The use and impact of novel technologies for risk assessment of pharmaceuticals
Ian Cotgreave, AstraZeneca
From in silico modeling to stem cells and multi-dimensional in vitro models
SFT Årsmöte, Nobel Forum KI 22 Maj 2013
Scope of the lecture
Comtempory socio-political demands on pharmaceutical(and chemical) risk assessment
Increasing need for early prediction of risk in the make-test cycle pharma industry.
The ”3Ms” of moden toxicology: Models, Mechanismsand Markers and demands on technological innovation
Contemporary predictive modelsIn silico screensIn vitro screens
Emerging technologies and modelsBiological innovation in stem cell based-modelsTissue engineering in multicellular/organotypic models
2
Name | Date
8–12 år från idé till färdigt läkemedel
The research process in drug discovery and development
Prekliniska studier Kliniska prövningar
KEMI/FARMAKO-LOGI
IND* FAS I FAS II FAS III NDA** FAS IV
Sökande efteraktiva substanser
Toxikologi, effektstudier på olika djurslag
Myndighets-behandling
Effektstudier på friska försöks-personer
Patientstudier i begränsad skala
Jämförande studier på stort antal patienter
Myndighets-behandling
Fortsatta jämförande studier
50–150individer
100–200patienter
500–5 000patienter
Registrering, introduktion på marknaden**New Drug
ApplicationAnsökan om att få marknadsföra ett nytt läkemedel
KUNSKAPS-NIVÅ
2–4 år 2–6 mån 3–6 år 1–3 år
TIDSÅTGÅNG
Discovery Development
KUNSKAPS-NIVÅ
*Investigational New DrugAnsökan om att få ge ett nytt läkemedel till människa
Productivitygap
Phase Preclinical ‘Nonclinical’ Phase IAZ
Phase I Phase I-III Phase I-III Phase III/ Marketing
Post-Marketing
Post-Marketing
Information: Causes of attrition
Causes of attrition
Dose-limiting ADRs
Serious ADRs
Causes of attrition
Causes of attrition
ADRs on label Serious ADRs Withdrawal from sale
Source: ABPI (2008) Car (2006) AstraZeneca (2000-2009)
Sibille et al. (1998)
ABPI (2008) Olson et al. (2000)
BioPrint® (2006)
Budnitz et al. (2006)
Stevens & Baker (2008)
Sample size: 156 CDs stopped
88 CDs stopped 14 CDs with dose-limiting ADRs
1,015 subjects
63 CDs stopped
82 CDs stopped
1,138 drugs 21,298 patients
47 drugs
Cardiovascular: 24% 27% 7% 9% 35% 21% 36% 15% 45%Hepatotoxicity: 15% 8% 0% 7% 29% 21% 13% 0% 32%
Haematology/BM: 3% 7% 0% 2% 3% 4% 16% 10% 9%Nervous system: 12% 14% 71% 28% 2% 21% 67% 39% 2%
Immunotox; photosensitivity:
7% 7% 0% 16% 10% 11% 25% 34% 2%
Gastrointestinal: 5% 3% 36% 23% 2% 5% 67% 14% 2%Reprotox: 9% 13% 0% 0% 5% 1% 10% 0% 2%
Musculoskeletal: 8% 4% 0% 0% 5% 1% 28% 3% 2%Respiratory: 1% 2% 0% 0% 2% 0% 32% 8% 2%
Renal: 6% 2% 0% 0% 5% 9% 19% 2% 0%
Genetic tox: 5% 5% 0% 0% 0% 0% 0% 0% 0%
Carcinogenicity: 0% 3% 0% 0% 3% 0% 1% 0% 0%
Other: 4% 0% 0% 0% 2% 4% 16% 2% 2%
Adapted from Redfern WS et al. SOT 2010 Poster 1081 1-9% 10-19% >20%0%
The various toxicity domains have been ranked first by contribution to products withdrawn from sale, then by attrition during clinical development. Note general agreement between pairs of equivalent studies.
So where does toxicology fit in the R+D process? Regulatory toxicology versus ”front-loading” toxicology
År
1 162 3 4 5 6 7 8 9 10 11 12 13 14 15
Förstapatentansökan
Ansökan omklinisk prövning
Ansökan omregistrering
Explorativ forskning (Discovery) Läkemedelsutveckling (Development) Identifiering avmåltavlor ochledsubstanser
Optimering avledsubstans
Test avterapikoncept
Klinisk utveckling Lansering
Kliniska prövningarFas I50-150personer
Fas II100-200personer
Fas III500-5 000personer
Fas IV, fortsatta studier
Forskningsstöd under livscykeln
Toxikologiska och farmakokinetiska studier(absorption, distribution, metabolism, utsöndring)
Farmaceutisk och analytisk utveckling
Processutveckling och tillverkning
Registreringsarbete
Försäljning och marknadsföring (inkl planering och förberedelser)
Antal substanserUpp till 1 000 000 10-15 1-8 1-3 1
Early prediction”front-loading”
Regulatory plus problem-solving
”Safety front-loading”: A truely Integtrated Testing Strategy
103 - 107
Computational predictions10 - 103
In Vitro toxicity Screening1-100
1-10In Vivo Confirmatory Assay
1Regulatory Safety Assessment
Synthesise
Test Evaluate
Drug Discovery: The make-test cycle
CandidateDrug
Drug Discovery:A process of iterative chemical synthesis, design and biological testing
Design
screening criteria•Potency•DMPK•Phys. Chem•Toxicity
Two weeks!!
Predictive toxicology: Animals, cells, computers or all three?
Is this safe??
The predicament:
The ”3 Ms” of the ”new toxicology” paradigmModels
Mechanisms
Markers
What is Computational Toxicology?
“Modeling and Informatics of Safety Endpoints”
Provide computational Safety support to project all
through the Discovery and Development phases – tools
for:
- Predictions
- Structuring what we already know
- Hypothesis generation
Tobias Noeske AZ Mölndal
Industrial Perspective – computational work
Hit ID Lead ID
DISCOVERY DEVELOPMENT
Lead Opt Cand Drug FTIM LaunchPhI,II&III
CLINICAL
IND NDA
Remove unwantedfragments
Prioritize for synthesis
Optimizeproperties
Problem solving
Assessment of Impurities and Regulatory usage
Molecular descriptor is any molecular property to characterize the molecule to search through a database, to calculate another molecular property, etc.
"The molecular descriptor is the final result of a logic and mathematical procedure which transforms chemical informationencoded within a symbolic representation of a molecule into a useful number or the result of some standardized experiment."Todeschini and Consonni, Handbook of Molecular Descriptors, Wiley-VCH, 2000.http://www.moleculardescriptors.eu/books/handbook.htm
Molecular Description
Molecular Description
Intinsic Molecular Properties(logP, charge)
Size/Shape/Flexibility
Surface Properties&
Pharmacophoric/Stuctural Features
Molecular Description
O
OH
Intrinsic Properties Size & Shape Surface & ’Features’
D e s c r i p t o r V e c t o r’Feature Similarity’ most common
Structure Activity Relationship (SAR)
15
History:
Galileo Galilei 1564-1642
In order to introduce order into the universe man must pay attention to the quantitative aspects of his surroundings and discover the underlying mathematical relationships that exist between them.
Once these relationships were known certain consequences could be deduced and then verified by experiment.
Focused Data Sets
Diverse Data Sets
Pattern RecognitionProfiling
Classification
Prediction
Clustering
Rule Extraction
QSAR
Com
poun
d gr
oupi
ng
A
naly
sis
SAR to QSAR
ta100 ta1535 ta98 ta1537
Stru
ctur
al
desc
ripto
rs
biological profile
Functional Group Association
Deaths
Drug Withdrawals
Label Warnings
hERG/QT/TdP - A Universal Problem
Strictly Regulated
13 May 2008
KEN460 - QSAR – Scott Boyer
AstraZeneca QSAR Modelling
PLS—linear
techniqueIf x <5 If x >=5
CART
+
Consensus(Average)
Model
QSARModelling
Techniques—robust methods
Used 2D descriptorsNR2
R3
X R1
LOGD65LOGD74AROMLOGPHMO_RESON__ENERGYNUM_RINGSNONPOLAR_COUNT
Docking Score
TraditionalDescriptors
PharmacophoreFeatures
HiddenInput Outputwij
f(Σsiwij)
Neural Networks
TerminalNode
INACTIVE
TerminalNode
INACTIVE
TerminalNode
INACTIVE
TerminalNodeACTIVE
Leaf Node
N = 154
Leaf Node
N = 274
Leaf NodeN = 810
TerminalNode
ACTIVE
Root NodeN = 1203
Decision Trees
-10
-5
0
5
-20 -10 0 10
t[3]
t[1]
PLS
ConsensushERG
Prediction
ConsensushERG
Prediction
AstraZeneca hERG QSAR:Diverse Molecular Descriptors and Statistical Methods to Generate a ’Consensus’ Prediction
POL_SURF_AREANEGCHARGE_GASTPOSCHARGE_GASTCHARGE_GASTDIPOLE_MOMENTMOL_VOLUMEEtc…………
2003 2005 2007
2009 2011
hERG IW global screening results
Real Impact?
No ’non-negative’ thorough QT Studies since 2007 in AZ!
General Screening Strategy Genetic Toxicology
103 - 107Computational
10 - 103SOS umu
1-100Ames
1-10In Vivo Confirmatory Assay
1Preclinical Safety Assessment
Testing Strategy
DISCOVERY DEVELOPMENT
LI FTIMLO
ComputationalFilters / QSAR
In VitroAmes
Cand
Potential Genotoxic Impurities
In Vitro/In Vivo Tests As appropriate
Carcinogenicity
Computational
In VitroMouse Lymphoma
In VivoMicro Nucleus
Application throughout the whole pipeline including post launch (degradation product).
QSAR Prediction
Tested Structural Near Neighbours
QSAR Interpretation
Structural Alerts
Consensus Model
60
16
5
119
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Active Inactive
Predicted Class
The Genetox ”Early Warning System”
Databases – The Raw Materials
Assays
Ames
Mouse Lymphoma
Lymphoma/Ovarian Cell
Chomosome Aberration
Micronucleus Assay
Carcinogenicity
Data Sources
AstraZeneca
NLM (CCRIS, GeneTOX)
NTP
Leadscope
MultiCASE
Genetic Tox Impact
?
Development of Computational Model sto Predict Genetic Toxicity
<10% of new molecules checked in experimental model
All new compounds checked in Computational model
2000-2005 2006-2011
Genetic Tox = 10-20% of all Preclinical Safety Failures
Genetic Tox = 0% of all PreclinicalSafety Failures
Translational InformaticsNauseaInvestigate if there are preclinical signals that can be linked to CLINICALnausea ?
Joanna Parkinson, Daniel Muthas, Matthew Clark, Scott Boyer, Jean-Pierre Valentin, and Lorna EwartToxicol. Sci. 126(1), 275–284 (2012)
abacavir
alosetron
alprostadil
altretamine
amcinonide
busulfan
calcitriol
capecitabine
certolizumab
chloramphenicol
cinacalcet
cisplatin
clofarabine
desonidedexamethasone
dexmedetomidinedihydroergotamine
dipyridamole
dorzolamide
duloxetine
enfuvirtide
epinastine
ertapenem
erythromycin
estazolam
etomidate
exenatide
famciclovirfelbamate
fesoterodine
finasteride
flavoxate
flecainide
flunisolide
fluticasone
fomepizolefomivirsen
fondaparinux
foscarnet
glucagon
guaifenesin
haloperidol
hexachlorophene
iloprost
irinotecan
isoniazid
lanthanum
lapatinib
lubiprostone
maprotiline
mebendazole
mefloquine
methoxsalen
micafungin
midodrine
nadolol
nicotine
nilotinib
nystatin
oxaliplatin
oxazepam
oxybate
palonosetron
paroxetine
peginterferon
phenylbutazone
pramlintide
ribavirin
rifaximin
riluzole
sapropterin
sorbitol
succimer
telithromycin
terconazole
thalidomide
theophylline
thiabendazole
tinidazole
topotecan
triamcinolone
trimetrexate
trovafloxacin
vareniclinevorinostat
warfarin
test
Compound1
Compound10
Compound11
Compound12
Compound13
Compound14
Compound15Compound16
Compound17Compound18
Compound19
Compound2
Compound20
Compound3
Compound4
Compound5
Compound6
Compound7
Compound8
Compound9
Vom
iting
Dia
rrho
ea
Saliv
ary
hype
rsec
retio
nVx
D
VxSH
DxS
HVx
DxS
HC
onst
ipat
ion
Gas
troin
test
inal
Ulc
erG
astri
tisG
astro
inte
stin
al d
isor
der
VxC
DxC
Panc
reat
ic d
isor
der
SHxC
Gas
troin
test
inal
hae
mor
rhag
e
Gas
tric
diso
rder
Saliv
ary
glan
d di
sord
er
Gas
troin
test
inal
infla
mm
atio
n
Abdo
min
al d
iste
nsio
n
Abno
rmal
faec
es
Oes
opha
geal
dis
orde
r
Faec
al v
olum
e de
crea
sed
Gas
troin
test
inal
toxi
city
Gas
troin
test
inal
ero
sion
Meg
acol
on
Gas
troin
test
inal
nec
rosi
s
Um
bilic
al h
erni
a
Gas
troin
test
inal
muc
osal
dis
orde
r
Perit
onitis
Inte
stin
al d
ilata
tion
Dendrogram
• 86 known drugs (nausea & non-nausea) clustered
• 20 AZD compounds that have reached Phase I used in blinded validation
• 90% positive and negative prediction
Bioinformatics Support: Problem solving and hypothesisgeneration from in-house data
”My compound causescholangitis in dog, has this lesion been observed beforeand is there any data on
recovery, internal or external?”
”Has the kidney lesion we’veobserved ever been seen in conjunction with our targetbefore? What about the
compound class?”
uBMO
Comprehensive list of search terms
PharmaConnect
T-lab
MedLine
Customized text mining
Compound Study
BMOTarget
FDA
EMEA EPAR
Elucidating the potential relations between a compound and a finding by adding the missing bits and pieces.
TargetGene
PathwayCompound Phenotype
Bioinformatics:Integrated searches of externalsources
Two examples of contemporary cellular screening paradigms aimed at the major organs of attritionfor safety:
Liver
Heart
What is the problem in addressing liver toxicity? DILI mechanisms are complex
These include:Reactive metabolite (RM) formation• Most often produced via CYP metabolism• Organ toxicity occurs via complex
mechanisms (glutathione depletion, oxidative stress, innate immune activation, adaptive immune activation etc.)
Mitochondrial impairment• Many options (e.g. uncoupling, altered lipid
metabolism, reactive oxygen species, mtDNA damage)• May be a primary or secondary
mechanismInhibition of biliary efflux (BSEP, Mrp2)• Impaired bile flow leads to intrahepatic
accumulation of toxic bile saltsImmune activation• Innate immune activation is required for
Type A DILI (e.g. acetaminophen)• Adaptive immune activation is a key
mechanism of idiosyncratic DILI: drug-specific antibodies & T cells (e.g. halothane, tienilic acid)
Lee; N Engl J Med 2003, 349: 474-85. Russmann et al., Curr Med Chem. 2009, 16: 3041-3053. Kaplowitz, Nat Rev Drug Discov 2005, 4: 489-99.
31
Role of RM in toxicity
Biotransformation
DNAProtein ROSe.g. carbamazepine oxide
Chemical reactivity
or instability
Macromolecular interaction
Reproducible and dose dependentpre-clinical tox
Genotoxicity
Yes
No
Idiosyncratic toxicity
Intrinsic toxicity
Reactive intermediate
R Thompson AZ Mölndal
Hepatic Screening Panel (=computational+ molecular + cellular screens)
Cellular Screen Endpoint Assessed Output
THLE* toxicityToxicity to THLE-Null (CYP independent) THLE Null EC50
THLE-3A4 (CYP potentiated) toxicity THLE 3A4/Null EC50 ratio
HepG2 MitoToxHepG2 toxicity in glucose (mito-independent) vs. galactose (mito-dependent) media
HepG2 glu/gal EC50 ratio
BSEP inhibition Inhibition of human BSEP transport activity hBSEP IC50
Mrp2 inhibition Inhibition of rat Mrp2 transport activity rMrp2 IC50
*SV40 - T antigen immortalised Human Liver Epithelial Cells
•Computational models:
Reactive Metabolite Warning System
Genetox Warning System
Molecular screens:Chemical trapping2ndary pharm
Reactive Metabolites
Disposition and tissue exposure
Mitochondrial Impairment
etc.
Immunemechanisms
Liver ToxicityTransporter interactions
Integrated in vitro Hepatic Signal Assessment
Estimated RM Body Burden (mg/day)
0
1
2
3
4
0,0050,01 0,05 0,1 0,5 1 5 10 25
Integrated in vitro hepaticsignal detected
Integrated in vitro hepatic signal detected
Integrated in vitro hepatic signal detected
Integrated in vitro hepatic signal not detected
Sign
als
in t
he H
eSP
Where is the industry going?AZ strategy concept presented in 2010
Reactive Metabolites
Disposition and tissue exposure
Mitochondrial Impairment
etc.
Immunemechanisms
Transporter interactions
Liver Toxicity
Mitotox
BSEP
MRP2
THLEs
TDI
AG t1/2CVB
‘RM Trapping’
BRI VIII, Barcelona, July 2010
Strategy concept published with hepatotoxicity focus
Has this been successful?
Type I hepatotoxicity: Emerging success, No surprises in current early phase Portfolio
Type II hepatotoxicity: Too early to demonstrate
Example 2: The heart: contractility changes‘Whole-Cell, Label-Free Assay Technology
Impedance-based
MorphologyCell-cell contact
Adhesion
Principles:
Cellular Events:
Instruments:ECIS- Applied BioPhysicsCellKey-Molecular DevicesxCELLigence- ACEA
High Sensitivity!
Giaever I, Keese CR.PNAS 1991, 88:7896 Nature 1993, 366: 591
Impedance-based Sensitivity (Applied BioPhysics ECIS)
• ATP & actin polymerization dependent → cellular micromotion
• 1 nanometer resolution (Light microscopy ~250nm, Cell membrane 3nm)
• 2+ orders of magnitude increase in sensitivity → new opportunities
Pharmacological data quality- Precision & Dynamic Range
Time (hr)
•Excellent precision• spanning wide potency range• including increase/decrease beat rate
(avg of 3 wells ± SEM)
BayK – Ca++ channel activator
DobutamineIsoproterenol
TTX– Na+ channel blocker
Carbachol– mAChR agonist
Adrenergic agonists
Impact??? We will see!
Emerging technologies: Innovations in stem cell biology andapplication to safety screening and risk assessment
Björquist et al, 2008, Drug Discovery World
hESC – A source of progenitors and specialized cells
Applications for:• Research• Drug Discovery• Toxicology• Therapies
Pluripotentstem cell colony
”Progenitor” cells
Functional cells
Phase IDefinitive Endoderm
(DE)Day 1-5
hESCPhase II
Hepatic Progenitor (DE-Hep prog)
Day 6-18
Activin A FGF2
FGF1&2BMP2&4
EGFHGF
growthfactors
FGF2EGFHGF
OsMDexM
Phase IIIHepatocyte
(DE-Hep)Day 19-28
Liver cells from stem cell origin?…..
QPCR:-All Phase I enzymes expressed-Phase II enzymes highly expressed-MRP2-OATP2-CK18-α1-antitrypsin-albumin-LFABP-CYP7A1 (Cholesterol-7α-hydroxylase)-glucose 6 phosphatase-alcohol dehydrogenase 1-tyrosine aminotransferase (TAT)-foxA2
Urea synthesis
Albumin secretion
DE-Hep
+ +
Phase III Mature hepatocytes
Day 18-25
Characterization of hES-Hep
Functional properties expressed
Cellartis /data in press
Some characteristics of the liver cells from hESC source
Another more exotic example: lung progenitor cells
• Isolated as colony-forming cells with co-culture of mouse embryonicfibroblasts
• Combined phenotypes with mesenchymal stem cells and Type II cells• Generating both mesenchymal cells (fibroblast, etc.) and Type II cells• Proliferating within Type II cell-hyperplasia in fibrotic lungs
Progenitor cells
SP-C CD90
SP-C CD90 a-SMA
Epithelialdifferentiation
Mesenchymaldifferentiation
Type II cells FibroblastsFujino N, et al. Lab Invest2011;91:363
Facts and figures
•Full project title: Stem cells for biological assays of novel drugs and predictive toxicology •Start date: 1st October 2012 •Duration: 5 years •Total cost: € 55.6 million •Project coordinator: F. Hoffmann-La Roche Ltd •Managing entity: University of Oxford
Patient-derived stem cells for drug discovery and safety
hESCs and iPSCs are Pluripotent: They have the potential to differentiate into all tissue of an adult.
MuscleHeart
Vessels
Blood
KidneyBone Skin
Nervous System
Liver
Pancreas
Emerging technologies: Tissue engineering and organotypic toxicity models
Tommy Andersson AZ Mölndal
http://wyss.harvard.edu/
Yvonne dragan AZ Boston
Every organ on a chip: Joined together!!
Seurat-1 Research Initiative
Towards the replacement of in vivo repeated dose systemictoxicity testing
Joint funding by the European Commission and a specific industrial sector (cosmetics industry / Colipa)
€ 25 million EC & € 25 million Colipa
OBJECTIVESDevelopment of an innovative concept for repeated dose systemic toxicity testing.
Proof of concept for a future full implementation of a mode-of-action strategy.
Development of innovative testing methods more predictive than existing testing procedures.
~ 70 research groups from European Universities, Public Research Institutes and Companies (more than 30% SMEs)
www.seurat-1.eu