Biomarkers  and  Companion  Diagnos1c  Applica1ons  

in  Oncology  Drug  Development  

Nicholas  C.  Dracopoli,  Ph.D.  

Vice  President,  Head  Oncology  Biomarkers  Janssen  R&D  

Johnson  &  Johnson  

Shanghai,  China  May  10,  2012  

Empirical  Drug  Development  Strategies  are  Unsustainable    

•  Overall  aLri1on  rates  are  too  high  during  development:  –  Poor  in  vivo  and  in  vitro  disease  models  lead  to  failure  early  in  

development  –  Too  many  compound  fail  for  lack  of  efficacy  late  in  development  

•  Disease  heterogeneity  means  too  few  pa1ents  respond  to  any  one  therapeu1c  approach:  –  Need  beLer  markers  to  monitor  status  of  the  drug  target  and  

cognate  pathway    •  Development  costs  for  novel  drugs  with  low  response  rates  

are  too  high:  –  Large  Phase  III  trials  required  to  demonstrate  clinical  benefit  –  High  risk  of  registra1onal  failure  –  Length  of  1me  required  to  show  overall  survival  benefit  

New  Drug  Approvals  in  US:  1996-­‐2010  

Mullard, A. (2011) 2010 FDA drug approvals, Nature Reviews Drug Discovery 10:82-85

ALri1on  in  Drug  Development:  2009  •   Overall  clinical  success  (Phase  I  entry  

to  approval)  has  risen:  •  2004  es1mate:  11%  •  2009  es1mate:  18%  

•   Companion  diagnos1cs  have  impacted  approval  of  some  kinase  inhibitors:  

–  cKIT  for  ima1nib  (GIST)  –  KRAS  for  panitumumab  (colorectal  cancer)  –  HER2  for  lapa1nib  (breast  cancer)  

•  Clinical  success  for  kinase  inhibitors  is  ~2.5-­‐fold  higher  than  the  overall  average:  

•  How  much  of  this  is  due  to  undifferen1ated  fast  follow  on  compounds?  

•  Has  the  transi1on  from  cytotoxic  to  targeted  therapies  reduced  overall  aLri1on?  

•  How  much  is  this  due  to  precedented  chemistry  and  biology  for  kinase  inhibitors?   Walker & Newell, 2009

Biomarkers  in  Drug  Development  Marker Func*on Test

PD/MOA •  Determine  whether  a  drug  hits  the  target  and  has  impact  on  the  biological  pathway  

•  Evaluate  mechanism  of  ac1on  (MOA)  

•  PK/PD  correla1ons  and  determine  dose  and  schedule  

•  Determine  biologically  effec1ve  dose

•  Research  test  used  during  drug  development  

•  Not  developed  as  companion  diagnos1c

Predic1ve •  Iden1fy  pa1ents  most  likely  to  respond,  or  are  least  likely  to  suffer  an  adverse  event  when  treated  with  a  drug.

•  Companion  diagnos1c  test  (e.g.  hercep1n,  EGFR)

Resistance •  Iden1fy  mechanisms  driving  acquired  drug  resistance •  Muta1on  analyses  (e.g.  Bcr-­‐Abl  muta1on  in  ima1nib  treated  CML)

Prognos1c •  Predicts  course  of  disease  independent  of  any  specific  treatment  modality

•  Approved  tests  (e.g.  CellSearch,  Mammaprint)

Surrogate •   Approved  registra1onal  endpoints •  Commercial  diagnos1c  tests  (e.g.  LDL,  HbA1c,  viral  load,  blood  pressure)

The  Biomarker  Hypothesis  

•  Biomarkers  will:  –  Reduce  development  1me  for  ac1ve  compounds  –  Accelerate  failure  of  unsafe  or  inac1ve  compounds  –  Reduce  average  development  costs  for  approved  compounds  

–  Lead  to  beLer  outcomes  for  cancer  pa1ents  •  The  costs  for  biomarker  research  will  be  more  than  compensated  by  increased  efficiency  of  the  drug  development  process:  –  Early  at-­‐risk  investment  in  biomarkers  leads  to  more  approved  compounds  with  beLer  pa1ent  outcomes  and  stronger  cases  for  reimbursement  

The  Biomarker  Paradox  

   There  are  11,166  biomarkers  listed  in  GOBIOM  database  (01/31/2011)    

-­‐  BUT  -­‐  only  32  valid  genomic  biomarkers  in  FDA    approved  drug  

labels  

-­‐  AND  -­‐  0  are  mul1plex  IVD’s    based  on  proteomic  or  genomic  profiles  

Protein  Kinase  Inhibitors:  A  Model  for  Biomarker  Development  in  Oncology  

•  216*  protein  kinase  drugs  in  Phase  II  or  III  for  cancer  indica1ons  (23%):  –  Most  common  cancer  drugs  in  oncology  development  (23%*)  

–  2nd  most  common  drug  class  aker  G-­‐protein  coupled  receptors  (GPCR)  in  all  indica1ons  

•  12  drugs  approved  by  FDA  for  cancer  indica1ons  that  target  receptor  tyrosine  kinases  (RTK):  

–  7  have  predic1ve  markers  in  the  drug  label  

–  No  other  cancer  drug  classes  have  predic1ve  markers  in  their  labels  when  launched  

•  Biomarkers  are  required  for  RTK  drug  development  to:  –  Predict  dependency  on  specific  signaling  pathways  

–  Screen  for  acquired  drug  resistance  

–  Monitor  pathological  changes  during  disease  progression  

*The Beacon Group, 2010

Targeted  Therapy  with  Tyrosine  Kinase  Inhibitors  

 Mul1ple  druggable  approaches  to  inhibi1ng  protein  kinase  signaling:  –  Reduce  ligand  –  bevacizumab  

(Avas1n)  binds  VEGF  and  reduces  ligand-­‐dependant  receptor  ac1va1on  

–  Block  receptor  –  cetuximab  (Erbitux)  blocks  EGFR  and  prevents  ligand-­‐induced  receptor  ac1va1on  

–  Inhibit  intracellular  kinase  –  erlo1nib  (Tarceva)  inhibits  the  intracellular  phosphoryla1on  of  EGFR  kinase    Ciardiello & Tortora, New Engl. J. Med. 358:1160, 2008

Signal  Transduc1on  Pathways  are  Ini1ated  by  Mul1ple  Pathological  Events    

A: Normal signal Transduction

B: Activate intracellular Kinase (mutation or translocation)

C: Mutate intermediate pathway member (e.g. KRAS)

D: Receptor gene amplification

E: Increase ligand expression

F: Utilize alternative Receptor (e.g. MET)

Approved  Companion  Diagnos1cs:  2011  Markers Direct  Markers Secondary  

Markers Molecular  Profiles*

Readout Drug  target  status Downstream  pathway

Consolidated  profiles

Examples HER2+  

ER+  

CD20+  

BCR-­‐ABL  (Ph+)  

KIT+  

EGFR+  

BRAF  

EML4-­‐ALK  

KRAS  wt

Companion  diagnos1cs:  KRAS  in  colorectal  cancer  

   Predic1ve  values  of  KRAS  muta1ons  in  colorectal  cancer  (Raponi  et  al.,  2008)  :  

–   35%  PPV  –  97%  NPV  

Karapetis et al., 2008

No  IVDMIA  Tests  Approved  as  Companion  Diagnos1cs  

Test   Company   Companion  Diagnos*c  

Prognos*c  Test  

Mammaprint   Agendia   No   Yes  

Tumor  of  Unknown  Origin  

Pathwork  Diagnos1cs  

No   Yes  

Allomap   XDx   No   Yes  

An IVDMIA is a device that combines the values of multiple variables using an interpretation function to yield a single, patient-specific result that is intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment or prevention of disease and provides a result whose derivation is non-transparent and cannot be independently derived or verified by the end user. Draft Guidance for Industry, Clinical Laboratories, and FDA staff – Multivariate Index Assays (Rockville, MD: FDA, Center for Devices and Radiologic Health, 2007)

Efficacy  Biomarker  Discovery  &  Valida1on  

Pre-­‐Clinical  

Phase  I  

Dose  Escala1on  

Phase  I  

Extension  at  MTD  

Phase  II   Phase  III   Post-­‐Launch  

in  vivo  &  

in  vitro  

models  

1st    

Training  

1st  

 Valida1on  

2nd  Valida1on  

&  Registra1on  

Simple Biomarker (e.g. BRAF V600E)

in  vivo  &  

in  vitro  

models  

1st  

Training  

1st  

Training  

1st  

Valida1on  

2nd  Valida1on  

&    

Registra1on  

Molecular Profile

>30 >80 >200 N 0 0

N: # patients treated at or above biological effective dose

Biomarkers  for  Oncology  Targeted  Therapies  

Ph+,  KRAS,  EGFR,  KIT,  HER2,  BRAF,  ALK  

Predictive Biomarkers

CD3,  CD4,  CD5,  CD8,  CD19,CD20,  CD41,  IgA,  IgM,  IgG,  Estradiol,  Estrone,  Estrone  sulfate,  soluble  HER2,  PET  tratsuzumab,  Testosterone,  Androstenedione,  SHBG,  plasma  HDL,  Albumin,  Treg,  CD8,  CBC,  CD4+,  Caspase  3-­‐9,  Bcl2,  PDGFR,  cKIT,  ER,  PR,  Ki67,  pS2,  IgA,  IgG,  IgM,  IgG,  IgA,  IgM,  20S  proteasome,  EGFR,  pEGFR,  Ki67,p27,  pMAPK,  AKT,  pAKT  ,  kera1n  1,  STAT3,  VEGF,  FDG-­‐PET,  CT,  DCE-­‐MRI,  plasma  PLG,  CECs,  EGFR,  pEGFR,  Ki67,p27,  TGFalpha  ,  amphiregulin,  epiregulin,  EGFRvIII,  MEK,  ERK1,  pERK1,  ERK2,  pERK2,  ac1n,  Acetylated  H3,  H4,  HDAC2-­‐6,  TopoIIa,  HP1,    KRAS,  SRC,  pSRC,  pBCR/ABL,  pCRKL,    IGFR1,  pS6,  TGF-­‐alpha,  p95,  4EBP1,  p4E-­‐BP1,  eIF-­‐4G,  S6,  pS6,  IDO,  TNFalpha,  ……………..  

PD/MOA Biomarkers

Oncology  CoDx:  Nine  Drugs  Against  Six  Targets  

0  

1  

2  

3  

4  

5  

6  

FDA  Oncology  Approvals  

No  CoDx   With  CoDx  

Date   Drug   Markers  

1998   trastuzumab   HER2  

2007   lapa1nib   HER2,  EGFR  

2001   ima1nib   BCR-­‐ABL,  KIT  

2006     dasa1nib   BCR-­‐ABL  

2007   nilo1nib   BCR-­‐ABL  

2004   cetuximab   KRAS  

2006   panitumumab   KRAS  

2011   crizo1nib   EML4-­‐ALK  

2011   vemurafenib   BRAF  

Oncology  Drug  Approvals:  Room  for  Improvement  

•  >500  targeted  therapies  in  clinical  development  

–  <10%  of  therapies  entering  Phase  1  tes1ng  will  eventually  achieve  regulatory  approval  

•  Most  recently  approved  Oncology  drugs  have  only  modest  improvements  in  hazard  ra1os  (HR)  

•  Effec1ve  targe1ng  of  tumors  with  predic1ve  markers  significantly  improves  HR  in  defined  subsets:  

–  BRAF  muta1on  in  melanoma  

–  EML4-­‐ALK  transloca1on  in  NSCLC  

Hazard Ratio (HR) in randomized, controlled trial supporting 1st approved indication (data from www.fda.gov)

0.00  

0.10  

0.20  

0.30  

0.40  

0.50  

0.60  

0.70  

0.80  

0.90  

1.00  

Gleevec  

Afin

itor  

Zactem

a  

Sutent  

Zelboraf  

Votrient  

Zy1ga  

Yervoy  

Hercep1

n  

Nexavar  

Tykerb  

Torisel  

Erbitux  

Proven

ge  

Avas1n

 

Tarceva  

Iressa  

Allcomers  

Marker  +’ve  only  

Biomarkers  Can  be  the  Difference  in  Eventual  Approval  of  New  Drugs  

MOA  poorly  understood  

MOA  well  understood  

Available  clinical  biomarker  

15%   75%  

No  clinical  biomarker   5%   35%  

Adapted from E. Zerhouni – with permission

Probability of Success

Conclusion  •  Clinical  innova1on  always  takes  longer  than  expected:  

–  Biomarkers  are  no  excep1on!  –  Diseases  are  complex  and  individual  biomarker  effect  sizes  are  oken  too  small  

•  Biomarker  science  is  the  major  cause  of  the  delay:  –  When  important  markers  emerge  (e.g.  crizo1nib,  vemurafenib),  regulatory  authori1es  have  adapted  

quickly  and  adjusted  previous  requirements  to  include  them  in  the  drug  labels  

–  We  have  been  much  more  successful  with  PD/MOA  than  predic1ve  biomarkers  –  To  date,  we  have  largely  failed  to  develop  complex  molecular  profiles  as  useful  predic1ve  markers  

•  Companion  diagnos1cs  will  remain  rare  un1l  we  can  develop  more  biomarkers  with:  –  Strong  predic1ve  values  –  Evidence  they  are  predic1ve  not  prognos1c  –  Available  “fit-­‐for-­‐purpose”  assays  –  Ac1onable  data  

•  To  be  successful,  we  must  change  the  way  we  implement  biomarker  research  in  pharmaceu1cal  development:  

–  Implement  biomarker  work  much  earlier  in  the  development  plan  –  Modify  clinical  trial  design  to  enable  biomarker  discovery  valida1on  –  Demonstrate  that  biomarker  data  improves  the  drug  development  process