC-Therapy-Ellis Camp et al, Clin Cancer Res 2005 Src Act PI-3K Akt Mut Nucleus EGFR MoAB EGFR Fak Ras Mut Raf Mut MEK Erk PTEN Mut /Inactivaed Survival Pathway 40% CRC mutated Proliferative Pathway 40% CRC mutated Mut

19. PTEN Overview PTEN is a phosphatase that blocks activation of the Akt/survival pathway –Loss of function of PTEN is associated with an increase in cell survival signaling Leslie, Downes Biochem J 2004

20. Mechanisms of Regulation of PTEN Protein Levels Mutation Loss of heterozygosity Methylation of promoter –decreases transcription Certain oncogenes can downregulate transcription

21. Important Findings in This Study PTEN protein expression in primary tumor does not correspond to PTEN expression in mets –PTEN in mets is predictive of response whereas PTEN in the primary is not Kras mutational status in the primary does correlate with mutational status in the metastasis The biology of these observations makes sense –Mutational status should not change (Ras) –Methylation status is subject to regulation by the microenvironment (PTEN)

22. Logrank Test: p=0.001 HR = 0.42 95% CI: 0.17-0.65 PTEN+/KRAS wt All others Logrank Test: p=0.005 HR = 0,49 95% CI: 0.20-0.75 PTEN+ PTEN- PTEN/RAS and PFS Can We Better Predict Efficacy with a Multifactorial Analysis? Sorry…slides from the investigators..need their permission to use

23. British Journal of Cancer (2007) 97, 1139-1145.

24. “Whereas neither EGFR protein/mRNA expression nor gene copy number correlated with cetuximab response, examination of the mutation status of signaling components downstream of EGFR showed that cell lines with activating PIK3CA mutations or loss of PTEN expression (PTEN null) were more resistant to cetuximab than PIK3CA wild type (WT)/PTEN-expressing cell lines”

25. Characteristics of a Good Predictive Biomarker Ras/EGFRPTEN/EGFR Consistent across numerous studies and numerous drugs within the same class YesProbably Relatively easy assay with rapid turnaround time Yes Makes biologic senseYes “Yes” or “no” answer Objective, not subjective Not subject to lab-to-lab variation YesNo (IHC staining) 123 Intensi ty 1+2+3+ % +cells 0-2525-50>50 >4 positive

26. Predictive Oncology in mCRC: What Have We Learned Today? VEGF/R Targeting Gene expression profiling of the HIF pathway may be predictive for efficacy of PTK/ZK + FOLFOX –“Hypothesis generating” study Divergent results in CONFIRM-1 and -2 are confusing –Investigations of gene expression profiling with other VEGF/R inhibitors are essential in determining the true value of this approach –We must continue the search for predictors of VEGF/R efficacy EGFR Targeting Ras continues to predict for EGFR MoAB efficacy PTEN levels in metastases predicts for efficacy of cetuximab –PTEN in primary tumors is not predictive Proteins that can be regulated by the tumor microenvironment may require biopsy when determining value as a predictive marker Multi-factorial analyses may better predict for efficacy

27. After ASCO 2008, it is time to update recommendations for CRC!!

28. Thank You for Your Interest! Enjoy ASCO 2008