Prsentation PowerPoint

1
Population pharmacokinetics of nelfinavir and M8
in HIV-patients treated with a stable
antiretroviral therapy
Panhard X1, Brendel K1, Legrand M1,Taburet AM2,
Goujard C3, Mentré F1 and the Cophar 1 ANRS 102
study group 1 INSERM E0357, Department of
Epidemiology and biostatistics, Bichat Hospital
2 Clinical Pharmacy, Bicêtre Hospital
3 Internal Medicine, Bicêtre Hospital Paris,
France
2
OBJECTIVES

• To model the PK of nelfinavir (NFV) and M8 in
  HIV- infected patients with sustained virological
  response
• To estimate inter and intra- patient variability
  of the PK parameters of these protease inhibitors
  (PI)
• To study the influence of covariates in the PK of
  these PI

3
MATERIAL

• COPHAR I - ANRS 102 TRIAL (Feb. to Oct. 2001)
• Prospective, open, multicenter trial in HIV1-
  infected patients with
• - HAART containing NFV
• - stable regimen for at least 6 months
• - viral load lt 200 copies/ml for at least 4
  months
• PLASMA PI CONCENTRATIONS
• 1st visit (V1) 5 samples (trough, 0.5, 1, 3 and
  6 h after dosing)
• 2nd visit (V2) 2 samples (trough and 3 h after
  dosing)
• 1 to 3 months later
• HPLC quantification with UV-detection (in each
  center, after validation
• by a blind quality control)
• NB data below LOQ fixed to LOQ/2

4
METHODS (1)

• PK model
• one compartment model with 1st order absorption
  and elimination
• Identifiable parameters for NVF/ M8 model
• - ka
• - for NFV Cl/F and V/F
• - for M8 kem and (Vm/F) / km
• Cl/F and V/F derived assuming that km(ke km)/2

NFV (V/F)
M8 (Vm/F)
ka
km
kem
ke
5
METHODS (2)

• Statistical Modelling
• Population analysis, exponential random effects
  and combined error model
• joint analysis of NFV and M8 concentrations
• Analysis of concentration at V1
• ? Best random-effects model based on
  goodness-of-fit plots and AIC
• Test of the effect of covariates on the
  individual random effects
• age, sex, body weight
• comedications with other ARV
• Selection of the best combination of covariates
  (AIC and LRT)
• Analysis of concentrations at V1 and V2
• Estimation and test of IIV and IOV in the model
  with covariates
• Backward elimination of covariates (LRT)
• Software

6
RESULTS (1)

• Patients
• 46 patients, 2 most frequent regimens - 750 mg
  tid (n16)
• 
  - 1250 mg bid (n27)

Comedications with NRTI and NNRTI
Patient characteristics at baseline
7
NFV Concentrations
8
M8 concentrations
9
Data set for nlme
10
Fonctions Splus NFV-M8
fClnfv et fClm8 décrivent respectivement les
concentrations de NFV et de M8 fClmet est la
fonction utilisée dans nlme pour analyser
conjointement les concentrations
11
Code nlme V1 sans covariables
Sortie correspondante
12
Sélection des effets aléatoires



Modèle de départ effet aléatoire sur les 5
paramètres
? on supprime leffet aléatoire sur Ka
? on supprime leffet aléatoire sur VF2
? on conserve le modèle avec de la variabilité
inter-patients sur VF, Cl et Cl2



13
RESULTS (2)

• asignificantly different with and
  without RTV b assuming km(ke km)/2
• Estimated variabilities (expressed as coefficient
  of variation)

Estimated mean parameter
14
RESULTS (3)
Predicted and observed concentrations at V1 (?)
and V2 (x)
Nelfinavir
M8
15
Sélection des covariables sur EBE
Covariables continues (Corrélation de Spearman)
Covariables binaires (Wilcoxon)
On construit donc le modèle de population avec
ZDV sur Cl et AgeZDV sur Cl2.
16
Tableau de AIC sélection des covariables à V1

On teste toutes les combinaisons de covariables
afin de choisir le modèle avec le plus petit AIC

On retient le modèle avec ZDV sur CL et age ZDV
sur CL2

17
RESULTS (4)
Significant effect of covariates on CL/F of
Nelfinavir

• CL/F ? by 1.24 fold in patients with ZDV
  (plt10-4)

Simulated trough NFV concentrations in patients
with 1250mg bid
Gender Female
Male ZDV
No Yes
ZDV No Yes No
Yes
18
GOF plots du modèle final - NFV
19
GOF plots du modèle final M8
20
 Validation 

• Simulation of 1000 patients for most standard
  regimen (1250 mg bid)
• - random effects
• - predicted concentrations
• - error ? pseudo-observed concentrations
• At each time point
• - computation of median, 10th and 90th
  percentiles
• - superposition with observed data

21
Validation NFV sujets en 1250 mg bid
22
Validation M8 sujets en 1250 mg bid
23
AKNOWLEDGMENTS
Clinical centers Pr Vildé, Pr Leport, Dr
Duval, Bichat, Paris Pr Salmon, Dr Spiridon,
Cochin, Paris Pr Katlama, Dr De Sa, S. Maury,
Pitié-Salpétrière, Paris Pr Bazin, Pr Gallais,
Dr Ravaux, La Conception, Marseille Pr
Delfraissy, Dr Goujard, M. Mole, Kremlin-Bicetre,
Bicetre Pr Raffi, M. Sicot, Hotel-Dieu, Nantes
Dr Simon, Dr Amirat, C. Brançon,
Pitié-Salpétrière, Paris P. Goubin, Côte de
Nacre, Caen Dr Poizot-Martin, Dr Marin, Ste
Marguerite, Marseille Pr Estavoyer, Pr Séréni,
Dr Prevoteau, St Louis, Paris Pr Laurent, Dr
Drobacheff, C. Roche, St Jacques, Besançon Pr
Lang, Dr Krant, Hôpital Civil, Strasbourg Pr
Reynes, C. Merle, Gui de Chauliac, Montpellier Pr
Hoen, Dr Achard, St Jacques, Besançon Pr
Rozembaum, Dr Naït-Ighil, Rothschild, Paris
Pr Yéni, Dr Bouvet, Bichat, Paris
Pharmacological centers Pr Diquet, Dr Aymard,
Pitié-Salpétrière, Paris Dr
Peytavin, Dr Lamotte, Bichat, Paris Dr Dailly,
Hôtel Dieu, Nantes Dr Taburet,
Dr Vincent, Kremlin-Bicetre, Bicetre Dr Royer,
Dr Muret, Jean Minjoz, Besançon
Pr Lacarelle, Dr Solas, Timone, Marseille Dr
Rey, C. Abbara, St Vincent de Paul, Paris
Dr Alvarez, Mignot, Versailles Dr
Hillaire, Institut de Myologie, Montpellier
Dr Tod, Avicenne, Bobigny Dr
Poirier, St Antoine, Paris Dr
Sauvageon, St Louis, Paris Monitoring,
statistical analysis and modelling Pr Mentré,
Dr Legrand, K. Brendel, X. Panhard, L. Joubin, G.
Baron, Bichat, Paris