I ntroduction to population PKPD modelling modelling I ntroduction - - PowerPoint PPT Presentation

1

I ntroduction to population PKPD I ntroduction to population PKPD modelling modelling in in paediatric paediatric clinical pharmacology clinical pharmacology

Catherijne Knibbe Catherijne Knibbe, Oscar , Oscar Della Pasqua Della Pasqua, , Meindert Danhof Meindert Danhof

Leiden/ Amsterdam Center for Drug Research Division of Pharmacology

2

What is the problem? What is the problem?

• Drugs dosing in children is largely empirical
• Frequent under-and overdosing problems
• Efficacy and safety of drugs, in particular in

(premature) newborns is largely unknown Body weight is used for dose adjustment Body weight is used for dose adjustment instead of the PKPD relationships instead of the PKPD relationships

3

PKPD MODELLI NG: What is it? PKPD MODELLI NG: What is it?

4

• How to identify a safe and effective dosing

regimen in children in different age groups?

– First time in kids (early drug development) – Change in indication or age group, including neonates (clinical practice)

• Which factor(s) should be used to adjust the

dose for the individual child in different age groups? – dosing recommendation in the label

Clinical Questions Clinical Questions

5

Paediatric Paediatric Research I ssues Research I ssues

Unbalanced Unbalanced vs vs balanced balanced designs: designs:

– 100 observations for subject A 100 observations for subject A – 1 observation for subject B 1 observation for subject B

Sparse vs. serial Sparse vs. serial data: data:

– 2 measurements per subject 2 measurements per subject

20 40 60 80 100 120 6 12 18 24

Population approach Population approach

Simultaneous analysis of all available data: PK and/or PD parameters are simultaneously estimated taking into account differences between patients

1. POPULATION PK and/or PD parameters (fixed effects) 2. Inter-individual variability 3. Residual error

20 40 60 80 100 120 6 12 18 24 Concentration (mg/L) TIme (hr)

Predicted Observed

20 40 60 80 100 120 6 12 18 24 ID=1 (pred) ID=1 (obs) ID=2 (pred) ID=2 (obs) ID=3 (pred) ID=3 (obs)

7

Population PKPD Population PKPD modelling modelling

Inter- individual variability

Residual error

8

• Applicable to sparse and unbalanced data

sets (neonates, children, etc)

• Scientific basis for study/trial simulations,

Scientific basis for study/trial simulations, Scientific basis for study/trial simulations, dose adjustment or labeling extensions in dose adjustment or labeling extensions in dose adjustment or labeling extensions in

• ther populations
• ther populations
• ther populations (intra and interspecies)

(intra and interspecies) (intra and interspecies)

• Covariate analysis for identification of

Covariate analysis for identification of Covariate analysis for identification of predictors of variability in PK and PD predictors of variability in PK and PD predictors of variability in PK and PD

(genetics, body weight, age, interactions etc) (genetics, body weight, age, interactions etc) (genetics, body weight, age, interactions etc)

Population PK/ PD Population PK/ PD modelling modelling

9

Propofol concentration (mg/l)

1 2 200 400 600 1 2 200 400 600 1 2 200 400 600 1 2 200 400 600 1 2 200 400 600 1 2 200 400 600

Best Worst Median

Ventilated children Ventilated children (1 (1-

• 5

5 yrs yrs) ) following following cardiac surgery cardiac surgery in the I CU in the I CU

Children Children Adults Adults

Cl (l/min) Cl (l/min) (ml/kg/min) (ml/kg/min) 35* 35* 2.3 2.3 28* 28* V1 (l) V1 (l) (l/kg) (l/kg) 12 12 0.78* 0.78* 21 21 0.26* 0.26* Q (l/min) Q (l/min) (l/kg/min) (l/kg/min) 0.35 0.35 23 23 1.4 1.4 18 18 V2 (l) V2 (l) (l/kg) (l/kg) 24 24 1.54 1.54 139 139 1.88 1.88 Knibbe et al., Br J Clin Pharmacol 2002

6 samples of 250 6 samples of 250 ul ul per child per child 6 children 6 children

10 10

• Applicable to sparse and unbalanced data

sets (neonates, children, etc)

• Scientific basis for study/trial simulations,

dose adjustment or labeling extensions in

• ther populations (intra and interspecies)
• Covariate analysis for identification of

Covariate analysis for identification of Covariate analysis for identification of predictors of variability in PK and PD predictors of variability in PK and PD predictors of variability in PK and PD

(genetics, body weight, age, interactions etc) (genetics, body weight, age, interactions etc) (genetics, body weight, age, interactions etc)

Population PKPD Population PKPD modelling modelling

11

Propofol Propofol in non in non-

• ventilated children

ventilated children

200 400 600 800 1000 1200 time (min) 0.0 0.5 1.0 1.5 2.0

Propofol concentration (mg/l)

Knibbe

• bserved conc

Rigby-Jones Schuttler current study

Peeters MYM et al., Anesthesiology 2006 ; 104(3):466-474

12

Propofol Propofol in in nonventilated nonventilated children children

200 400 600 800 1000 1200 time (min) 0.0 0.5 1.0 1.5 2.0

Propofol concentration (mg/l)

Knibbe

• bserved conc

Rigby-Jones Schuttler current study

Peeters MYM et al., Anesthesiology 2006 Mar; 104(3):466-474

13

COMFORT-B 6 behaviour items

• alertness
• Calmness/agitation
• Respiratory response /

crying

• Physical movement
• Muscle tone
• Facial tension

14

Non Non-

• agitated children

agitated children

200 400 600 800 1000 1200 1400

time (min)

6 12 18 24 30

COMFORT-B

0.2 0.8 0.6 0.4

Propofol concentration (mg/l)

non-agitated, median performance no propofol

1.0 20.00 h 07.00 h

Peeters et al., Anesthesiology, March 2006

15

200 400 600 800 1000 1200 1400

time (min)

6 12 18 24 30

COMFORT-B

0.2 0.8 0.6 0.4

Propofol concentration (mg/l)

B) agitated, median performance start propofol 18 mg/h

8.6 kg

24 mg/h

1.0

Peeters et al., Anesthesiology, March 2006

propofol propofol

Agitated children Agitated children

16

Model based advised Model based advised propofol propofol dose dose 30 mg/h for a postoperative child of 10 kg 30 mg/h for a postoperative child of 10 kg

200 400 600 800 1000 1200 1400

time (min)

6 10 14 18 22

COMFORT-B

0.0 0.2 0.4 0.6 0.8 1.0

Propofol concentration (mg/l)

Peeters et al., Anesthesiology, March 2006

17 17

• Applicable to sparse and unbalanced data

sets (neonates, children, etc)

• Scientific basis for study/trial simulations,

dose adjustment or labeling extensions in

• ther populations (intra and interspecies)
• Covariate analysis for identification of

predictors of variability in PK and PD

(genetics, body weight, age, interactions etc)

Population PKPD Population PKPD modelling modelling

18

Body weight or age? Body weight or age?

400 800 1200 PCA 5000 10000 15000 BWS

age age

19 19

Identification of potential covariates Identification of potential covariates

(body weight, gender, age, renal function, PGx etc). Graphical evaluation of each covariate versus

• The individual post-hoc PK or PD parameter estimate
• the weighted residuals

Statistical evaluation using standard techniques Statistical evaluation using standard techniques

• 1. Change in objective function
• 2. Standard error of the additional parameter
• 3. Improvement of individual fits
• 4. Diagnostics: B) observed versus model-predicted

Covariate analysis Covariate analysis

Krekels et al, Expert Opin. Pharmacother. (2007) 8(12):1787-1800 Peeters MY et al., Anesthesiology, March 2006 and Dec 2006, CP&T March 2008

20 20

When more than one significant covariate for the simple model is found, the covariate-adjusted model with the largest decrease in objection function is chosen as a basis to explore the influence of additional covariates sequentially with the use of the same criteria Forward inclusion and backward deletion

Covariate analysis Covariate analysis

Krekels et al, Expert Opin. Pharmacother. (2007) 8(12):1787-1800 Peeters MY et al., Anesthesiology, March 2006 and Dec 2006, CP&T March 2008

21 21

Nature of the influen Nature of the influence of t ce of the covariat he covariate – preferably non-empirical (mechanism/physiologically based) – Consider the possibility of potential extrapolation or interpolation

Validation Validation confirms the influence of the covariates

Covariate analysis Covariate analysis

Krekels et al, Expert Opin. Pharmacother. (2007) 8(12):1787-1800 Peeters MY et al., Anesthesiology, March 2006 and Dec 2006, CP&T March 2008

22

Morphine Morphine PK in PK in Children Children

Supported by a grant of the Sophia Stichting voor Wetenschappelijk Onderzoek

500 3000 5500 8000 10500 13000 15500 18000 BWS 0.0 0.4 0.8 1.2 CL

Body weight Clearance

• 250 children:

– 70 premature neonates, 70 premature neonates, – 60 neonates, 60 neonates, – 60 < ½ 60 < ½ yr, r, – 30 < 1 yr, 30 < 1 yr, – 30 < 3 yr 30 < 3 yr

• 1-4 samples/24 h/pt
• BW median 2.8 kg

23

10-1.0 100.0 101.0 102.0 103.0

2 3 4 5 6 2 3 4 5 6 2 3 4 5 6 2 3 4 5 6 2 3

PNA

• 1.0
• 0.5

0.0 0.5 ET1

I nfluence of post natal age > / < 10 d I nfluence of post natal age > / < 10 d

Independent of gestational time or body weight at birth

24

Formation clearance to M3glucuronide Formation clearance to M3glucuronide

• bserved versus model
• bserved versus model-
• predicted

predicted

Post natal age < 10 d Post natal age < 10 d Post natal age > 10 d Post natal age > 10 d

2500 5000 7500 10000 12500 15000 17500 BWS 0.0 0.1 0.2 0.3 0.4 0.5 0.6

PNA > 10 days

1000 2000 3000 4000 BWS 0.00 0.02 0.04 0.06

25

Validation of sparse data studies Validation of sparse data studies

• Diagnostics

(e.g. observed versus model-predicted)

26

Validation of sparse data studies Validation of sparse data studies

• Diagnostics (e.g. observed versus model-predicted)
• Bootstrap resampling

– repeated random sampling to produce another data set (same size but different combination of individuals) – Compare parameters (250 times) with estimates from the original data set

• Visual predictive check

– Simulation with final estimates and compare the distribution of the

• bservations with the simulated distribution

– Plot of the time course of the observations and prediction interval for the simulated values

27

Validation of sparse data studies Validation of sparse data studies

• Diagnostics (e.g. observed versus model-predicted)
• Bootstrap re-sampling

– repeated random sampling to produce another data set (same size but different combination of individuals) – Compare parameters (250 times) with estimates from the original data set

• Visual predictive check

– Simulation with final estimates and compare the distribution of the

• bservations with the simulated distribution

– Plot of the time course of the observations and prediction interval for the simulated values

• Normalised Prediction

Discrepancy Errors (NPDE)

1) Brendel et al. Pharm. Res. 23(9); 2036-49 (2006)

Points to consider Points to consider

• Use of the population approach (nonlinear

mixed effects modelling) in all phases of the investigation

• Validation of population PKPD models
• Infrastructure for data sharing
• Neonates deserve further attention

28

29

Neonates Neonates, , young infants young infants are different! are different!

1 10 100

2 3 4 5 6 7 8 2 3 4 5 6 7 8 2 3 4 5 6 7

Bodyweight (kg)

0.001 0.010 0.100 1.000

9 2 3 4 5 7 9 2 3 4 5 7 9 2 3 4 5 7 9 2 3 4

Clearance (L/min)

R R R R R R R R R R R R R R R R R R R R R R R R C C C C C C A A A A A A A A A A A A A A AA A A A A A A A A

N N N N N N N N N N N D D D D D D D D D D D D D D I I I I I I I I I II I I I I I I II I I I N N N N N N N N N N N N NN

30

• Population PK-PD modelling (or non linear mixed

effects modelling) should be the PRIMARY ANALYSIS METHOD in paediatric drug development and dosing studies

• Population PK-PD models can be also be

developed based on data from PREVIOUS CLINICAL STUDIES (retrospective studies/meta analyses)

• Dosing regimen based on VALIDATED

POPULATION PK-PD MODELS should be included in the LABEL of drugs

Conclusions Conclusions

• University-Industry consortium with 6 industrial partners

(Eli Lilly, GSK, Johnson & Johnson, Organon, Nycomed, Pfizer)

• Unique infrastructure for data management, data analysis

and reporting: sharing of data, models and biological system specific information

• Emphasis on key factors in drug discovery and

development

– Translational pharmacology (efficacy and safety) – Developmental pharmacology (pediatrics, elderly) – Disease system analysis

Mechanism-based PK-PD modeling platform

32

Multidisciplinary Multidisciplinary, , multicentre multicentre research research

• Dr. M. van
• Dr. M. van Dijk

Dijk

• Dr. R.N. van
• Dr. R.N. van Schaik

Schaik

• Prof. dr. J.N. van den
• Prof. dr. J.N. van den Anker

Anker

• Prof. dr. D.
• Prof. dr. D. Tibboel

Tibboel E.

• E. Krekels

Krekels

• Dr. J. De
• Dr. J. De Jongh

Jongh

• Dr. O. Della
• Dr. O. Della Pasqua

Pasqua

• Prof. Dr. M.
• Prof. Dr. M. Danhof

Danhof

• Dr. M.Y.M.
• Dr. M.Y.M. Peeters

Peeters

• Dr. C.A.J.
• Dr. C.A.J. Knibbe

Knibbe Supported by NWO/ Supported by NWO/ Veni Veni

Leiden/ Amsterdam Center for Drug Research Division of Pharmacology