Role of PK/PD in Evidence based Medicine. Tausif... · Role of PK/PD in Evidence based Medicine...
Transcript of Role of PK/PD in Evidence based Medicine. Tausif... · Role of PK/PD in Evidence based Medicine...
Role of PK/PD in Evidence based Medicine
Tausif Ahmed, PhD
Asst. Director,
Modeling & Simulation, GLP-BA and Met-ID
Piramal Enterprises Ltd, Mumbai
Third Annual Conference, Indian Association for Statistics in Clinical Trials
Agenda
� Introduction
� Historical perspectives
� PK-PD Models
� Applications of PK/PD
� Challenges in PK/PD modeling
� Summary
Is it safe? Does it work?
Does it work in double blind trials?
KNOWLEDGE
LEVEL
Drug Development Cycle
Drug Discovery & Development - Attrition Rate
Reasons for Phase 2 Failures 2008-2010
Nature Reviews: Drug Discovery, May 2011
Reasons for Phase 3 Failures 2008-2010
Nature Reviews: Drug Discovery, Feb 2011
Agenda
� Introduction
� Historical perspectives
� PK-PD Models
� Applications of PK/PD
� Challenges in PK/PD modeling
� Summary
– Pharmacometrics – still an emerging science
– Science that quantifies drug, disease, and trial information to aid efficient drug
development and/or regulatory decisions
– Pharmacometrics - a collection of model-based approaches used to
• extract from data & organize our understanding of a system’s behavior in a
concise manner
• do so in a language (i.e. mathematics) that allows simulation of the system
output
– Pharmacometric Models - three broad classes:
• Exposure-Response Models- specifically describe the relationships
among dose, drug concentration in blood (or another matrix), and clinical
response (effectiveness and undesirable effects)
• Disease Models- aim to describe disease progression
• Clinical Trial Models- describe patient demographics, adherence, dropout
rates, trial structure, and so on
Pharmacometrics defined
A Brief History of Pharmacometrics
THE LATE 60’S:
A PREMATURE BIRTH ?
1968: The Birth of PK/PD
• Presence of a delay between norepinephrine concentration-time
profiles and the kinetics of pharmacological response, i.e. blood
pressure-time data
• Gino Segre introduced the concept of a hypothetical effect
compartment to account for this delay
• This allowed an empirical description of time-dissociated kinetics
Segre G. Kinetics of interaction between drugs and biological systems. FarmacoSci. 1968
Oct;23(10):907-18
• NONLIN software only introduced in a year later by Carl Metzler
• It was written in FORTRAN-66 programming language for mainframe
computers
• Long gap of PK/PD publications until 1979
• CM Metzler. A Users Manual for NONLIN. Technical Report 7297 69 7292 005. Upjohn
Co., Kalamazoo, Michigan (1969)
1968: Was it a Premature Birth?
THE LATE 70’S:
A REBIRTH
A Brief History of Pharmacometrics
1979: Rebirth
• Lewis Sheiner and coworkers made Segre’s model more popular. They were
the first to formalize this concept into a model to describe hysteresis caused
by distribution to the biophase
• It was reborn as the ―Link Model
• SheinerLB, Stanski DR, Vozeh S, Miller RD, Ham J. Simultaneous modeling of
pharmacokinetics and pharmacodynamics: application to d-tubocurarine Clin Pharmacol
Ther. 1979 Mar;25(3):358-71
A Brief History of Pharmacometrics
THE LATE 80’S:
TODDLING
80’s: Growing Application
• Growing use of PK/PD modeling, with applications to diverse
therapeutic areas (mainly cardiovascular)
• Source: Pubmed search (Key-words: ―pharmacodynamic AND modeling)
80’s: Growing Application
• Growing use of PK/PD modeling, with applications to diverse
therapeutic areas (mainly cardiovascular)
A Brief History of Pharmacometrics
THE LATE 90’S:
A STEEP LEARNING CURVE
90’s: The Advent of Mechanism-Based PK/PD
• In so called – Indirect Physiological Response (IPR) models, the drug
concentration is no longer related to the PD variable itself. Instead, it
is assumed to modulate upstream and/or down stream regulation
mechanisms
90’s: The Advent of Mechanism-Based PK/PD
• Significant increase in the number of publications. Predominantly
theoretical until 1993. Growing number of increasingly complex
applications afterwards
• Source: Pubmed search (Key-words: ―pharmacodynamic and modeling)
90’s: The Advent of Mechanism-Based PK/PD
A Brief History of Pharmacometrics
TODAY:
A MATURE DISCIPLINE
• Application of integrated drug-disease-trial models to optimize clinical
development programs with respect to therapeutic potential, R&D
productivity and commercial value
Today: A Mature Discipline
Predict, LearnConfirm, Save
Predict, LearnConfirm, Save
Preclinical Phase
Clinical Phase
Drug Model: PK/PD
Post-NDA Phase
eIND IND EOP2a EOP2 NDA 6 mo safetypreIND VGDS
����Quantitative Analysis &/or Simulation
Safety Model: learn ‘at risk’ population, detect early or avoid risk
PK/PD Bridging
• Pediatrics
• Elderly
• Dosage forms
����
����
Disease Model: detect change, qualify new biomarkers, simulate trial design����
Label Update
BenefitRisk
����
Efficacy/Safety Benefit/Risk
Approval• Drug• Label
����
Individual Dosing
Cross-trial analysis: dose-response (efficacy/safety)
����Dose
RangingConfirming
S S����PK/PD
Dose-escalationPOP
S����Human PK/PD Prediction
Simulate (S) ����
• Dosing
• Human proof of principle
• Phase 3 trial design
• Value
Target Product Profile
Today: A Mature Discipline
1999 2003
• ...............as well as regulatory decisions about labeling and approval
Today: A Mature Discipline
Confidential
Operation of M&S
Modeling & Simulation
DMPK
Pharmacology
Biomarkers
Clinical ResearchData Management & Biostatistics IT
Imaging
Software's: WinNonlin,
NONMEM, R, S-plus,
ADAPT, SAS
• Preclinical & in-vitro studies
• Single and multiple dose pharmacokinetics
• Absolute bioavailability & dose proportionality
• Metabolism and drug interactions
• Food effects studies; Bioequivalence studies
• Special population studies – age, gender, race
• Pharmacokinetics in the target population
• Disease states such as renal and liver impairment
• Pharmacokinetic (PK) Modeling
• In vitro-in vivo correlation (IVIVC)
• Pharmacodynamic (PD) Modeling
• Population Pharmacokinetic (PopPK) Modeling
• Pharmacokinetic/ Pharmacodynamic (PK/PD) Modeling
• Physiology Based Pharmacokinetic (PBPK) Modeling
• Clinical Trial Simulation
Domain
Agenda
� Introduction
� Historical perspectives
� PK-PD Models
� Applications of PK/PD
� Challenges in PK/PD modeling
� Summary
PK/PD Modeling
Pharmacokinetics(PK)
Pharmacodynamics(PD)
Effe
ct
Concentration
Con
cent
ratio
n
Time
Effe
ct
Time
Pharmacokinetics/Pharmacodynamic Modeling(PK/PD)
What the body does to the drug
What the drug does to the body
MODELSimplified description of some aspect of realityHELPS IN PREDICTION
- Linear model
- Log-linear model
- Emax - model
- Sigmoid Emax – model
- Inhibitory models
- Effect Compartment
- Indirect Models
- Tolerance Models
Pharmacodynamic Models
� Drug must “interact” with a “receptor substance” to elicit an
activity
� Drug(D)+Receptor(R)↔ [DR] →Effect
� Rearrangement leads to Michaelis-Menten Equation
- D - Free drug concentration, Emax - Maximal effect,
- KD - Binding constant
PD Models: Basic Principles
EffectE D
DDK=
•+
max
Maximum Effect Model
Inhibitory Effect Model
Direct Effect Model
Biophase Distribution Model
Indirect Link Model
Analgesic effect of 400 mg oral ibuprofen quantified by
subjective pain intensity rating
Suri et al., Int J Clin Pharmacol Ther 1997, 35, 1-8 Effect
Ck10
D
k1
e
ka
Ce
D DoseC Plasma concentrationCe Effect compartment
concentrationk10, ka, k1e, ke0First-order rate constants
Emax-model
ke0
b
� Distributional delay between plasma and effect site concentration
� Dissociated time courses of concentration and effect
− Concentration maximum before effect maximum− Effect intensity increasing despite decreasing plasma
concentrations− Effects persist beyond the time plasma concentrations
are detectable
Counterclockwise hysteresis loop
Indirect Link Models
� The general indirect response model assumes that the change in the response parameter, which is related to the effect, is governed by an input or production process (zero-order rate constant kin) and an output or degradation process (first-order rate constant kout)
� Temporal dissociation between the concentration time course and the effect-time course (hysteresis) due to synthesis of a protein, reduction in a synthesis rate (reduction in hormonal levels)
� Thus, the rate of change in response (R) is described by
Indirect Response Models
Response (R)ki
n
kout
Production Degradation
Rkkdt
dRoutin
×-=
Dayneka et al., J Pharmacokinet Biopharm 1993, 21, 457-78
Indirect Response Models
Pharmacokinetics Pharmacodynamics
k21Ck10
Modulation of Input or Production Process
D
ka
Response
k0in
kout
+-
RkCIC
CIk
dt
dRoutin
⋅−
+⋅−⋅=
50
max0 1
RkCSC
CSk
dt
dRoutin
⋅−
+⋅+⋅=
50
max0 1
-
+
100 maxmax≤<> IS
+
-
Agenda
� Introduction
� Historical perspectives
� PK-PD Models
� Applications of PK/PD
� Challenges in PK/PD modeling
� Summary
Example 1: Dose-response Relationship
� Artemisinin derivatives commonly used to treat drug resistant falciparum malaria
� Doses of artesunate used in mono therapy and combination treatment- derived
empirically
� PD end point- PCT (Parasite clearance time)
� 47 adult patients with acute uncomplicated falciparum malaria and parasitemia were
randomized to receive a single oral dose of artesunate: 0 - 250 mg
� Inhibitory sigmoid Emax model fitted to dose vs shortening of PCT
� Emax was estimated as 28.6 h, and the 50% effective dose was 1.6 mg/kg bw
� No reduction in PCTs with the use of single oral doses of artesunate higher than 2 mg/kg,
and this reflects the average lower limit of the maximally effective dose
Example 1: Dose-response Relationship
Example 1: Dose-response Relationship
0 4 8 12 16 Time (hr)
Pla
sm
a C
oncentr
ation
AUC
Time above MIC
MIC
Peak
Trough
AUC / MIC
Efficacy of Aminoglycosides
Safety of Aminoglycosides
Efficacy of New Quinolones
Efficacy of b-lactam, macrolides, glycopeptides
Example 2: PK-PD Modelsin Preclinical Drug Development- Antimicrobials
Example 2: PK/PD Modelsin Preclinical Drug Development- Antimicrobials
� Telithromycin (Tel)- belongs to the class of antimicrobial, ketolides
� Effective against penicillin and macrolide resistant gram +ive Streptococcus pneumoniae
� Thigh infection model: CD-1 mice rendered neutropenic by ip injection of cyclophosphamide
� Colonies of S. pneumoniae (106- 107 CFU/mL) appx. 0.1 mL inocculum injected in thigh of
mice, 2h prior to initiation of antimicrobial therapy
� 2h post-infection, Tel 50 or 100 mpk dose administered
� PK- blood collected at regular intervals till 24 hours post-dose
� PD- 2h- post infection, Tel doses ranging from 25-200 mg/kg administered at different
dosing regimens
� After 24h of treatment, mice were sacrificed and thighs removed and CFU counted vs control
� PK-PD analysis done using Inhibitory Emax model
Example 2: PK/PD Modelsin Preclinical Drug Development- Antimicrobials
� AUC/MIC- strong determinant of the response against S. pneumoniae
� Maximal efficacy and bacterial inhibition against S. pneumoniae strains were predicted
by AUC/MIC and Cmax/MIC ratios of appx 1000 and 200, respectively
Example 2: PK/PD Modelsin Preclinical Drug Development- Antimicrobials
Confidential
Example 3: Preclinical Development-Diabetes
� xx (antidiabetic drug)
� PK-PD Analysis of data from the study in rat, hamsters and ob/ob mice
� Key results:
� PD Analysis: IC50 for xx ranged from 100-300 ng/mL across 3
different preclinical disease models (concn. vs biomarker levels
� Similar concn. expected in clinic for efficacy
� Good correlation between biologic response and biomarker (r = 0.85)
� Biomarker focus right from preclinical stage
� Compare and combine with data from all the preclinical studies : Build
knowledge-base for extrapolation to clinical trials
Confidential
Preclinical Data
� xx in Hamsters
y = 0.0218x + 103.38
R2 = 0.40
80
90
100
110
120
130
140
0 200 400 600 800 1000
Day 21 TG Level
Day
21
Bo
dy
Wt
Response vs Biomarker model
Disease Progression model
PK-PD (Response) model
PK-PD (Biomarker) model
Confidential
Prediction of FIH Dose
CL*MLP y = 1.3113x + 1.6455R2 = 0.9746
-1.00
0.00
1.00
2.00
3.00
4.00
5.00
-2.0 -1.0 0.0 1.0 2.0 3.0
log BW (kg)
log
CL
(m
L/m
in)
� Use of allometric scaling in predicting FIH dose� Other alternative approaches for FIH dose prediction: NOAEL from preclin. Species� FIH dose prediction based on efficacious doses in preclin. disease models� In vitro-In vivo correlation: Prediction human clearance from human hepatocyte intrinsic
clearance data� Simulated human PK profiles and correlation with efficacy/toxicity- Integrated approach
NOAEL Cmax= 11, 000 ng/mLToxic dose Cmax: 42, 000 ng/mL
Efficacious conc.
Confidential
Disease Progression Model- Diabetes
• Change in fasting plasma glucose (FPG) concentrations modeled as a function of Cp via an indirect-effect model on the assumption that drug xx reduces glucose by increasing theremoval rate of glucose from the plasma compartment
� Models developed based on phase I/IIa data help make valid predictions for larger phase II/III trials
FPG
HbA1c
)1(50
max
CEC
CEKout +
⋅+⋅inK
inK ' outK '
cHbAKFPGKdt
cdHbAoutin 1''
1 ⋅−⋅=
HbAlc
FPG
Dru
g C
onc.
Time (Week)
FPGCEC
CEKK
dt
dFPGoutin ⋅
+⋅+−= )1(
50
max
Cmt 1 Cmt 2
1st order Oral Absorption
Disease Progression Model- Diabetes
Disease Progression Model- Diabetes
Confidential
� Simulation of multiple dose profile of drug x based on single dose PK
� Correlate exposure to efficacy in deciding the proposed doses for RMD study
� Decide dosing regimen (QD vs BID) based on efficacy and safety
Design of Phase I RMD Study- QD Dosing
SS reached by day 4
Target Cmax= 80900 ng/mL
Agenda
� Introduction
� Historical perspectives
� PK-PD Models
� Applications of PK/PD
� Challenges in PK/PD modeling
� Summary
Current Problem in PK/PD
� Today, we do not have an adequate understanding of the clinicalefficacy/ MOA for most disease states
� Do not have an adequate understanding of the MOA for clinical toxicity
� This is the reason for the lack of suitable biomarkers and surrogatemarkers
� Validation of PD biomarkers
� Correlation of PK/PD model with safety or efficacy outcomes- Need todevelop disease progression models
� Validated Assay (reproducible, high precision….)
� Understanding of pharmacologic behavior of the drug andpathophysiology of the disease
Future research needs to address above areas
(Colburn, Washington, 1999)
Practical Aspects
CORPORATE PRESSURES
• mergers, consolidation,
small start-up
• Changing corporate
philosophy and structure is
changing the development
process
• Increased Productivity
THUSDevelop Innovative Drugsfaster with reduced risk, more effective cheaper
New Discovery Approaches:
� e.g combinatorial chemistry
� computational approaches
� robotic systems
More drug candidates in shorter period of timechallenges and opportunities in the PK/PD area
Pharmacometrics Status in India
� Search of website http://www.ctri.nic.in (clinical trials registry- India, Indian Council
of Medical Research) with key words “phase II/III trials 2010” reveals: None of the
trials have pharmacometrics (PM) component
� PM is at infancy stage in India
� Efforts are made to impart trainings in the fields of PK/PD data analysis and clinical
protocol writing
� Preconference workshop entitled “Pharmacokinetics: protocol development, conduct
and analysis” organized by South Asian chapter of ACCP at PLSL in Aug, 2011
� Trained (hands-on/didactic lectures) about 50 medical and pharmacy grad and post
graduates
� PAGIN formed in 2008: Formed to provide PM training in India
� ICMR grant to Sponsor “Research Methodology Workshop on PK/PD” held at ACTREC
(Advanced Centre for Training Research and Education in Cancer), Navi Mumbai from
May 1-7, 2012 to train post graduate students in the field of PM (co hosted by
Piramal)
Pharmacometrics Status in India
Summary
� Increasing awareness and understanding of PK-PD in drug discovery and dev
� Emphasis on biomarkers relating to target modulation for novel targets
� Strong collaboration between discovery DMPK, biology, IT, statistician and clinical
biomarker groups
� Translation of preclinical biomarkers to the clinical setting
� Start early and transfer PK-PD knowledge from discovery to development; refine model
as more data becomes available
� PK-PD provides a scientific basis for optimal FIH dose
� Optimal use of PK-PD modeling and simulation- fewer failed compounds, fewer study
failures and smaller number of studies needed for registration: Save time and money
Pharmacokinetics/Pharmacodynamics an uphill climb – but nice view!
Thanks