DRUG EVALUATION IN PEDIATRICS USING K-PD MODELS: PERSPECTIVES Michel TOD, PhD Pascal Girard, PhD EA3738, Faculté Médecine 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. EMEA 14 April 2008 Tod / Girard 2
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. EMEA 14 April 2008 Tod / Girard 4
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 Time EMEA 14 April 2008 Tod / Girard 6
COMPONENTS OF A K-PD MODEL (2) � Effect model: links IR(t) to E(t) Emax E E max . IR ( t ) = E ( t ) 50 + EDK IR ( t ) EDK 50 = CL.CE 50 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 ) E max . IR ( t ) = − − Kin .( 1 ) Kout . R ( t ) production + dt EDK IR ( t ) 50 EMEA 14 April 2008 Tod / Girard 8
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 25 mg BID 80 20 120 mg BID 120 mg BID 15 60 360 mg BID 1200 mg BID 10 40 5 20 0 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 )] B k + EDK IR ( t ) 50 EMEA 14 April 2008 Tod / Girard 10
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 PK–PD 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 K–PD Model. P. Jacqmin et al. J. Pharmacokin Pharmacodyn 2007 NEFA plasma concentration–time profiles after IV infusion of N 6-( 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. EMEA 14 April 2008 Tod / Girard 14
K-PD SET-POINT MODEL (1) A kinetic-pharmacodynamic model for clinical trial simulation of antidepressant action: Application to clomipramine–lithium 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 clomipramine–lithium interaction. B.Gruwez et al. , Contemp Clin Trials, 2007. Oscillating profile… MADRS scores of patients treated with clomipramine and placebo or lithium. (clinical data) EMEA 14 April 2008 Tod / Girard 16
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: catégorical data 1.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 2 2 2 2 2 2 2 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. Hénin et al, 2007 0.25 5 0 0 0 0 0 0 0 0 0 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.0 0 Observed HFS 2 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. EMEA 14 April 2008 Tod / Girard 18
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. EMEA 14 April 2008 Tod / Girard 20
Recommend
More recommend
