Predictive Biomarkers: Using What’s Learned from Past ... · Thought Process for PresentationThought Process for Presentation Predictive biomarkers Predictive biomarkers ––

Predictive Biomarkers: Using What’s Predictive Biomarkers: Using What’s Learned from Past Regulatory Decisions Learned from Past Regulatory Decisions to Inform Future Development Plansto Inform Future Development Plans

Bi k W ld CBi k W ld CBiomarker World CongressBiomarker World CongressMay 27, 2009May 27, 2009Philadelphia, PennsylvaniaPhiladelphia, Pennsylvania

Lawrence J. Lesko, Ph.D., Lawrence J. Lesko, Ph.D., F.C.P.F.C.P.Di ectoDi ecto Office of ClinicalOffice of Clinical Pha macologPha macolog

p , yp , y

DirectorDirector, Office of Clinical , Office of Clinical PharmacologyPharmacologyFood Food and Drug Administrationand Drug Administration

Thought Process for PresentationThought Process for PresentationThought Process for PresentationThought Process for Presentation

Predictive biomarkers Predictive biomarkers –– provide information that provide information that can lead to good decisioncan lead to good decision--making or the best making or the best course of treatmentcourse of treatmentcourse of treatmentcourse of treatment

Industry and regulatory agencies use biomarkers Industry and regulatory agencies use biomarkers y g y gy g y gdifferently in many cases differently in many cases –– decisions related to decisions related to approval and labeling are common groundapproval and labeling are common ground

Some biomarkers are validated as diagnostic tests Some biomarkers are validated as diagnostic tests for clinical practice for clinical practice –– precision medicineprecision medicine

The Risk of Biomarker FatigueThe Risk of Biomarker Fatigue

Oh no!! Not another biomarker meeting

The Scream – Edward Munch (1893)

Holy Grail of Biomarkers: To Identify Holy Grail of Biomarkers: To Identify y yy yWho Will and Will Not Respond to DrugsWho Will and Will Not Respond to Drugs

MAY 27, 2009

Industry Leaders Stressed Efficiency in Industry Leaders Stressed Efficiency in y yy y2008 and Individualization in 20092008 and Individualization in 2009

“The model of drugs combined with diagnostics is good for pharmaceutical companies, regulators, payers and patients…..the landscape is ripe for us to develop more p p p ppersonalized treatments.”

Paul Stoffels, J&J, March 2009

“Our goal, despite the overall objective of being able to stratify patients where its necessary and appropriate, given the choice our preference is to find a therapy that will treat achoice…..our preference is to find a therapy that will treat a disease that many people will have that doesn’t require stratification…..and come up with a drug that will work in 80% of the people ”of the people.

Reid Leonard, Merck, May 2009

2009: Drug Development Enterprise in 2009: Drug Development Enterprise in g p pg p pTransformation in FDA and CongressTransformation in FDA and Congress

Patients will expect new drugs to be wonderful, affordable, and will be angry when they aren’t much better than existing drugs ..or havemuch better than existing drugs…..or have unanticipated side effects, especially severe ones

Specific areas – drug safety – are changingSpecific areas drug safety are changing rapidly…..”Safety First”, “Sentinel Initiative”, PMC, OSE expanded role, OND safety teams, OCP safety program, OB drug safety division, new FDA CRCp g , g y ,

Obama’s 2009 economic stimulus bill contains $1.1B for comparative effectiveness studies for NIH $ pand AHRQ …..IOM will recommend priorities to HHS by June 30th

OverOver--Arching Goal for Biomarkers Seems Arching Goal for Biomarkers Seems ggto be Stratification and Individualizationto be Stratification and Individualization

R l t h i i t t i hRegulators have a growing interest in how responses to drugs differ among individuals or subsets we’re already seeing this in geneticsubsets…..we re already seeing this in genetic

differences in metabolism and targeted therapies in oncology and infectious diseasesp gy

Reducing Variability in Platelet Function: Reducing Variability in Platelet Function: g yg yThe Sum of the PartsThe Sum of the Parts

Antiplatelet DrugIdentify patients at high risk

for non-response

161820

ts

Resistance=31% 2C19 gene variantsP2Y12 gene variantsBaseline platelet functionCompliance

Redu

ce to P

468

101214

Num

ber o

f Pat

ient Compliance

DoseDrug interactions

Practice

Platelet Function

02

-30--20 -20--10 -10-0 0-10 .10-20 20-30 30-40 40-50 50-60 <60

Change in Aggregation

Distribution of adjusted platelet Functionj presponse in 544 patients after 30 days

of clopidegrel

Example: Genotyping Information Now Example: Genotyping Information Now p yp gp yp gin US Clopidegrel Labelin US Clopidegrel Label

May 21, 2009

• Large section of label on CYP 2C19 pharmacogenetics• Phenotype and genotype distribution in populations• Effect of genotype on active metabolite exposure• Differences in antiplatelet response between IMs and PMs

CV t t t t th b i li k d ith t• CV event rates or stent thrombosis linked with genotype• PGx testing can identify genotypes• Omits recommendation of optimal doses for PMs• New advice on PPI inhibition of CYP2C19• New advice on PPI inhibition of CYP2C19

Unifying Concept for Biomarkers: Unifying Concept for Biomarkers: y g py g pHeterogeneity Among PeopleHeterogeneity Among People

Live in a world of variability and uncertainty: accept and deal with it at the outset using predictive biomarkers of dosing, efficacy and risk. This is fully aligned with what physicians need in dealing with one patient at a time

The Case for BiomarkerThe Case for Biomarker--Driven Precision Driven Precision Medicine: The PatientMedicine: The Patient--Doctor InterfaceDoctor Interface

“When prescribing a treatment cliniciansWhen prescribing a treatment, clinicians should ask not does it work for most patients, but rather does it work for this patient”p

Mant D, Lancet 353:743-746, 1999

Example: Drug Response Rates in a Clinical Example: Drug Response Rates in a Clinical p g pp g pPhenotype with Different EtiologiesPhenotype with Different Etiologies

RCT looked at anxiety disorders common to children and RCT looked at anxiety disorders common to children and adolescents between 7 and 17 yrsadolescents between 7 and 17 yrsParticipants (n=488) randomized to (1) Pbo, (2) cognitiveParticipants (n=488) randomized to (1) Pbo, (2) cognitiveParticipants (n 488) randomized to (1) Pbo, (2) cognitive Participants (n 488) randomized to (1) Pbo, (2) cognitive behavior therapy, (3) sertraline and (4) combination of (2) behavior therapy, (3) sertraline and (4) combination of (2) and (3) for 12 weeksand (3) for 12 weeksPrimary endpoint was clinical global improvementPrimary endpoint was clinical global improvement--impressionimpressionPrimary endpoint was clinical global improvementPrimary endpoint was clinical global improvement impression impression scorescore

Treatment % Patient Very Much Improved

Pb 24%Pbo 24%

Cognitive Behavior 60%

Sertraline 55%

Combination 81%

Walkup et al, NEJM 359: 2753-2766, 2008

How Can Biomarkers Be Used to Tweak or How Can Biomarkers Be Used to Tweak or Overhaul the Drug Development Process?Overhaul the Drug Development Process?

NMEs/BLAs per $B R&D spent*

> 6400 active INDs (50% rise)• > 6400 active INDs (50% rise)

• > 2500 meeting requests (2X)

• > 30% of approvals have PMRs (30%)

• > 400,000 ADRs reports per year (2X)

* Measured 5 years after initial R&D investment$0.12

• 24 NMEs in 2008 ~ an i t 3

Source: EvaluatePharma, FDA website (PDUFA)

increase over past 3 yrs• 5O% from big PhRMA• As many as 45 NMEs could be approved in 2009

“These trends are not sustainable. A new development model or paradigm is needed”

Dr. Janet Woodcock, CDER DirectorNIST Talk September 24 2007NIST Talk, September 24, 2007

Attrition: Analyze Through the Noise to Attrition: Analyze Through the Noise to y gy gUnderstand Where Biomarkers Can ImpactUnderstand Where Biomarkers Can Impact

55% 45% 69% 1997-2002

Success Rates Out of Each Phase: Small Molecules

2/3 of drugs in

51% 33% 56% 2002-2007Cumulative

Success Rate~ 11%

2/3 of drugs in pipeline in late

stage development

ImmuneEndocrine

GIOncology

GIImmune

MusculoskeletalOncology40

50

60

OncologyHeme

MusculoskeletalOther

OncologyCardioHeme

Respiratory10

20

30

401997-20022002-2007

Pharmaprojects, 2008, McKinsey & Company, London Biotechnology Network and Kola and Landis, Nat Rev Drug Disc 3, 711-716, 2004

0E ffic a c y vs P bo S a fe t y C o n firme d S a fe t y S u sp e c t e d E ffic a c y D iffe re n t S a fe t y D iffe re n t

Example: Study Design Issues ~ Stimulation of Example: Study Design Issues ~ Stimulation of p y gp y gCholecystokininCholecystokinin--A Receptors with GI181771XA Receptors with GI181771X

• New drug for obesity; two month phase 2A POC study (n = 40/arm)• Two SNPs from 2 candidate genes relate to MOA (appetite suppressant)• “Failed” study by comparing mean wt loss between drug and Pbo• Clinical endpoint was weight loss on an ad lib diet• Clinical endpoint was weight loss on an ad lib diet

Confirms drug has efficacyas appetite suppressant

Gain 1.3 kg:1:1 genotype Loss 3.3 kg

Phase 2B/3 trials might be enriched with 2-2 and 1-2 genotypes

Test other molecules for 1-1 genotype

Hyper-responder subgroup (20%):Lose up to 10% of BW and

homozygous for single alleleRoses, Nat Rev Genet 5, 645-656, 2004

So What Happened? Trial Doomed to Show So What Happened? Trial Doomed to Show ppppNo Population Differences in OutcomeNo Population Differences in Outcome

Phase 2B/3 RCTs Phase 2B/3 RCTs compared drug vs Pbo using a compared drug vs Pbo using a rigorous hypocaloric diet rigorous hypocaloric diet over 24 weeksover 24 weeks

Did not replicate design of phase 2A studyDid not replicate design of phase 2A study–– Did not replicate design of phase 2A studyDid not replicate design of phase 2A study–– Clinical approvable weight loss of 5%Clinical approvable weight loss of 5%

Results: all participants lost weight on hypocaloric Results: all participants lost weight on hypocaloric p p g ypp p g ypdiet, so drug was no different than Pbo (p>0.05)diet, so drug was no different than Pbo (p>0.05)–– Trial design did not allow Trial design did not allow appetite suppression effect appetite suppression effect to be to be

demonstrated in alldemonstrated in all--comers or genotype subsetscomers or genotype subsetsdemonstrated in alldemonstrated in all comers or genotype subsetscomers or genotype subsets

Lesson learned: a “failed” trial is not always the Lesson learned: a “failed” trial is not always the fault of the drug but the design of the studyfault of the drug but the design of the study

Jordan et al, Clin Pharmacol Ther 83, 281-287, 2008

The Case for Developing New Biomarker The Case for Developing New Biomarker p gp gStrategiesStrategies

Imagine This:

The biomarker gods or gurus descend and say: “we have heard you and we agree – we would be happy to fundamentally change the current drug development process with biomarkers. JUST TELL US HOW AND WE WILL DO ITTELL US HOW AND WE WILL DO IT.

What would we do?

We need to have an answer. We need a biomarker development process that captures the value of prediction -- to reduce uncertainty and heterogeneity of response -- and as diagnosticuncertainty and heterogeneity of response and as diagnostic tests, allow for the improvement of individual patient care as a result.

What’s New Is Old: Landmark Advances What’s New Is Old: Landmark Advances in Biomarker Sciencein Biomarker Science

• CAST trial ~ fallibility of biomarkers (VPC) as surrogates (survival)

• 21 CFR 314.50 ~ accelerated approval regulations allowing approval on surrogatesHIV/AIDS f CD4 ll t d i l l d• HIV/AIDS ~ use of CD4 cell counts and viral load as surrogates

• FDAMA Sec 115A ~ one adequate and well-controlled trial and confirmatory evidence to establish efficacy

• FDA Guidance ~ Providing Clinical Evidence of

1990 - 1999

Effectiveness for Human Drug and Biological Products

• NIH/FDA publication of workshop proceedings on biomarkers/surrogates ~ definition and terminology

• Peck/Rubin/Sheiner publication on a single clinical trial• Peck/Rubin/Sheiner publication on a single clinical trial plus causal evidence of effectiveness for drug approval

• FDA CPI and Opportunities List ~ Over 60 references to biomarkers as better evaluation tools

• Coming of age of critical path-driven public-private t hi FDA/FNIH FDA/NCI SAE d PSTC

2000 -2009partnerships ~ FDA/FNIH, FDA/NCI, SAE and PSTC

• Biomarker qualification process (MaPP) and future guidance

Break Down Root Cause of Low and High Break Down Root Cause of Low and High ggAttrition By Therapeutic Area Attrition By Therapeutic Area

Attributes of Drugs With Low Attrition Rates

Attributes of Drugs With High Attrition Rates

Biomarker Strategies to Improve Attrition Rates

Predictive animal models (T2diabetes bioma ke s)

Useless animal models (Al heime ’s disease no

Transgenic animals, cultured cell lines in silico modelsdiabetes biomarkers) (Alzheimer’s disease; no

known predictive biomarkers)cell lines, in silico models, biomarker panels

Precise diagnostic criteria; credible prognostic and/or predictive biomarkers

Imprecise diagnostic criteria; lack of biomarkers for broad indications

Biomarker-based stratification of phenotype; enrichment in POC trials; comparepredictive biomarkers indications POC trials; compare subgroups genomically

Understanding of MOA; good D/R optimize dosing, less off-target effects

Uncertain MOA; small signal size; large Pbo effect; incomplete D/R > safety

Carry D/R into phase 3; collect DNA for safety signals; run-in phase to manage Pbooff target effects incomplete D/R, > safety

risks run in phase to manage Pbo

Objective clinical endpoints; solid analytic validation; reliable sample size

Less objective endpoints; highly variable measurementsrelated to PI/site; often

Add more secondary endpoints; drug-disease-Pbo models with M/S to evaluate p

estimates; low drop-out rates; reliance on single biomarkers

/ ;underpowered; high drop-out rates; single biomarker not predictive of clinical endpoint

/probabilities of success of trial designs; adaptive designs – all comers/subsets

Biomarkers Need to DeBiomarkers Need to De--Lump Diagnostic Lump Diagnostic p gp gCategories and Be Integrated into RCTsCategories and Be Integrated into RCTs

The body has a limited vocabulary to express itself when The body has a limited vocabulary to express itself when something is wrong; there are not enough symptoms something is wrong; there are not enough symptoms (biomarkers) to go around for all diseases(biomarkers) to go around for all diseases(biomarkers) to go around for all diseases(biomarkers) to go around for all diseases

High blood pressure Heart FailureHigh cholesterolg p High cholesterol

Chest Pain ArrthymiasKidney damage

Same disease phenotypes are composed of different pathologiesDrug’s MOA becomes unrelated to etiology of patient’s syndrome

Adapted from “The Innovator’s Prescription” , Christensen, Grossman and Hwang (2009)

g gy p yNet result is low efficacy rates and skewed B/R ratio

Need Better Biomarkers To Define Need Better Biomarkers To Define Homogenous Patient SubsetsHomogenous Patient Subsets

Likely

CertainThe image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it again.

The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it again.

The image cannot be displayed. Your computer may not have enough memory to open the image, or the i h b t d R t t t d th th fil i If th d till







The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted Restart your computer and then open the file again If the red x still









The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted Restart your computer and then open the file again If the red x stillff

icac

y

Precise diagnosis more likelyBiomarkers provide pattern recognitionM d t tt iti d t ffi t

Strep ThroatType I DiabetesHER2 Breast Cancer

Possible

LikelyThe image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted Restart your computer and then open the file again If the red x still












Th i t b di l d Y t t h h t th i t

The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still














The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still …



The image cannot be displayed Your computer may not have enough memory to open the image or teatm

ent

Ef

AsthmaProstate CancerO t i

Modest attrition, moderate efficacy rates HIV/AIDS

Precise diagnosis possibleNumerous predictive biomarkers

Unlikely

















The image cannot be displayed Your computer may not have enough memory to open the image or t





The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still …




The image cannot be displayed Your computer may not have enough memory to open the image or thetent

of

Tre Osteoporosis

Osteoarthritis

Alzheimer’s DiseaseDepression

b i Precise diagnosis isn’t possible

Numerous predictive biomarkersLow attrition, high efficacy rates

Intuitive Empirical Precise

Uncertain






Ext Obesity

MigrainePrecise diagnosis isn t possibleLacking predictive biomarkersHigh attrition, low efficacy rates









The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrup














The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been co








































The image cannot be displayed. Your computer may not have enough memory to open the i






The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it again.Extent of Understanding of Disease

Adapted from “The Innovator’s Prescription” , Christensen, Grossman and Hwang (2009)

Finding Biological Pathways Related to Disease Finding Biological Pathways Related to Disease g g yg g yEtiology and Clinical Phenotype CriteriaEtiology and Clinical Phenotype Criteria

• Two SNPS on chromosome 12p13 are associated with an increased risk of stroke (HR = 1.42)• Stratification of stroke into SV, LV, atherosclerotic and ischemic subgroupsischemic subgroups

• Web-based database of all molecular alterations in

ti

APRIL 2009

DISEASE SIGNATURES

pancreatic cancer• Compendium of 2,516 genes (potential biomarkers) that are over-expressedp• Identified 60 most promising drug targets

Common Molecular Pathways of Disease Common Molecular Pathways of Disease yyRequires Enhanced Diagnostic PrecisionRequires Enhanced Diagnostic Precision

Incontinence

Duloxetine

A mechanistic common thread: treatment increases the activity of serotonin and norepinephrine in brain and spinal cord by reducing absorption into cells: B/R unique for each disease

This has led many to a pathway-based biomarker strategy to develop drugs for cancer and HIV/AIDs with higher efficacy rates

Beachhead For BiomarkerBeachhead For Biomarker--Based Based Precision Medicine: OncologyPrecision Medicine: Oncology

More than 50% of industry pipeline products are in oncology More than 50% of industry pipeline products are in oncology ~ 400+ cancer molecules~ 400+ cancer moleculesFDA approvals based on biomarkers for “boutique”FDA approvals based on biomarkers for “boutique”FDA approvals based on biomarkers for boutique FDA approvals based on biomarkers for boutique malignancies starting with Gleevec in 2001malignancies starting with Gleevec in 2001–– Increased chance of early approval, greater efficacy, Increased chance of early approval, greater efficacy,

higher penetration and adherence and commercialhigher penetration and adherence and commercialhigher penetration and adherence and commercial higher penetration and adherence and commercial opportunities to expand label indicationsopportunities to expand label indications

Top picks for ASC0 2009Top picks for ASC0 2009

Phase 1 study in NSCLC cancer irrespective of tumor mutations with little response at MTD. Additional cohort of patients with Met or ALD mutations had striking response, tumor shrinkage in 50% of patients. Mutations occur in 3-5% of patients, a $500M market.

BiomarkerBiomarker--Driven Pathway Approaches Driven Pathway Approaches y ppy ppLeads to Lower Attrition RatesLeads to Lower Attrition Rates

Phase Attrition Probabilities

Walker and Newell, Nature Rev Drug Disc (2009) 8(15), 15-16

Counterintuitive Drug Development Counterintuitive Drug Development g pg pParadigmParadigm

RIS

K

PrecisionMedicine

EmpiricalMedicine

ENEF

IT/R

BE

TIME

Continuous reduction in uncertainty and reassessment of benefit/risk using linkages to predictive biomarkers at key

milestones in early and late phase clinical studies

Evolving Biomarker DirectionsEvolving Biomarker DirectionsEvolving Biomarker DirectionsEvolving Biomarker Directions

Strategies available to address development:

1 Staggered approvals: initial approval in small indication and1. Staggered approvals: initial approval in small indication and limited patient subset with later expansion to additional diseases and wider populations (Ex: Kit+ GIST indication for GleevecR and later success in BCR-ABL-CML, SCLC, , ,glioblastoma and prostate cancer

2. Biomarkers to define patient subsets in POC studies p(enrichment guidance), in prospective-retrospective studies (adaptive design guidance), in postmarketing studies (observational and retrospective analyais) and in label updates of approved drugs (PK, PD, dosing)

Valid Genomic Biomarkers That Should Valid Genomic Biomarkers That Should Be Considered in Drug DevelopmentBe Considered in Drug Development

http://www.fda.gov/cder/genomics/default.htm

Established Biomarkers (Tests): Established Biomarkers (Tests): ( )( )Driving Drug Treatment Decisions TodayDriving Drug Treatment Decisions Today

Imatinib and Kit + GIST (prospective, Imatinib and Kit + GIST (prospective, preapprovalpreapproval))Dasatinib and PH + ALL (prospective, Dasatinib and PH + ALL (prospective, preapprovalpreapproval) ) Maraviroc and CCR5 + (tropic) HIVMaraviroc and CCR5 + (tropic) HIV--1 (prospective, 1 (prospective, preapprovalpreapproval))( p )( p ) (p p ,(p p , p ppp pp ))Tetrabenazine and 2D6 dosing (prospective, Tetrabenazine and 2D6 dosing (prospective, preapprovalpreapproval))Erlotinib and EGFR + lung cancer (retrospective, Erlotinib and EGFR + lung cancer (retrospective, preapprovalpreapproval))Nilotinib and UGT hyperbilirubin (retrospective, Nilotinib and UGT hyperbilirubin (retrospective, preapprovalpreapproval))yp ( p ,yp ( p , p ppp pp ))

Abacavir and HLAB*5701 HAS (prospective, Abacavir and HLAB*5701 HAS (prospective, postpost--approvalapproval))Clopidegrel and 2C19 “resistance” (prospective, Clopidegrel and 2C19 “resistance” (prospective, postpost--approvalapproval))Codeine and 2D6 toxicity (retrospective, Codeine and 2D6 toxicity (retrospective, postpost--approvalapproval))Panitumamab and KRAS (retrospective, Panitumamab and KRAS (retrospective, postpost--approvalapproval))Carbamazepine and HLAB*1502 SJS (retrospective, Carbamazepine and HLAB*1502 SJS (retrospective, postpost--approval)approval)Warfarin and 2C9/VKORC1 dosing (retrospective, Warfarin and 2C9/VKORC1 dosing (retrospective, postpost--approvalapproval))

Safety Monitoring Continues Over Drug’s Safety Monitoring Continues Over Drug’s y g gy g gLifeLife--Cycle: Biomarkers Added to LabelsCycle: Biomarkers Added to Labels

Stratification of patients Stratification of patients to optimize dose selection on to optimize dose selection on the basis of pharmacogeneticsthe basis of pharmacogenetics–– Tetrabenazine (XenazineTetrabenazine (XenazineRR) and CYP2D6 polymorphisms) and CYP2D6 polymorphismsTetrabenazine (XenazineTetrabenazine (Xenazine ) and CYP2D6 polymorphisms) and CYP2D6 polymorphisms–– Irinotecan (CamptosarIrinotecan (CamptosarRR) and UGT1A1 gene variants) and UGT1A1 gene variants–– Warfarin (CoumadinWarfarin (CoumadinRR) and CYP2C9/VKORC1 genotypes) and CYP2C9/VKORC1 genotypes–– 66--MP (PurinetholMP (PurinetholRR) and TPMT polymorphisms) and TPMT polymorphisms66 MP (PurinetholMP (Purinethol ) and TPMT polymorphisms) and TPMT polymorphisms

Identification of patients Identification of patients likely to receive no benefit or at likely to receive no benefit or at risk for severe adverse eventsrisk for severe adverse eventsrisk for severe adverse eventsrisk for severe adverse events–– Tamoxifen (NolvadexTamoxifen (NolvadexRR) and CYP2D6 PMs) and CYP2D6 PMs–– Codeine (Generic) and CYP2D6 UMsCodeine (Generic) and CYP2D6 UMs

CBZ (TegretolCBZ (TegretolRR) and HLA) and HLA B*1502 allelesB*1502 alleles–– CBZ (TegretolCBZ (TegretolRR) and HLA) and HLA--B*1502 allelesB*1502 alleles

–– Abacavir (ZiagenAbacavir (ZiagenRR) and HLA) and HLA--B*5701B*5701 allelesalleles

Biomarker Data Useful to Drug Biomarker Data Useful to Drug ggDevelopment, Labels and Patient CareDevelopment, Labels and Patient Care

Reduce heterogeneity in the population by defining Reduce heterogeneity in the population by defining disease subsets disease subsets and not clusters of symptomsand not clusters of symptoms–– HerceptinHerceptinRR for breast cancer with HER2 overexpressionfor breast cancer with HER2 overexpressionpp pp

Identify population subsets Identify population subsets capable of responding capable of responding to to treatmenttreatment–– Tropism testing in HIV/AIDS before maraviroc (SelzentryTropism testing in HIV/AIDS before maraviroc (SelzentryRR))Tropism testing in HIV/AIDS before maraviroc (SelzentryTropism testing in HIV/AIDS before maraviroc (Selzentry ))

Identify Identify patients unlikely to benefitpatients unlikely to benefit but subject to but subject to unnecessary adverse eventsunnecessary adverse events

P it b ill t k i l t l ti t ith t t dP it b ill t k i l t l ti t ith t t d–– Panitumamab will not work in colorectal patients with mutated Panitumamab will not work in colorectal patients with mutated KRAS geneKRAS gene

Find Find high risk patients high risk patients to facilitate rapid and timely to facilitate rapid and timely l f t i k d ti f dl f t i k d ti f daccrual of events or risk reduction for new drugsaccrual of events or risk reduction for new drugs

–– Rosuvastatin prevents CV events in healthy people when hsRosuvastatin prevents CV events in healthy people when hs--CRP CRP > 3 mg/dl > 3 mg/dl

Lessons LearnedLessons LearnedLessons LearnedLessons Learned

What this suggests to me:

U d t d l ti h t it i PK d PD• Understand population heterogeneity in PK and PD enzymes and receptors early in development

C d t l li i l t di t th t f• Conduct early clinical studies at the extremes of metabolic and receptor genotypes (so called “outliers”)

f f• Define upper and lower limits of systemic exposure to calculate benefit/risk ratio in biomarker-defined subsets

• Consider using biomarkers as stratification criteria in pivotal clinical trials for dosing, efficacy or safety reasons

Retrospective Subset Analysis: Summary of Retrospective Subset Analysis: Summary of p y yp y yKey Points from the ODAC on KRASKey Points from the ODAC on KRAS

Forest plots are frequently used in labels to show Forest plots are frequently used in labels to show treatment effects (CI) in unplanned subsetstreatment effects (CI) in unplanned subsetsKRAS t ti f t i l t l tKRAS t ti f t i l t l t llKRAS testing of tumor in colorectal cancer to KRAS testing of tumor in colorectal cancer to rule rule outout patients who can’t respond to antipatients who can’t respond to anti--EGFR EGFR treatment (40% of patients with mutated KRAS)treatment (40% of patients with mutated KRAS)( p )( p )

Conditions for retrospective biomarker analysis

1. AWC trial2. Large sample size3. High biomarker ascertainment4. Analytically valid assay5. Prespecified analysis plan

Summary: OptimismSummary: OptimismSummary: OptimismSummary: Optimism

“If we can recognize that change and uncertainty are basic principles, we can greet the future and the transformation we are undergoing with the understanding that we do not know enough to beunderstanding that we do not know enough to be pessimistic.”

Hazel HendersonHazel HendersonFuturist, Economist and Columnist

Thank you for your timeand attention

l k l@ d [email protected]

Documents

Predictive Biomarkers: Using What’s Learned from Past ... · Thought Process for PresentationThought Process for Presentation Predictive biomarkers Predictive biomarkers ––