beyond PKPD CPTR Workshop October 2 4 2012 Pentagon City MampSWG Objective For Preclinical Phases Deliver Quantitative PKPD models to help sponsors select therapeutic combinations ID: 340125
Download Presentation The PPT/PDF document "Modeling and Simulation" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Modeling and Simulation beyond PK/PD
CPTR Workshop October 2 – 4, 2012
Pentagon CitySlide2
M&S-WG Objective
:
For Preclinical Phases: Deliver Quantitative PKPD models to help sponsors select therapeutic combinations For Phase I: Deliver PBPK models to help sponsors predict first-in-human results for combination regimens (Pulmosim/SIMCYP) For Phase II & III: Deliver clinical trial simulation tools (based on quantitative drug-disease-trial models) to be used to help design TB drug regimen development studies Here a more in-depth look at the clinical setting
Mission and GoalsSlide3
SIMCYP Grant Application (CPTR+U of F)
Pulmosim
tool from PfizerDeveloped TB modeling inventoryDevelop drug-disease-trial model for TB White papersFDA qualificationData standards
Data sources
Database
3
Hollow Fiber modelSlide4
PBPK
Complex ADME processes: PBPK models account for anatomical, physiological, physical, and chemical mechanisms.
Multi-compartment approach to account for organs or tissues, with interconnections corresponding to blood, lymph flows and even diffusions.Develops a system of differential equations for drug concentration on each compartment as a function of timeIts parameters represent blood flows, pulmonary ventilation rate, organ volumes etc., for which information is reliable known[Enter Presentation Title in Insert Tab > Header & Footer4Slide5
PBPK Integrates the
Complex
Process of DistributionNormal lung tissue[Enter Presentation Title in Insert Tab > Header & Footer5
Inflamed lung tissue
Granulomatous tissue
CPTRSlide6
PBPK
6
PulmoSim: Framework for inhaled drugs that can serve as a foundation for orally administered antibiotics systemically distributed to the lungsSlide7
Clinical Trial Simulation Tools
Integrate the disease with pharmacology models
Takes into account design considerationsGobburu JV, Lesko LJ. Annu Rev Pharmacol Toxicol. 2009;49:291-301.Slide8
8
Trial Simulations Optimize Design Based on Quantitative Principles
Test Multiple Replications of
Trial Design Assumptions
Modify Design
0.4
0.5
0.6
0.7
0.8
0
10
20
30
40
50
60
Effect of Dose and Number of Subjects on Power to
Estimate Significant Effect of Drug vs Placebo
1 mg
2 mg
5 mg
10 mg
20 mg
30
4.5
6.5
18
48.5
73.5
40
13
29
76
87
91
50
27.5
52
85
95
99
60
40.5
62
90
97
100
70
55.5
71
94
99
100
N
Drug/Disease Model
Trial Designs
X possible doses
Different N
Sampling time
Inclusion criteria
Range of Outcomes
Analytics/Statistics
CFU
Trial Simulations Optimize Design Based on Quantitative PrinciplesSlide9
For Predictions
the Top-Down Approach is
Too LimitingDescribes existing data, lacks mechanistic insights, limited to explore new scenarios.Davies GR, et al. Antimicrob Agents Chemother. 2006;50(9):3154-6.Slide10
But the Bottom-up Approach is too expansive
Requires detailed mechanistic understanding, makes models more “portable”, limited by unverifiable assumptions.
Wigginton JE, et al. A model to predict cell-mediated immune regulatory mechanisms during human infection with Mt. J Immunol. 2001;166:1951-67Slide11
Intermediate Approach: Mechanistically-Inspired
Retains key mechanistic verifiable components, allows for parameter estimations and is fit for simulation purposes
Marino S et al. A hybrid multicompartment model for granuloma formation and T-cell priming in TB. J of Theor Bio. 2011:280:50-62Slide12
Leverage can be Obtained F
rom
Other AreasPredator-Prey models in viral infections such as with HCV may provide useful insights for TB modeling and simulationGuedj J. et al. Understanding HCV dynamics with direct-acting antiviral agents due to interplay between intracellular replication and cellular infection dynamics. J Theor Bio 2010;267:330-40Slide13
The Path F
orward
to a Successful M&S Platform in TBObtain the right datasets to model the dynamics of CFU as a function of drug exposure/dose and disease progression in a mechanistically-inspired settingLongitudinal dataDifferent combination therapiesDrug susceptible, MDR and XDR strain dataDevelop model that is predictive of CFU and linked to outcome taking into account appropriate other factors as co-therapy, demographics etcTest and validate the model(s) with regulatory buy-inDevelop tool that can interrogate the model to aid in trial design of compounds under investigation or in development[Enter Presentation Title in Insert Tab > Header & Footer13Slide14
Regulatory Review Process: What’s success?
Consultation and
Advise Process
14
Regulatory decision qualifying or endorsing the submitted tools
Success!!!Slide15
Modeling and Simulation beyond PK/PD
CPTR Workshop October 2 – 4, 2012
Pentagon CitySlide16
WHAT PREDICTIVE MODELING SHOULD DO
A DISEASE MODEL AND A MATHEMATICAL MODEL SHOULD GIVE A QUANTITATIVE PREDICTION:
HOW MUCH RESPONSE?
WITH WHAT DOSE?
ACCURACY SHOULD BE JUDGED BASED ON CLINICAL EVENT RATES and NOT another model or CONSESUSS
ACCURACY SHOULD BE BASED ON HOW ACCURATE CLINICAL PREDICTIONS ARE, NOT ON LACK OF COMPLEXITY OF THE MODELINGSlide17
M. tuberculosis
in the hollow fiber system
Gumbo T
, et al. (2006)
J Infect Dis
2006;195:194-201
Slide18
HFS:
Moxifloxacin
Concentration-Time ProfileSlide19
HFS, Simulations
and
Predictions Later on “Validated with CLINICAL Data” Efflux pump & cessation of effect of antibioticsThe rapid emergence of quinolone resistanceThe potency & ADR of Cipro/Orflox versus MoxiThe “biphasic” effect of quinolonesThe exact dose of Rifampin associated with optimal effectThe population PK variability hypothesis, and the rates of ADR arising during DOTSThe role of higher doses of pyrazinamide The “breakpoints” that define drug resistanceSlide20
The HFS in Quantitative
P
redictionHFS quantitative output on the relationship between changing concentration and microbial effect Human pharmacokinetics and their variability MODELING & SIMULATIONS Predictive outcome: dose, breakpoints, microbial effect, resistance emergence, regimen performanceSlide21
Gumbo T, et al.
Antimicrob
. Agents. Chemotherapy. 2007: 51:2329-36
Slide22
ISONIAZID HFS: Monte Carlo Simulations
INH inhibitory sigmoid
Emax based on hollow fiber studies% patients with nat-2 SNPs associated with fast acetylation versus slow acetylation in different ethnic groups: Cape Town, Hong Kong, ChennaiM. tuberculosis MICs in clinical isolatesPopulation PK data from (Antimicrob.Agents Chemother. 41:2670-2679) input into the subroutine PRIOR of the ADAPT II9,999 Monte Carlo simulation for different ethnic groups to sample distributions for SCL→AUC→AUC/MIC→EBA
Gumbo T, et al. Antimicrob. Agents. Chemotherapy. 2007: 51:2329-36
Slide23
PK-PD PREDICTED vs
OBSERVED EBA IN CLINICAL TRIALS
Gumbo T, et al. Antimicrob. Agents. Chemotherapy. 2007: 51:2329-36 Slide24
PREDICTION
PREDICT:Etymology
via
Latin
:
præ
-, "before"
dicere, "to say".“PREDICT” to say BEFORE
QUALITATIVE: Predict an event in terms of whether it occursQUANTITATIVE: Predict
extent and values prior to the event
ORACLES AND DEVINING THE FUTURE
http://www.crystalinks.com/delphi.htmlSlide25
If MDR-TB
Does
N
ot
A
rise
F
rom
P
oor
C
ompliance, W
hy
D
oes
I
t
?
Hypothesis: Perhaps the PK system (i.e., patient’s xenobiotic metabolism) is to blame
HFS output: kill rates, sterilizing effect rates (i.e., log10 CFU/ml/day)
Known clinical kill rates, sterilizing effect rates (i.e., log10 CFU/ml/day)
Performed MCS in 10,000 Western Cape Patients on the FULL REGIMEN
Srivastava S, et al. J. Infect. Dis. 2011; 204:1951-9
.
Slide26
Sputum conversion rate predicted = 56% of patients
Sputum conversion rate from prospective clinical studies in WC= 51-63%
External
Validation
of
Model
:
S
putum
C
onversion
R
ates
in
10,000
Patients
Srivastava S, et al. J. Infect. Dis. 2011; 204:1951-9
.
Slide27
Many (simulated) patients had 1-2 of the 3 drugs at very low concentration throughout, leading to
monotherapy
of the remaining drug
Drug resistance predicted to arise in 0.68% of all
pts
on therapy in first 2 months despite 100% adherence
Srivastava
S, et al. J. Infect. Dis. 2011; 204:1951-9
.
Slide28
Prospective
study of 142 patients in the Western Cape province of South Africa
Jotam
Pasipanodya
, Helen
McIlleron
*, André Burger, Peter A. Wash, Peter Smith,
Tawanda
Gumbo
Pasipanodya
J, et al. Submitted.Slide29
What
W
as Done
All patients hospitalized first 2
months
All had 100% adherence first 2
months
Drug concentrations measured at 8 time points over 24hrs in month
2
Followed for 2 years, 6% non-adherence
Pasipanodya J, et al. Submitted.Slide30
CART ANALYSIS: Top 3 predictors of Long term outcomes
Pasipanodya J, et al. Submitted.
0.7% patients developed ADR in 2 months versus 0.68% we predicted IN THE PAST from
modeling and simulations
: All ADR had low concentrations of at least one drugSlide31
Thank you!
[Enter Presentation Title in Insert Tab > Header & Footer
31Slide32
Identifying sources of variability
Individual variability in blood/air flow with body positions may affect drug distribution and elimination in different parts of the lung
http://www.nigms.nih.gov/NR/rdonlyres/8ECB1F7C-BE3B-431F-89E6-A43411811AB1/0/SystemsPharmaWPSorger2011.pdf32Slide33
Identifying sources of variability
Dormant and active bacterial populations may exhibit different effect sizes, even at saturation concentrations
http://www.nigms.nih.gov/NR/rdonlyres/8ECB1F7C-BE3B-431F-89E6-A43411811AB1/0/SystemsPharmaWPSorger2011.pdf33Slide34
Identifying sources of variability
Levels of resistance may explain a drug’s varying IC
50 magnitudeshttp://www.nigms.nih.gov/NR/rdonlyres/8ECB1F7C-BE3B-431F-89E6-A43411811AB1/0/SystemsPharmaWPSorger2011.pdf34Slide35
Identifying sources of variability
Additional factors that induce variability in a defined population?
http://www.nigms.nih.gov/NR/rdonlyres/8ECB1F7C-BE3B-431F-89E6-A43411811AB1/0/SystemsPharmaWPSorger2011.pdf35Slide36
Identifying sources of variability
Deeper mechanistic understanding of the disease processes
http://www.nigms.nih.gov/NR/rdonlyres/8ECB1F7C-BE3B-431F-89E6-A43411811AB1/0/SystemsPharmaWPSorger2011.pdf36Slide37
The new CPTR modeling and simulation work group
Integrating quantitative systems pharmacology, spanning different stages of the combination drug development process for TB
Leveraging previous work to advance existing drug development tools and develop new ones for specific contexts of useData-driven modeling and simulation tools: data standards and databases from available and relevant studiesSpearheading regulatory review pathways with FDA and EMA, to facilitate the applicability of those drug development toolsAligning and cross-fertilizing with other work groups to increase efficiency[Enter Presentation Title in Insert Tab > Header & Footer37