systems pharmacology; an industrial perspective … · , systems pharmacology of the ngf pathway;...
TRANSCRIPT
-
SYSTEMS PHARMACOLOGY; AN INDUSTRIAL
PERSPECTIVE
(Warwick school of Engineering vacation school)
Dr Neil Benson. Director, Xenologiq Ltd.
www.xenologiq.com Quantitative drug discovery solutions
Interested in PKPD modelling, dose prediction, PBPK and systems pharmacology
mailto:[email protected]
-
Why should drug discovery be interested in systems pharmacology? Attrition
www.xenologiq.com Quantitative drug discovery solutions
Attrition – a small proportion of the ideas worked on in drug discovery become drugs -> significant $ spent on ideas that never become products
-
10 years of evaluation & proposals and arguably nothing has changed
www.xenologiq.com Quantitative drug discovery solutions
Kola & Landis, Nature Rev DD (2004); PHII attrition due to lack of efficacy or safety key issue
Paul et al, Nature reviews, DDT, (2010) – PHII attrition by far largest contributor to productivity
Hay et al (2014); Nature biotech; The attrition statistics showed a decrease in probability of success compared to previous evaluations
2004 2010 2014
-
Why drug discovery needs mathematical models •Kola & Landis, Nature Rev DD (2004); Paul et al, Nature reviews, DDT, (2010), Hay et al (2014) – PHII attrition by far largest contributor to productivity
•Mainly due to lack of efficacy or safety (ie we didn’t understand the biology -> consequences perturbing a complex system)
Mathematical modelling & simulation proven approach for tackling complexity in many areas of science and engineering
-
What is systems pharmacology ?
Time
Effect
Concen
tration
PD
PK
PKPD
Understanding of behaviour of a drug in a disease www.xenologiq.com Quantitative drug discovery solutions
Requires mathematical modelling approach
-
Example 1 – PKPD & systems pharmacology approach to Toll-like 7 receptor agonists for Hepatitis C virus
•HCV unmet medical need
•Hypothesis - TLR7 agonism -> interferon response -> improved anti HCV effect
www.xenologiq.com Quantitative drug discovery solutions
-
Project questions at preclinical stage
www.xenologiq.com Quantitative drug discovery solutions
1. Can we achieve a dose that will deliver efficacy ?
2. What is the minimum anticipated biological effect level ?
Benson et al Antimicrobial Agents and Chemotherapy, March 2010, p. 1179-1185, Vol. 54, No. 3
-
Toll-like 7 receptor agonists for Hepatitis C virus; aims
Challenge:
No animal model for HCV Link between TLR7 agonism and antiviral effect not established Dose prediction for interferon inducers for hepatitis C ?
Opportunity:
IFNα available as translatable biomarker Clinical PK and PKPD data for recombinant IFNα Published math model for antiviral efficacy of IFNα
Solution:
Integrated mechanism-based PKPD approach linking exposurebiomarkeroutcome
www.xenologiq.com Quantitative drug discovery solutions
-
www.xenologiq.com 9
Very complex – mathematical modelling solution required
Predicting dose/ antiviral response in man
-
TLR7 – projecting human interferon induction
• Data generated in the mouse
• Rapid and robust induction of IFNa observed from zero baseline
5 15 25
5 15 25
TIME
100
300
100
300
100
300
DV
AMT: 100 AMT:300
AMT: 500 AMT1000
AMT: 5000 AMT:1000.00
5 15 25
5 15 25
TIME
100
300
100
300
100
300
DV
AMT: 100 AMT:300
AMT: 500 AMT1000
AMT: 5000 AMT:1000.00
Plasma IFNa (IU/ml)
www.xenologiq.com Quantitative drug discovery solutions
How do we use this data in our model ?
-
TLR7 – indirect PKPD model of IFNa production
.RCSC
.CS
dt
dR
p50
pmax
outk
Data described using a modified indirect effect eqn
www.xenologiq.com Quantitative drug discovery solutions
Where; dR/dt = rate of production IFNa (IUml-1h-1), Smax = max rate of production IFNa ( IUml
-1h-1) SC50 (ng/ml) = conc drug for ½ max production IFNa kout = elimination rate constant IFNa (h
-1), Cp = predicted plasma concentration (ng/ml) R = plasma IFNa conc (IUml-1)
5 15 25
5 15 25
TIME
100
300
100
300
100
300
DV
AMT: 100 AMT:300
AMT: 500 AMT1000
AMT: 5000 AMT:1000.00
5 15 25
5 15 25
TIME
100
300
100
300
100
300
DV
AMT: 100 AMT:300
AMT: 500 AMT1000
AMT: 5000 AMT:1000.00
Fit to PKPD data
Mathematical model of time & dose dependence
PK
tktkea
a ae ee)kV(k
Dose.kC
+
PD
-
TLR7 agonist PKPD fit results
Parameter Estimate CV(%)
kout (h-1) 0.958 0.1
SC50 (ng/ml) 135 24
Smax (IU/ml/h) 294 8
IIV Smax (%) 70 26
Residual error
(IU/ml)
65 19
Individual predicted IFNa in plasma (IU/ml)
0 200 400 600O
bse
rve
d I
FN
a in p
lasm
a (
IU/m
l)
0
200
400
600
www.xenologiq.com Quantitative drug discovery solutions
0 10 20 30 40 50 60 70 80
101
102
103
104
105
Time (hours)
Co
nce
ntr
atio
n (
dru
g (
ng
/ml IF
Na
pg
/ml)
PK
PD
Example individual
-
TLR7 – linking IFNa to antiviral effect
VTT ).1(dsdt
dT IVT ).1(
dt
dI
Parameter estimates from clinical data (Neumann, A.U., et al, Science, 1998. 282(5386): p. 103-7)
HCV = hepatitis C viral load
c the viral decay rate constant,
δ the infected hepatocyte decay
rate constant.
ε = inhibition of viral release (p)
= inhibition of viral infection s&d = Target cell
production/degradation rates
II TT
Infection Rate
c
Clearance
p
Virions/d
Target Cell
Infected
Cell
Loss
cVpI ).1(dt
dV
IFNa blocks viral release
www.xenologiq.com
Can be represented as a set of equations
IFNIC
IIFN
50
max
-
Modelling enabled an integrated mechanism-based PKPD approach for predicting antiviral efficacy of TRL7 agonists
Rat PK
Mouse IFNα induction
PKPD
Human IFNα PKPD Human PK
Human IFNα induction
PKPD
Human antiviral PKPD
PBPK
Clinical
Literature
Viral load
PKPD model
Experimental
Literature
In silico
0 20 40 60
time (days)
101102103104105106107
101102103104105106107
101102103104105106107
vir
al lo
ad
(R
NA
co
pie
s/m
L)
dose: 1 dose: 3
dose: 10 dose: 30
dose: 100
Trial simulation
Human dose
prediction
www.xenologiq.com
-
Conclusions
• Collected & integrated all available data for decision making
• Allowed conversion of information into knowledge for; 1. Dose prediction – likely to achieve efficacy at a practical dose
2. Minimum anticipated biological effect level (MABEL) ID
3. Sensitivity analysis – what is important? (compound potency <
system properties)
www.xenologiq.com Quantitative drug discovery solutions
-
Example 2; systems pharmacology & technology feasibility
Drug molecule – binds the target but has poor pharmacokinetics (the ‘war head’)
Molecule coupled to the drug; improves the PK whilst retaining the drug effect
General outline of the approach;
Acknowledgement;Tomomi Matsuura & Piet van der Graaf, Pfizer
Gives improved PK
‘War head’ binds the target
www.xenologiq.com Quantitative drug discovery solutions
+
Pharmacophore + Linker Antibody CovX- body
-
Technology feasibility
Project question for target Z
• The coupled drug exhibits pharmacological action, but...
• The war head strongly binds human serum albumin (>>90% bound in plasma)
• Given dose of the new drug is limited by cost & what can be delivered...
• Can the required receptor occupancy be achieved via this approach in this case?
www.xenologiq.com Quantitative drug discovery solutions
-
Model of the system
Serum albumin
Drug
R= receptor
Parameters measured (eg surface plasmon resonance) or derived from pharmacokinetic pharmacodynamic analysis of in vivo experiments
www.xenologiq.com Quantitative drug discovery solutions
-
Summary
•FR: free receptor concentration/initial receptor concentration
•Green is good, red is bad
•At higher ppb require v low target Kd –project terminated 19
-
Perspective
• A plausible scenario is that this molecule would have exhibited;
Good in vitro efficacy
Clean pre-clinical safety profile (no pharmacology)
Good PK prediction
Good safety & PK in Phase 1 trials (no pharmacology)
Phase II negative (no pharmacology)
www.xenologiq.com Quantitative drug discovery solutions
• Model based approach - terminating this early saved $$’s
-
Example 3; NFKB pathway
Ihekwaba AE, Broomhead DS, Grimley RL, Benson N, Kell DB.(2004). Systems Biology 1(1): 93-103.
& Science, 306, Oct 2004, 704 - 708.
• Project question – what are the optimal targets in the pathway and what % effect/inhibition is required?
• Use published ODE model; 64 parameters & 26 variables
0
0.4
0.8
1.2
1.6
2
0 100 200 300 400 500 600
Time (min) after TNFa
Am
plitu
de
B
215 350 440 (min)
0 50 170
A
www.xenologiq.com Quantitative drug discovery solutions
NFKBn
Acknowledgement; L Cucurull-Sanchez, Karen Spink
-
Further detail
www.xenologiq.com Quantitative drug discovery solutions
Cucurull-Sanchez, K.S Spink & S. Moschos, Drug Discovery Today, 17, 13-14, 665-, 2012
-
Project approach – Small interfering RNA (siRNA)
Challenge: IL-1
Target: IKK-2
Biomarker: IL-8
www.xenologiq.com Quantitative drug discovery solutions
Max attenuation [IKK] – 80% ?
-
Sensitivity analysis identified IKK as a potential target
IKK would appear to be a good potential target
dk
tdx )( Change in the species (eg [NFKBn])
Change in variable (eg parameter or reactant conc)
www.xenologiq.com Quantitative drug discovery solutions
-
But high % inhibition required to fully dampen NFKB oscillations..
0 1 2 3 4 5 6 7 8 9 10
x 104
0
0.01
0.02
0.03
0.04
0.05
0.06
Time
Sta
tes
States versus Time
IkBa
NF_kB
IkBa_NF_kB
IkBb
IkBb_NF_kB
IkBe
IkBe_NF_kB
IKKIkBa
IKKIkBa_NF_kB
IKK
IKKIkBb
IKKIkBb_NF_kB
IKKIkBe
IKKIkBe_NF_kB
NF_kBn
IkBan
IkBan_NF_kBn
IkBbn
IkBbn_NF_kBn
IkBen
IkBen_NF_kBn
source
IkBa_t
sink
IkBb_t
IkBe_t
Basal 80% knock-down of IKK
0 1 2 3 4 5 6 7 8 9 10
x 104
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
Time
Sta
tes
States versus Time
IkBa
NF_kB
IkBa_NF_kB
IkBb
IkBb_NF_kB
IkBe
IkBe_NF_kB
IKKIkBa
IKKIkBa_NF_kB
IKK
IKKIkBb
IKKIkBb_NF_kB
IKKIkBe
IKKIkBe_NF_kB
NF_kBn
IkBan
IkBan_NF_kBn
IkBbn
IkBbn_NF_kBn
IkBen
IkBen_NF_kBn
source
IkBa_t
sink
IkBb_t
IkBe_t
Time (s) Time (s)
[NF-
B
(nu
c)]
(M
)
[NF-
B
(nu
c)]
(M
) •Assuming ~remove oscillation to suppress IL8
•and given technology typically gives maximum 80% decrease in target
•Risk that strong attenuation of biomarker may not be achieved
www.xenologiq.com Quantitative drug discovery solutions
Maximum achievable inhibition via siRNA
-
Validation of model prediction
Thanks; Sterghios Moschos Karen Spink for data
•80% inhibition of IKK enzyme
•~30% attenuation of IL8 production
•Project terminated
www.xenologiq.com Quantitative drug discovery solutions
-
Perspective
• Laboratory costs to test siRNA hypothesis considerable
• Failure of outcome could have been predicted before any work was done
• Model based approach - $$ savings and focus on a less risky target/approach
www.xenologiq.com Quantitative drug discovery solutions
-
Example 4; systems pharmacology of the nerve growth factor pathway
•Extracellular Regulated Kinase (ERK) activation controls pain response (eg trka levels, ion channel activity etc)
•We assume dppERKnuc is proportional to pain
•Team question – what are the best pain targets ?
0 20 40 60 80 1000
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Time (mins)
dppE
RK
nuc(µ
M)
Acknowledgement; Cesar Pichardo, Pinky Dua & Piet van der Graaf (Pfizer Neusentis) and at ISB Oleg Demin)
www.xenologiq.com Quantitative drug discovery solutions
Nat Cell Biol
-
Models in literature Fujioka et al, J Biol Chem, 2006
0 2 4 6 8 10 12 14 16 18 200
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Pfizer model Sasagawa et al
ppnuc
ERK
Time (mins)
Sasagawa et al, Nat Cell Biol, 2005
Benson & Dua et al, Systems pharmacology of the NGF pathway; Interface Focus, 6 April 2013 vol. 3 no. 2, 20120071.
www.xenologiq.com Quantitative drug discovery solutions
• One compartment integrated ‘systems biology’ model constructed (59 molecular species and 233 parameters)
-
0 10 20 30 40 50 60
10-2
100
102
104
106
Species
Tim
e in
teg
ral d
[dp
pE
RK
nu
c]/d
[X]
Finding optimal targets; sensitivity analysis of the model
www.xenologiq.com Quantitative drug discovery solutions
dk
tdx )(Change in the species (eg [dppERKnuc])
Change in variable (eg parameter or reactant conc)
Rank
(dppERKnuc)
Species Initial
value
Unit Description
1 NGFext 30 pM NGF concentration in extracellular matrix
2 Grb2_SOS_pShc_pTrkA 0 µM Downstream multi-protein complex
3 pShc_pTrkA 0 µM Downstream multi-protein complex
4 Shc_pTrkA 0 µM Downstream multi-protein complex
5 pTrkA 0 µM Phosphorylated TrkA concentration
6 L_NGFR 0 µM NGF NGFR complex
7 FRS2_pTrkA 0 µM Downstream multi-protein complex
8 pFRS2_pTrkA 0 µM Downstream multi-protein complex
9 Crk_C3G_pFRS2_pTrkA 0 µM Downstream multi-protein complex
10 Grb2_SOS_pShc_pTrkA_endo 0 µM Downstream multi-protein complex
NGF
By calculating integrals -> rank of importance
-
Finding optimal targets; sensitivity analysis (SA) of the model
www.xenologiq.com Quantitative drug discovery solutions
• SA provides a way to rank targets & further triage by druggability criteria -> interesting targets;
1. NGF (Consistent with NGF mAb efficacy in pain; eg Lane et al, 2010 )
2. RAS activity (Mutations in NF1 gene associated with a chronic pain phenotype (J Neurophysiol 94: 3659-3660, 2005)
3. Trka kinase activity (eg Expert Opin Ther Pat. 2009 Mar;19(3):305-19. Trk kinase inhibitors as new treatments for cancer and pain).
-
Systems biology -> systems pharmacology
www.xenologiq.com Quantitative drug discovery solutions
Systems biology model; One compartment No physiological data No easy way to compare eg
mAb’s and small molecules Gives large dose over-
prediction
Systems pharmacology model; Two (n) compartments Physiological information (eg
compartment volume) Separate compartments enable
comparison of small molecules and mAb’s
Can be used for dose prediction
-
But what about dose? The NGF Systems pharmacology model
www.xenologiq.com Quantitative drug discovery solutions
2 compartments
1. Extracellular body water (~15L)
2. Neuronal compartment (~0.001 L)
•Signal transduction elements in neuronal compartment
•Enables dose prediction for mAbs and small molecules
•Model predicted max efficacious dose hypothetical mAb at ~ 1000xKD (a non-intuitively high ratio)
Benson & Dua et al, Systems pharmacology of the
NGF pathway; Interface focus, 2013.
-
Model predictions quantitatively concordant with data for prototype NGF mAbs eg tanezumab
www.xenologiq.com Quantitative drug discovery solutions
http://www.fda.gov/downloads/AdvisoryCommi
ttees/CommitteesMeetingMaterials/Drugs/Arthr
itisAdvisoryCommittee/UCM301305.pdf
• Cmax ~ 30 nM (=3000 x KD) • -> Model prediction consistent with clinical pain data • -> efficacy & dose could have been predicted from ‘05 model
http://www.page-
meeting.org/pdf_asset
s/7203-
PageAthensCleton.pdf
• Model predicts higher than expected drug concentration (>1000 x KD) for maximum efficacy
-
Perspective
• ‘05 published model predicted efficacy & (non-intuitive) dose for NGF mAb well ahead of PHII data
• Model provides mechanistic insight into why this is so;
ie due to a negative feedback loop in the NGF pathway (Interface focus 6 April 2013 vol. 3 no. 2, 20120071.)
www.xenologiq.com Quantitative drug discovery solutions
-
Example 5; application of systems pharmacology to the clinical development of Fatty acid amide
hydrolase (FAAH) inhibitors for pain.
www.xenologiq.com Quantitative drug discovery solutions
‘A systems pharmacology perspective on the clinical development of Fatty acid
amide hydrolase (FAAH) inhibitors for pain’. CPT Pharmacometrics Syst. Pharmacol. (2014) 3, e91; doi:10.1038/psp.2013.72
-
Fatty acid amide hydrolase (FAAH) inhibitors for pain.
• Cannabinoid receptors CB1&2 target for THC – active ingredient of cannabis
• Endogenous Anadamide (AEA) is an agonist of CB1&2
• Wealth of interesting biological data eg increased AEA attenuates nerve firing
(eg Bacci et al, Nature 431,312–6).
• Some evidence of efficacy of cannabis preparations in pain
(eg Nurmikko et al., 2007,Pain, 133, 210-220)
www.xenologiq.com Quantitative drug discovery solutions
Acknowledgements; van der Graaf, PV D. Nichols & G. L. Li (Pfizer) & Demin,O (Institute for SB, Moscow).
-
FAAH • AEA degraded by FAAH
• Project hypothesis;
• FAAHi PF-04457845 animal model data positive
• PF-04457845; good pharmacology & PK
• Appears to be very promising drug discovery target
www.xenologiq.com Quantitative drug discovery solutions
-
Aim for systems pharmacology approach ‘09
• Project questions;
1. will we test the pharmacology fully?
2. can we predict receptor occupancy and level required for efficacy?
• Systems pharmacology approach with available internal & external data prior to PII data
• Collaboration with Demin & Metelkin, ISB in 2009 as PF-4457845 –> phase 1
www.xenologiq.com Quantitative drug discovery solutions
-
Literature review; main conclusions 1. Some useful data (eg AEA disposition, enzyme kinetics to
enable model construction)
2. Endocannabinoid biology complex & not elucidated eg with
apparent contradictory impacts for eg AEA on CB1 & TRPV1 ( Piscitelli and Di MarzoChem. Neurosci. 2012, 3, 356−363)
? Steady state impact
3. Link between occupancy & nerve firing attenuation – no quantitative data
Can’t specify what success would be
in receptor occupancy vs time in man www.xenologiq.com Quantitative drug discovery solutions
Presynaptic neuron
Reaction
or transport
Activation
Inhibition
Reaction
or transport
Activation
Inhibition
signal
CB1RCa2+
Ca2+
med
iato
rs
Ca2+
AEA
NA-PE
PECa2+
?
AEA
?
AEACOX-2, LOX FAAH
NAT
NAPE-PLD
G proteins
G proteins
signal
2-AG
DAG
PI
?
2-AG
PLC
DAG lipase
degradation
?
rece
pto
rs
COX-2, LOX MGL
degradation
2-AG
-
Summary of model construction
1. XEA disposition; inter-
compartmental equilibration of
XEA’s, location of enzymes
involved, tissue binding
• Large ODE model variables – 39, reactions – 75. 5 main components…
The model
FAAH + AEA FAAH_AEA FAAH + P
+
XEA
FAAH_XEA
FAAH + P
.XEAK
KAEAK
Vmax.Sv
xeaFAAH,m,
aeaFAAH,m,
FAAHm,
2. Multi-substrate binding of FAAH competitive
3. 2 step bio-synthesis of XEA’s
FAAH + I FAAH_I*
kinh
4. Irreversible inhibition of FAAH
& PK;clinical data parameters
5.Ligand partitioning &
receptor occupancy
theoretical model KD,obs=Kp
-1*KD
-
An additional clearance mechanism?
400 350 300 250 200 150 100 50 0
10
9
8
7
6
5
4
3
2
1
400 350 300 250 200 150 100 50 0
260
240
220
200
180
160
140
120
100
80
60
40
20
0
With unknown enzyme Without unknown enzyme
Plateau
No plateau
Pla
sm
a A
EA
, n
M
Pla
sm
a A
EA
, nM
Time, h Time, h
Modelling results Modelling results
Enzymes
• FAAH-2
• NAAA N-acylethanolamine
hydrolysing acid amidase
• COX-2 cycloxygenase-2
• 12-LOX 12-lipoxygenase
• 15-LOX 15-lipoxygenase
• CYP2D6 P450
• CYP4F2 P450
• CYP3A4 P450
Consume
all XEA’s Incorporating quantitative perspective -> hypothesis NAAA most likely
Dosage: 10 mg PF
OEA
PEA
LEA
AEA
TIME, hours
FA
As
co
nc
en
tra
tio
n,n
M
350300250200150100500
50
45
40
35
30
25
20
15
10
5
0
-
The model simulated observed clinical biomarker
data
-
0 50 100 150 200 250 300 350
Time after dose (h)
0
5
10
15
20
25
CB
1 r
ecepto
r o
ccupancy (
%)
0.1
0.3
1
3
10
20
40
PF-04457845 (mg)
Simulation of CB1 RO
• FAAH inhibition >> 96%
•Receptor occupancy limited at ~ 25% (on basis of model hypothesise this is; N-acylethanolamine hydrolysing acid amidase (NAAA))
•Is 25% enough to give pharmacological effect?
-
Pre phase II clinical feedback 2010
• Model suggested at doses planned we will test pharmacology as far as possible (FAAH inhibition >96%).
• But alternative clearance of AEA exists that limits effect – hypothesis NAAA
• No quantitative data to link receptor occupancy and pain outcome; success criteria open question
• Model suggests some receptor occupancy, but data to confirm or rule this out desirable; ‘Pillar 2’ (Morgan et al, Drug Discovery Today Volume 17, Issues 9–10, May 2012, Pages 419–424)
• High risk project & will we learn from the experiment?
-
Epilogue
www.xenologiq.com Quantitative drug discovery solutions
•2012 PF-04457845 failed in OA
•? Confidence in animal model translation
•? Did we express the pharmacology
•Systems pharmacology perspective –> build knowledge & technology before any further phase II
-
Perspective
• Model contains numerous assumptions –> ‘wrong’
• Value not ‘description’ but as a tool to ask better questions
• This process highlighted risks that were not apparent via other means –> ‘useful’
www.xenologiq.com Quantitative drug discovery solutions
-
www.xenologiq.com Quantitative drug discovery solutions
A vision for the future of drug discovery; systems models central in all projects as a tool to support decision making
-
Conclusions
• PKPD & Systems pharmacology models can help identify optimal targets & approaches and estimate dose at very early stage
• Growing body of evidence that systems models can add value in drug discovery (eg Benson et al, Proceedings of the 11th International Conference on Systems Biology Advances in Experimental Medicine and Biology. 2012, Volume 736, Part 5)
• Can be applied to efficacy and safety toxicity/questions -> make better decisions in drug discovery & tackle attrition
www.xenologiq.com Quantitative drug discovery solutions
-
Questions/ comments?
www.xenologiq.com Quantitative drug discovery solutions
-
Backups
www.xenologiq.com Quantitative drug discovery solutions
-
Initial clinical data; how could AEA exhibit pharmacology?
Dosage: 10 mg PF
OEA
PEA
LEA
AEA
TIME, hours
FA
As
co
ncen
tra
tio
n,n
M
350300250200150100500
50
45
40
35
30
25
20
15
10
5
0
•AEA binds strongly to HSA (fu ~ 0.0001)
•Primary pharmacology ~300 nM •Peak total AEA ~ 10nM
•Peak free = 1 pM
-
How could AEA exhibit pharmacology?
www.xenologiq.com Quantitative drug discovery solutions
Some evidence AEA binds within membrane (THE JOURNAL OF BIOLOGICAL CHEMISTRY, Vol. 280, 2005).
ie ~ order of magnitude for the hypothesis that we can drive CBR pharmacology
Log D7.4 anandamide ~5.67^ KD = 2.1e-6*300nM = ~1pM
^brain:plasma rat Kp= 7.2 Thanks Katie Critchell PDM Pfizer.