Workshop WS6 Modelling and simulation to help MABEL definition

Workshop WS6
Modelling and simulation to help MABEL definition
Introduction: The MABEL concept
Steven Martin, Pfizer
Bart Laurijssens, BEL Pharm Consulting
Joint Conference of European Human Pharmacological Societies
31 March & 01 April 2011, Berlin, Germany
Slide 1
What is MABEL?
Minimum Anticipated Biological Effect Level
In Humans!
“[For a biological product], the estimation of the dose or exposure
required at the bottom end of the dose response curve in man is more
important than NOAEL. This might be termed the Minimum Anticipated
Biological Effect Level (MABEL).”
Early Stage Clinical Trial Taskforce, Joint ABPI/BIA Report, July 2006
Slide 2
Slide 3
What constitutes Safety?
Primary Pharmacology
? AE
PREDICTABLE
Dose
Conc
Secondary Pharmacology
Non-specific Effects
Slide 4
? AE
? AE
Pharmacology & Toxicology
[Jennifer Sims, ABPI/BIA Early Clinical Trials Taskforce, slideset]
Slide 5
What about AABEL?
Acceptable Anticipated Biological Effect Level
In Humans! of course
•  Oncology
•  Clinical experience with target pharmacology
Slide 6
How to calculate?
•  Is there a specific methodology?
•  Is there a specific experiment?
–  The “MABEL” experiment?
–  in vitro or in vivo?
•  Is there a target value?
–  say, 10%?
•  No, no and no!
•  It is a judgement call based a quantitative integration of all available
information.
–  considering uncertainty of the estimate
–  risks (tox observed, systems targeted, agonist vs antagonist)
–  transparent communication
Slide 7
Mechanistic Classification of Biomarkers
Slide 8
How to Predict a Dose with Minimal
Pharmacology?
•  Pharmacokinetics
•  Physicochemical properties
•  ADME
•  Distribution to target(s) in Humans:
–  Transporters (eg PgP)
–  Extracellular vs Intracellular target
•  Interaction with the Human Target(s)
–  Affinity (in vitro, ex vivo)
–  Efficacy (agonism vs antagonism)
•  Human pharmacology
–  In vitro, ex vivo
–  Animal models of physiology (or Disease)
–  Direct vs Indirect effects
•  Knowledge
–  Experience with mechanism in Humans
–  Human physiology
–  General Pharmacological Theory
Slide 9
So what about Animal Models of Disease?
•  Face Validity
–  Phenomenological Similarities with the disorder
•  Predictive Validity
– 
– 
– 
– 
Need drugs that work
Quantitative
False positives/negatives
Mechanism specific?
•  Construct Validity
–  Sound theoretical rationale
–  Need to understand disease and animal
Slide 10
Biologics vs NCEs
•  In general well understood human Pharmacokinetics
•  No liver metabolism (hence no active metabolites)
•  Highly (species) specific for the target
•  High affinity (target mediated disposition)
•  Limited off target effects
•  Most AEs through Primary Pharmacology!
Slide 11
Mechanism of Action of TGN1412
Slide 12
Dose Rational from the Protocol
Slide 13
Receptor Occupancy of TGN1412 at starting
Dose
[Jennifer Sims, ABPI/BIA Early Clinical Trials Taskforce, slideset]
Slide 14
Conclusion
•  Pharmacology
•  Human! Pharmacology
•  Be clear and transparent about the underlying
assumptions
•  Manage uncertainty
•  Case by Case
Slide 15
Workshop WS6
Modelling and simulation to help MABEL definition
An Example
Steven Martin, Pfizer
Bart Laurijssens, BEL Pharm Consulting
Joint Conference of European Human Pharmacological Societies
31 March & 01 April 2011, Berlin, Germany
Slide 16
Your Data: What is the MABEL?
RAT 1 mg/kg
n=3
400
300
!
50
!
!
!
!
!
!
0
5
30
100
20
Response, mean ± sd
40
200
!
10
15
20
DOG 1 mg/kg
n=3
25
10
Time [hrs]
0
400
Concentration [ng/ml], mean ± sd
Concentration [ng/ml], mean ± sd
RAT Translatable Pharmacology
n=8
300
!
0
10
Dose [mg/kg]
!
200
!
!
100
!
Slide 17
2
!
!
!
0
5
10
Time [hrs]
15
20
25
30
Step 1: rat and dog PK (CL and Vd)
RAT 1 mg/kg
DOG 1 mg/kg
400
!
!
!
400
!
!
Concentration [ng/ml]
Concentration [ng/ml]
300
!
!
200
!
!
300
!
!
!
!
!
!
!
!
200
!
!
!
!
!
!
!
100
!
!
100
!
!
!
!
!
!
0
!
0
5
!
!
10
Time [hrs]
Slide 18
0
!
15
20
25
!
!
!
0
5
10
Time [hrs]
15
20
25
Step 2: Predict human PK
Allometric Scaling Clearance
3.5
!
3
2.5
!
!
2
1.5
CL
1
Human CL = 12 L/hr, Human Vd = 1
L
0.5
!
!
!
2
4
6
8
10
BW
Allometric Scaling Volume of Distribution
16
14
12
10
!
!
!
8
6
4
V
2
!
!
!
Slide 19
2
BW
4
6
8
10
ASSUMPTIONS:
1.  CL mechanisms are similar across
species
2.  The physiology underlying the CL
and V scale allometrically
3.  Bio-availability is 1 (to be safe)
4.  Absorption is fast and similar
across species
5.  1 compartmental PK model
Step 3: Rat Pharmacology
Response was measured at 2 hrs,
PK sample taken.
Concentration−Response relationship
!
!
!
!
!
40
!
!
!
E max* conc
EC50 + conc
!
!
!
!
!!
!
Response
!
!
!
RESP = base +
!
!
!
20
!
€
!
!
!
!
!
!
0
!
!
−20
0
2000
4000
6000
8000
Concentration [ng/ml]
Slide 20
10000
12000
ASSUMPTIONS:
1.  No delay between plasma
compartment and response!
2.  No experimental artifacts (ie max
response is not assay limit)
Step 4: Predict Human PKPD
•  Human EC50 = rat EC50, Emax=100%
ASSUMPTIONS:
1.  Rat pharmacology is relevant
2.  Total plasma concentrations drives effect (or free fraction rat and
human the same)
3.  Distribution to target site similar rat and human
4.  Similar Affinity for target rat and human
5.  Target distribution and physiology the same (incl time component)
Slide 21
Step 5: Simulate!
•  And decide what is “minimal” or “acceptable”!
–  which is dependent on the particular pharmacology endpoint
simulated.
Slide 22
M&S
•  It will remain a prediction, confidence dependent on:
– 
– 
– 
– 
quality of the data
risks
strength of assumptions made
inability to predict the unpredictable
•  Sensitivity analysis, worst case scenarios, and robust designs
(safety margins, proper monitoring) can address these.
–  Thus merely “calculating” a MABEL is not sufficient!
•  Planning!
•  Resource:
–  Time
–  Money
–  People
Slide 23
Workshop WS6
Modelling and simulation to help MABEL definition
Case: small molecule CNS indication
Steven Martin, Pfizer
Bart Laurijssens, BEL Pharm Consulting
Joint Conference of European Human Pharmacological Societies
31 March & 01 April 2011, Berlin, Germany
Slide 24
The case:
•  Target Indication: Alzheimer’s disease
•  Novel Mechanism
•  Target Receptor is in the CNS (G-protein coupled
Receptor)
•  Small molecule, antagonist
Slide 25
Your data.
•  Target in CNS
–  physicochemical properties (BCS class 1)
–  transport (PgP mouse knockout 30:1)
•  What do we know about pharmacology of the receptor
–  Little.
•  Animal model of cognition (!?)
•  Antagonist of a G-protein coupled receptor
–  High Occupancy likely needed (>80%) for any pharmacological
response.
•  Affinity to human receptor
–  (we have ex vivo binding in the rat)
•  Other receptors? CNS, periphery
•  Pre-clinical tox
Slide 26
Animal model
Rat RO
Human RO
Human PET
Animal PK Human PK Human PK
+
Animal Tox
Human safety/tolerability
Therapeutic window
Slide 27
Human PET
Doses for PoC
(Pharmacol)
Therapeutic window
!"#$%&'()*+,*-#.'+&*
Blood Brain Barrier
Fu
PgP
Endogenous neurotransmitter (NT)
Competitive inhibition
Fu, Ki
X
affinity
Target Receptor
Fu, Ki s
Compound
Fu
Ki s
transduction
/.$"0*1"#"2.+0(
?
/.$"0*",,"#.(
“NT” response
?
/.$"0*1"#"2.+0(
?
/.$"0*",,"#.(
Slide 28
Clinical response
!"#$%$&#'()*+'#,%-.%-*
+"/+)%0+-'#%-#'1+#2('
Ex vivo binding experiment in the rat
5000
Counts = counts 0 * (1 −
cpm/mg
4000
conc
) + countsRES
IC 50 + conc
3000
,3&&.#4567#)('#8#69:6#-*;03
2000
1000
0
0.01
0
0.1
1
10
100
Blood concentration (ng/ml)
observed data
NONMEM fit
Slide 29
1000
10000
5#$617%89:;4:<#"#2=
Predicted Rat plasma concentration - receptor occupancy relationship
120
100
!"##$%&'()%*%(+,(%-.+)%/%,+,0%1234"
RO at time/Dose
when in vivo pharmacology
was observed
%binding
80
60
40
Predicted brain unbound IC50 = 12-60 pg/ml
Compare with in vitro ki = 88 pg/ml
20
0
0.01
0.1
1
10
100
1000
blood concentration (ng/ml)
Correcting
Point estimate
95% CI
Slide 30
10000
Blood IC50 for:
•Blood:plasma ratio
•Plasma protein binding
•PgP induced gradient (use mouse knock-out value)
!789:;<89&(=4;84<7#<:=4&1 >?(@A
78"#<:=4$B:C&:4&DE%#4$
120
100
!"#$%#&'()*&+&,-.&/0-,&1&2.-,3&456%"
%binding
80
60
40
20
0
0.01
0.1
1
10
100
1000
10000
Plasma concentration (ng/ml)
Point estimate
95% CI
Correcting Rat blood IC50 for species differences in:
•Target receptor affinity
•Blood:plasma ratio
•Plasma protein binding
PgP induced gradient assumed to be the same
Slide 31
!"#$%&'(')%*+,-"#.'#%*%/&+#'
"001,1&1%)'2345",6
Table 3.2 Binding affinity of Compound in A-PT panel
Receptor
RecA
RecB
RecC
RecD
RecE
RecF
RecG
Target
RecH
RecI
RecJ
RecK
RecL
Slide 32
pKI
<5
6.3
6.9
8.0
7.4
7.3
5.5
9.6
6.4
<5.2
<6
5.3
5.2
!"#$%&%$%'()
*+#,%$'#,-%.-/%/0-12-%.-3456.7
CNS
Predicted Therapeutic window
100
Predicted Therapeutic window
100
Target
CNS
RecD
3
4
5
6
7
80
70
60
50
40
30
20
10
0
1e-2
RecD
3
4
5
6
7
90
Predicted Receptor Occupancy (%)
90
Predicted Receptor Occupancy (%)
Systemic
80
70
60
50
40
30
20
10
1e-1
1e+0
1e+1
1e+2
1e+3
1e+4
Compound plasma concentration (ng/ml)
1e+5
1e+6
0
1e-2
1e-1
1e+0
1e+1
1e+2
1e+3
1e+4
Compound plasma concentration (ng/ml)
Assumptions:
•All receptors within a region are exposed to similar concentrations
•For CNS, relative affinities are same in vitro as in vivo
Slide 33
1e+5
1e+6
9%:#83$;#&'3)%()*&68+((')#,!".#<+&7+&/#(%#(;'#
"*)/'(#)'$'1(%)#+6#&''7'7#=%)#*&#'=='$(>#,?.
100
100
100
@'$%&7#8'66'&/')#)'61%&6'
90
90
80
70
70
60
60
50
50
"*)/'(#02
40
40
30
30
20
20
10
10
0
0
80
70
)*$3154"-51$8>:
80
6'$%&7#8'66'&/')#)'61%&6'#,5.
"*)/'(#0'$'1(%)#2$$31*&$4#,5.
90
!"#$%&'($)*$+,-.,-/$0"$0(1$023/10$
31'140"3$,5$-11.1.$6"3$2-$1661'07$89:
60
50
Near maximal response
with only 20% RO!
40
30
20
0.001
0.01
0.1
1
10
100
1000
!"#$%&$'&()*(+%&#,&-.
A*#§ DE?#&-
Slide 34
1e+4
10
0
0
A+ § ?BC#&-
10
20
30
40
50
60
70
80
*23/10$;1'140"3$<''&42-'=$+=$)*$8>:
90
100
?(@&%1'A&7#,4*+&8*'*0+(,&%*2-#+*2&B=7&*..*'+C&
'#)&D*&D"('E*2&D:&+A*&'(%0(1)2F
90
90
80
70
70
60
60
50
50
=*1,(+,#)$%-++*,&89
40
40
30
30
20
20
10
10
<=7&>..*'+<&3;6
7#,4*+&8*'*0+(,&9''10#)':&3;6
2] Do not study doses with
/00
<80% RO
/(%0(1)2&89
B=7&*..*'+
G-#&+#,4*+&,*'*0+(,C
80
Minimal “response” during dosing interval at steady state
:0
:0
B0
123-+4(5+6+$4"3(E66%$2&67
0
0.01
0.1
1
10
!"#$%#&'()'*)+,#+-()&(.&/(%0(1)2&3)45%"6
Slide 35
100
1000
B0
A0
3] Doses that hardly
separate
based on RO, potentially
@0
separate in efficacy
;0
A0
@0
;0
?0
C1(3+D24+'(3+*$"&*+
?0
>0
>0
=0
=0
/0
/0
0
/
0
/00
C1(3+D24+'(3+*$"&*+(,8.
100
123-+4(5+6+$4"3("66%$2&67(,89(:;8(!<.
100
)"*+(*+D+64F"&(G"3(H3""G("G(!"&6+$4(*4%'7
1] =theoretical range of efficacy:
/0
No suppression – No effect
!"#$"%&'()"*+(,#-.
Max suppression
– Max effect
0
/00
Workshop WS6
Modelling and simulation to help MABEL definition
Case: small molecule Inflammation
Steven Martin, Pfizer
Bart Laurijssens, BEL Pharm Consulting
Joint Conference of European Human Pharmacological Societies
31 March & 01 April 2011, Berlin, Germany
Slide 36
The case:
•  Target Indication: Inflammation
•  Semi-novel mechanism, competitors are in the clinical
development
•  Target receptor widely spread, also on monocytes
•  Small molecule, antagonist, inhibitor of induced mediator
release
Slide 37
Your data.
•  Target everywhere (including CNS):
–  physicochemical properties (BCS class 1)
–  transport (no PgP substrate)
•  Including blood:
–  ex vivo Human Whole blood pharmacology! (CNS risk?)
•  Affinity for the receptor (not feasible)
•  Other receptors? (clean)
•  Competitor data
–  dosed well beyond “100% inhibition”
•  Knowledge regarding down stream cascade
–  other compounds interfering there available
Slide 38
Biomarkers
Markers of
pharmacology
Markers of
disease
Slide 39
inhibition of Mediator
Assumptions
Calculations
and Assumpions
ƒ Pharmacokinetics:
ƒ Low Exposures: CLb like dog/monkey (40% LBF), F=60%
ƒ High Exposures: CLb like rat (10% LBF), F=100%
ƒ Pharmacodynamics:
ƒ Inhibition in Human Whole blood
ƒ pKi=7.12 which translates to ICb50 of about 30 ng/ml
ƒ Imax=100% and slope=1
ƒ Free concentration (Cu) in plasma responsible for pharmacology
ƒ Relevant tissues Cu = Cu plasma (CNS)
ƒ Direct effect of inhibition.
ƒ Safety:
ƒ NOAEL
Slide 40
Predicted inhibition over time for the planned doses
Predicted Target inhibition over time
Slide 41
Safety
FIH table
For High predicted exposures
Slide 42