1. Preclinical Models to Support Dosage Selection
Lisa Benincosa, Ph.D. Pfizer Global Research & Development Groton, CT

2. Objectives of Early Drug Development
Identification of critical risk factors prior to investment in full clinical development
select most promising compounds
Provide critical data to identify safe and effective dose and dose regimens
more efficient development

3. Continuum of PK/PD Modeling in Drug Development
Validation
Of PK/PD
Relationship
Preclinical PK/PD in experimental models
PK/PD in healthy subjects
PK/PD in dose-ranging study in patients
using efficacy & safety endpoints (POC)
Confirm efficacy & safety in the
pivotal studies
NDA

4. Preclinical Models Animal models used at Pfizer Advantages Challenges
Murine pneumonia model, thigh model and peritonitis model
Gerbil otitis media model
Advantages
Explore the in vivo exposure response relationship
Explore hypothesis
Assessment of PD at suboptimum doses
Supra-therapeutic doses to explore full dose range
Assessment of tissue distribution
Challenges
Validation of animal models for extrapolation of results to clinical setting

5. PK/PD Indices for Efficacy
Global Indices
AUC/MIC (e.g. fluoroquinolones, macrolides)
Cmax/MIC (e.g. aminoglycosides)
Time above MIC (e.g. b-lactams, carbapenems)
New Approaches
Incorporating PK and PD time course using mechanism based PK/PD models

6. Example
Comparisons of Different Dosing Regimens of Azithromycin

7. Pharmacodynamic Results
Gerbil otitis media model infected with H. influenzae (MIC = 1.6 mg/mL)
Azithromycin (200 mg/kg) 9 B
1-Day Therapy J
2-Day Therapy H
3-Day Therapy F
Infected Control
Limit of Detection 8 H J F B F F F 7 6
Bacterial Recovery (log10 CFU per ml) H 5 4 J 3 2 B H J B J H B 1 24 48 72 96
Time in Hours Post-challenge
Girard et al. ASM A-57, 2002

8. Hypothesis
For azithromycin, front-loading (1-day) appears to be more effective although 2- and 3-day regimens were also effective
Next Step:
PK/PD model for azithromycin to quantitate the effect of front-loading the dose and to differentiate from 3-d and 5-d regimens

9. Study Design Gerbil Otitis Media Model with H. influenzae
Threshold oral doses (~ED50) of azithromycin were selected for comparison: 1-day vs. 3-day vs. 5-day regimens
Humanized PK profiles were generated using adaptive design
Two H. influenzae strains were tested: 54A1100 and 54A1325 (MIC 0.5 and 2 mg/mL)
Plasma concentrations and CFU were determined pre-dose, 1,2,3,4,5,6,12,24,48, and 72 hr (n=3 animals/time point)
One group of 33 drug free controls were also evaluated at the above time points for each strain

10. Rationale of Dose Selection
Emax
Most informative region is between ED20 and ED80
Doses < ED20 all result in a high probability of failure
Doses > ED80 all result in a high probability of cure
% of Cure
ED80 Eo
ED20
ED50
Exposure/MIC

11. Results: Global PD Blue: H. influenzae MIC 2 mg/mL1 1
1 Day
3 Day
5 Day
Blue: H. influenzae MIC 2 mg/mL
Red: H. influenzae MIC 0.5 mg/mL
Log10(AUC/AUCGC ratio) -1 -1
— The line was the fitted curve for
1-day (•) at different doses -2 -2 -3 -3 -4 -4
100
100
200
200
300
300
400
400
Dose/MIC

12. Azithromycin Pharmacokinetic Profiles
1 Day Equivalent Regimens
in Humans & Gerbils
3 Day Equivalent Regimens
in Humans & Gerbils
0.4
0.4
0.3
0.3
AZM Concentration (mg/L)
0.2
0.2
AZM Concentration (mg/L)
0.1
0.1
0.0
0.0 24 24 48 48 72 72 24 24 48 48 72 72
Time (hours)
Time (hours)
Time (hours)
Time (hours)
— human; ---- gerbils

13. Dynamics of Bacterial Growth and Death
Time course of total bacteria growth is a result of a mixture of homogenous sub-populations (mixture model)
Model incorporates bacterial replication modelled as a capacity limited function
1st order rate constant for death
Drug effect enhancing bacterial death or inhibiting replication
Bacteria
CFU/mL
Pop 1
Pop2
Pop3 KD
Replication
IC50
Drug
(+)
(-)
δCFUi/δt = Vgmax.CFUi/[CFUM + CFUTOT] – kd.I(t).CFUi
I(t) = 1± [Emax.(C/MIC)H]/[SITMiH + (C/MIC)H]

14. Pharmacokinetic Results
H. Influenzae (strain 54A1325)
404.8 mg/kg total dose – 1 day PK
H. Influenzae (strain 54A1325)
404.8 mg/kg total dose – 3 day PK

15. Pharmacodynamic Results
H. Influenzae (strain 54A1325)
404.8 mg/kg total dose – 1 day PD
H. Influenzae (strain 54A1325)
404.8 mg/kg total dose – 3 day PD

16. Simulation based on PK/PD model
1-day Concentration vs Time
1-day % of Baseline Kd
3-day Concentration vs Time
3-day % of Baseline Kd
Strain 54A1325

17. Summary
Preclinical PK/PD modeling provided a means to quantitate the effect of front-loading the dose of azithromycin
Front-loading the AUC of azithromycin results in a more rapid and complete bacterial kill
Concentration related amplification of bacterial death (Kd)
Having the highest AUC at the time of greatest bacterial count results in the greatest kill possible for both the sensitive and resistance strain
Optimizes the likelihood of positive clinical outcome

18. Conclusions of Preclinical PK/PD Modeling
Preclinical PK/PD models are useful for the selection of clinical dosing regimens
Best surrogate of efficacy should be identified using mechanism based PK/PD models
Global PK/PD indices of anti-bacterial efficacy may not be optimal
Preclinical PK/PD models can be used to support the overall clinical benefit of the proposed clinical dosing regimen

19. Acknowledgements Pfizer Colleagues SUNY at Buffalo Cognigen
Dennis Girard
Amar Sharma
Ping Liu
Barbara Kamicker
Mary Lame
Steve Finegan
Scott Seibel
Judy Hamel
L Dean Kendall
Phil Inskeep
SUNY at Buffalo
Alan Forrest
Lanre Okusanya (Pfizer Fellow)
Brent Booker
Cognigen
Paul Ambrose
Sujata Bhavnani