Presented by Neil Parrott, Pharmaceutical Sciences

1. Mechanistic Oral Absorption Modelling
An update on cross-industry activities
Neil Parrott, Pharmaceutical Sciences,
Roche Pharma Research and Early Development, Roche Innovation Center Basel
1
Paris, April, 2019

2. Roche Group
Roche pRED is one of three fully independent research hubs
2

3. Overview
• Food Effects – Review Recent Activities
• Predicting the Effect of Acid Reducing Agents
– A Roche Case Study
• Future Directions
3

4. 4
Confidence in MAM : Industry

5. Confidence in MAM : Regulators
5
AAPS webinar Sept 2017. First-In-Class Regulatory PBPK Modeling
Guidelines from both Sides of the Pond – Ping Zhao, Anna Nordmark.
https://www.pathlms.com/aaps/events/643/video_presentations/80736
“Very low confidence”
“Not scientifically there yet”.

6. Confidence in MAM : Regulators
6
“The large knowledge gaps in product, API, and physiology hinder the ability of
PBPK to prospectively predict the food effect”
48 food effect predictions
~50% within 1.25-fold
75% within 2-fold

7. Regulatory Guidance Prior to 2019
7
Food effect bioavailability studies are needed to support global filings of NDAs

8. Regulatory Guidance 2019
8
No mention of MAM
A missed opportunity to encourage this to
streamline and enhance food effect
assessments. May effectively discourage
sponsors from investing in this approach
Mentions possible consideration of BCS
category to waive FE studies specifically for
BCS1 without high first pass metabolism.

9. New from the GastroPlus User Group
9
Journal of Pharmaceutical Sciences Volume 108, Issue 1, January 2019, Pages 592-602

10. New from the GastroPlus User Group
10
• Consideration of molecule type to set level
of confidence
• Workflow with standardized inputs and a
model validation step with clinical data in
one prandial state
• Model must be validated against clinical
food effect data before prediction of food
effect (e.g., for new formulations, API
polymorphs, or change of dose)
• Mechanistic predictions of food effect
could substitute for unnecessary clinical
studies during late-stage clinical
development or life cycle management

11. PBPK Food Effect Working Group
January 2018 – Dec 2019
Chair – Arian Emami Riedmaier (AbbVie)
Co-chair – Neil Parrott (Roche)
Goals
• Assess performance of mechanistic model predictions of
food effect using a consistent strategy and input data
• Provide an industry best practice, categorizing molecules
according to prediction confidence

12. Timeline - 2019
Jan Feb Mar Apr May Jun Jul Aug Sep NovOct Dec
• Evaluate modeling
outcome and
progress
• Finalize any
outstanding
modeling work
• Compile information
on modeling success
based on criteria
Manuscript compilation and
writing
• Post-modeling
evaluation
(e.g. sensitivity
analysis)
• Reach out to
regulatory
authorities for
input
Review and submit
manuscript

13. The European Network on Understanding
Gastrointestinal Absorption-related Processes
• COST ACTION CA16205
• SPRING MEETING Sofia, 12-13 Feb 2019
• Presentations available at
https://gbiomed.kuleuven.be/english/research/
50000715/50000716/ungap

14. Predicting the Effect of Acid Reducing Agents
• pH-dependent DDI may occur in the stomach when a poorly soluble
weakly basic drug with pH dependent solubility is co-administered with an
acid reducing agent (ARA) e.g. proton pump inhibitor (PPI), histamine 2
receptor antagonist (H2RA) or antacid
• Many weakly basic compounds show reduced exposure (Cmax and AUC)
which can lead to significant impact on efficacy of these compounds

15. PBBM for effect of acid reducing agents

16. MAM for ARA
• Data on pH-dependent solubility can be integrated
• Measured data on the physiological changes due to ARA can be integrated
– PPIs increases fasting gastric pH from ~1.3 to ~4.5
– Postprandial gastric pH increases from ~4.5 to ~6.5
– Decreased gastric emptying rate
FastedFed
4 - 56.5 1.8

17. A Case Study
• Erlotinib EGFR inhibitor used to treat
patients with locally advanced or
metastatic non-small cell lung cancer
• Lipophilic with high permeability and
low solubility
• CYP3A4 & CYP1A2 substrate
• The effect of omeprazole and
ranitidine on erlotinib has been
studied clinically and this modelling
was done retrospectively
Parameter
logP (O/W) 2.7
pka 5.65
fu 0.046
B/P 0.55
Permeability (cm/s) caco-2 33.6x10-6 -> human Peff 4.3x10-4
Buffer solubility of
HCl salt at different
pH (mg/ml)
pH mg/mL
2.5 0.6
3.4 0.32
5 0.0145
6.5 0.0058
Biorelevant solubility
at 37°C (mg/ml)
Media start pH end pH mg/mL
FaSSIF 6.5 6.4 0.0085
FeSSIF 5 5 0.0533

18. Step 1: Disposition Model
• Mean Cp(t) for IV and PO crossover
study 150-mg tablet vs 25-mg 30
minute intravenous infusion in 20
healthy mainly female subjects
• 2 compartmental model with nonlinear
clearance fit gives best fit
• Bioavailability estimated with saturable
clearance is 59% vs 106% based on a
simple non-compartmental analysis
Vc/kg= 0.826 L/kg CV= 26%
CL2/kg= 0.150 L/h/kg CV= 54%
V2/kg= 1.138 L/kg CV= 33%
Vmax = 4.47E-4 mg/s CV= 53%
Km = 0.232 µg/mL CV= 77%
K12 = 0.182 1/h CV= 60%
K21 = 0.132 1/h CV= 63%
Tlag = 0.228 h CV= 19%
Ka = 0.731 1/h CV= 53%
F = 59.27 % CV= 19%
nonlinear model fit in PKPlus

19. Step 2: Fasted State Simulation
• Vmax and Km transferred to
the enzyme table accounting
for changed units and free
fraction in plasma
• Default model simulation over
estimates observed Cp(t)
• Reduction in %fluid colon
improves match
10% fluid in colon
25% absorption from the large intestine.
1% fluid in colon
8% absorption from the large intestine.

20. Step 3: Fasted State with/without ARA
• Stomach pH changed from 1.3 to 4.0
• Gastric transit increased from 0.25h to 0.5hWithout omeprazole
Without ranitidine
With omeprazole
With ranitidine
AUCinf omeprazole ranitidine
Observed 54% 67%
Simulated 51% 60%
Sensitivity to gastric pH

21. Role of MAM in Managing the Effect of ARA
• MAM should play a role in integrating physicochemical, in vitro and in vivo data into a
mechanistic framework which can yield a fuller understanding of pH dependent DDIs
• A bottom-up approach assumes that all relevant factors are captured in the model and that in
vitro to in vivo translation is accurate.
• Therefore verification of simulations with clinical data is recommended before application to
predict untested situations
• PBPK simulations should be used to guide study design appropriately and allow exploration
of different scenarios (e.g. staggering of dosing with regard to the two interacting drugs)
• Collaborative cross-industry efforts are need to build confidence and extend the utility of this
approach. Further work is needed to support more detailed models for physiological
changes due to different types of ARA and in different populations.

22. Conclusion
• There is wide recognition of the
potential for MAM to streamline
development of oral formulations
• Increased confidence in MAM is
needed to expand the impact
with the regulators
• Collaborative efforts are ongoing
to address this and we can
confidently expect progress in the
near future PBPK in IND/NDA submissions to US
FDA OCP from 2008 to 2017
Grimstein et al JPS 2019

23. Coming in 2019
• FDA hosted workshop to take place in
Silver Spring this September
• Physiological Based Biopharmaceutics Modeling
(PBBM) to Support Pharmaceutical Quality
• Day 1 in vitro,
• Day 2 model verification
• Day 3 case studies.
• J. Dressman, Uni Frankfurt
• Xavier Pepin, AstraZeneca
• FDA
• Sandra Suarez, Andrew Babiskin, Poonam Delvadia,
Vidula Kolhatkar, Xinyuan Zhang,

24. Acknowledgements
• Colleagues from Roche pRED Pharmaceutical Sciences
• Colleagues from the GastroPlus User Group
• Colleagues from the IQ Food Effect Working Group
24

25. Questions raised by FDA
• What are the characteristics of drugs that are susceptible to pH-dependent
DDIs? Can a stepwise approach be applied to evaluate the interaction
potential?
• When conducting pH-dependent DDI assessments:
– What are the utilities and limitations of different approaches to evaluating DDIs (e.g., in silico,
in vitro, and dedicated clinical studies, as well as population pharmacokinetic analyses)?
– What are the study design considerations (e.g., study population, choice of ARAs, dosing
regimen and administration, and pharmacokinetic sampling) for the in vivo assessments
discussed in 2a above?
– Can we extrapolate the findings from a clinical DDI study with one ARA drug (a PPI, H2
blocker, or antacid) to anticipate the DDI potential for other ARAs in the same class or in a
different class?