When and how to implement biomarker for cancer care from a French perspective. Fabrice ANDRE Institut Gustave Roussy Villejuif, France

Transcription

1 When and how to implement biomarker for cancer care from a French perspective Fabrice ANDRE Institut Gustave Roussy Villejuif, France

2 Outline Levels of evidence before implementing a biomarker to the daily care Technical issues Models for implementation of genomic tests at the nationwide level Conclusion / perspectives: Learning cohorts Implementation of NGS

3 EGAPP criteria Has the use of biomarker changed the outcome? Randomized trials

4 Prospective trials testing biomarkers in the breast cancer field With the exception of few illustrative trials (TailorX, MINDACT, RxPONDER) Clinical trials testing biomarkers are not existing in the field of breast cancer What are the other options to validate biomarkers? Dogan, Breast cancer Res Treat, 2013

5 Retrospective studies: Simon-Hayes level of evidence scale Level 1 Consistent Retrospective analyses could be acceptable to define a Level Ib evidence and biomarker implementation RM Simon, S Paik, DF Hayes. JNCI, 201:1-7, 2009

6 Limitations of retrospective studies: Publication biais Prognostic Value P53 in H&N cancers «consistent studies suggesting prognostic Value for p53» Lack of statistical significance when unpublished data are added Need to capture and collect all biomarker Studies: WIN registry (Andre, Nature Rev Clin Oncol, 2011) Kyzas P A et al. JNCI J Natl Cancer Inst 2005;97:

7 Limitations of retrospective studies: Lack of precise quantification of biomarker effect Retrospective analyses are not powered enough to detect small differences

8 Early integration of biomarkers during phase I/II trials Dogan, Breast Cancer Res Treat, 2013

9 Take Home message Randomized trials testing medical usefulness of biomarker is the gold standard but they are not feasible in most of the cases Retrospective analyses of randomized trials provide information about clinical validity but not clinical utility Some biomarkers for targeted therapies are developed as companion diagnostic as soon as phase I trials, but this remains rare Need to find a new model for biomarker validation, when not feasible in phase I/II trials: Quick access to personalized medicine for patients Precise evaluation of the medical usefulness

10 Solutions: Temporary use of biomarkers in validation cohorts Large cohorts of prospective validation and implementation with reimbursement of the test Registration of patients with registries Light CRFs Coordination to assure high quality Allows: Quick access to the innovation for patients Validation of the test without requiring randomized trials De-reimbursement if the test is not valid Inclusion in databases for research use

11 Outline Levels of evidence before implementing a biomarker to the daily care Technical issues Models for implementation of genomic tests at the nationwide level Conclusion / perspectives: Learning cohorts Implementation of NGS

12 Issues in testing and interpretation Are preanalytical steps compatible with the test? Need to homogenize preanalytical steps (MINDACT trial) Analytical validity? Across time? Antibodies specificity, Batch effect (PgR, Ibrahim, Am J Clin Pathol 2008) Issues around cut-off Should grey zones be more frequently used for biomarkers with normal distribution? (Goubar, ASCO, 2012) Inter-center reproducibility / centers of expertise Coordination by authorities Centralized analysis Need to homogenize processes and limit the variability: Centers of expertise, coordination and single stable technology

13 Issues in logistics Too many tests using too many technologies Need to develop «all-in-one» solutions

14 Solution: Automated, stable high through technologies Next generation sequencing NanoString technologies

15 Outline Levels of evidence before implementing a biomarker to the daily care Technical issues Models for implementation of genomic tests at the nationwide level Conclusion / perspectives: Learning cohorts Implementation of NGS

16 Genomic tests: Models of implementation Private Biomarker companies Test reimbursed (ex: Genomic Health) Public Hospitals Funding based on nb of tests (ex: French model) Private Biomarker companies Test not reimbursed (ex: NGS facilities)

17 Implementation of genomics in France: A network of 28 public genomic centers 28 genomic centers Deliver results for clinical decision tests in 2010 Yearly funding from French NCI and Ministry of Health (12 M Euros in 2010) Coordination of the 28 centers for quality Control, SOP Molecular Epidemiology through single database Added value: equal access to genomic tests for the whole population Keep the «know how-to» in the hospitals Generate large molecular epidemiology database Andre F, Clin Cancer Res, 2012, Nowak Nature Rev Clin Oncol, 2012

18 Private companies and genomic testings for targeted therapies Three prerequisites / issues: Transparent information about level of evidence for efficacy Data sharing and owning Drug access is a major issue Risk: generates a model where patient will have to pay for investigational drugs

19 Conclusion: Biomarker Implementation Need for new models of biomarker validation / implementation that would allow quick access to innovation while validating biomarker: COHORTS Need for coordination and stable technology Multiplicity of tests is not sustainable Need to develop «all-in-one» approaches: High throughput technologies Model of genomic test implementation at the level of society should: Allow access to innovation for the majority of citizens While not neglecting the freedom to access newer innovation to individuals Nationwide genomic programs allow: Wide access and Molecular Epidemiology database NGS by private companies without coordination: Is acceptable, in societies where health is part of the market, pending transparency about the level of evidence Does not allow datasharing, research information from patient Could distort the field of drug market