European Regulatory Viewpoint on PBPK in Support of ...€¦ · European Regulatory Viewpoint on...
-
Upload
nguyennguyet -
Category
Documents
-
view
220 -
download
0
Transcript of European Regulatory Viewpoint on PBPK in Support of ...€¦ · European Regulatory Viewpoint on...
European Regulatory Viewpoint on PBPK in
Support of Regulatory Decision Making
Susan Cole1,4, Anna Nordmark2,6, Ine Rusten3,5,6, Theresa Shepard1,6
2014 MISG New Technologies Forum on PBPK, June 30th, London
Affiliations: Medicines and Healthcare products Regulatory Agency, London, UK (1), Medical
Products Agency, Uppsala, Sweden (2), The Norwegian Medicines Agency, Oslo, Norway (3),
EMA Pharmacokinetics Working Party (4), EMA Paediatric Committee (5), EMA Modelling and
Simulation Working Group (6)
Disclaimer
The views expressed in this presentation are those
of the speaker, and are not necessarily those of
MHRA, MPA, NOMA or EMA.
30th June 2014
PBPK seen as a valuable tool: Encouraged in numerous European guidelines
“…Physiological based pharmacokinetic models may
be used as a tool… .” (Hepatic impairment guideline)
“PK/PD modelling techniques, using age appropriate and validated biomarkers,
need to be considered to find the optimal dose. … physiologically based
pharmacokinetic models to predict PK characteristics in the neonatal population
may be considered if appropriate.” (Medicinal products in term and preterm neonates)
“PBPK simulations may be used to evaluate the in vivo relevance of competitive or
mechanism based inhibition (MBI) observed in vitro… . If the results of the simulation
with appropriate sensitivity analyses are negative and the modelling is acceptable, no
in vivo study is required,” (DDI guideline)
“For some enzyme systems, where well validated in silico PBPK models
have been developed, these can be used to predict pharmacogenetic
differences… . “It may also be acceptable to use PBPK simulations to
predict the interaction effect in the subpopulation if the simulation is
qualified for this purpose.”(Use of pharmacogenetic methodologies)
3
PBPK seen as a valuable tool: Important potential value for benefit:risk decisions
From EMA-EFPIA Modelling and Simulation Workshop, December 2011
PBPK provides a mechanistic basis to reduce/quantify uncertainty in
extrapolation and to identify “at risk” populations.
Extrapolation - always a component of
benefit:risk decisions and can be an
important contributor to uncertainty.
Examples: elderly, polypharmacy (DDI),
critically ill, obese patients, paediatric, ethnic
groups (pharmacogenetics), … .
4
PBPK in European Procedures: When are regulatory decisions based on M&S made?
Drug development and model building
Learning and confirming
Preclinical Phase I Phase IIb Phase III Registration/ Labelling
(MAA/SmPC)
Phase IIa Phase IV
Continuum of learn/confirm/predict at each decision point
M&S M&S M&S M&S M&S
Uncertainty Confidence in drug and disease
Adapted from Lalonde RL et al., Model-based drug development. Clin Pharmacol Ther 2007;82:21-32
First presented at the EMA/EFPIA Modelling and Simulation Workshop, 2011
Anytime Scientific Advice
Clinical Trial Applications (some National Agencies), Qualification of Novel Methodologies
Early Paediatric Investigation Plan
Late MAA + post-lic.
M&S: modelling and simulation
MAA: marketing authorisation application
SmPC: summary of product characteristics 5
European regulatory experience: PBPK in Paediatric Investigation Plans
• Mandatory procedure,
binding decision on
development plan
• To support “paediatric use” in
all subsets of paediatric
population
• Data on efficacy, safety and
age-appropriate formulation
• M&S to support dose
selection, study design,
analysis plan
Analysis technique
Descriptive analyses
73 (90.1%) summary statistics including confidence intervals; graphics; summaries of PK or PD parameters
PK modelling 41 (50.6%) fixed effect or population PK models
PK-PD modelling
17 (21.0%) including exposure-response, PK-response models
Dose-response modelling
10 (12.3%) including dose-PD (eg, ANCOVA model), dose-toxicity, dose-PK-PD models
Physiologically-based PK modelling 3 (3.7%)
Dose-exposure modelling 3 (3.7%)
Other 22 (27.2%) Formal hypothesis testing on efficacy or PD endpoints; non-parametric
time-to-event analyses; other types of models not captured above
Hampson et al, Bridging the gaps: A review of dose-investigations in paediatric
investigation plans. Br J Clin Pharmacol. 2014 Apr 10. doi: 10.1111/bcp.12402.
6
European regulatory experience: PBPK in EMA Scientific Advice Procedures
• Optional procedure, written
advice, not binding
– Prospective input on
development
– Focus on development strategies
(not pre-evaluation of MAA data)
• Advice provided by SAWP (chair =
Rob Hemmings)
• MSWG opinion when M&S
involved (chair = Terry Shepard)
• Highly encourage for M&S with high
regulatory impact
59 procedures referred to MSWG in 2013
5 involving PBPK
(1 adult, 4 paediatric)
CHMP: Committee on Human Medicinal Products
SAWP: Scientific Advice Working Party
MSWG: Modelling and Simulation Working Group
Note: Can also seek advice directly with national
authorities (e.g. MHRA, MPA, NOMA, etc)
MSWG report and work plan:
http://www.ema.europa.eu/ema/index.jsp?curl=pages/contacts/CHMP/people_listing_000122.jsp&mid=WC0b01ac058063f485 7
European regulatory experience: PBPK in Marketing Authorisation Applications
During
MAA?
Drug as enzyme substrate +
Drug as enzyme perpetrator +
Transporter-based: substrate or perpetrator +
Organ impairment
(Hepatic and renal)-
Paediatrics +
Pregnancy, ethnicity, geriatrics, obesity,
disease states-
Biopharmaceutics
Food effect, formulation change, pH effect
(including DDIs on gastric pH), other dose
route
+
Miscellaneous Tissue concentration +
Drug-drug
interactions
Specific
populations
Potential Applications
8
European regulatory experience: PBPK in Marketing Authorisation Applications (cont’d)
Example SmPC text:
“In the presence of ketoconazole 400 mg once daily, …, exposure to …
increased approximately 2-fold. The predicted increase was approximately
3-fold in the presence of ketoconazole 200 mg twice daily using
physiologically based pharmacokinetic (PBPK) modelling. The uncertainties
of such modelling should be considered.”
Example Decision
27-fold increase in AUC when co-administered with
200 mg ketoconazole BID. Used to predict impact of
moderate and weak CYP3A inhibitors.
Did not agree that model was sufficiently
verified. Asked to do in vivo interaction study
with weak inhibitor.
Waiver of in vivo study for CYP inhibitor, not
meeting in vitro criteria.
Agreed that in vivo study unnecessary and lack
of interaction could be included in SmPC
Worst case evaluation when study not ideal (400
mg ketoconazole QD for substrate with long half-
life).
Avoided need for a repeated DDI study.
Simulated maximum interaction included in
SmPC.
Simulation of DDI with simultaneous inhibition of
two CYP pathways to support contradindication.In progress.
Waiver of in vivo study for inhibitor of drug
transporters.In progress, but unlikely to be endorsed.
9
Our approach to regulatory
review
Question submitted as part of feedback:
“I would appreciate a discussion on how the PBPK submission will be reviewed by the HA. Will
reviewers be hands-on experienced modellers? Will there be any possibility for discussion/clarification
of the model or follow up questions/requests for additional simulations? Any chance to supply model
files as is done with FDA?”
10
European philosophy:
1. All regulatory documents should be constructed with clarity of explanations to allow
high quality, efficient assessment. a. Emphasis on how supports quality, safety and efficacy for benefit:risk decision
b. Applies equally to 1°, 2° documents (MAA, response to questions; briefing documents, list of issues)
c. Applies equally to M&S (including PBPK)
2. Reviewers a. Process ensures that high impact M&S seen by relevant experts (e.g. MSWG)
b. PBPK guideline will prompt assessor training (expected 2015/16)
c. Many complementary skill sets are involved in review of PIPs, scientific advice, MAA module
2.7.2 (e.g. plausibility of assumptions, therapeutic context, clinical implications of uncertainty).
Our approach to regulatory
review
Question submitted as part of feedback:
“I would appreciate a discussion on how the PBPK submission will be reviewed by the HA. Will
reviewers be hands-on experienced modellers? Will there be any possibility for discussion/clarification
of the model or follow up questions/requests for additional simulations? Any chance to supply model
files as is done with FDA?”
11
European philosophy (cont’d):
3. Model files a. Seen as best practice to request these
b. Used to provide confidence that the company conclusions can be endorsed or to inform
remaining uncertainties to be addressed through questions to the company
4. Discussion/clarification a. Integral part of EMA qualification procedure
b. Integral part of national scientific advice, not guaranteed in EMA scientific advice. More likely
if highlight PBPK within questions
c. For MAA, request clarification TC (e.g. Day 120)
Three concepts applied to PBPK
modelling and regulatory review
Value
Uncertainty (opposite: confidence)
Regulatory
Impact
12
Medium impact
High impact
Low impact
Impact o
n re
gula
tory
decis
ion
+++ Scientific Advice, Supporting Documentation, Regulatory Scrutiny
++ Scientific Advice, Supporting Documentation, Regulatory Scrutiny
+ Scientific Advice, Supporting Documentation, Regulatory Scrutiny
Framework for M&S in Regulatory Review According to impact on regulatory decision
From EMA-EFPIA Modelling and Simulation Workshop, December 2011
Justify
Describe
Replace
13
Regulatory impact framework: PBPK applications
To support waiver of an in vivo study
for substrate of CYP enzymes.
To support waiver of an in vivo study
for inhibitor of CYP enzymes.
To predict optimal doses in different
age and weight categories of
children. To support SmPC statements
regarding the need to adjust dosage
for drug combinations not tested.
To provide quantitative evidence of
the plausibility of mechanisms
important for the disposition of the
drug.
High
High
Medium to High
High
Low to Medium
Key point: Regulatory Impact ≠ Value
14
IQ White Paper: Level of confidence in PBPK according to application
Application Confidence Comments
P450 + passive processes High -Moderate Intestinal metabolism challenging
Non-P450 + passive processes
Moderate -Low Hepatocytes predictive for some non-P450 processes. Expression levels & scaling factors
unclear.
Clearance/absorption by active
transport
Low Activity scaling factors poorly understood. Interplay of transporters and metabolic enzymes
challenging.
Reversible CYP inhibition or
induction alone High -Moderate
Accurate fm when non-P450 involved challenging. IV CL and mass balance not available at
early stages. Must account for experimental variability in Ki.
Time dependent CYP inhibition Moderate -Low Trend to over-prediction from in vitro data
Combined reversible, TDI &
induction
Low Difficult to evaluate mechanisms
Involving active transport Low to Moderate Predicting transport inhibition possible but intracellular concentrations challenging
B(DD)CS I drugs High No significant limitations or challenges in fasted or fed states
B(DD)CS II drugs
Moderate Need to ensure that in vitro data and / or in vivo models for solubility, dissolution and
precipitation are relevant for human
B(DD)CS III drugs Low
B(DD)CS IV drugs Low
Paediatrics, ethnic variations,
smokers, pregnancy, obese, elderly
Moderate -Low Abundance of enzymes and transporters limited or lacking. Changes in gut physiology
limited.
Organ impairment (renal and
hepatic)
Low Limited verification vs clinical data. Impact of renal/hepatic impairment on CYP expression
and transporter activities not fully clear.
Absorption, food effect & formulation prediction
Special population PK prediction
Preclinical & Clinical PK prediction
DDI prediction
Complex interplay between multiple factors including transporter interactions
15
IQ White Paper: Level of confidence in PBPK according to application
Application Confidence Comments
P450 + passive processes High -Moderate Intestinal metabolism challenging
Non-P450 + passive processes
Moderate -Low Hepatocytes predictive for some non-P450 processes. Expression levels & scaling factors
unclear.
Clearance/absorption by active
transport
Low Activity scaling factors poorly understood. Interplay of transporters and metabolic enzymes
challenging.
Reversible CYP inhibition or
induction alone High -Moderate
Accurate fm when non-P450 involved challenging. IV CL and mass balance not available at
early stages. Must account for experimental variability in Ki.
Time dependent CYP inhibition Moderate -Low Trend to over-prediction from in vitro data
Combined reversible, TDI &
induction
Low Difficult to evaluate mechanisms
Involving active transport Low to Moderate Predicting transport inhibition possible but intracellular concentrations challenging
B(DD)CS I drugs High No significant limitations or challenges in fasted or fed states
B(DD)CS II drugs
Moderate Need to ensure that in vitro data and / or in vivo models for solubility, dissolution and
precipitation are relevant for human
B(DD)CS III drugs Low
B(DD)CS IV drugs Low
Paediatrics, ethnic variations,
smokers, pregnancy, obese, elderly
Moderate -Low Abundance of enzymes and transporters limited or lacking. Changes in gut physiology
limited.
Organ impairment (renal and
hepatic)
Low Limited verification vs clinical data. Impact of renal/hepatic impairment on CYP expression
and transporter activities not fully clear.
Absorption, food effect & formulation prediction
Special population PK prediction
Preclinical & Clinical PK prediction
DDI prediction
Complex interplay between multiple factors including transporter interactions
16
Add value even where confidence is moderate to low?
Key point: Confidence ≠ Value, but uncertainty (i.e.
lack of confidence) must be managed
Prerequisites for regulatory
applications of PBPK modelling: Verification of drug model
PBPK Model
System model Anatomy
Biology
Physiology
Pathophysiology
Patient/disease extrinsic factors
Drug specific parameters ADME, PK, PD and MOA Metabolism
Active transport/Passive diffusion
Protein binding
Drug-drug interactions
Receptor binding
Weakest link, but important (DDI risk, paediatric ontogeny,
pharmacogenetics, etc)
17
ADME in typical dossier
In vitro metabolism studies • Human microsomes, hepatocytes
• Purified enzymes
• Addition of specific inhibitors
• Transporter interactions
In vivo studies • Excretion balance study (radioactivity in
excreta, metabolic profiling)
Often not integrated in a useful way
In addition • Correlation of in vivo metabolites to in vitro
pathways
• Co-administration of enzyme inhibitors
• Studies in extensive and poor metabolisers
• Other sources: ethnic differences,
extrapolation discrepancies
KEY: Quantitative integration across studies • Scaling of in vitro data to man + verification of in vivo data (with and without inhibitors,
extensive and poor metabolisers, etc)
• If can’t predict → understand drug disposition mechanisms? → confidence for extrapolation
• PBPK with bottom up, top down (+middle out) most reassuring
18
Optimal ADME in dossier
In vitro metabolism studies • Human microsomes, hepatocytes (follow
metabolite formation)
• Purified enzymes
• Addition of specific inhibitors
• Transporter interactions
In vivo studies • Excretion balance study (radioactivity in
excreta, metabolic profiling) design informed
by PBPK*
• With and without inhibitors, extensive and poor
metabolisers, etc
• IV study (dose)
19 * “virtual mass balance study” (Steve Hall, FDA PBPK meeting, March 2014)
Plausible Range: Transparency around uncertainty
Plausible Range
20
1. Description of M&S and role in development
(including regulatory/company impact).
2. M&S Assumptions.
3. Model building methodology and model evaluation.
Simulation methodology and good practices.
4. Issues for discussion in MSWG.
5. Answers to specific M&S questions. Other
comments/questions to SAWP/CHMP/PDCO.
MSWG Template: Highlighting M&S assumptions
21
E F P I A M I D 3 P A G E 2 0 1 4 A L I C A N T E
MID3: Assumptions
Assumptions Justification New/
Established
Testable/
Not-Testable
Test/Approach
to assess
impact
Evaluation
Pharmacological assumptions
Physiological assumptions
Disease assumptions
Data assumptions
Mathematical and statistical assumptions
Part 2:
Document & Reporting
Current Limitations
Good Doc Practice
Assumptions
Components
& Considerations:
Analysis Plan
Simulations Plan
Report
Slide from Scott Marshall, PAGE 2014 22
Preclinical Phase I Phase IIb Phase III Registration/ Labelling
(MAA/SmPC)
Phase IIa Phase IV
Anytime Scientific Advice
Clinical Trial Applications (some National Agencies), Qualification of Novel Methodologies
Early Paediatric Investigation Plan
Late MAA + post-lic.
PBPK in European Procedures Opportunities for regulatory review – drug models
Drug specific parameters ADME, PK, PD and MOA Metabolism
Active transport/Passive diffusion
Protein binding
Drug-drug interactions
Receptor binding
23
Can be considered in any of
Paediatric Investigation
Plan, Scientific Advice, MAA
Recommend/highly recommend to
seek scientific advice for medium/high
regulatory impact applications
Preclinical Phase I Phase IIb Phase III Registration/ Labelling
(MAA/SmPC)
Phase IIa Phase IV
Anytime Scientific Advice
Clinical Trial Applications (some National Agencies), Qualification of Novel Methodologies
Early Paediatric Investigation Plan
Late MAA + post-lic.
PBPK in European Procedures Opportunities for regulatory review – system models
System model Anatomy
Biology
Physiology
Pathophysiology
Patient/disease extrinsic factors
Independent of drug, common
across many drugs → can dissociate
from MAA
24
Consider qualification procedure
(specific conditions of use)
Plausible Range for System Models: Transparency around uncertainty
Plausible Range
Similar approach for systems
models with uncertainty? (e.g. plausible range for CYP ontogeny)
Transparency is key!
25
Preclinical Phase I Phase IIb Phase III Registration/ Labelling
(MAA/SmPC)
Phase IIa Phase IV
Anytime Scientific Advice
Clinical Trial Applications (some National Agencies), Qualification of Novel Methodologies
Early Paediatric Investigation Plan
Late MAA + post-lic.
PBPK in European Procedures Opportunities for regulatory review – drug models (cont’d)
Drug specific parameters ADME, PK, PD and MOA Metabolism
Active transport/Passive diffusion
Protein binding
Drug-drug interactions
Receptor binding
26
May be appropriate to consider
qualification procedure (e.g. model
substrates/inhibitors)
Qualification opinion – published
Qualification advice – not published
Benefits of “PBPK-thinking” in drug
development
1. Forces integration of physicochemical data and in vitro and in vivo ADME data
→ mechanistic understanding
2. Raises red flag when in vivo profiles are not predicted → identify gaps in
understanding of ADME
3. Likely to conduct more informative studies; and not to conduct uninformative
studies
4. Complementary to other M&S approaches
a. Dose selection, optimal study design for unstudied populations (e.g. FIH, paediatrics)
b. Anticipate important/informative covariates
5. “Chain of evidence” (idea from small populations guideline)
6. Build confidence for extrapolation (NCE)
7. Continued development of system models and increased confidence when
systematically applied over many NCEs and many applications
27
Challenges for qualification of PBPK
models
28
Characteristic Consequence
Lack of structural identifiability of system
models.
Very careful optimisation of drug-specific parameter,
hypothesis generation and testing, extensive supportive
data (in vitro, in vivo, pre-clinical).
Sensitivity analyses should include important tissue levels
(organs of elimination, site of action for efficacy and
toxicity)
Plethora of published data for probe
substrates, perpetrators.
Possibility of bias in data selection. Criteria for inclusion
and exclusion of studies (meta-analysis guideline).
System models continually evolve with new
systems data (e.g. transporter expression, GI
pH in children, covariance of CYP enzymes),
new predictive methodologies (e.g. Orbito
project).
Literature citations can never be accepted as adequate
demonstration of predictive performance of current
version of a software. Need mechanism for continuous
demonstration of continued system validity.
System models qualified using proprietary
data.
Possible to use in system qualification data sets for
demonstration of continued system validity of new
software versions? Need creative solution to deal with
this?
Limited systems data for certain processes,
subpopulations (transporter abundance,
ontogeny of enzymes, GI pH in children,
impact of aging)
Limited confidence in predictions. Driver for data
gathering, new methodology to derive systems
parameters, etc. Opportunity for EU projects? (e.g. Orbito)
Identify gap
Experience, expertise
Develop new
methodology
Acceptance
Regulatory standards,
guidelines, practice Cycle of
Innovation
The Future: PBPK in regulatory decision making is evolving
29
Regulatory mechanisms that support
continual evolution of models, while
maintaining documentation of continued
system validity
PBPK Concept paper
Released for public consultation (clarity around expectations for system
model and drug-specific parameters)
Key points
Endorse and support growth of PBPK applications to • provide mechanistic understanding;
• support dose selection, study design;
• inform SmPC for unstudied interactions;
• identify “at risk” populations;
• reduce or waive in vivo studies.
Confidence ≠Value ≠Regulatory Impact
• Where there is value, manage uncertainty and communicate openly → drive regulatory decision (more conservative)
• EFPIA assumption framework useful? • Regulatory impact drives assessment, expectations of
documentation • Seek scientific advice, include PBPK explicitly in questions • Modeller should attend discussion meeting
Maturity of PBPK and regulatory experience →
time is right to develop and communicate standards
Refusal or
Withdrawal
Approval
Benefit
Risk
CHMP Opinion + Annexes (SmPC, Conditions)
Indication Specific Obligations, RMP
Refusal or
Withdrawal
Approval
Benefit
Risk
CHMP Opinion + Annexes (SmPC, Conditions)
Indication Specific Obligations, RMP
30