DRUG EVALUATION IN PEDIATRICS

USING K-PD MODELS:

PERSPECTIVES

Michel TOD, PhD

Pascal Girard, PhD

EA3738, Facult Mdecine Lyon Sud

Lyon I University, France
Tod / Girard 1 EMEA 14 April 2008

USES OF MODELS

Describe quantitatively drug kinetics.

Simulate and predict.
Plan and design clinical trials.
Bayesian adaptation of drug dosing.

Tod / Girard 2 EMEA 14 April 2008

 DRAWBACKS OF PK-PD MODELS

Invasiveness:
blood samples for PK.

Logistic and cost

associated with samples
and measurements

Tod / Girard 3 EMEA 14 April 2008

A SOLUTION: THE K-PD MODEL

Kinetic Pharmaco-Dynamic
model
Drug concentrations are not
measured
Only the kinetics of response is
measured.
A simple model is used to
describe drug concentration
kinetics.

Tod / Girard 4 EMEA 14 April 2008

 LESS INVASIVE MEASUREMENTS OF
THERAPEUTIC RESPONSE

Body temperature, heart rate, blood pressure,

respiratory peek flow
Scores for depression (HAMD,), pain (VAS),
Frequency of seizures, emesis,
ECG, EEG
Bone density, tumor size,

Tod / Girard 5 EMEA 14 April 2008

COMPONENTS OF A K-PD MODEL (1)

Simplified PK model:
variable of interest: Input Rate (t) in mg/h
IR(t) = Ke.A(t)
5 Log2 / Ke
A(t)

Dosing A(t)
history Ke
Biophase
Tod / Girard 6 Time EMEA 14 April 2008
 COMPONENTS OF A K-PD MODEL (2)

Effect model: links IR(t) to E(t)

Emax
E max .IR( t ) E
E( t ) =
EDK 50 + IR( t )

EDK50 = CL.CE50 in mg/h

EDK50

IR
Tod / Girard 7 EMEA 14 April 2008

COMPONENTS OF A K-PD MODEL (3)

Model for a continuous response: links E(t) to R(t)

Example:
inhibition of dR( t )
= Kin.(1
E max .IR( t )
) Kout.R( t )
production dt EDK 50 + IR( t )
Tod / Girard 8 EMEA 14 April 2008
 TYPICAL CURVES OF 2 K-PD MODELS

K-PD MODEL (INHIBITION OF PRODUCTION) K-PD MODEL (INHIBITION OF LOSS)

35 120
30
100
25

RESPONSE
RESPONSE

25 mg BID 80 25 mg BID
20
120 mg BID 120 mg BID
15 1200 mg BID 60 360 mg BID
10
40
5
0 20
0 12 24 36 48 60 0 12 24 36 48 60
TIME (H) TIME (H)

Tod / Girard 9 EMEA 14 April 2008

COMPONENTS OF A K-PD MODEL (4)

Model for categorical response: links E(t) to probability to

observe score k of response R(t)

E max .IR( t )
log it[P(R( t ) k )] = Bk
EDK 50 + IR( t )

Tod / Girard 10 EMEA 14 April 2008

 TYPICAL CURVES FOR A K-PD MODEL OF
CATEGORICAL RESPONSE

K-PD MODEL (DIMINUTION OF PROBABILITY)

0.9
0.8
0.7
PROB. RESPONSE

0.6
Score from
P(R = 0)
0.5
P(R <= 1)
0 to 3 0.4 P(R <= 2)
0.3
0.2
0.1
0
0 12 24 36 48 60
TIME (H)

Tod / Girard 11 EMEA 14 April 2008

K-PD versus PK-PD MODELS

A semimechanistic and mechanistic population PKPD model for biomarker response

to ibandronate, a new bisphosphonate for the treatment of osteoporosis. G. Pillai, 2004
K-PD PK-PD

Typical fits for the K-PD and the PK-PD models in arbitrarily chosen subjects.
Observation (o), individual prediction (), population prediction (----) EMEA 14 April 2008
Tod / Girard 12
 K-PD MODEL: continuous response

Modelling Response Time Profiles in

the Absence of Drug Concentrations:
Definition and Performance Evaluation
of the KPD Model.
P. Jacqmin et al.
J. Pharmacokin Pharmacodyn 2007

NEFA plasma concentrationtime

profiles after IV infusion of N6-(p-
sulfophenyl) adenosine in Wistar rats.

Tod / Girard 13 EMEA 14 April 2008

K-PD MODEL: continuous response

Pharmacokinetic/Pharmacodynamic and
Time-to-Event Models of Ribavirin-
Induced Anaemia in Chronic Hepatitis C
M. Tod et al.
Clin. Pharmacokinet. 2005

Prediction of the K-PD model

for a typical patient.

Tod / Girard 14 EMEA 14 April 2008

 K-PD SET-POINT MODEL (1)

A kinetic-pharmacodynamic model for

clinical trial simulation of antidepressant
action: Application to clomipraminelithium
interaction.
B. Gruwez et al., Contemp Clin Trials, 2007.

Non-invasive
measurements

Box-plot of MADRS scores of patients

treated with clomipramine and placebo
or lithium. (clinical data)
Tod / Girard 15 EMEA 14 April 2008

K-PD SET-POINT MODEL (2)

A kinetic-pharmacodynamic model for

clinical trial simulation of antidepressant
action: Application to clomipraminelithium
interaction.
B.Gruwez et al., Contemp Clin Trials, 2007.

Oscillating
profile

MADRS scores of patients treated

with clomipramine and placebo or
lithium. (clinical data)
Tod / Girard 16 EMEA 14 April 2008
 KPD MODEL: biphasic kinetics

PKPD Modeling of the Effect of

Triamcinolone Acetonide on Central
Macular Thickness in Patients with
Diabetic Macular Edema
F. Audren et al.,
Invest Ophthalmol Vis Sci. 2004

Non-invasive
measurements
Examples of individual central macular
thickness (CMT) curves calculated from
individual CMT values (circles).
Tod / Girard 17 EMEA 14 April 2008

K-PD MODEL:
catgorical data
1.00

0 0 0 0 0 2 2 2 2 2
0 0 0 0 0 0 0 0 0 0 0 0 2 2 2
0 0 0 0 2
25
0.75

Weekly drug rate (g/week)

Predicted probability

20

A predictive model of Hand-and-

Foot Syndrome dynamics in
0.50

15

patients receiving capecitabine.

10

E. Hnin et al, 2007

0.25

1 2
1 2 1 2
1 2 1 2
2 2 1 2
1 2
1 2 0 0
1 0 0 0 0 0 1 0 0
1 2
1 2
2 1 2
1 2
1 2 1 2
1 2
1 2 1 2
1 2 1 1
1 1 1 1 1 1 1 1
0.0

0
Observed HFS

1
Evolution with time of observed 0
HFS scores, weekly drug rate and
predicted probabilities of grade 0, 0 5 10 15 20 25 30

1 and 2 predicted by the model Week

in a patient.
Tod / Girard 18 EMEA 14 April 2008
 LIMITS OF THE K-PD MODEL

The drug PK in biophase is handled as monocompartmental:

- PK is actually monocompartmental, or :
- Effect kinetics is slow compared to drug kinetics (Kout < Ke)
Complicated response models may be handled if correctly
specified.
K-PD models for drug-drug interaction are merely
identifiable.

Tod / Girard 19 EMEA 14 April 2008

CONCLUSIONS

K-PD models have been useful for modelling animal or

human data in adults.
Well suited if effect kinetics is rate limiting
Might be used in pediatrics to reduce experimental workload.
More useful if coupled with a minimally invasive
measurement of response.

Tod / Girard 20 EMEA 14 April 2008