Diagnostics and personalized medicine


Published on

Lecture given for the biotechnology program at the Kellogg Graduate School of Business (NWU)

1 Like
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Diagnostics and personalized medicine

  1. 1. Diagnos(cs  and  Personalized   Medicine   Dynamics  in  the  Biotechnology  and   Life  Science  Industry   Tuesday,  February  6,  2007  
  2. 2. Objec(ves   •  Challenge  common  wisdom:  help  you  think   •  Prepare  you  for  pitches   •  Provide  basic  background   ©  2013  Winton  Gibbons   2  
  3. 3. Topics  to  Cover   •  Overview  of  diagnos(c  market   •  Nomenclature   •  Marker  mining  and  valida(on   •  Personalized  medicine   •  Miscellaneous  and  Q  &  A   – Recent,  major  acquisi(ons   – Point-­‐of-­‐care   ©  2013  Winton  Gibbons   3  
  4. 4. Overall  Market  Size  and  Structure   Source:  BBC,  Amersham  and  WG  analysis   In-vitro 81% In-vivo 19% 100%=$41.6  billion   ©  2013  Winton  Gibbons   4  
  5. 5. IVD  Market  Size  and  Structure   $7,997 $6,582 $6,034 $1,861 $1,827 $1,740 $1,295 $1,217 $1,228 $4,088 Diabetes Infectious Disease Clinical Chemistry Hematology Immunology Endocrinology Coagulation Cancer Cardiac Other Source:  BBC  and  WG  analysis   ©  2013  Winton  Gibbons   5  
  6. 6. IVD  Market  Size  and  Structure   Source:  BBC  and  WG  analysis   US 37% Europe 35% Japan 10% ROW 18% ©  2013  Winton  Gibbons   6  
  7. 7. IVD  Market  Size  and  Structure   Source:  BBC  and  WG  analysis   Lab 75% PST 18% Ambulatory 7% ©  2013  Winton  Gibbons   7  
  8. 8. Roche  Dominates  the  IVDs,  Especially   A`er  GE’s  Move   Roche 21% Abbott (pre GE) 12% J&J 10% Bayer (Siemens) 9% Beckman 7% Dade 6% Other 35% Source:  BBC  and  WG  analysis   ©  2013  Winton  Gibbons   8  
  9. 9. Other  Thoughts  on  Industry  Structure   •  Top  4  Diagnos(cs  players  part  of  Larger  Medical  Product  Firm  (Roche,  GE,  J&J  and   Siemens)   –  Compe((ve  Informa(on  Spoey   •  Overlap  with  Life  Science  firms   –  Diagnos(cs  uses  much  of  the  same  technology  as  Life  Sciences,  so  a  number  of   companies  straddle  both  (Beckman,  BioRad,  Cepheid,  Celera  and  even  Roche).   –  However,  Diagnos(cs  is  different  due  to  regulatory,  medical  prac(ce,  reimbursement,   razor  /  razor  blade  and  larger,  diversified  players.   •  In-­‐vitro  Diagnos(cs  is  a  large  ($34  billion),  but  generally    grows  about  the  same   rate  as  nominal  GDP;  however,  there  are  a  few  fast-­‐growing  sub-­‐sectors  and  some   niche  opportuni(es   –  Molecular  diagnos(cs  (e.g.,  DNA)   –  Blood  Glucose   –  Novel  protein  markers  (e.g.,  BNP  and  others)   ©  2013  Winton  Gibbons   9  
  10. 10. M.D.s Rx firms Device firms Dx Hospitals Pharmacies Distribution Stronger Poli(cal  Power  for  IVD  Firms  Typically   is  not  Strong   ©  2013  Winton  Gibbons   10  
  11. 11. Dimension RxL Max Chemistry/Immunochemistry Analyzer GeneXpert Triage Diagnos(c  Instruments  Vary  in  Size   and  Complexity     ©  2013  Winton  Gibbons   11  
  12. 12. Large  System  Purchases  Typically   Don’t  Depend  on  Single  Markers   • 5- to 6-year repurchase cycle • Labor savings (2/3 of cost) – Laboratory automation • 12- to 24-month selling cycle – Ease of use – Easy maintenance / reliability •Important analytes on the menu: Troponin I, HbA1c, BNP/NT-proBNP • Menu should cover 90%+ of volume high-sensitivity TSH and HCG Source: William Blair & Company, L.L.C. analysis Purchasing Behavior for Mainframe Immunodiagnostic Analyzers ©  2013  Winton  Gibbons   12  
  13. 13. Some  Myths  in  Diagnos(cs   •  Best  test   –  Standardiza(on  /  installed  based—VHS  versus  Betamax  (e.g.,  Troponin  I  versus  T;  BNP  versus  NT-­‐proBNP?)   –  Plaoorm  migra(on  (NA  to  IA  to  CC)   –  Trial  and  error  (e.g.,  sta(ns)   •  POC   –  Cost  center  versus  total  cost   –  Lab  Director  power   –  MD  office   •  Work  flow   •  Profit  (Stark  II—July  26)   •  Pharmacogenomics   –  Metabolizing  enzymes  (CYP450s)   •  Yes   •  Drug-­‐drug  interac(ons   –  Individualized  medicine   •  Not  always   •  Except  certain  cancers  or  orphans   •  Drugs  to  target  big  markets,  just  using  new  biology   ©  2013  Winton  Gibbons   13  
  14. 14. Nomenclature   •  Sensi(vity   –  Percent  with  disease  who  test  posi(ve   •  Specificity   –  Percent  of  without  disease  who  test  nega(ve   •  Posi(ve  Predic(ve  Value   –  Prevalence*Sensi(vity/(Prevalence*Sensi(vity+(1-­‐Prevalence)*(1-­‐Specificity))   •  Nega(ve  Predic(ve  Value   –  (1-­‐  Prevalence)*Specificity/((1-­‐  Prevalence)*Specificity+Prevalence*(1-­‐Sensi(vity)   •  Odds  Ra(o   –  Odds/Odds   –  Odds=p/(1-­‐p)   •  ROC  Curve   –  True  Posi(ve  Frac(on  versus  False  Posi(ve   Disease Present Disease Absent Positive Test A B A+B Negative Test C D C+D A+C B+D Sensitivity A/A+C Specificity D/B+D ©  2013  Winton  Gibbons   14  
  15. 15. Reading  List   •  Believability  of  rela(ve  risks  and  odds  ra(os  in  abstracts:  cross  sec(onal  study.   –  BMJ,  Jul  2006;  333:  231  -­‐  234   •  Evidence  of  bias  and  varia(on  in  diagnos(c  accuracy  studies   –  Can.  Med.  Assoc.  J.,  Feb  2006;  174:  469  -­‐  476.   •  Tips  for  learners  of  evidence-­‐based  medicine:  5.  The  effect  of  spectrum  of  disease  on  the  performance  of  diagnos(c  tests   –  Can.  Med.  Assoc.  J.,  Aug  2005;  173:  385  -­‐  390   •  Predic(on  of  cancer  outcome  with  microarrays:  a  mul(ple  random  valida(on  strategy.   –  Lancet.  2005;365:488-­‐92.   •  Can  Genentech  Double  Its  NHL  Franchise?  Focus  on  Fc  Receptors   –  William  Blair  &  Company  Research  Note.  December  2,  2004   •  Limita(ons  of  the  Odds  Ra(o  in  Gauging  the  Performance  of  a  Diagnos(c,  Prognos(c,  or  Screening  Marker   –  Am.  J.  Epidemiol.,  May  2004;  159:  882  -­‐  890.   •  When  can  a  risk  factor  be  used  as  a  worthwhile  screening  test?   –  BMJ,  Dec  1999;  319:  1562.   •  Drug  Metabolism  and  Variability  among  Pa(ents  in  Drug  Response   –  N.  Engl.  J.  Med.,  May  2005;  352:  2211  -­‐  2221.   •  Codeine  Intoxica(on  Associated  with  Ultrarapid  CYP2D6  Metabolism   –  N.  Engl.  J.  Med.,  Dec  2004;  351:  2827  -­‐  2831.   •  Developmental  Pharmacology  —  Drug  Disposi(on,  Ac(on,  and  Therapy  in  Infants  and  Children   –  N.  Engl.  J.  Med.,  Sep  2003;  349:  1157  -­‐  1167.   ©  2013  Winton  Gibbons   15  
  16. 16. One  Week’s  Worth  of  Gene(c   Biomarker  Discovery   •  “Gene(c  fingerprints  iden(fy  brain  tumors'  origins”  (Feb  1)     •  “Mayo  Clinic  Research  Shows  35  Percent  Of  49  Young  People  Who  Died  Suddenly  And  Inexplicably  Had   Gene(c  Heart  Defects”  (Jan  31)   •  “UCLA  Researchers  Discover  Genes  Linked  To  Lymphoma,  Opening  Way  For  New  Targeted  Drugs”  (Jan  31)   •  “Study  finds  genes  that  predict  transplant  rejec(on”  (Jan  30)     •  “A  Form  Of  The  Alcohol  Dehydrogenase  Gene  May  Protect  Afro-­‐Trinidadians  From  Developing   Alcoholism”  (Jan  30)   •  “Autoimmune  Disease  Breakthrough  Gained  By  Iden(fica(on  Of  30  Errant  Genes”  (Jan  29)     •  “Gene  'could  predict  ADHD  drug  reac(on'”  (Jan  29)     •  “50%  of  Americans  have  gene  that  affects  how  body  burns  sugar”  (Jan  28)   •  “Scien(sts  link  paternal  gene,  au(sm”  (Jan  26)     •  “Gene(c  Risk  Factor  For  Parkinson's  Found”  (Jan  25)   •  “Calculated  Risk:  Scien(sts  Discover  Gene(c  Risk  Factor  For  Smoking-­‐linked  Head  And  Neck  Cancer”  (Jan   25)   Source: National Office of Public Health Genomics (NOPHG) ©  2013  Winton  Gibbons   16  
  17. 17. Senator  Barack  Obama  Introduces  the  Genomics  and  Personalized   Medicine  Act     The  Personalized  Medicine  Coali(on  welcomes  the  introduc(on  of   S.3822,  the  Genomics  and  Personalized  Medicine  Act,  and  looks   forward  to  working  with  Senator  Barack  Obama,  the  bill's  author,   and  his  colleagues  in  Congress,  to  hasten  the  introduc(on  of   personalized  medicine.  The  legisla(on,  among  other  things,  aims  to   improve  the  coordina(on  of  public  and  private  efforts  to  facilitate   the  development  of  safer  and  more  effec(ve  drugs,  create  a   biobanking  ini(a(ve,  expand  the  genomics  workforce,  and  improve   the  quality  of  clinical  gene(c  tes(ng.   ©  2013  Winton  Gibbons   17  
  18. 18. The  Genomics  and  Personalized   Medicine  Act  of  2006     •  Sponsoring  Research.    The  bill  sets  aside  $150  million  to  sponsor  research  on  genomics.    It  enables   a  na(onal  biobanking  ini(a(ve  and  sets  up  a  system  to  pool  and  synthesize  genomic  data  from   local  sources.  This  act  establishes  an  interagency  task  force  to  accelerate  the  transla(on  of   research  into  medical  prac(ce.    Finally,  the  legisla(on  invests  in  the  next  genera(on  genomics   workforce  by  encouraging  the  recruitment  and  reten(on  of  genomic  professionals,  and  promotes   the  integra(on  of  genomics  across  all  clinical  and  public  health  disciplines.   •  Encouraging  InnovaAon.    The  legisla(on  provides  a  100  percent  tax  credit  for  the  development  of   companion  diagnos(c  tests  that  can  improve  the  effec(veness  or  safety  of  certain  drugs.      In   addi(on,  the  Na(onal  Academies  will  conduct  a  study  to  determine  what  addi(onal  incen(ves  are   needed,  and  how  they  should  be  structured.   •  Modernizing  the  FDA  and  CMS.    The  bill  requires  that  FDA  and  CMS  study  and  update  regulatory   processes  to  assure  the  quality  of  genomic  tests  through  improved  oversight  and  regula(on.   •  ProtecAng  Consumers.    The  legisla(on  protects  consumers  by  reaffirming  Congress  commitment  to   stopping  gene(c  discrimina(on  and  protec(ng  gene(c  privacy.  In  addi(on,  direct-­‐to-­‐consumer   marke(ng  of  gene(c  tests  would  receive  greater  scru(ny  and  regula(on.     ©  2013  Winton  Gibbons   18  
  19. 19. Cytochrome  p450  Enzymes   •  The  superfamily  has  undergone  divergent  evolu(on,  and   the  ancestral  gene  is  likely  2  to  3  1/2  billion  years  old.   •  The  recent  'burst'  in  new  P450  genes,  par(cularly  in  the  II   family  during  the  past  800  million  years,  appears  to  be  the   result  of  'animal-­‐plant  warfare'.   •  Due  to  the  presence  or  absence  of  a  par(cular  P450  gene  in   one  species  but  not  the  other,  it  may  not  be  correct  to   extrapolate  toxicity  or  cancer  data  from  rodent  to  human.   •  Increases  in  the  P450  gene  product  (enzyme  induc(on)   almost  always  reflect  an  elevated  rate  in  gene   transcrip(on,  although  there  are  several  excep(ons.   ©  2013  Winton  Gibbons   19  
  20. 20. Posted  on:  Monday,  22  January  2007,  21:00  CST   GENETIC  MEDICINE  ;  Some  Heart  PaAents  Get  DNA  Tests  to  Determine  Correct  Drug  Dose     By  Linda  A.  Johnson     Personalized  medicine,  the  tailored  treatments  that  a  few  pa(ents  now  get  based  on  their  own   DNA,  is  finally  headed  for  the  masses:  the  many  heart  pa(ents  at  risk  of  deadly  blood  clots.     At  least  2  million  Americans  with  an  abnormal,  clot-­‐triggering  heart  rhythm  take  the  pill  warfarin,   also  sold  as  Coumadin.     Gewng  too  liele  can  lead  to  a  stroke,  and  too  much  can  cause  life-­‐threatening  bleeding.  To  find   the  right  dose  for  each  pa(ent,  doctors  use  trial  and  error  -­‐-­‐  and  the  errors  lead  to  tens  of   thousands  of  hospitaliza(ons  and  deaths  every  year.     Star(ng  this  month,  about  1,000  pa(ents  who  have  a  condi(on  known  as  atrial  fibrilla(on  will   take  part  in  a  project  that  will  match  their  Coumadin  dose  to  their  specific  gene(c  needs.     This  gene(c  fingerprin(ng  should  single  out  the  many  people  whose  bodies  break  down  warfarin   faster  or  slower  than  normal,  and  their  doctors  can  immediately  adjust  their  dosage  to  prevent   dangerous  complica8ons.     "Twenty  percent  to  30  percent  of  people  are  either  very  fast  or  very  slow"  to  metabolize  many   drugs  but  don't  know  it,  said  Dr.  Robert  Epstein,  chief  medical  officer  at  prescrip(on  benefit   manager  Medco  Health  Solu(ons  of  Franklin  Lakes,  N.J.,  which  is  collabora(ng  in  the  effort  with   the  Mayo  Clinic,  based  in  Rochester,  Minn.   ©  2013  Winton  Gibbons   20  
  21. 21. Effect  of  CYP450  Muta(ons   •  Rapid  metabolizers   –  Carry  mul(ple  copies  (3-­‐13)  of   func(onal  alleles  and  produce  excess   enzyma(c  ac(vity   •  Normal  metabolizers   –  Possess  normal  func(onal  alleles   •  Intermediate  metabolizers   –  Possess  one  reduced  ac(vity  allele  or   one  null  allele   •  Poor  metabolizers   –  Carry  two  mutant  alleles  which  result   in  complete  loss  of  enzyme  ac(vity   •  2D6  gene  duplica(ons   –  Ethiopians  16.0%   –  Saudi  Arabians  10.4%   –  Spaniards  10%   –  Italians  8.3%   –  Zimbabweans  2%   –  Germans  1.8%   –  Chinese  1.3%   •  2D6  Intermediate  and  poor  metabolizers   –  Caucasians  7-­‐8%   –  Japanese  ~1%   –  Chinese  ~1%   –  African  Americans~6%   Source:  Roche  ©  2013  Winton  Gibbons   21  
  22. 22. CYP450  Substrates  (drugs)   1A2 2B6 2C8 2C19 2C9 amitriptyline bupropion paclitaxel Proton Pump Inhibitors: NSAIDs: caffeine cyclophosphamide torsemide lansoprazole diclofenac clomipramine efavirenz amodiaquine omeprazole ibuprofen clozapine ifosfamide cerivastatin pantoprazole lornoxicam cyclobenzaprine methadone repaglinide rabeprazole meloxicam estradiol E-3810 S-naproxen=>Nor fluvoxamine piroxicam haloperidol Anti-epileptics: diazepam=>Nor suprofen imipramine N-DeMe phenytoin(O) mexilletine S-mephenytoin Oral Hypoglycemic Agents: naproxen phenobarbitone tolbutamide olanzapine glipizide ondansetron amitriptyline phenacetin=> carisoprodol Angiotensin II Blockers: acetaminophen=>NAPQI citalopram losartan propranolol clomipramine irbesartan riluzole cyclophosphamide ropivacaine hexobarbital Sulfonylureas: tacrine imipramine N-DeME glyburide/ theophylline indomethacin glibenclamide tizanidine R-mephobarbital glipizide verapamil moclobemide glimepiride (R)warfarin nelfinavir tolbutamide zileuton nilutamide zolmitriptan primidone amitriptyline progesterone celecoxib proguanil fluoxetine propranolol fluvastatin glyburide teniposide nateglinide R-warfarin=>8-OH phenytoin=>4-OH rosiglitazone tamoxifen torsemide S-warfarin Source:  CYTOCHROME  P450  DRUG-­‐INTERACTION  TABLE-­‐-­‐Last  Updated:  Tue  Oct  17  2006-­‐-­‐Indiana University Department of Medicine, Division of Clinical Pharmacology ©  2013  Winton  Gibbons   22  
  23. 23. CYP450  Substrates  (drugs)-­‐-­‐con(nued  2E1 Beta Blockers: alprenolol Anesthetics: Macrolide antibiotics: Steroid 6beta-OH: carvedilol amphetamine enflurane clarithromycin estradiol S-metoprolol aripiprazole halothane erythromycin (not 3A5) hydrocortisone propafenone atomoxetine isoflurane NOT azithromycin progesterone timolol bufuralol methoxyflurane telithromycin testosterone chlorpheniramine sevoflurane Antidepressants: chlorpromazine Anti-arrhythmics: alfentanyl amitriptyline codeine (=>O-desMe) acetaminophen quinidine=>3-OH (not 3A5) aprepitant clomipramine debrisoquine =>NAPQI aripiprazole desipramine dexfenfluramine aniline Benzodiazepines: buspirone imipramine dextromethorphan benzene alprazolam cafergot paroxetine duloxetine chlorzoxazone diazepam=>3OH caffeine=>TMU encainide ethanol midazolam cilostazol Antipsychotics: flecainide N,N-dimethyl formamide triazolam cocaine haloperidol fluoxetine theophylline codeine- N-demethylation perphenazine fluvoxamine =>8-OH Immune Modulators: dapsone risperidone=>9OH lidocaine cyclosporine dexamethasone thioridazine metoclopramide tacrolimus (FK506) dextromethorphan zuclopenthixol methoxyamphetamine docetaxel mexilletine HIV Antivirals: domperidone minaprine indinavir eplerenone nebivolol nelfinavir fentanyl nortriptyline ritonavir finasteride ondansetron saquinavir gleevec oxycodone haloperidol perhexiline Prokinetic: irinotecan phenacetin cisapride LAAM phenformin lidocaine promethazine Antihistamines: methadone propranolol astemizole nateglinide sparteine chlorpheniramine odanestron tamoxifen terfenidine pimozide tramadol propranolol venlafaxine Calcium Channel Blockers: quetiapine amlodipine quinine diltiazem risperidone felodipine NOT rosuvastatin lercanidipine salmeterol nifedipine sildenafil nisoldipine sirolimus nitrendipine tamoxifen verapamil taxol terfenadine HMG CoA Reductase Inhibitors: trazodone atorvastatin vincristine cerivastatin zaleplon lovastatin ziprasidone NOT pravastatin zolpidem simvastatin 2D6 3A4,5,7 Source:  CYTOCHROME  P450  DRUG-­‐INTERACTION  TABLE-­‐-­‐Last  Updated:  Tue   Oct  17  2006-­‐-­‐Indiana University Department of Medicine, Division of Clinical Pharmacology ©  2013  Winton  Gibbons   23  
  24. 24. CYP450  Inhibitors   Source:  CYTOCHROME  P450  DRUG-­‐INTERACTION  TABLE-­‐-­‐Last   Updated:  Tue  Oct  17  2006-­‐-­‐Indiana University Department of Medicine, Division of Clinical Pharmacology 1A2 2C19 2C9 2D6 3A4,5,7 amiodarone chloramphenicol amiodarone amiodarone HIV Antivirals: cimetidine cimetidine fenofibrate bupropion delaviridine ciprofloxacin felbamate fluconazole celecoxib indinavir fluoroquinolones fluoxetine fluvastatin chlorpheniramine nelfinavir fluvoxamine fluvoxamine fluvoxamine chlorpheniramine ritonavir furafylline indomethacin isoniazid chlorpromazine interferon ketoconazole lovastatin cimetidine amiodarone methoxsalen lansoprazole phenylbutazone citalopram aprepitant mibefradil modafinil omeprazole probenicid clemastine NOT azithromycin oxcarbazepine sertraline clomipramine chloramphenicol 2B6 probenicid sulfamethoxazole cocaine cimetidine thiotepa ticlopidine sulfaphenazole diphenhydramine clarithromycin ticlopidine topiramate teniposide doxepin diethyl- dithiocarbamate voriconazole doxorubicin diltiazem 2C8 2E1 zafirlukast duloxetine erythromycin trimethoprim diethyl- dithiocarbamate escitalopram fluoxetine fluconazole quercetin disulfiram halofantrine fluvoxamine glitazones histamine H1 receptor antagonists gestodene gemfibrozil hydroxyzine grapefruit juice montelukast levomepromazine imatinib trimethoprim methadone itraconazole metoclopramide ketoconazole mibefradil mifepristone midodrine nefazodone moclobemide norfloxacin paroxetine norfluoxetine perphenazine mibefradil quinidine star fruit ranitidine verapamil red-haloperidol voriconazole ritonavir sertraline terbinafine ticlopidine tripelennamine ©  2013  Winton  Gibbons   24  
  25. 25. CYP450  Inducers   Source:  CYTOCHROME  P450  DRUG-­‐INTERACTION  TABLE-­‐-­‐Last  Updated:  Tue  Oct  17  2006-­‐-­‐Indiana University Department of Medicine, Division of Clinical Pharmacology 1A2 2C19 3A4,5,7 broccoli carbamazepine HIV Antivirals: brussel sprouts norethindrone efavirenz char-grilled meat NOT pentobarbital nevirapine insulin prednisone methyl cholanthrene rifampin barbiturates modafinil carbamazepine nafcillin 2C9 efavirenz beta- naphthoflavone rifampin glucocorticoids omeprazole secobarbital modafinil tobacco nevirapine 2D6 phenobarbital 2B6 dexamethasone phenytoin phenobarbital rifampin rifampin rifampin St. John's wort 2E1 troglitazone 2C8 ethanol oxcarbazepine rifampin isoniazid pioglitazone rifabutin ©  2013  Winton  Gibbons   25  
  26. 26. Personalized  Medicine-­‐Pros  and  Cons   •  Desire  for  efficacy   •  Desire  for  safety   •  Because  we  can…   •  Clinical  study  size  and   cost   –  Efficacy   –  Safety   •  COGS   •  Marke(ng  expense   •  Design  around   –  Dosing  (case  studies)   –  Ligand  binding  (case  study)   –  Alterna(ve  target  or  MOA   ©  2013  Winton  Gibbons   26  
  27. 27. Future Rx Targets Not Likely; Orphan Drugs Existing Rx Today Practical with Proper Data Only Dosing Modifications for Metabolism One-Size Rationalize Personalize OneDrug AFew Drugs Many Drugs Economic Pressure Scientific Pressure Outlook  for  Personalized  Medicine   ©  2013  Winton  Gibbons   27  
  28. 28. Cases   •  Rituxan  and  NHL   •  MRI  for  stroke  (Lancet  2007)   •  Gene  signature  for  breast  cancer  (NEJM  2007)   ©  2013  Winton  Gibbons   28  
  29. 29. Thoughts  on  the  Cases   •  Read  the  literature  and  do  your  homework   •  Design  around  personaliza(on   –  Genitope  and  Favrille  s(ll  have  a  chance   •  Understand  the  clinical  environment   •  Odd  Ra(os   •  Comparison  against  normals  versus  mimics  or  common   differen(al  (all  comers)   •  Over-­‐fiwng   –  Degrees  of  freedom   –  Algorithms   –  Sta(s(cal  tests   •  Valida(on,  valida(on,  valida(on   ©  2013  Winton  Gibbons   29  
  30. 30. Performance  of  Rituxan  Varies  by  Fc   Polymorphism   ©  2013  Winton  Gibbons   30  
  31. 31. An(-­‐idiotype  Vaccines  S(ll    Appear  to   Perform  Best     ©  2013  Winton  Gibbons   31  
  32. 32. High Heaven Med Orphan Drugs Low Hell Low Med High (Number and Effect) Power of Target Degree of Polymorphism Efficacy Higher Throughput Screening Needed Target  Selec(on  Strategy   ©  2013  Winton  Gibbons   32  
  33. 33. Almost  20%  of  ERs  May  Not  Have  Any   Access  to  MRI     Source:  Biosite  Investor  R&D  Day  2006   ©  2013  Winton  Gibbons   33  
  34. 34. Available  MRIs  Take  Long  and  Are  Not   Available  All  Shi`s   Source:  Biosite   Investor  R&D  Day   2006   ©  2013  Winton  Gibbons   34  
  35. 35. Odds  Ra(os  Must  Be  Quite  High  to  Be   Useful   Limitations of the Odds Ratio in Gauging the Performance of a Diagnostic, Prognostic, or Screening Marker. Am. J. Epidemiol., May 2004; 159: 882 - 890. ©  2013  Winton  Gibbons   35  
  36. 36. Sources  of  Bias  in   DiagnosAcs   ©  2013  Winton  Gibbons   36   Evidence of bias and variation in diagnostic accuracy studies. Can. Med. Assoc. J., Feb 2006; 174: 469 - 476.
  37. 37. Q  &  A  
  38. 38. •  LinkedIn   – hep://www.linkedin.com/in/wintongibbons/   •  Twieer   – @wingibbons   •  Blog   – hep://www.wingibbons.wordpress.com     ©  2013  Winton  Gibbons   38