Methodological issues ( som e) relating to patient selection- CHMP - PowerPoint PPT Presentation

Methodological issues ( som e) relating to patient selection- CHMP Reflection paper K Prasad, MD, FRCP MHRA Sc Scienc ience to Clinic workshop; Oc e to Clinic workshop; Oct 2012 t 2012

Overview • Background • Personalised medicine and Regulation • Why do we need a Guidance /RP? • So what does the paper say? • What next? Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Biomarkers, patients and Regulation BMs Com m only asked Questions repres represent an additional test burden nt an additional test burden •  How do we qualify/ validate the How do we qualify/ validate the Need qualificatio Need qualification and validatio n and validation • marker marker  Level of evidence required for a Level of evidence required for a • Need context of the BM. Need context of the BM. BM/PGBM in regulatory te BM/PGBM in regul tory terms? rms? ( Why, When and How to replace ‘old’ Why, When and How to replace ‘old’ with ) ith )  Are RCT’s Are RCT’s mandatory andatory for BM ?? for BM ?? “new” ? “new” ?  When will retrospective data be When will retrospective data be • Need ‘reliability’ Need ‘reliability’ and nd acceptable? acceptable? ‘discriminatory ability’ ‘discriminatory ability’  BMs as Surrogate end points BMs as Surrogate end points • Need cut off points?! Need cut off points?! Scienc Sc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Regulatory Objective! Regulatory Objective! Evaluation & recommendations need to be Evaluation & recommendations need to be Evaluation & recommendations need to be Evaluation & recommendations need to be Scientifically robust • • Clinically relevant • Practical to implement …. ….to enable Safe Safe and Effective Effective use of the medicinal i.e., influence clinical practice influence clinical practice product i.e., Scienc Sc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

W hat are w e need w ith BMs No biomarker effect Predictive biomarker probability Survival probability T Survival Predic tive TT C effect C T C C Survival time Survival time Blue Line --Marker (–) Blue Line --Marker (–) Red Line --Marker (+) Red Line --Marker (+) Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Genomic BMs, scope of inclusion in labelling. Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Reflection Paper Aims; • to identify products where genomic biomarkers influenced the development or raised questions • to identify and streamline regulatory expectations • to discuss the CHMP experience and concerns Scope; Restricted mainly to centralised MAAs and other major safety issues discussed at CHMP relating to genomic biomarkers So; It is not a detailed review It is not a detailed review of literature on method of literature on methodologies of genomi ologies of genomic m c marker arker application application Scienc Sc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Clinical Trial Designs ( 1 ) Clinical Trial Designs ( 1 ) May test all or Drug some Interaction Interac n analysis between analysis between ALL subjects treatment and test treatment and test Placebo Such a design may be Such a design may be adopted if the results of the adopted if the results of the testing will not be readily testing will not be readily available at the clinical sites available at the clinical sites for informing randomisation for informing randomisation Scienc Sc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Importance of samples/ BM data from Importance of samples/ BM data from majority of subjects in a CT majority of subjects in a CT Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

KRAS story oWild type v Wild type vs mutant KRAS tant KRAS, I , Impa pact on ct on PFS; PFS; o Retrospective analysis; (prosp Retrospective analysis; (prosp defined) defined) Several Interesting issues; Several Interesting issues; • Prognostic or Predictive Prognostic or Predictive • 2nd Renewal- 2nd Renewal- suggested that suggested that use use of of Vectibix in mutated KRAS may be Vectibix in mutated KRAS may be detrimental – a marker of safety. a marker of safety. detrimental • Relationship with B RAF, Relationship with B RAF, NRAS, NRAS, Not defined Not defined De Rook W et al; De Rook W et al; Lancet oncol, 11:753-62, 2010 Lancet oncol, 11:753-62, 2010 Scienc Sc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Clinical Trial Designs ( 2 ) Clinical Trial Designs ( 2 ) Enriched or targeted design Enriched or targeted design Enriched or targeted design Enriched or targeted design No information on No information on Test - efficacy or safety efficacy or safety ALL subjects ALL subjects ALL tested ALL tested Test + Drugs would be integrally Drugs would be integrally Drug Placebo linked with availability and linked with availability and use of the biomarker test at use of the biomarker test at the time of approval. the time of approval. Scienc Sc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Do we need marker –ve subjects?? Breast cancer and Herceptin Vem urafenib in m elanom a • ~ 30% of all Br Ca are Her-2 (+) • IHC staining 1+ to 3 + is not infallible (~20% discordance between CTA & test) BM+ or negative BM+ or negative freq frequency is not ency is not • • Reports of response in Her (-) ve such an issue such an issue subjects? 12

Marker by RX- I nteraction Design MARVEL Trial-- erlotinib Answ ers lim itations of unselected or enriched designs ; but impacts sample size and expense.. Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Observations on Trial designs  Predictive values need to be generated in populations that reflect anticipated clinical use.  Enriched design often does not validate the marker (!) or define its utility in the broad population. Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Safety m arkers and clinical trials Abacavir Abacavir Carbemazepine Carbemazepine Warfarin Warfarin Marker Marker HLA-B*5701 HLA B* 1502 CYP 450 SNPs+ VKORC1 Discove Discovery Retrospective Retrospective Retrospective Confirmatory Confirmatory Prospective Prospective Retrospective Retrospective Variable Variable evidence evidence study study Prospective study Prospective study not feasible not feasible Replication Replication Yes Yes ongoing Validity Validity Clear evidence Clear evidence of Evidence of harm association Dosing alteration?? Utility Utility Clear Clear Clear Clear Contradictory Contradictory Scienc Sc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Population and predictive ability; recent exam ple Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Population and predictive ability; ‘ Old’ exam ple Predictive ability in a particular population will be event/ incidence dependent. Rarer the event, more difficult to establish evidence. Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Retrospective data/ analyses  Often based on associations..  Inability to replicate the results  Winner’s Curse Winner’s Curse (first result is usually best)  often -overestimation of Effect size  underpowered FU studies (sample sizes)  Phenotype heterogeneity..  Bias (various) Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Caveats for Retrospective data analysis • Ideally, data from well conducted RCT • Biological sample or BM status availability from all or majority of the subjects (from RCT) • Prospectively stated hypothesis & appropriate analysis plan • Replication • Biological plausibility & • difference between BM+ vs BM- is clear Scienc Sc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Experience at CHMP: ‘A lot of post-hoc analyses’ and “Pharm acoepi-PGx” experience is lim ited A Number of open questions Answers (May be !?) • When is a RCT mandatory? • Most of the time ( Most of the time ( alm ime ) almost st everyt everytim • Is retrospective identification • Yes –depends on the Yes –depends on the and validation BM acceptable? circumstances circumstances. • Are BM (-) patients be required • Clearly situation dependent Clearly situation dependent in trials ? • Can we use Adaptive designs?? Can we use Adaptive designs?? Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Adaptive designs Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012

Sc Scienc ience to Clinic workshop; London Oc e to Clinic workshop; London Oct 2012 t 2012