Kassir Nastya 2012 These
Kassir Nastya 2012 These
Kassir Nastya 2012 These
par
Nastya Kassir
Mars, 2012
Présentée par :
Nastya Kassir
Résumé
Ce travail de thèse porte sur l’application de la pharmacocinétique de population
dans le but d’optimiser l’utilisation de certains médicaments chez les enfants
immunosupprimés et subissant une greffe. Parmi les différents médicaments utilisés chez
les enfants immunosupprimés, l’utilisation du busulfan, du tacrolimus et du voriconazole
reste problématique, notamment à cause d’une très grande variabilité interindividuelle de
leur pharmacocinétique rendant nécessaire l’individualisation des doses par le suivi
thérapeutique pharmacologique. De plus, ces médicaments n’ont pas fait l’objet d’études
chez les enfants et les doses sont adaptées à partir des adultes. Cette dernière pratique ne
prend pas en compte les particularités pharmacologiques qui caractérisent l’enfant tout au
long de son développement et rend illusoire l’extrapolation aux enfants des données
acquises chez les adultes.
Les travaux effectués dans le cadre de cette thèse ont étudié successivement la
pharmacocinétique du busulfan, du voriconazole et du tacrolimus par une approche de
population en une étape (modèles non-linéaires à effets mixtes). Ces modèles ont permis
d’identifier les principales sources de variabilités interindividuelles sur les paramètres
pharmacocinétiques. Les covariables identifiées sont la surface corporelle et le poids. Ces
résultats confirment l’importance de tenir en compte l’effet de la croissance en pédiatrie.
Ces paramètres ont été inclus de façon allométrique dans les modèles. Cette approche
permet de séparer l’effet de la mesure anthropométrique d’autres covariables et permet la
comparaison des paramètres pharmacocinétiques en pédiatrie avec ceux des adultes. La
prise en compte de ces covariables explicatives devrait permettre d’améliorer la prise en
charge a priori des patients.
Ces modèles développés ont été évalués pour confirmer leur stabilité, leur
performance de simulation et leur capacité à répondre aux objectifs initiaux de la
modélisation.
iv
Dans le cas du busulfan, le modèle validé a été utilisé pour proposer par simulation
une posologie qui améliorerait l’atteinte de l’exposition cible, diminuerait l’échec
thérapeutique et les risques de toxicité.
Abstract
This thesis deals with the application of population pharmacokinetics in order to
optimize the use of certain medications in immunocompromised children undergoing
transplantation. Among the various drugs used in immunocompromised children, the use of
busulfan, tacrolimus and voriconazole remains problematic, particularly because of high
interindividual variability in their pharmacokinetics necessitating individualized doses
based on therapeutic drug monitoring. In addition, these drugs have not been studied in
children and the doses are adapted from adults. This practice does not take into account the
pharmacological characteristics of pediatrics throughout their development and makes
illusory the extrapolation of data acquired in adults to children.
The work done in this thesis studied sequentially the pharmacokinetics of busulfan,
voriconazole and tacrolimus by a population approach (non-linear mixed effects models).
The developed models have identified the main sources of interindividual variability in the
pharmacokinetic parameters of these drugs. The identified covariates are body surface area
and weight. These results confirm the importance of taking into account the effect of
growth in children. These parameters were allometrically included in the models. This
approach allows separating the effect of size from other covariates and enables the
comparison of pediatric pharmacokinetic parameters with those of adults. The inclusion of
these explanatory covariates should improve the management a priori of patients.
The developed models were evaluated to confirm their stability, performance, and
their ability to answer the original objectives of modeling.
In the case of busulfan, the validated model was used to simulate dosing regimens
that improve reaching the target exposure, reduce treatment failure and toxicity episodes.
3 Results....................................................................................................................75
4 Discussion ..............................................................................................................77
5 Acknowledgements .................................................................................................80
6 Conflict of Interest/Disclosure .................................................................................81
7 References..............................................................................................................82
8 Figure Legends .......................................................................................................87
9 Tables.....................................................................................................................88
10 Figures ...................................................................................................................92
CHAPITRE III ...............................................................................................................96
Population pharmacokinetic of intravenous and oral voriconazole in children ...................97
Abstract .........................................................................................................................98
1 Introduction ............................................................................................................99
2 Methods ............................................................................................................... 101
3 Results.................................................................................................................. 103
4 Discussion ............................................................................................................ 105
5 References............................................................................................................ 108
6 Figure Legends ..................................................................................................... 110
7 Tables................................................................................................................... 111
8 Figures ................................................................................................................. 114
CHAPITRE IV............................................................................................................. 116
Population pharmacokinetics and bayesian estimation of tacrolimus exposure in pediatric
liver transplant recipients .............................................................................................. 117
Abstract ....................................................................................................................... 118
1 Introduction .......................................................................................................... 119
2 Methods ............................................................................................................... 121
3 Results.................................................................................................................. 125
4 Discussion ............................................................................................................ 128
5 Acknowledgements ............................................................................................... 133
6 References............................................................................................................ 134
x
ASC0-t Aire sous la courbe calculée entre t=0 et le dernier temps de prélèvement
CL Clairance systémique
D Dose
F Biodisponibilité
GST Glutathion-S-transférases
xiv
P-gp P-glycoprotéine
pH Potentiel hydrogène
PK Pharmacocinétique
QH Débit hépatique
R0 Vitesse de perfusion
τ Intervalle d’administration
V Volume de distribution
WT Poids corporel
xv
Dédicace
Remerciements
Je tiens à remercier les personnes, qui de près ou de loin, ont contribué à la
réalisation de ce travail de thèse durant les années qui viennent de s’écouler.
Ma gratitude s’adresse aussi à Dr. Catherine Litalien et Dr. Yves Théorêt pour leur
confiance et leur soutien. Merci pour votre passion communicative, vos talents de
pédagogues, vos encouragements et le temps consacré à l’accomplissement de ce travail.
Merci au Dr. Philippe Ovetchkine pour m’avoir donné la chance de collaborer à ses
projets. Merci d’avoir partagé votre expertise et pour nos échanges intéressants.
Je remercie les membres du jury de m’avoir fait l’honneur de juger ce travail et tout
particulièrement au Dr Jacques Turgeon pour tout le temps que vous avez consacrez à cette
thèse.
À Samer Mouksassi, mon collègue et mon ami, qui m’a supportée au quotidien et
m’a transmis sa passion scientifique. Merci pour tous les moments et souvenirs que nous
partageons.
À Yana Kanaan, 19 ans d’amitié ça ne s’invente pas! Un peu comme une sœur; tu as
toujours été là dans les bons comme dans les mauvais moments.
À tous mes autres amis, c’est grâce à vous que j’arrive à garder mon équilibre au
quotidien. Vous comptez énormément pour moi.
À mes parents et mes frères, c’est grâce à chacun de vous si aujourd’hui j’en suis
arrivée là. Merci!
Enfin, je tiens à remercier Walid. Merci de m’avoir épaulée tout le long de cette
épreuve. Merci d’être toujours là pour moi.
1
CHAPITRE I
2
ÉTAT DE L’ART
Ce travail de thèse porte sur l’application de la pharmacocinétique de population dans
le but d’optimiser l’utilisation de certains médicaments chez les enfants immunosupprimés
et subissant une greffe.
La greffe d’organe est une pratique largement utilisée depuis plus d’un demi-siècle. La
transplantation de parties du corps d'un individu à un autre fascine l'être humain depuis des
siècles comme le prouve les traces laissées par ce sujet dans les récits mythologiques.
L'avènement de la médecine moderne a été accompagné de nombreuses tentatives dans ce
domaine et des progrès impressionnants ont été réalisés. Tous ces essais ont contribué à
mettre en lumière les barrières techniques et les limites biologiques fondamentales des
transplantations. L'histoire de la transplantation peut être comprise comme la découverte et
le dépassement de ces obstacles. En résultat, la transplantation est de mieux en mieux
maîtrisée.
Malgré les succès chirurgicaux que la transplantation a connus avant les années 70, le
rejet des greffons restait l’obstacle majeur et la principale complication des
transplantations. La découverte d'un principe actif, la ciclosporine, en 1970, a constitué une
nouvelle étape. Le développement des médicaments immunosuppresseurs a toutefois
amélioré le pronostic des greffes durant ces dernières années et a été une des conditions
essentielles pour que les transplantations d'organes, de tissus et de cellules deviennent le
traitement standard pour de nombreuses maladies graves. Ces médicaments préviennent le
rejet de greffe. L’objectif principal de l’immunosuppression est de maintenir le greffon
fonctionnel en ciblant l’activation des cellules T, la production de cytokines et l’expansion
clonale (prolifération), tout en limitant les risques de toxicité. Cependant pour des doses
excessives, l’utilisation de ces médicaments est associée à la survenue d’effets indésirables,
parfois toxiques; et à faible exposition systémique, elle est associée à un risque élevé de
rejet de greffe. En pratique clinique, le suivi thérapeutique pharmacologique (STP) est
utilisé pour optimiser les effets thérapeutiques de médicaments tout en diminuant les effets
indésirables.[1] Cette méthode d’optimisation et d’individualisation de la dose consiste à
3
1.1 Généralités
La pharmacologie est la "science des médicaments". Elle s’intéresse à l’interaction qui
a lieu entre du matériel vivant (cellules isolées, animal, et ultimement homme) et du
matériel chimique ou biochimique (principe actif). Il s’agit en effet d’une interaction
puisque le principe actif agit sur le vivant, définissant ainsi la pharmacodynamie et puisque
le vivant agit sur le principe actif, définissant ainsi la pharmacocinétique.[3]
Les vertus thérapeutiques de nombreuses plantes sont connues depuis la nuit des
temps, mais l’identification des principes actifs responsables de leurs effets
pharmacologiques date des XIXe et XXe siècles. Rudolf Buchheim est considéré comme le
fondateur de la pharmacologie en tant que discipline indépendante pour son entreprise en
1847 de l’étude expérimentale et systématique du mode d’action des médicaments.[3] Dans
les années 1930, le professeur Harry Gold, est parmi les premiers à utiliser le terme
pharmacologie clinique dans ses études sur la pharmacologie humaine du glycoside
digitalis.
Les enfants sont très différents des adultes en termes de perspectives sociales,
psychosociales, comportementales, et médicales. Contrairement à l’adulte, l’enfant est un
être en croissance et cette croissance n’est pas un processus linéaire. Son développement
débute dès la conception et se poursuit jusqu’à la fin de la puberté. Il s’agit d’un processus
dynamique qui s’accompagne de modifications importantes de la composition corporelle,
de la maturation fonctionnelle des organes impliqués notamment dans le devenir des
médicaments ainsi que de la maturation des récepteurs impliqués dans leurs effets (Tableau
1-1). Ces changements continus du corps et des organes influencent les effets des
médicaments et leur disposition et doivent être pris en compte dans la détermination de
schémas posologiques pédiatriques afin d’éviter les accidents toxiques et/ou l’inefficacité
thérapeutique.[5-7]
La population pédiatrique regroupe des enfants âgés entre 0 et 18 ans et elle est
généralement divisée en groupes d’âge. Une classification, internationalement acceptée,
comportant 5 groupes d’âge a été proposée lors de la Conférence internationale sur
l’harmonisation (ICH) tenue en 2000 [8]:
x les prématurés (<37 semaines de gestation)
x les nouveau-nés à terme (0 – 28 jours)
x les nourrissons (>28 jours – 23 mois)
x les enfants (2 – 11 ans)
x les adolescents (12 à 16 ou 18 ans, selon les pays)
6
Tableau 1-1 Différences physiologiques entre les enfants et les adultes influençant la
pharmacocinétique
Les pédiatres essayent d’adopter des approches pour déterminer la première dose
pédiatrique basées sur les doses chez l’adulte. Les différentes approches comportent [11]:
x les catégories d'âge
x la normalisation de la dose par le poids (WT)
x la normalisation de la dose par la surface corporelle (BSA)
x les principes d’allométrie
La détermination des doses basée sur les catégories d’âge peut paraître raisonnable vu
qu’elle prend en considération les différences physiologiques entre les nouveau-nés,
nourrissons, enfants et adolescents. Cette méthode est facile à appliquer, par contre elle ne
prend pas en considération l’effet de la maturation sur la PK des médicaments au sein d’un
même groupe. De plus, cette méthode considère un enfant n’ayant pas de pathologies.[11]
L’âge et le poids sont deux paramètres très corrélés, mais les paramètres PK
normalisés par le poids, peuvent varier en fonction de l’âge. La clairance normalisée par le
poids de certains médicaments peut être plus élevée chez les enfants que chez les adultes
suggérant une augmentation de la dose chez les enfants. Des limites de poids (minimum –
maximum) doivent être définies pour l’application de cette méthode.[11]
l’adulte (<0.4 L/kg), donc distribution dans les fluides extracellulaires qui sont corrélés
avec la surface corporelle chez l’enfant; 2) médicament métabolisé par les enzymes matures
du foie chez les enfants à partir de 6 mois car la clairance est corrélée au BSA. Cependant,
des études de pharmacocinétique de certains médicaments métabolisés par l’enzyme uridine
diphosphate glucuronosyltransférase et le CYP2D6 montrent que leurs clairances semblent
être plutôt corrélées au poids.[12-14] Les désavantages de cette approche sont la difficulté
de calculer la BSA, la disponibilité de plusieurs formules et la possibilité de surdoser les
nouveau-nés (par exemple, surdosage de valganciclovir [15]).[11]
En conclusion, la plupart des approches de dosage selon l'âge sont fondées sur
l'influence connue de l'ontogenèse sur l'élimination des médicaments. Les lacunes actuelles
dans les connaissances (par exemple, profils incomplets de maturation des enzymes
hépatiques et extra-hépatiques, le manque de connaissances en ce qui concerne l'expression
des transporteurs de médicaments) empêchent l'utilisation de formules simples pour la
prédiction de la dose efficace pédiatrique, surtout chez les nouveau-nés, où la relation entre
9
1.2.1 Absorption
Différentes voies sont utilisées pour administrer les médicaments aux enfants, la plus
utilisée étant la voie extravasculaire. L’absorption d’un médicament administré par cette
voie est influencée par des facteurs chimiques, physiques, mécaniques et biologiques
(Tableau 1-2).[9]
10
Tableau 1-2 Ontogenèse des principaux facteurs impliqués dans l’absorption gastro-
intestinale des médicaments
La plupart des médicaments sont administrés par voie orale chez l’enfant. En général,
la bouche est un site peu important pour l’absorption des médicaments car ces derniers ont
besoin d’être dissous en premier dans la salive et à cause du court séjour dans la bouche.
L’anatomie du pharynx et de l’œsophage est similaire aux adultes avec la seule différence
de taille. Les mouvements coordonnés oraux et pharyngaux nécessaires pour avaler des
matières solides se développent durant les 2 premiers mois d’un enfant né à terme.[9] Les
facteurs les plus importants qui influencent l’absorption des médicaments sont liés à la
physiologie de l’estomac, de l’intestin, et du conduit biliaire.
11
Dans les premières heures et jours qui suivent la naissance, le poids intestinal et la
masse muqueuse doublent presque pour tenir en compte le passage du cordon ombilical à
une alimentation orale, ce qui est similaire au passage de l’alimentation parentérale à
l'alimentation orale.[25] Cette croissance intestinale rapide et ce développement fonctionnel
sont stimulés par l'alimentation par le colostrum, qui est riche en facteurs de croissance
peptidiques tels que EGF, TGF alpha et l'IGF-1. La grande quantité d'IgA sécrétoires dans
le colostrum, qui a une haute biodisponibilité systémique, peut également jouer un rôle
dans le développement intestinal en plus de fournir une protection immunitaire.[25]
L’intestin proximal est le site principal d’absorption des médicaments et joue un rôle
important dans l’effet de premier passage de certains médicaments administrés par voie
orale. Le CYP3A4 est l’enzyme majoritairement exprimée dans la muqueuse intestinale.
Une faible expression de cette enzyme entraîne une augmentation de la biodisponibilité des
substrats CYP3A4 chez les nouveau-nés.[5, 24] La P-glycoprotéine (P-gp) est un élément
important qui limite la biodisponibilité des médicaments. La P-gp est une protéine de la
membrane cellulaire exprimée chez l’homme par le gène MDR1. Elle est présente au
niveau de l’intestin au pôle apical des entérocytes, où elle expulse les médicaments vers la
lumière intestinale, ce qui diminue leur biodisponibilité. Au niveau de la barrière
hématoencéphalique, elle diminue le passage des xénobiotiques dans le cerveau. Aux
niveaux hépatiques et rénaux, elle contribue à l’excrétion des substances xénobiotiques
dans la bile et l’urine, respectivement.[9] Cependant, les données sur l’ontogenèse de la P-
gp intestinale sont très limitées et son rôle chez les enfants n’a pas encore été décrit.[30]
D’autres transporteurs exprimés au niveau intestinal, comme le BCRP (ABCG2), limitent
la biodisponibilité des médicaments. Des biopsies duodénales des enfants montrent
l’expression très variable de l’ARN messager du gène MDR1. L’expression des
transporteurs P-gp et BCRP a été détectée dans les membranes de l’épithélium intestinal à
partir de 12 semaines de gestation.[31] Cependant, l'effet réduit de premier passage
13
1.2.2 Distribution
Une fois le médicament dans le sang "compartiment central", il peut se distribuer
dans différents autres compartiments du corps selon un processus indépendant de la voie
d’administration.[31] Les changements âge-dépendant dans la composition corporelle
altèrent la distribution des médicaments (Figure 1-1). Ces changements sont en partie dus à
des modifications physiologiques de la composition corporelle, à la liaison aux protéines
plasmatiques et tissulaires, au rôle des transporteurs ainsi qu’à la perméabilité
membranaire. D’autres facteurs qui varient au cours du développement tels que
l’augmentation du débit cardiaque et de la perfusion des organes peuvent également avoir
une influence sur la distribution. Le volume de distribution est un marqueur des
changements âge-dépendant dans la distribution des médicaments.[5, 9, 11, 24, 31]
à la naissance (10 – 15%), augmente chez le nourrisson et le petit enfant pour atteindre des
valeurs proches de celles des adultes à l’âge d’un an. La teneur en lipide diminue de
nouveau jusqu’à l’âge de préadolescence (11 ans) pour recommencer à augmenter. En effet,
les proportions plus grandes d’eau totale et d’eau extracellulaire chez le nouveau-né et les
jeunes enfants, avec le ratio eau/lipide plus important chez le nouveau-né que chez l’adulte
(57% vs. 25%) résulte en des concentrations plasmatiques plus faibles quand les doses sont
normalisées par le poids.[5, 24]
Figure 1-1 Changements physiologiques influençant le devenir des médicaments chez les
enfants
Perméabilité membranaire
1.2.3 Élimination
La clairance systémique est le paramètre pharmacocinétique le plus important qui
décrit l’efficacité avec laquelle les médicaments sont éliminés de l’organisme. La vitesse
d’administration et la clairance systémique déterminent la concentration moyenne à l’état
d’équilibre qui ensuite est liée à la réponse pharmacologique et/ou toxicologique.[2]
R0 FD/τ
ୱୱ ൌ ou
CL CL
La clairance systémique est la somme de toutes les voies d’élimination, mais chez
les enfants, les médicaments sont principalement éliminés par le foie et/ou par les reins.[35]
où QH est le débit sanguin hépatique, EH est le coefficient d’extraction hépatique qui dépend
de la fraction libre (flibre), de la clairance intrinsèque de la fraction libre (CLint,libre) et du
débit hépatique.
19
Pour les médicament avec un coefficient d’extraction hépatique faible (EH ≤0.3), le facteur
flibre *CLint,libre est négligeable devant QH et l’équation peut être simplifiée à:
Biotransformation hépatique
Enzymes de phase I
Parmi les enzymes de phase I, celles appartenant à la famille des CYP450 sont les
plus importantes dans l’élimination de nombreux médicaments.
x Les enzymes CYP450 fœtales (CYP3A7 et CYP4A1), qui sont fortement exprimées
au niveau du foie fœtal et qui sont impliquées principalement dans la
biotransformation des substances endogènes.
Tableau 1-3 Activité enzymatique in vitro des cytochromes P450 au niveau de microsomes
hépatiques chez le fœtus et l’enfant de différents âges, exprimés sous forme de fraction de
l’activité adulte (nmol/min/mg microsomal protein)
CYP1A2: L’activité du CYP1A2 est celle qui se développe en dernier, avec une
activité significative qui apparaît à partir de 1-3 mois [44, 45] et qui atteint 50% de celle de
l’adulte à l’âge d’un an.[46]
CYP2C: L’activité du CYP2C est absente chez le fœtus. Elle est limitée chez le
nouveau-né de moins de huit jours de vie et augmente considérablement à l’âge d’un mois
pour atteindre 50% de l’activité adulte.[47, 48] Par la suite, cette activité demeure
globalement stable au cours de la première année de vie.
atteint ~ 30% de l’activité adulte à l’âge d’un mois. Sa maturation se complète vers l’âge
d’un an.[46]
CYP2E1: L’expression fœtale du CYP2E1 est incertaine mais, si elle existe, c’est en
très faible quantité. Son activité augmente immédiatement après la naissance pour atteindre
la valeur adulte vers l’âge d’un an.[47]
Bien que la vitesse de maturation de ces enzymes soit différente, il est connu
qu’elles sont toutes, à l’exception du CYP3A7, immatures à la naissance et qu’elles arrivent
à maturité durant la première année de vie. Par conséquent, la clairance hépatique des
médicaments éliminés par les enzymes CYP450 est faible chez le prématuré, le nouveau-né
et le jeune nourrisson, se traduisant par une demi-vie d’élimination plus longue et une
nécessité de réduire la posologie.
x Une activité CYP450 hépatique plus grande et une masse hépatique plus importante
chez l’enfant s’accompagnant d’une capacité de biotransformation absolue plus
élevée
Enzymes de phase II
L’ontogenèse des réactions impliquant des enzymes de la phase II est moins bien
définie que celles des réactions impliquant les enzymes de phase I. Les voies de
glucuroconjuguaison et d’acétylation sont inefficaces et immatures chez le nouveau-né et le
jeune enfant comparativement à l’adulte, alors que les sulfotransférases et les glutathion-
transférases sont bien développées et qu’elles fonctionnent efficacement dès la
naissance.[35]
hépatocytes entre les semaines 16 et 24 de la gestation, alors que la classe alpha prédomine
dans le foie des enfants et adultes. Les données sur le développement de l'expression des
enzymes GST demeurent confuses, mais suggèrent que le développement est substrat-
dépendant et relativement bien développé chez le nourrisson.[35, 55]
Sulfotransférase (SULT) : Les SULT peuvent être décrits dans 4 catégories. Les
données sur les différents sous-types sont limitées. Les SULT1A1 sont présents au même
niveau à l’âge fœtal, néonatal et adulte. Les SULT1A3 diminuent après la naissance. Les
SULT1A2 sont faibles à l’âge fœtal et augmentent de façon stable jusqu’à atteindre le
niveau adulte. Une nouvelle famille avec expression prédominante dans le cerveau,
SULT4A, a été récemment décrite mais ses fonctions sont encore inconnues.[35, 55]
N-acétyltransférase (NAT) : Les données sur l’ontogenèse des NAT sont limitées.
Deux gènes ont été identifiés, NAT1 et NAT2. NAT1 est plus faible dans les échantillons
fœtaux que chez les adultes.[3, 55]
25
Pour les médicaments principalement éliminés inchangés dans l’urine par le rein, leur
clairance chez les nouveau-nés et nourrissons est limitée par l’immaturité de la filtration
glomérulaire et la sécrétion tubulaire. Cependant, pour les nourrissons plus âgés et les
enfants, une clairance rénale plus rapide est observée comparativement aux adultes. Ceci a
été observé surtout pour la digoxine, phénytoine, carbamazepine, clindamycine, cimétidine,
morphine, chlorpheniramine, cetirizine, etc.[9] En conséquence, des doses en mg/kg
relativement plus élevées sont nécessaires pour atteindre les mêmes niveaux plasmatiques
que les adultes. Ce ratio concentration-dose plus faible chez les enfants peut être dû à
plusieurs phénomènes liés à la fonction rénale (augmentation de la capacité de sécrétion
tubulaire) ou indépendant de la fonction rénale (liaison aux protéines plus faible, liaison
tissulaire plus forte, augmentation du métabolisme hépatique). L’explication est parfois
plus difficile à cause de la maladie sous-jacente chez les enfants.[9]
27
Dans les essais de phase I réalisés habituellement chez le volontaire sain, les études
de pharmacométrie permettent de décrire la relation concentration-réponse, de simuler pour
déterminer le design de l’étude et choisir les doses pour évaluation ultérieure. Dans les
essais de phase II réalisées chez le patient avec un nombre plus important de sujets, les
études de pharmacométrie permettent d’identifier et de confirmer les covariables
prédictives et de simuler pour déterminer le design de l’étude. Dans les essais de phase III,
les études de pharmacométrie permettent de confirmer les covariables prédictives, les
ajustements de dose et d’extrapoler à d’autres populations ou indications.
Cette analyse est effectuée chez chaque sujet de l’étude séparément pour estimer les
paramètres PK individuels. Cette méthode permet une estimation rapide et simple des
paramètres PK pour une dose donnée. Cependant, la fiabilité repose sur la qualité, la
quantité des données pour une bonne estimation de l’ASC et de ke. Idéalement, un
minimum de 3-4 demi-vies devrait être écoulées pour obtenir une estimation fiable de ke. Il
est aussi suggéré d’avoir 3-4 observations pour estimer ke même si deux points sont
suffisants pour faire passer une ligne. De plus, avec cette méthode il n’est pas possible de
prédire les concentrations en fonction du temps, de caractériser la PK des principes actifs
ayant une élimination non-linéaire ou de faire des corrélations PK/PD directes.
50
40
C o n ce n tr a tio n
30
20
10
0 2 4 6 8 10 12 14
Te mp s (h e u re s )
Les cercles vides représentent les concentrations observées, les cercles pleins représentent la
concentration moyenne à chaque temps d’observation, et la ligne représente le profil moyen.
25
20
C o n c e n tra tio n
15
10
0
0 2 4 6 8 10 12
T e m p s (h e u re s )
Les lignes bleues représentent les profils individuels et la ligne rouge représente le profil moyen.
La variabilité interindividuelle (η i) suit une loi gaussienne ayant une moyenne nulle
et une variance Ω (ω2) qui est une matrice définie positive:
La variance des erreurs résiduelles peut être constante et assume donc que la
précision est égale pour toutes les mesures. Ceci est souvent le cas pour les modèle s
concentrations-effet. Dans certaines circonstances, l’erreur peut changer avec les
différentes mesures et la variance est alors proportionnelle. L’erreur proportionnelle est
utilisée si les mesures sont toujours positives et si leur précision diminue avec
l’augmentation de la valeur de la mesure. Une combinaison des deux modèles d’erreur peut
être utilisée pour améliorer la prédiction aux limites inférieures des essais où l’erreur est
considérée constante, et une erreur proportionnelle.
37
Le but de l’étude PKPOP est d’estimer les valeurs des paramètres PK typiques, des
effets aléatoires et des erreurs résiduelles afin que les concentrations prédites aient le
maximum de vraisemblance avec les concentrations observées.
Dans les modèles non-linéaires à effets mixtes, la vraisemblance n’a pas de solution
analytique. Plusieurs algorithmes d’estimation par maximum de vraisemblance spécifique
ont donc été développés. Les premiers algorithmes étaient basés sur une linéarisation du
modèle au premier ordre autour des effets fixes comme dans l’algorithme First-Order
(FO)[70] ou autour des effets aléatoires comme dans l’algorithme First-Order Conditional
Estimation (FOCE)[82] ce qui permet de se ramener à un modèle linéaire mixte dont la
fonction de vraisemblance a une forme explicite. Par la suite, une méthode d’estimation où
la vraisemblance est calculée par une approximation de Laplace a été proposée.
Récemment, l’algorithme Maximum Likelihood Expectation Maximization (EM) a été
développé.[83] Contrairement aux algorithmes FO et FOCE, cette méthode d’estimation
exacte ne requiert pas la linéarisation des équations non-linéaires. L’étape d’estimation se
décompose en une estimation de paramètres individuels suivie de la maximisation qui
permet d’obtenir de nouvelles valeurs des paramètres, réutilisés à la première étape de
l’itération suivante jusqu’à convergence de l’algorithme.
données et d’avoir une idée de leur structure. Cette analyse permet d’identifier les
valeurs aberrantes, la corrélation entre les variables et leur distribution. Ceci est
important car la plupart des modèles sont basés sur des assomptions sur les
distributions sous-jacentes. Les représentations graphiques des données permet
d’accomplir l’étape de l’exploration des données.
x Modèle stochastique: les variabilités sur les paramètres PK peuvent être de nature
interindividuelle, résiduelle, et parfois inter-occasionnelle.[79]
La variabilité interindividuelle: Elle quantifie l’écart entre un paramètre
typique d’une population et celui d’un individu. Cette variabilité peut être liée
à des facteurs démographiques (âge, sexe, race, etc.), environnementaux
(diète, etc.), génétiques (phénotype d’une voie métabolique), facteurs
fonctionnels (maladie concomitante, insuffisance rénale ou hépatique…),
facteurs biologiques ou chronobiologiques. Ces différents facteurs sont
appelés covariables et sont à l’origine d’une variation des paramètres
pharmacocinétiques entre individus ou, parfois même, chez un même individu
au cours du traitement.
La variabilité résiduelle: Elle fait référence à la variabilité non expliquée d’un
paramètre associée par exemple aux erreurs de mesure (variabilité due aux
méthodes analytiques, à l’imprécision des temps de prélèvements…), erreurs
du modèle, entre autres. Elle quantifie l’écart entre la valeur prédite par le
modèle et la valeur observée pour un individu.
Il est aussi possible de considérer une variabilité intraindividuelle (ou
variabilité inter-occasionnelle) qui quantifie l’écart des valeurs des paramètres
PK d’un même individu au cours du temps.
x Analyse de covariable: L’identification des covariables qui peuvent expliquer les
variabilités interindividuelles des paramètres PK, permet d’individualiser les
posologies.
prédites et les concentrations observées est estimée par la valeur de la fonction objective
(OFV) qui est égale à -2*log (vraisemblance). En utilisant la même méthode d’estimation
avec la distribution χ 2 et pour un degré de liberté (modèles hiérarchiques), une diminution
de OFV de 3.84 et 10.8 témoigne d’une différence significative avec p=0.05 et 0.01,
respectivement. Dans certaines situations spéciales comme un petit nombre d’individus, les
conditions asymptomatiques ne sont pas réalisées et le test de rapport de vraisemblance
n’est plus valide. On pourrait alors utiliser des tests de permutations ou autres distributions
de la différence de -2 log-likelihood (-2LL).[84] La qualité de l’ajustement du modèle est
déterminée aussi par une distribution homogène des résidus en fonction du temps et des
prédictions, une faible erreur résiduelle du modèle et une bonne précision de l’estimation
des paramètres PK.
initial qui a été réellement prélevé dans la population, l’inférence statistique étant basée sur
les résultats des échantillons ainsi obtenus. La technique de bootstrap est itérative. La
première étape consiste à tirer aléatoirement avec remise n-1 individus parmi les n formant
l’échantillon initial. A chaque tirage, chaque individu possède une même probabilité de
figurer dans l’échantillon de bootstrap égale à 1/n. La seconde étape consiste à calculer la
moyenne empirique sur l’échantillon de bootstrap. Ces deux étapes sont ensuite répétées un
grand nombre de fois, typiquement 1000 et plus (N). Les N moyennes empiriques
fournissent alors une approximation de la distribution de la vraie moyenne, dont la valeur
estimée correspond à la moyenne empirique des N moyennes issues des échantillons de
bootstrap. L’estimateur ainsi obtenu est alors plus robuste que la moyenne empirique sur
l’échantillon initial. Les distributions statistiques des paramètres estimés dans le modèle
initial et dans les échantillons bootstrapés (donc N jeux de valeurs) sont comparées. Le
biais sur chaque paramètre est calculé en faisant la différence entre la valeur obtenue par le
modèle final et celle obtenue par le bootstrap. Si le biais est faible le modèle est dit robuste
aux variations aléatoires des données.
information Matrix».[92] Malheureusement, les enfants ou leurs parents peuvent refuser les
prélèvements répétitifs. Cependant, les données manquantes peuvent toujours être utilisées
dans une étude de population pédiatrique.
Les simulations basées sur les paramètres PK et les variabilités estimées permettent
d’améliorer les études en pédiatrie. Le choix des doses pour les études cliniques peut être
fait par les simulations (clinical trial simulations).
Les enfants sont différents des adultes par deux aspects principalement: la
croissance et le développement. Ces aspects propres aux enfants sont à prendre en
considération lors de la modélisation.
Dans les essais de phase I réalisés habituellement chez le volontaire sain, les études
pharmacocinétiques et pharmacodynamiques permettent de déterminer la dose maximale
tolérée. Dans les essais de phase II réalisées chez le patient avec un nombre plus important
de sujets, les études pharmacocinétiques et pharmacodynamiques permettent d’évaluer
l’efficacité et l’influence de covariables sur la relation dose-réponse. Les études
pharmacogénétiques dans ces premières phases du développement permettent d’évaluer,
pour des protéines clés, l’impact de polymorphismes génétiques sur la réponse au
médicament et le cas échéant d’influer sur la sélection des patients dans les essais de phase
III. Ces dernières sont des étapes clés du développement du médicament qui impliquent de
très larges effectifs. A ce stade, les études pharmacogénétiques permettent de considérer un
plus grand nombre de polymorphismes et de quantifier la part de la variabilité
interindividuelle associée à ces polymorphismes. Après la mise sur le marché, les études
pharmacogénétiques permettent d’améliorer la compréhension de la relation entre
l’efficacité et/ou la toxicité du médicament et les polymorphismes.
est appelé CYP3A4*1B. La fréquence de cette allèle varie selon les ethnies, elle est de 0%
chez les Asiatiques, 2-10% chez les Caucasiens et 35-67% chez les Africains-
américains.[100] L’allèle CYP3A4*1B a été associé à une augmentation de la transcription
du gène in vitro,[101] mais ces observations n’ont pas été confirmées par d’autres
investigateurs.[102, 103]
est de 2.2% avec absence de l’enzyme.[105] Les allèles CYP2C19*2 et CYP2C19*3 sont
responsables de la majorité des allèles retrouvés chez les métaboliseurs lents; chez les
asiatiques, le CYP2C19*3 est particulièrement présent.[106]
Le busulfan, à forte dose, induit la mort cellulaire en alkylant les acides nucléaires pour
inhiber la transcription et la réplication des régions d’ADN concernées.[115] Le caractère
lipophile du busulfan lui confère sa forte myélotoxicité.
Le busulfan était initialement disponible juste sous forme orale. En 1999, la FDA a
approuvé la formulation intraveineuse.
4.1.1 Absorption
Le busulfan est rapidement absorbé après administration orale. La concentration
maximale est atteinte entre 1.5 et 2.5 heures.
La biodisponibilité du busulfan est très variable, surtout chez les enfants.[116] Elle
est estimée entre 22% et 120%, chez les enfants de 1.5 à 6 ans. Chez les patients de 13 à 60
ans, la biodisponibilité varie de 47% à 103%.[117] Cette grande variabilité est liée en partie
à la forme d’administration (comprimés, comprimés écrasés et mis en solution ou
48
L’absorption du busulfan a été décrite par un modèle d’ordre zéro[118] ainsi que
par un modèle de premier ordre.[119, 120] Certains patients expérimentent une absorption
retardée ou une élimination prolongée.[118]
4.1.2 Distribution
Le busulfan est faiblement lié aux protéines plasmatiques. Moins que 5% du
busulfan est réversiblement lié aux protéines plasmatiques alors que 30% sont
irréversiblement liés.[121]
4.1.3 Élimination
Le busulfan est métabolisé principalement par conjugaison au glutathion. Cette
réaction est catalysée par les glutathion-S-transférases hépatiques. Aucun des métabolites
ne semble contribuer significativement à l’efficacité et à la toxicité. Les métabolites sont
éliminés par voie rénale alors que la molécule mère est éliminée par voie biliaire.
49
L’acétaminophène peut diminuer les taux de glutathion dans le sang et les tissus, et
par conséquent diminuer la clairance du busulfan.
Tableau 4-1 Résumé des paramètres PK du busulfan chez les enfants greffés du foie
BSV BSV
Âge
Référence n Paramètres Pharmacocinétiques CL V
(années)
(%) (%)
Modèle avec BSA: CL = 4.2 (l/h/m2)
Tram et Modèle avec WT allométrique : CL = 4.1 (l/h/kg 0.75)
94 0.4-18.8
al.[132] V = 18.4 (l/kg)
F = 93 – 99 (%)
Nguyen et CL = 4.57 + 2.97*(LOG(WT) - 3) (l/h)
24 0.45-16.7 19 12
al.[133] V ~ WT
Zwaveling et CL = 4.8*(WT/19)0.84
77 0.2-23 24 28
al.[125] V = 15*(WT/19)0.98
Age ≤ 4 ans : CL = 0.288*WT (l/h)
Johnson et Age > 4 ans : CL = 0.173*WT (l/h)
29 0.08-18.3 24.7 11.1
al.[126] GSTA1*B: CL *0.704 (l/h)
V = 0.682*WT (l)
Booth et CL = 4.04 *(WT/20)0.742 (l/h)
24 -- 23 10.9
al.[134] V = 12.8*(WT/20)0.843 (l)
Nakamura et CL/F = 3.25*(AGE/18)0.376 *(AST/29)-0.161 *0.787DIS (l/h)
103 0.2-11 31.2 13.6
al.[135] V/F = 9.07*(1+0.0736*(WT-10)) (l)
Schiltmeyer CL/F = 4.13*BSA (l/h/m2)
48 0.4-18.1 26 31
et al.[136] V/F = 21.3*BSA (l/m2)
AST = alanine aminotransférase; BSA = surface corporelle; BSV = variabilité interindividuelle; CL =
clairance systémique; CL/F = clairance orale apparente; DIS = maladie maligne (oui=1, non=0); F =
biodisponibilité; GSTA1*B = génotype de la glutathion-S-transférase; n = nombre de patients; V = volume de
distribution; WT = poids corporel.
52
Des études in vitro ont démontré une activité du voriconazole contre les Candida spp,
incluant Candida albicans, Candida glabrata, Candida krusei, Candida neoformans,
Candida parapsilosis, et Candida tropicalis.[142]
Une exposition faible au voriconazole est associée à un échec thérapeutique chez les
patients atteints d’une aspergillose, alors qu’une exposition élevée est associée à une
augmentation du risque de toxicités oculaires et des encéphalopathies.[147, 149, 150] Le
suivi thérapeutique pharmacologique est actuellement pratiqué pour optimiser les thérapies
avec voriconazole. L’ASC est le paramètre PK le plus utilisé pour caractériser l’exposition
totale au médicament.
5.1.1 Absorption
Le voriconazole est rapidement absorbé après administration orale et la
concentration maximale est atteinte en 2 heures.[143] La biodisponibilité du voriconazole
est estimée à plus que 90% et n’est pas affectée par le pH gastrique. Cette biodisponibilité
élevée permet le passage de la forme IV à la forme orale chez les patients.
5.1.2 Distribution
Le voriconazole se lie à 58% aux protéines plasmatiques.[151] Il a un grand volume
de distribution ce qui suggère une distribution dans les compartiments intra- et
extracellulaires. Le volume de distribution du voriconazole à l’état d’équilibre est de 4.6
L/kg.[144]
5.1.3 Élimination
Le voriconazole est éliminé par métabolisme : moins que 2% d’une dose
intraveineuse sont éliminés sous forme inchangée dans les urines. Il est métabolisé par N-
oxydation dans le foie par le CYP2C19, majoritairement, et à moindre taux par les
55
restent élevées (>50%) ce qui implique le besoin de plus d’études pour mieux l’expliquer.
Cependant, selon les études publiées, le polymorphisme du CYP2C19 affecte la clairance
du voriconazole.
59
6.1.1 Absorption
Le tacrolimus est rapidement absorbé après administration orale. La concentration
maximale est atteinte en 2 heures chez les enfants. La biodisponibilité du tacrolimus est
généralement faible. Chez les enfants ayant subi une transplantation hépatique, elle a été
estimée en moyenne à 25% et varie de 3 à 77%.[175]
6.1.2 Distribution
Le tacrolimus se lie fortement aux érythrocytes et la fraction sang/plasma varie de 4
à 114[7, 152] et aux protéines, surtout l’albumine et l’α1-glycoprotéine. La distribution
entre le plasma et les érythrocytes dépend de la concentration du tacrolimus et est affectée
par plusieurs paramètres: l’hématocrite et la concentration des protéines plasmatiques.[177,
178] La concentration des protéines et le nombre des érythrocytes augmentent durant les
premiers mois après la transplantation ce qui pourrait affecter la distribution et l’élimination
du tacrolimus.[179] En transplantation hépatique, le volume de distribution du tacrolimus
chez les enfants peut atteindre 1.8 fois celui des adultes.[180]
61
6.1.3 Élimination
Le tacrolimus est éliminé de façon linéaire par métabolisme: moins que 1% d’une
dose intraveineuse est éliminé sous forme inchangée dans les urines.[7, 152] Il est
métabolisé par le CYP3A4 et CYP3A5 dans le foie et dans la paroi intestinale pour former
au moins 10 métabolites inactifs.[32, 181-183] Les voies métaboliques sont la O-
déméthylation principalement, puis l’hydroxylation et la glucuronidation. La faible affinité
et la capacité du CYP3A7 pour le tacrolimus suggèrent que cette enzyme ne jouerait aucun
rôle dans le métabolisme du tacrolimus in vivo.[52]
La voie oxydative catalysée par les enzymes CYP450 est immature durant les
premiers mois après la naissance et atteint une activité importante entre 6 et 12 mois, qui
peut être plus élevée que l’activité moyenne chez l’adulte. Cette maturation graduelle de
l’activité enzymatique durant les premiers mois après la naissance contribue
significativement à la variabilité interindividuelle du taux d’élimination du tacrolimus chez
les enfants.
La clairance du tacrolimus augmente avec le temps chez les enfants ayant subi une
transplantation hépatique[7] et diminue en cas d’infection par le virus de l’hépatite C.[178]
Des ajustements de la dose en découlent.
62
Le tacrolimus est métabolisé par les CYP3As. Les inducteurs ou les inhibiteurs de ces
enzymes causent une diminution ou une augmentation de la concentration de tacrolimus.
Les conséquences sont un rejet de greffe dans le premier cas (phénytoine, phénobarbital,
carbamazépine, et primidone)[184] et une augmentation du risque de toxicité dans le
deuxième cas (érythromycine,[185] clarithromycine,[186] clotrimazole,[187]
fluconazole,[188, 189] kétoconazole,[173] itraconazole,[190] nifédipine,[191]
diltiazem,[192] chloramphénicol,[193] et corticostéroïdes[184]). L’association de la
rifampine avec le tacrolimus provoque une diminution des concentrations du dernier.[56]
Tableau 6-1 Résumé des paramètres PK du tacrolimus chez les enfants greffés du foie
BSV BSV
Référence n Paramètres Pharmacocinétiques CL V
(%) (%)
CL= (0.0749+0.000457*POD)*(15*(WT/15)0.290) (l/h)
Yasuhara et
33 V= 2.76*(15*(WT/15)0.290) (l) 52.1 27.4
al.[7]
F= 19% (fixed)
CL= 1.46*(1+0.339*(AGE–2.25) (l/h)
Sam et V= 39.1*[1+4.57*(BSA–0.49)] (l)
20 33.5 33.0
al.[173] BILI <200 μmol/l: F= 0.197*(1+0.0887*WT–11.4) (%)
BILI ≥200 μmol/l: F= 0.197*(1+0.0887*WT–11.4)*1.61 (%)
CL/Fwhole liver = 44 (l/h)
Staatz et
35 CL/Fcut-down liver = 5.75 (l/h) 110 297
al.[194]
V/F= 617 (l)
Garcia
Sanchez et 18 CL/F= 10.4*(WT/70)0.75*e(–0.00032 T)*e(–0.057 BILI)*(1–0.079*ALT) (l/h) 24.3 --
al.[193]
Wallin et CL/F= 0.148 + (1.37*POD3.78)/(5.383.78+POD3.78) (l/h/kg 0.75)
73 NR 90
al.[195] V/F= 27.2 (l/kg)
développés à partir de creux. Le poids corporel a été inclus dans la plupart des équations
pour décrire la cinétique tu tacrolimus. L’inclusion de paramètres décrivant la fonction
hépatique dans le calcul de la clairance semble raisonnable vu que le tacrolimus est
métabolisé par les enzymes CYP3A. Enfin, plusieurs études ont confirmé l’effet du
polymorphisme du CYP3A5 sur l’absorption et le métabolisme du tacrolimus. Certaines
études ont rapporté un changement de la cinétique du tacrolimus avec le temps post-
transplantation alors que d’autres n’ont pas trouvé d’effet. Enfin, l’étude par Staatz et.
al.[194], décrit une cinétique différente selon le type de foie transplanté. Basé sur ces
résultats très divers et sur la variabilité interindividuelle, large dans certaines études, plus
d’études pourraient améliorer la compréhension de variables affectant la cinétique du
tacrolimus.
65
7 Objectifs de la thèse
L’utilisation de médicaments chez les enfants subissant des greffes et
immunosupprimés n’est pas toujours optimale. Pour certains de ces médicaments,
caractérisés par une fenêtre thérapeutique étroite et une grande variabilité
pharmacocinétique inter- et intraindividuelle au cours du temps, le suivi thérapeutique
pharmacologique est utilisé pour optimiser les effets thérapeutiques tout en diminuant les
effets indésirables. Cette méthode d’optimisation consiste à surveiller les concentrations
sanguines du médicament et à adapter les posologies de façon à maintenir les
concentrations dans une zone thérapeutique optimale. Le suivi thérapeutique
pharmacologique est aujourd’hui considéré comme indispensable à la bonne utilisation de
certains médicaments comme le tacrolimus, le busulfan et le voriconazole. Cette méthode
n’est pas convenable chez les enfants, n’est pas pratique pour les patients externes et coûte
cher à l’hôpital.
Dans le but d’améliorer l’utilisation des médicaments en pédiatrie, les données de suivi
thérapeutiques à l’hôpital Ste-Justine, ont été utilisées pour développer des modèles de
pharmacocinétique de population, identifier les facteurs responsables des variabilités
interindividuelles et proposer un ajustement de doses à partir d’un nombre limité de
prélèvements sanguins.
Le troisième article de cette thèse a pour but de développer, à l’aide d’une analyse de
population, un estimateur Bayesien permettant d’estimer l’exposition individuelle au
66
tacrolimus (AUC0-12) sur la base d’un nombre limité de prélèvements. Cet estimateur est
caractérisé par la facilité d’usage en routine hospitalière et devrait améliorer le suivi
thérapeutique pharmacologique et l’adaptation individuelle des doses de tacrolimus chez les
enfants greffés de foie.
67
CHAPITRE II
68
Correspondence to:
Line Labbé, PhD
Faculty of Pharmacy, Université de Montréal
C.P.6128, Succursale Centre-ville
Montreal, Quebec, Canada, H3C 3J7
Telephone number: 514-343-6111 ext. 2781
Fax number: 514-343-2102
Article à resoumettre
69
Abstract
A retrospective population pharmacokinetic analysis was performed on 41 children
receiving intravenous busulfan as part of their conditioning regimen for hematopoietic stem
cell transplantation. Initial dose of busulfan was 0.8 or 1 mg/kg, in a 2 hour infusion, and
then every 6 hours, for a total of 16 doses. Pharmacokinetics was determined on the first
dose, and subsequent doses were adjusted, if needed. Data analysis was carried out by
NONMEM. Clearance (CL) and volume of distribution (V) were found to be function of
body surface area (BSA) and BSA + SEX, respectively. The interindividual variability was
decreased from 77.5% to 18.5% in CL and from 76.4% to 7.9% in V by including BSA and
SEX in the model. The final population estimates were: CL = 101 ml/min/m2 (92.6 – 109.4
ml/min/m2) and V = 21.2 l/m2 (19.4 – 22.4 l/m2). Different simulated dose regimens were
tested. The percentage of patients reaching target AUC after optimization of the first dose
was 71% versus 42% with the original administered dose. Based on this, the suggested
dosing regimen would be 25 mg/m2 if BSA ≤ 1 m2 and 32 mg/m2 if BSA > 1 m2.
Individualization of busulfan dosage is required and until our results are confirmed
therapeutic drug monitoring is mandatory to target 900 to 1500 μM.min.
1 Introduction
Busulfan is an alkylating agent used in combination with other chemotherapy agents
as a substitute to radiotherapy in preparative regimens needed for hematopoietic stem cell
transplantation (HSCT), both in adults and children (1, 2).
The most common busulfan dosing regimen in children is 0.8 or 1 mg/kg/dose for
the i.v. formulation (3, 4); and 1 mg/kg/dose for the oral formulation, every 6 hours for 16
consecutive doses (1). Busulfan has a narrow therapeutic window that ranges from 900 to
1500 μM.min (3.7 – 6.2 μg.h/ml) (5, 6). At a dosage exceeding the therapeutic levels, veno-
occlusive disease and seizures are seen (7-9). A low systemic exposure is associated with
increased risk of engraftment failure and, in certain patients, leukemia relapse.
2 Methods
Patients and dosing schedule
Forty-one consecutive patients (24 girls and 17 boys), under the age of 21 years,
undergoing HSCT at our tertiary care center, and who received i.v. busufan (Busulfex, PDL
pharma) as part of their preparative regimen, were included in this retrospective study.
Initial dose of i.v. busulfan was usually prescribed based on actual weight, as per the
following: 0.8 mg/kg/dose for 33 patients and 1 mg/kg/dose for 8 patients. Sixteen doses
were given in a 2 hour infusion every 6 hours for all patients. Pharmacokinetics was
determined on the first dose, and subsequent doses were adjusted, if needed, based on target
busulfan concentration at steady state for each patient. Therefore, 5 patients had their
subsequent doses decreased (0.3 – 0.7 mg/kg), while 20 patients needed an increase of their
doses (0.9 – 1.5 mg/kg).
Blood samples for drug monitoring were collected in heparinised tubes from the
central venous catheter using a different line from the one used for busulfan administration.
One to 3 ml were withdrawn immediately before the administration of busulfan, 5 minutes
before the end of infusion, at the end of infusion and 15, 30, 60, 120, 180, and 240 minutes
after the end of infusion. Whether dose adjustment was made or not, confirmatory steady
state level was determined from blood sampling done in each patient immediately before
72
and at the end of 6, 7, 8, and/or 9th infusions. Plasma samples obtained by centrifugation
were stored frozen until analysis the same day or day after.
where CLij is the hypothetical clearance value of the ith individual on the jth occasion, CL is
the typical value of population clearance, η iCL and κij consist of independent normally
distributed random variables with mean zero and variance ω2 and π 2, respectively. The
same modeling was used for the V.
To describe the residual unexplained variability, the additive, the proportional and
the combined (proportional and additive) error models were tested.
Once the basic pharmacokinetic model was defined, the influence of the covariates
age, sex, actual body weight (ABW) and body surface area (BSA) was studied. The BSA
was calculated using the Dubois formula (17). The generalized additive modeling (GAM)
approach in the Xpose 3.104 program (18) was used as an exploratory tool for covariate
selection. Covariates were sequentially tested into the population PK model to assess the
effect on between-subject variability of PK parameters of interest. A decrease in the
minimum objective function value of 3.84 (p ≤ 0.05) was considered significant to include
the covariate in the final model. The objective function value (OFV) is proportional to -2
log-likelihood and the difference of OFV between two models is approximately Chi Square
distributed with n degrees of freedom; n denoting the difference in the number of
parameters between the two models.
Sex and age were introduced in term of linear relationship; ABW and BSA were
introduced on the basis of allometric scaling (19, 20). Continuous covariates were centered
on the mean.
74
The area under the plasma concentration-time curve in each patient was calculated
according to the linear trapezoidal rule using S-plus (25). Using the final population
pharmacokinetic model, different dose regimens were studied to optimize the number of
patients achieving the target AUC after the first dose using the approach of Karlsson et al.
(26) and to estimate the BSA cutoffs. The target AUC range chosen was from 900 to 1500
μM.min (3.7 – 6.2 μg.h/ml) (5, 6). The different tested dose regimens were: a) 30 mg/m2; b)
25 mg/m2 if BSA ≤ 1 m2 or 32 mg/m2 if BSA > 1 m2 c) 22 mg/m2 if BSA ≤ 0.65 m2 or 28
mg/m2 if 0.65 < BSA ≤ 1.3 m2 or 35 mg/m2 if BSA > 1.3 m2. BSA cutoffs were the mean (1
m2) and the tertiles (0.65 and 1.3 m2) of BSA. The dose regimens were chosen empirically
by ranging doses from 5 to 60 mg/m2. Then we tested the best regimen with simulation
(Subproblems = 1000) incorporating parameter uncertainty in NONMEM by sampling
from the posterior parameter distributions. Including uncertainty allows for a quantitative
evaluation of the state of knowledge.
75
3 Results
Plasma concentration data of busulfan from a total of 41 children, median age 7.92
years (range 0.21-20 years), were used in this analysis. Demographics of these patients are
summarized in Table 1. Thirty three children received an initial busulfan dose of 0.8 mg/kg,
while eight patients received 1 mg/kg. The analyzed dataset consisted of 648 plasma
concentrations. Observed concentrations versus sampling time after the first dose of
busulfan are represented in Figure 1.
A one compartment model was found to be more suitable than a two compartment
model to describe our data. IIV and IOV were modeled on both CL and V, their inclusion
was very significant and lead to a large drop of the OFV and markedly improved the quality
of fit of busulfan (p < 0.001). The combined (proportional and additive) error model was
used to describe the residual unexplained variability. Therefore, the combined error model
with IIV and IOV on CL and V was selected as the optimum base model and was used to
test the different covariates.
The most important pharmacokinetic models with their corresponding OFV and IIV
on CL and V are shown in Table 2. While building the model of covariates in NONMEM,
we found that CL and V were functions of BSA. As a consequence of this, CL and V
increase with BSA (9.9% and 10.7% per 0.1 m2, respectively; p < 0.05). V was also found a
function of sex. The drop in the OFV, when introducing the covariate sex on the parameter
V, was significant but borderline (ΔOFV = 9.29). We used a statistical test, the permutation
test, with 1000 replicates to confirm this effect. This test was realized using the program
Wings for NONMEM 409 (27). Since P value was 0.005, the covariate sex was kept in the
model, indicating that boys had a volume of distribution 12% higher as compared to girls.
Parameters obtained from the final model are listed in the Table 3. The final
population estimates were: CL = 101 ml/min/m2 (92.6 – 109.4 ml/min/m2) and V = 21.2
l/m2 (19.4 – 22.4 l/m2). The performance of the final population PK model including
76
A summary of the average parameter values from the 2000 bootstrap replicates are
listed in the Table 3. The bootstrap results indicated a relatively low bias in the mean
population parameters and they were within 7% of those obtained from the final model.
Confidence intervals were also calculated from the bootstrap replicates with the percentile
method.
For the qualification of our model, we tested it to produce reliable results by using
the method of predictive check. The result was obtained by comparing the calculated AUC,
after the first dose was administered to all patients from the original data and the calculated
AUC from the simulated data. The simulated AUC’s density was centered on the median of
the original data and this proves that this model can generate data similar to the original
ones. Figure 4 shows the result of the predictive check.
Based on our final model, we tested different dose regimens. The dose regimens
were a) 30 mg/m2; b) 25 mg/m2 if BSA ≤ 1 m2 or 32 mg/m2 if BSA > 1 m2 c) 22 mg/m2 if
BSA ≤ 0.65 m2 or 28 mg/m2 if 0.65 < BSA ≤ 1.3 m2 or 35 mg/m2 if BSA > 1.3 m2 (data not
shown). Among these regimens, the 25 mg/m2 if BSA ≤ 1 m2 or 32 mg/m2 if BSA > 1 m2
was the best one to obtain the target AUC. The percentage of our patients within the target
AUC after the first dose was 42%, but it would have been 71% if busulfan would have been
administered at a dose of 25 mg/m2 if BSA ≤ 1 m2 or 32 mg/m2 if BSA > 1 m2. The
percentage of patients having their AUC in the toxic range after the first dose was 4.9% in
the former and would have been 2.4% in the latter.
77
4 Discussion
A population pharmacokinetic analysis was carried out using retrospective data
from 41 consecutive children who were given i.v. busulfan as part of their preparative
regimen for hematopoietic stem cell transplantation. The objectives of this analysis were to
characterize the relationship between pharmacokinetic parameters of i.v. busulfan and
covariates, and to explain interindividual and interoccasion variabilities. In this study, we
have evaluated a novel dosing to optimize safety and efficacy for HSCT.
Overall, individual concentrations of busulfan were well fitted with the final
population PK model with allometric scaling on CL and V as well as sex effect on V. This
was demonstrated by the LOESS curve fitted between the observed concentrations and the
predicted concentrations values, which was very close to the identity line (Figure 2) and the
results of the bootstrap (Table 3).
78
The predicted CL derived with the final population PK model was 0.1 l/min (0.09 -
0.11 l/min). This CL’s range corresponds to 2.91 – 3.56 ml/min/kg. This value was very
consistent to CL values previously reported by pediatric and adult studies (1.80 to 5.58
ml/min/kg) (29, 31-37). Likewise, the total volume of distribution of busulfan in the current
study 21.2 l (19.4 - 22.4 l) was consistent with values found in the literature for children
(12.5 – 21.3 l) (28, 30, 32, 33, 35, 37).
In the final model, IIV on CL and V were found to be 18.5% and 7.9%,
respectively. In a study conducted by Nath et al. (37), the IIV on CL (35%) and IIV on V
(22%) were higher than in our study. Moreover, in Booth’s study (28), IIV on both CL and
V were higher (23% and 11%, respectively). Therefore, our developed pharmacokinetic
model with the covariates BSA and sex reduces more efficiently IIV in pharmacokinetic
parameters.
The interoccasion variability was found to have a large impact on CL and V (13.3%
and 9.4%, respectively). Comparing our findings with those reported by Takama et al. (29)
and Nguyen et al. (30), they had just modeled IOV on clearance and it was found to be
equivalent to 6.6% and 9%, respectively. In Booth’s study (28), IOV on CL and V were
9.5% and 6.1%, respectively. Overall, results derived with the current population PK model
were very consistent with those reported by other groups including studies in adult and
pediatric patients, oral and intravenous busulfan.
First, we proved that our model is able to produce reliable results by conducting a
visual predictive check and then we studied different dose regimens. From these regimens,
we suggest an administration based on BSA that would have resulted in 71% of patients
reaching the therapeutic range after the first dose and a very low percentage of AUCs in the
79
toxic range (2.4%). This dosing regimen is 25 mg/m2 if BSA ≤ 1 m2 or 32 mg/m2 if BSA >
1 m2.
Our study had some limitations. There are some functional polymorphisms of
genes, which control important enzymes in busulfan metabolism that contribute to the
observed IIV in the pharmacokinetic of this drug. Busulfan is metabolized principally by
the GSTA1, as well as by other GST enzymes like the GSTM1 and GSTP1 (38). These
enzymes are present in the liver as well as in the intestinal cells and are up regulated in the
digestive system of young children. In our retrospective study, information about GSTA1,
GSTM1 and GSTP1 genes’ polymorphism was not available. However, it is unknown if
including it would have added more information to the model since recent studies had
reported contradictory results about the impact of polymorphisms. Johnson et al. (39) and
Gaziev et al. (36) demonstrated that carriers of GSTA1*B had reduced busulfan clearance
by 30% and 10 %, respectively. Moreover, Ansari et al. (40) found that GSTA1 and
GSTM1 polymorphisms seemed to modify busulfan pharmacokinetics. On the other hand,
Zwaveling et al. (41) observed that variability in the pharmacokinetics of busulfan was not
related to polymorphism in GST. Underlying disease has also an influence on busulfan
disposition and affects its CL and AUC (35), but it was not taken in consideration in our
model. Another limitation of the study is the limited number of patients (n= 41) and the fact
that it is a retrospective study.
To conclude, the pharmacokinetics of busulfan varies with BSA and sex. Moreover,
our model suggests that a dosing of 25 mg/m2 in children with BSA ≤ 1 m2 and 32 mg/m2 if
BSA > 1 m2 would result more often in first dosing busulfan within therapeutic range than
our current per recipient body weight (71% versus 42%, respectively). However, the
individualization of busulfan doses remains required to control risks of toxicities or
treatment failure pending prospective documentation of our findings.
80
5 Acknowledgements
We thank the staff of the HSCT, and Ms MF Vachon, nurse specialist, for their
support in the pharmacokinetic studies. N.K. received a studentship from the Faculty of
graduate studies of Université de Montréal. L.L. was the recipient of scholarship from
Health Research Foundation of Rx&D.
81
6 Conflict of Interest/Disclosure
The authors declare no conflict of interest.
82
7 References
1. Santos GW. The development of busulfan/cyclophosphamide preparative regimens.
Semin Oncol 1993; 20(4 Suppl 4): 12-6; quiz 17.
2. Santos GW, Tutschka PJ, Brookmeyer R, Saral R, Beschorner WE, Bias WB et al.
Marrow transplantation for acute nonlymphocytic leukemia after treatment with
busulfan and cyclophosphamide. N Engl J Med 1983; 309(22): 1347-53.
4. Wall DA, Chan KW, Nieder ML, Hayashi RJ, Yeager AM, Kadota R et al. Safety,
efficacy, and pharmacokinetics of intravenous busulfan in children undergoing
allogeneic hematopoietic stem cell transplantation. Pediatr Blood Cancer; 54(2):
291-8.
5. Bolinger AM, Zangwill AB, Slattery JT, Glidden D, DeSantes K, Heyn L et al. An
evaluation of engraftment, toxicity and busulfan concentration in children receiving
bone marrow transplantation for leukemia or genetic disease. Bone Marrow
Transplant 2000; 25(9): 925-30.
6. McCune JS, Gooley T, Gibbs JP, Sanders JE, Petersdorf EW, Appelbaum FR et al.
Busulfan concentration and graft rejection in pediatric patients undergoing
hematopoietic stem cell transplantation. Bone Marrow Transplant 2002; 30(3): 167-
73.
7. Tran HT, Madden T, Petropoulos D, Worth LL, Felix EA, Sprigg-Saenz HA et al.
Individualizing high-dose oral busulfan: prospective dose adjustment in a pediatric
population undergoing allogeneic stem cell transplantation for advanced
hematologic malignancies. Bone Marrow Transplant 2000; 26(5): 463-70.
83
8. Grochow LB, Jones RJ, Brundrett RB, Braine HG, Chen TL, Saral R et al.
Pharmacokinetics of busulfan: correlation with veno-occlusive disease in patients
undergoing bone marrow transplantation. Cancer Chemother Pharmacol 1989;
25(1): 55-61.
9. Dix SP, Wingard JR, Mullins RE, Jerkunica I, Davidson TG, Gilmore CE et al.
Association of busulfan area under the curve with veno-occlusive disease following
BMT. Bone Marrow Transplant 1996; 17(2): 225-30.
12. Andersson BS, Kashyap A, Gian V, Wingard JR, Fernandez H, Cagnoni PJ et al.
Conditioning therapy with intravenous busulfan and cyclophosphamide (IV BuCy2)
for hematologic malignancies prior to allogeneic stem cell transplantation: a phase
II study. Biol Blood Marrow Transplant 2002; 8(3): 145-54.
15. Boeckman A, Beal S, Sheiner L. NONMEM user's guide San Francisco: Globomax.
In, 1998.
16. Karlsson MO, Sheiner LB. The importance of modeling interoccasion variability in
population pharmacokinetic analyses. J Pharmacokinet Biopharm 1993; 21(6): 735-
50.
84
17. Du Bois D, Du Bois E. A formula to estimate the approximate surface area if height
and weight be known. Archives of Internal Medicine 1916; 17: 863-871.
19. Holford NH. A size standard for pharmacokinetics. Clin Pharmacokinet 1996;
30(5): 329-32.
20. Meibohm B, Laer S, Panetta JC, Barrett JS. Population pharmacokinetic studies in
pediatrics: issues in design and analysis. Aaps J 2005; 7(2): E475-87.
21. Mathsoft. S-PLUS 7 User’s Guide Seattle: Data Analysis Products Division
Mathsoft. In, 2005.
23. Henderson AR. The bootstrap: a technique for data-driven statistics. Using
computer-intensive analyses to explore experimental data. Clin Chim Acta 2005;
359(1-2): 1-26.
25. Proost JH. Wagner's exact Loo-Riegelman equation: the need for a criterion to
choose between the linear and logarithmic trapezoidal rule. J Pharm Sci 1985;
74(7): 793-4.
26. Jonsson S, Karlsson MO. A rational approach for selection of optimal covariate-
based dosing strategies. Clinical pharmacology and therapeutics 2003; 73(1): 7-19.
27. Holford PN. Wings for NONMEM program version 409. In.
85
28. Booth BP, Rahman A, Dagher R, Griebel D, Lennon S, Fuller D et al. Population
pharmacokinetic-based dosing of intravenous busulfan in pediatric patients. J Clin
Pharmacol 2007; 47(1): 101-11.
31. Sandstrom M, Karlsson MO, Ljungman P, Hassan Z, Jonsson EN, Nilsson C et al.
Population pharmacokinetic analysis resulting in a tool for dose individualization of
busulphan in bone marrow transplantation recipients. Bone Marrow Transplant
2001; 28(7): 657-64.
32. Schiltmeyer B, Klingebiel T, Schwab M, Murdter TE, Ritter CA, Jenke A et al.
Population pharmacokinetics of oral busulfan in children. Cancer Chemother
Pharmacol 2003; 52(3): 209-16.
36. Gaziev J, Nguyen L, Puozzo C, Mozzi AF, Casella M, Perrone Donnorso M et al.
Novel pharmacokinetic behavior of intravenous busulfan in children with
thalassemia undergoing hematopoietic stem cell transplantation: a prospective
evaluation of pharmacokinetic and pharmacodynamic profile with therapeutic drug
monitoring. Blood; 115(22): 4597-604.
37. Nath CE, Earl JW, Pati N, Stephen K, Shaw PJ. Variability in the pharmacokinetics
of intravenous busulphan given as a single daily dose to paediatric blood or marrow
transplant recipients. Br J Clin Pharmacol 2008; 66(1): 50-9.
39. Johnson L, Orchard PJ, Baker KS, Brundage R, Cao Q, Wang X et al. Glutathione
S-transferase A1 genetic variants reduce busulfan clearance in children undergoing
hematopoietic cell transplantation. J Clin Pharmacol 2008; 48(9): 1052-62.
41. Zwaveling J, Press RR, Bredius RG, van Derstraaten TR, den Hartigh J, Bartelink
IH et al. Glutathione S-transferase polymorphisms are not associated with
population pharmacokinetic parameters of busulfan in pediatric patients. Ther Drug
Monit 2008; 30(4): 504-10.
87
8 Figure Legends
Figure 1 Plasma concentrations (observed individual concentrations) of busulfan
versus time with loess of predicted concentrations versus time after the first dose.
Figure 3 Maximum a posteriori Bayes individual CL (left panel) and V (right panel)
estimates versus BSA before inclusion in the model. A smooth line “LOESS” is drawn to
show the trend in the data.
Figure 4 Predictive check results. The bold black and gray lines represent the original
simulated area under the plasma concentration time curve (AUC 0-∞) from the final model,
respectively. The median of the observed AUC 0-∞ and the individual predicted AUC0-∞ are
shown as a black dashed line and a gray dashed line, respectively.
88
9 Tables
Table 1 Population characteristics
ABW: actual body weight, BSA: body surface area, SD: standard deviation
89
BSA: body surface area, CL: total clearance, IIV: interindividual variability, IIV~CL:
interindividual variability on clearance, IIV~V: interindividual variability on volume of
distribution, IOV: interoccasion variability, OFV: objective function value produced by
NONMEM, V: volume of distribution.
90
Table 3 Parameter estimates from the final model and the bootstrap, confidence
interval and stability of the model
Mean
Typical population
population Difference 95% confidence
Parameters parameter estimate
1
parameter (%)3 interval2
(precision CV %)
estimate2
1 2
Obtained from the original data; Mean and confidence interval calculated from the
3
bootstrap; (Bootstrap mean value – typical value from original data)/bootstrap mean value
x 100
θCL, BSA
CL = θCL x (BSA/Mean)
91
θV, BSA
V = θV x (BSA/Mean) x (1- θV, SEX x SEX)
θCL and θV are the typical values of population clearance and volume of distribution,
respectively; θCL,BSA and θV,BSA are the BSA effect on CL and V, respectively; θ V,SEX is the
difference in V between boys and girls. Dichotomous variable sex was coded as 1 if girls
and 0 otherwise.
BSA: body surface area, CL: clearance, CV: coefficient of variation, ERR ADD: additive
component of the residual error, ERR PROP: proportional residual error, IIV:
interindividual variability; IOV: interoccasion variability, Mean: BSA was centered on the
mean (1 m2), V: volume of distribution.
CV was calculated as the ratio of the asymptotic standard error given by NONMEM on the
parameter estimate.
92
10 Figures
Figure 1
93
Figure 2
94
Figure 3
0.2
40
0.15
30
Volume (l)
Clearance (l/min)
0.1
20
0.05 10
0 0
Figure 4
96
CHAPITRE III
97
Nastya Kassir1,2,3, Philippe Ovetchkine 1,4, Yves Théorêt1,5,6, Catherine Litalien1,4,6, Bruce
Tapiero4, Michel Duval4, Line Labbé 2
(1) Clinical Pharmacology Unit, CHU Ste-Justine, Montreal, Quebec, Canada; (2) Faculty
of Pharmacy, Université de Montréal, Montreal, Quebec, Canada; (3) Pharsight - A Certara
Company, Montreal, Quebec, Canada; Departments of (4) Pediatrics, (5), and
Biochemistry, CHU Ste-Justine, Montreal, Quebec, Canada; (6) Department of
Pharmacology, Université de Montréal, Montreal, Quebec, Canada.
Correspondence to:
Line Labbé, PhD
Faculty of Pharmacy, Université de Montréal
C.P.6128, Succursale Centre-ville
Montreal, Quebec, Canada, H3C 3J7
Telephone number: 514-343-6111 ext. 2781
Fax number: 514-343-2102
Abstract
Background. Voriconazole pharmacokinetics (PK) is known to have highly interindividual
variability in adults. Limited pediatric PK data are available. The aim of this study was to
characterise the PK of voriconazole in children.
Results. A total of 135 voriconazole concentrations from 14 patients (aged 2-18 years)
were analyzed. A one-compartment model with linear elimination best described the data.
Population estimates of clearance and volume of distribution after intravenous and/or oral
voriconazole administration were 7.9 l/h and 84.3 l, respectively. Oral bioavailability was
estimated to be 67%. Interindividual variability on clearance, volume of distribution and
bioavailability were 84.5%, 56.5% and 78.5%, respectively. None of the tested covariates
explained the variability in PK parameters of voriconazole. Bootstrap results showed that
bias in the model was less than 17%.
1 Introduction
Invasive fungal infections have been an increasing and severe problem in
immunocompromised patients and especially in bone marrow transplant recipients. 1 2
Voriconazole is a second generation antifungal triazole with potent activity against a broad
spectrum of clinically significant fungal pathogens such as Aspergillus, Candida and
emerging and less common pathogens including Scedosporium and Fusarium.3 Clinical
efficacy of voriconazole has been well documented in immunocompromised adults with
severe invasive fungal infections. 3, 4
The drug is approved for primary treatment of invasive
aspergillosis and for salvage therapy in cases of severe infections caused by unusual
moulds. Both intravenous (IV) and oral formulations of voriconazole are available. In
adults, oral voriconazole is absorbed within two hours of intake and oral bioavailability is
excellent (>90%) allowing switching between IV and oral formulations without major dose
modifications.5 Voriconazole is extensively metabolized, primarily by hepatic CYP2C19,
CYP2C9 and CYP3A4 isoenzymes, to inactive metabolites. 3, 6
In adults and children,
contradictory results are reported about the linearity of voriconazole pharmacokinetics. 7-13
Interindividual variability in plasma concentrations of voriconazole is high regardless of the
route of administration. 5, 14, 15
2 Methods
Study data
One hundred thirty five (135) blood samples (0.5 ml) for determination of
voriconazole levels from complete pharmacokinetic profiles (following dosing
modification) were available for the analysis. Plasma concentrations were measured using
high performance liquid chromatography.
Pharmacokinetic analysis
The population PK analysis was performed using NONMEM Version 7, level 1.2
(ICON Development Solutions, Ellicott City, Maryland, USA) with the first-order
conditional estimation (FOCE) and the INTERACTION option. Nonlinear mixed-effects
models were fitted to the concentration-time data of voriconazole following multiple IV
and/or oral administrations described by typical pharmacokinetic compartmental models
(e.g. 1- or 2-compartmental models with linear or non-linear elimination). Interindividual
variability on PK parameters and the form of the residual error model were evaluated.
102
Covariate analysis was carried out using visual inspection followed by a formal
evaluation in NONMEM. The latter consisted of a stepwise forward additive approach (P ≤
0.05) followed by a backward elimination (P ≤ 0.01). Potential covariates were: body
weight, age, sex, underlying disease, and liver function tests (AST and ALT).
The performance of the final population PK model was evaluated with diagnostic
plots and shrinkage of population PK parameters.
In order to assess whether the final model could be used to estimate individual PK
parameters based on population means and sparse PK data, changes in the estimates of IIV,
residual variability, and shrinkage in interindividual random effects were evaluated. 16
Shrinkage values of ≤ 0.2 indicate good individual estimates of a parameter of interest,
while larger shrinkage values indicate that the individual Bayesian estimates “shrunk”
towards the population mean values.
In addition, the stability and precision of the model were evaluated using a
nonparametric bootstrap within Perl-Speaks-NONMEM (PsN V3.4.2).17 The bootstrap
technique involves resampling from the original data with each individual subject
considered as a sampling unit. One thousand replicates of the data were generated by
bootstrap to obtain the median and 95% percentile of PK parameters and the fixed- and
random-effect parameters. The bias of each parameter was calculated by computing the
difference between the median value derived from the bootstrap and the final parameter
estimate.
103
3 Results
A total of 14 patients were included in this study. Table 1 summarizes the
characteristics of the patients, including body weight, age, sex, and liver function tests
(ALT and AST). Patients had a wide variety of diagnoses including invasive aspergillosis
(n = 8), invasive candidiasis (n = 2), febrile neutropenia (n = 2), and other (n = 2).
Population Pharmacokinetics
A 1-compartment model with first-order absorption and elimination with lag time
adequately fit the concentration data of voriconazole. More complex elimination models
were not supported by the data. The population estimates of IIV on clearance (CL), volume
of distribution (Vc) and bioavailability (F) were 84.5%, 56.5%, and 78.5%, respectively. A
proportional error model was used for residual unexplained variability (51%).
Population parameter estimates are reported in Table 2. For the final model, typical
population CL and Vc were 7.93 l/h and 84.3 l, respectively. The absorption rate constant
was estimated to be 1.94 h−1 and the extent of oral bioavailability was 66.9%.
104
Model Evaluation
The performance of the final population PK model was evaluated with diagnostic
plots and shrinkage of population PK parameters. The overall fit of observed and predicted
voriconazole concentrations based on individual parameter estimates is presented in Figures
2A and 2B. Shrinkages to the mean of the individual random effects of CL and Vc were
acceptable (less than 28%). Conditional weighted residuals were homogeneously
distributed around 0 (Figures 2C and 2D), suggesting no bias in the prediction of
voriconazole concentrations. Bootstrap resampling strategies were used to validate the
population PK model. A total of 912 runs were successfully minimized with covariance
step. Medians of PK parameters derived with the bootstrap resampling analysis were
consistent with those derived from the original analysis with bias <17% (Table 2),
demonstrating accuracy of predictions as well as model stability.
105
4 Discussion
The present study used a population pharmacokinetic approach to describe the PK
of intravenous and oral voriconazole and to identify potential sources of variability in
immunocompromised pediatric patients (2 to 18 years of age).
The estimated PK parameter values using the final model are slightly different than
those reported in literature for children. A typical pediatric patient would have clearance
and volume of distribution of 7.9 l/h and 84.3 l, respectively. While Karlsson et al. 12
reported that clearance and total volume of distribution were 13.3 l/h and 67.9 l,
respectively, Walsh et al.13 estimated them to 9.3 l/h and 58.5 l, respectively.
The stability and precision of the population PK model were evaluated using a
nonparametric bootstrap. The results of this analysis demonstrated the model stability
(91.2% successful convergence) as well as the accuracy of the estimated parameters (bias
<17%).
The current study has some limitations. The major ones were the limited number of
patients (n= 14) and the fact that it is a retrospective study. Moreover, the group of patients
had a wide variety of diagnoses adding another level of variability to the study.
5 References
1. Hovi, L., et al., Invasive fungal infections in pediatric bone marrow transplant
recipients: single center experience of 10 years. Bone Marrow Transplant, 2000. 26(9): p.
999-1004.
6. Johnson, L.B. and C.A. Kauffman, Voriconazole: a new triazole antifungal agent.
Clin Infect Dis, 2003. 36(5): p. 630-7.
9. Lutsar, I., et al., Safety of voriconazole and dose individualization. Clin Infect Dis,
2003. 36(8): p. 1087-8.
10. Han, K., et al., Population pharmacokinetic evaluation with external validation and
Bayesian estimator of voriconazole in liver transplant recipients. Clin Pharmacokinet, 2011.
50(3): p. 201-14.
109
12. Karlsson, M.O., I. Lutsar, and P.A. Milligan, Population pharmacokinetic analysis
of voriconazole plasma concentration data from pediatric studies. Antimicrob Agents
Chemother, 2009. 53(3): p. 935-44.
14. Smith, J., et al., Voriconazole therapeutic drug monitoring. Antimicrob Agents
Chemother, 2006. 50(4): p. 1570-2.
15. Pascual, A., et al., Voriconazole therapeutic drug monitoring in patients with
invasive mycoses improves efficacy and safety outcomes. Clin Infect Dis, 2008. 46(2): p.
201-11.
16. Karlsson, M.O. and R.M. Savic, Diagnosing model diagnostics. Clin Pharmacol
Ther, 2007. 82(1): p. 17-20.
17. Henderson, A.R., The bootstrap: a technique for data-driven statistics. Using
computer-intensive analyses to explore experimental data. Clin Chim Acta, 2005. 359(1-2):
p. 1-26.
18. Desta, Z., et al., Clinical significance of the cytochrome P450 2C19 genetic
polymorphism. Clin Pharmacokinet, 2002. 41(12): p. 913-58.
20. Hyland, R., B.C. Jones, and D.A. Smith, Identification of the cytochrome P450
enzymes involved in the N-oxidation of voriconazole. Drug Metab Dispos, 2003. 31(5): p.
540-7.
110
6 Figure Legends
Figure 1 Plasma concentrations (observed individual concentrations) of intravenous
and oral voriconazole versus time after the last dose with LOESS. LOESS is a locally
weighted scatterplot smoothing line.
7 Tables
Table 1: Patient Characteristics at Baseline
Sex
Female 3 (21.4)
Male 11 (78.6)
Underlying Disease
Other 2 (14.3)
Drug Formulation
Intravenous 7 (50.0)
Oral 7 (50.0)
Table 2. Parameter Estimates for the Final Model with Bootstrap Validation
Parameter Bootstrapb
Parameter Estimates
Median 95% CI Bias (%)
(RSE%)a
Pharmacokinetic Parameter
CL (l/h) 7.9 (32.4) 8.0 4.3, 17.4 1.1
Vc (l) 84.3 (36.7) 78.1 44.1, 198 -7.4
Ka (1/h) 1.9 (28.9) 2.2 1.1, 6.6 11.2
Tlag (h) 0.243 (4.6) 0.248 0.221, 0.499 2.2
F 0.669 (45.1) 0.664 0.289, 1.9 -0.75
Between-Subject Variability (BSV)
BSV CL (%) 84.5 (19.2) 80.2 46.9, 126.5 -9.9
BSV Vc (%) 56.5 (25.9) 41.1 14.1, 87.1 2.4
BSV F (%) 78.5 (14.7) 73.0 32.5, 99.3 16.2
Residual Variability
Residual
Proportional 51.0 (10) 48.1 23.0, 79.4 -5.9
Error (%)
CL= systemic clearance; F= bioavailability; Ka= absorption rate; Tlag= lag time; Vc=
apparent central volume of distribution.
a
Relative standard error calculated as the standard error of parameter estimate/parameter
estimate x 100%.
b
Median of 912 successful bootstrap samples from the 1000 runs with prediction intervals
(CI) calculated as the 2.5th and 97.5th percentiles.
114
8 Figures
Figure 1
115
Figure 2
116
CHAPITRE IV
117
(1) Clinical Pharmacology Unit, CHU Ste-Justine, Montreal, Quebec, Canada; (2) Faculty
of Pharmacy, Université de Montréal, Montreal, Quebec, Canada; (3) Pharsight - A Certara
Company, Montreal, Quebec, Canada; Departments of (4) Surgery, (5) Pediatrics, and (6)
Biochemistry, CHU Ste-Justine, Montreal, Quebec, Canada; (7) Department of
Pharmacology, Université de Montréal, Montreal, Quebec, Canada.
Corresponding author:
Dr. Catherine Litalien
Division of Pediatric Critical Care, Department of Pediatrics
Centre Hospitalier Universitaire Sainte-Justine
3175 chemin de la Côte Sainte-Catherine
Montréal (Québec), Canada H3T 1C5
Tel: (514) 345-4931 extension 2571
FAX: (514) 345-7731
Abstract
Aims: Objectives of this study were to develop a population pharmacokinetic (PopPK)
model for tacrolimus in pediatric liver transplant patients and determine optimal sampling
strategies (OSS).
Methods: Twelve-hour intensive pharmacokinetic profiles from 30 patients (age 0.4 – 18.4
years) receiving tacrolimus orally were analyzed. The PopPK model explored the
following covariates: weight, age, sex, type of transplant, liver function tests, albumin,
hematocrit, drug interactions, and time post-transplantation. OSS were developed, and
validated with jackknife.
Results: A 2-compartment model with first-order absorption and elimination and lag time
described the data. Weight was included on all pharmacokinetic (PK) parameters. Typical
apparent clearance and central volume of distribution were 12.1 l/h and 31.3 l, respectively.
The PopPK approach led to the development of an OSS, which allowed estimation of
tacrolimus PK and area under the concentration-time curve (AUC) on the basis of a C0–
C0.5h–C2h sampling schedule with ± 20% prediction error limit. The mean bias and precision
of the Bayesian versus reference (trapezoidal) AUCs were -2.6% and 12.5%, respectively.
1 Introduction
Tacrolimus (Prograf®, Fujisawa Healthcare Inc.) is a first-line immunosuppressive
agent widely used in pediatric and adult solid organ transplant recipients. Since it has a
narrow therapeutic index and significant inter- and intraindividual pharmacokinetic (PK)
variability, dosing individualization is essential to ensure graft survival and limit associated
toxicities including life-threatening complications.[1]
In this context, therapeutic drug monitoring has become a standard of care for
tacrolimus dosing optimization. Trough concentration (Ctrough) is commonly used to guide
tacrolimus dose individualization despite its inadequacy in reflecting total drug exposure as
was demonstrated by numerous studies involving solid organ transplant recipients.[4-6]
Although the relationship between tacrolimus exposure and patient outcome has not been
precisely defined, the last consensus report on twice-daily tacrolimus concluded that there
was an urgent need to evaluate alternative strategies to C trough such as area under the
concentration-time curve (AUC(0-12)), which is widely considered as the best marker for
drug exposure.[1] However AUC-based monitoring implies the measurement of multiple
concentration-time points over the entire dosing interval and is time-consuming, expensive,
and often impractical for routine clinical practice, especially in the pediatric population.[7]
Alternatively, AUC can be predicted using optimal sampling strategy (OSS).
120
The OSS using maximum a posteriori Bayesian estimators (MAP-BE) can predict
individual PK parameters using a limited number of concentration-time points with
flexibility in sampling time and consideration of patient characteristics. In addition, the
MAP-BE approach can be made simple to use by clinicians with appropriate computer
technology.[8] At present, OSS using MAP-BE has been developed for tacrolimus in adult
liver and kidney transplant recipients but not in pediatrics.
Therefore, the aims of this study were to develop: 1) a PopPK model for tacrolimus
in pediatric liver transplant patients using rich sampling; and 2) an OSS using MAP -BE
that accurately predicts tacrolimus PK parameters and AUC (0-12).
121
2 Methods
Patients and Study Design
Medical charts were reviewed and the following data collected: body weight,
height, age, sex, type of transplant (full or cut-down liver), age of liver donor, underlying
diagnosis, liver function tests (ALT, AST, GGT, albumin, and total bilirubin), serum
creatinine, hematocrit, time post-transplantation, drug formulation, use of steroids, and
presence of clinically relevant CYP3A4 inhibitors.
The PopPK analysis was performed using NONMEM Version 7, level 1.2 (ICON
Development Solutions, Ellicott City, Maryland, USA) with the first-order conditional
estimation (FOCE) and the INTERACTION option. Nonlinear mixed-effects models were
used to fit the concentration-time data of tacrolimus described by typical PK compartmental
models (e.g. 1- or 2-compartmental models, first-order absorption, with linear or non-linear
elimination). For a 2-compartment model, PK parameters were apparent oral clearance
(CL/F), apparent volume of distribution (V 1/F), Intercompartmental clearance (Q2/F),
apparent peripheral volume (V 2/F), and first-order absorption rate constant (ka). Between
subject variability (BSV) and between occasion variability (BOV) in PK parameters were
modeled as exponential random effect models in order to positively constrain the individual
parameter values, which were thus assumed to follow a log-normal distribution. Covariance
between parameters was also examined. Additive, proportional and combined error
structures were tested during modeling of residual random error. Model evaluation and
selection were based on pertinent graphical representations of goodness of fit and on the
minimization of -2 × Log (Likelihood), which was presented as the objective function value
(OFV). A decrease in the OFV of 3.84 (P=0.05) was considered significant to include a
parameter in the model. Allometric scaling was applied to the base model [15, 16] that is:
ka= θ3*(WT/WTmedian)-0.25
123
where θ is the typical value of the parameter in a child with median weight (WT) of the
study population.
Covariate analysis was carried out using visual inspection followed by a formal
evaluation in NONMEM. The latter consisted of a stepwise forward additive approach
(P=0.05) followed by backward elimination (P=0.01). Potential covariates that were
evaluated were: age, sex, type of transplant (full or cut-down liver), age of liver donor, time
post-transplantation, liver function tests (ALT, AST, GGT, and total bilirubin), albumin,
renal function (serum creatinine and creatinine clearance), hematocrit, use of steroids,
presence of clinically relevant CYP3A4 inhibitors, and drug formulation. Some of the
continuous covariates were also tested as categorical variables (Table 2).
Model Evaluation
The performance of the final PopPK model was evaluated with diagnostic plots and
shrinkage of PopPK parameters. In order to assess whether the final model could be used to
estimate individual PK parameters based on population means and sparse PK data, changes
in the estimates of BSV, residual variability, and shrinkage in individual random effects
were evaluated.[17] Shrinkage values of ≤20% indicate good individual estimates of a
parameter of interest, while larger shrinkage values show that individual Bayesian estimates
“shrunk” towards the population mean values.
In addition, the stability and precision of the model were evaluated using a
nonparametric bootstrap within Perl-Speaks-NONMEM (PsN V3.4.2).[18] The bootstrap
technique involves resampling from the original data with each individual subject
considered as a sampling unit. One thousand replicates of the data were generated by
bootstrap to obtain the median and 95% percentile of PK parameters and the fixed- and
random-effect parameters. The bias of each parameter was calculated by computing the
124
difference between median value derived from the bootstrap and the final parameter
estimate.
Using the final model, the initial time points for the OSS were obtained with the
WinPOPT® (Version 1.2) software. Limited sampling strategies among combinations of a
maximum of four sampling time points, including C trough, and up to 4 hours post-drug
administration were tested. The selection of the best strategy used the determinant of the
Fisher information matrix as a measure of the informativeness of the design. In addition,
the best 3 combinations of sampling time points previously identified (C trough-C1-C4, Ctrough-
C0.5-C2-C4, Ctrough-C1-C2-C4) using MRA in a subset of the actual data were also selected
for further evaluation.[19]
ͳ Ǧ
ሺΨሻൌͳͲͲȗ ൬ ൰
ͳ Ǧ ʹ
ሺΨሻ ൌͳͲͲȗ ቌඨ ൬ ൰ ቍ
where Pred is the Bayesian AUC (0-12) estimate, Obs is the reference AUC (0-12) value
obtained using the linear trapezoidal method applied to the 12-hour intensive PK profiles,
and n is the number of patients.
125
3 Results
Thirty-eight 12-hour intensive PK profiles obtained from 30 liver transplant
recipients were available for this study. Patients’ characteristics are summarized in Tables 2
and 3. Patients had a wide variety of diagnoses leading to whole or cut-down liver
transplantation, including biliary atresia (n = 12), tyrosinemia (n = 8), North American
Indian childhood cirrhosis (n = 3), fulminant hepatitis (n = 2), Alagille syndrome (n = 2),
histiocytosis X (n = 1), sclerosing cholangitis (n = 1), and auto-immune hepatitis (n = 1).
A total of 341 plasma samples were available for the population PK analysis. A
median of nine concentration-time points were obtained for each patient (range: 8 - 10).
Observed concentration-time profiles of tacrolimus are plotted in Figure 1.
Population Pharmacokinetics
where CL/F is apparent oral clearance, V 1/F is apparent volume of distribution, Q 2/F is
intercompartmental clearance, V 2/F is apparent peripheral volume of distribution, ka is first-
order absorption rate constant, tlag is lag time, and WT is body weight.
Population parameter estimates along with between subject and residual variabilities
are reported in Table 4. Typical population CL/F and V1/F for a child weighing 20 kg were
12.1 l/h and 31.3 l, respectively.
Model Evaluation
The performance of the final population PK model was evaluated with diagnostic
plots and shrinkage of population PK parameters. The overall fit of observed and predicted
tacrolimus concentrations based on population and individual parameter estimates are
presented in Figures 2A and 2B, respectively. Shrinkages to the mean of the individual
random effects of CL/F and V1/F were low (less than 12%). Conditional weighted residuals
were homogeneously distributed around 0 (Figures 2C and 2D), suggesting no bias in the
prediction of tacrolimus concentrations. Bootstrap resampling strategies were used to
validate the PopPK model. A total of 939 runs were successfully minimized with
covariance step. Medians of PK parameters derived with the bootstrap resampling analysis
were consistent with those derived from the original analysis with bias <9% (Table 4),
demonstrating accuracy of predictions as well as model stability.
127
The PopPK parameters obtained from the final model were used as priors for the
development of the MAP-BE. The OSS schedule, selected using the Fisher information
matrix, was 0, 0.5, and 2h post-dose with relative standard errors of 5.8% and 22.6% on
CL/F and V1/F, respectively. AUC(0-12) bias and precision obtained during the validation
procedure were -2.57% and 12.49%, respectively (Table 5). Typical fits for the estimation
of reference parameters (considering all concentrations) and Bayesian estimation (with
sampling times at 0, 0.5, and 2h post-dose) are illustrated in Figure 3. Moreover, the
predictive performance of the MAP-BEs using multiple regression analysis (MRA) derived
combinations of sampling times are reported in Table 5. Their relative bias and precision
were below ±10%. AUC estimation with these sampling times seemed to be more accurate
than the schedule identified with WINPOPT, however the MRA-derived sampling times
only targeted CL/F prediction while those identified by WINPOPT optimized for all PK
parameters.
4 Discussion
To the best of our knowledge, this study is the first maximum a posteriori Bayesian
estimator for the prediction of tacrolimus exposure in pediatric liver transplant recipients.
A 2-compartment model with first-order absorption and elimination best fitted the
concentration-time profiles of tacrolimus at steady state in the study population. This is in
contrast with other PopPK studies in pediatric liver transplant recipients where tacrolimus
PK followed a 1-compartment model. The use of trough concentrations in these studies
precluded the development of multi-compartment models whereas in the current analysis,
intensive PK profiles were available.[9-13] After inclusion of weight in the base model,
none of the tested covariates significantly described the variability in PK parameters. This
model allowed a precise estimation of tacrolimus clearance with a relative standard error of
10%.
Growth and development usually described by demographic factors such as size and
age, respectively, contribute most of the variance to drug PK through the pediatric age
span. In the present study, body weight was included on all PK parameters as an allometric
fixed term to account for the effect of growth.[16, 21] The impact of development on
tacrolimus PK in pediatric liver transplant recipients is complex to evaluate as its total
clearance is influenced by both recipient’s and donor’s characteristics. These include
among others: 1) recipient’s (i.e., patient’s) intestinal metabolism and transporter activity
maturation and CYP3A5 genotype; 2) donor’s hepatic metabolism which depends at least
on the donor’s age and CYP3A5 genotype; and 3) the ratio of liver transplant size to
recipient’s liver volume or recipient’s body weight. Also, the influence of recipient’s age
on the metabolic function of the transplant liver is currently unknown. In addition, the
maturation of intestinal and hepatic CYP3A is expected to be essentially completed by the
end of the first year of life and considering that only 4 patients less than 1 year were
included, it was not possible to study the developmental effects using the sigmoid E max
model (e.g., Hill equation) with postmenstrual age (PMA) which allows a description of the
gradual maturation of clearance during the first year of life. As such, the impact of hepatic
129
Donor’s age and transplant type were also evaluated, as they may influence
tacrolimus PK in pediatric liver transplant recipients. In the present study, both covariates
were not found to be significant. In the contrary, Staatz et al. [10] found that children who
received cut-down liver from an adult exhibited an average 7-fold lower tacrolimus
clearance compared to those who received a whole liver from a child donor. The authors
postulated that the transplant organ retains the metabolic characteristics of the donor and
that an adult donor liver has lower drug clearance than a child donor liver. If age of the
donor had been included in the model by Staatz et al., type of transplant may have not been
identified as a significant covariate.
The concomitant use of steroids has the potential to increase tacrolimus elimination
by inducing CYP3A4.[22] In this sense, the results reported in the literature are
contradictory. While some authors found that steroids may induce CYP3A4, other studies
have suggested decreased tacrolimus metabolism. The effect of concomitant steroids
administration could not be demonstrated in this study which is in agreement with the
130
findings from other pediatric studies.[12, 23] Hematocrit and albumin concentrations were
also tested as potential covariates as tacrolimus accumulates in erythrocytes and is highly
bound to plasma proteins. As such, low hematocrit and albumin concentrations are
expected to result in a reduction in total drug concentration in whole blood and an increase
in total clearance. Similar to Garcia Sanchez et al., hematocrit and albumin were not found
to be significant covariates. This is in opposition to results found by Zhao et al. where
clearance of tacrolimus was significantly higher in patients with low levels of hematocrit (<
33%). Albumin was not tested in the latter study.
The estimated PK parameter values using the final model are close to those
previously reported in pediatric liver transplant recipients.[9, 10, 12, 13] A typical pediatric
patient (WT = 20 kg) would have CL/F and total volume of distribution (V d/F), which is the
sum of V1/F and V2/F, of 12.1 l/h and 321.3 l, respectively. These values are similar to
those previously reported for CL/F (4.6 – 13.1 l/h) [9, 12] and Vd/F (196.1 – 617 l).[9, 10,
13].
Between subject variabilities were 55.6% and 126.1% for CL/F and V 1/F,
respectively, and were in the range of previously reported variabilities for CL (24.3 – 110
%) and V1/F (27.4 – 297 %).[9-14] Several reasons could potentially explain the remaining
BSV. It is now well recognized that CYP3A5 polymorphism plays an important role in the
absorption and metabolism of tacrolimus.[9, 24] Unfortunately, the impact of this genetic
variation could not be tested as CYP3A5 genotypes of neither donors nor recipients were
available. Moreover, BSV on CL/F could be attributed to the extremely variable tacrolimus
bioavailability (3 to 77%).[2] Furthermore, half of the patients received the drug as an
extemporaneously compounded oral suspension prepared by CHU Ste-Justine hospital [25]
or community drugstores (possibly using a different reconditioning method), for which
bioavailability is unknown.
As recommended by the U.S. Food and Drug Administration [26], the stability and
precision of the PopPK model were evaluated using a nonparametric bootstrap. The results
131
of this analysis demonstrated the model stability (93.9% successful convergence) as well as
the accuracy of the estimated parameters (bias <9%).
Since the best marker for drug exposure is AUC, which is related to clearance,
different OSS were evaluated to optimize the Bayesian estimation of individual CL/F. In
this context, the sampling combinations previously reported based on MRA (C trough-C1-C4,
Ctrough-C0.5-C2-C4, Ctrough-C1-C2-C4) for the estimation of tacrolimus AUC in pediatric liver
transplant recipients were tested with MAP-BE.[19] The limited sampling strategy using
MAP-BE based on predose, 1h and 4h post-dose estimated tacrolimus AUC with ±20%
prediction error limit. The mean bias and precision (95% confidence interval) of this design
were -1.93% (-14.90 to 15.34 %) and 8.44% (0.60 to 17.59 %), respectively, which is
clinically acceptable predictive performance. Inclusion of a fourth sample resulted in
further improvement of both bias and precision but impairs practical application.
The comparison of the predictive performance of both MAP-BE and MRA for the
same sampling times shows that both approaches can accurately estimate tacrolimus
exposure in pediatric liver transplant recipients. However, even though precision was
132
similar, MRA had the tendency to underestimate AUC (0-12) with a 95% confidence interval
for bias excluding zero.
Although MRA results in a simple equation, it can only be used for the estimation
of a single PK parameter with a near exactly similar dosing schedule and requires precise
sampling times. In contrast, MAP-BE allows the description of the PK profile and the
estimation of PK parameters, considering patients characteristics, with flexibility of blood
sampling times and number making this method compatible with real-life situations.[29]
Moreover, Bayesian predictions can improve as more patient-specific data is added to the
population model.
The optimal sampling strategies developed in this study were based on data from
pediatric liver transplant recipients and as such can only be applied to the same population.
Transferability of these models to other transplant types (kidney, heart…) could be
envisioned following proper validation in these groups. Furthermore, these models should
not be applied for the estimation of tacrolimus exposure in young infants (< 1 year), as
there were only 4 patients in this age group among the study population, and in early post-
transplantation, as no included patients were studied during the first two weeks after
transplantation. In conclusion, PopPK of tacrolimus and empirical Bayesian estimates
represent an accurate and convenient method to predict tacrolimus AUC (0-12) in pediatric
liver transplant recipients, despite high between subject variability in PK and patient
demographics. The developed OSS will allow the undertaking of prospective trials to
define tacrolimus target AUCs and dosing guidelines in this population and to evaluate the
efficacy of alternative strategies to Ctrough-based monitoring in order to prevent graft
rejection while minimizing toxicity events.
133
5 Acknowledgements
Financial support came from the Research Centre of Centre Hospitalier
Universitaire Ste-Justine. L. Labbé was the recipient of an Rx&D/CIHR Health Research
Foundation Career Award in health sciences. We thank Carole Viau for her technical
assistance. We also thank the nursing staff of the Multispeciality Clinic and medical
technologists at CHU Ste-Justine for their help with sample collection and analysis,
respectively.
134
6 References
1. Wallemacq P, Armstrong VW, Brunet M, Haufroid V, Holt DW, Johnston A,
Kuypers D, Le Meur Y, Marquet P, Oellerich M, Thervet E, Toenshoff B, Undre N, Weber
LT, Westley IS, Mourad M. Opportunities to optimize tacrolimus therapy in solid organ
transplantation: report of the European consensus conference. Ther Drug Monit.
2009;31:139-52.
10. Staatz CE, Taylor PJ, Lynch SV, Willis C, Charles BG, Tett SE. Population
pharmacokinetics of tacrolimus in children who receive cut-down or full liver transplants.
Transplantation. 2001;72:1056-61.
11. Sam WJ, Aw M, Quak SH, Lim SM, Charles BG, Chan SY, Ho PC. Population
pharmacokinetics of tacrolimus in Asian paediatric liver transplant patients. Br J Clin
Pharmacol. 2000;50:531-41.
14. Wallin JE, Bergstrand M, Wilczek HE, Nydert PS, Karlsson MO, Staatz CE.
Population pharmacokinetics of tacrolimus in pediatric liver transplantation: early
posttransplantation clearance. Ther Drug Monit. 2011;33:663-72.
17. Karlsson MO, Savic RM. Diagnosing model diagnostics. Clin Pharmacol Ther.
2007;82:17-20.
136
18. Henderson AR. The bootstrap: a technique for data-driven statistics. Using
computer-intensive analyses to explore experimental data. Clin Chim Acta. 2005;359:1-26.
19. Delaloye JR, Kassir N, Lapeyraque AL, Alvarez F, Lallier M, Beaunoyer M, Labbe
L, Theoret Y, Litalien C. Limited sampling strategies for monitoring tacrolimus in pediatric
liver transplant recipients. Ther Drug Monit. 2011;33:380-6.
20. Sheiner LB, Beal SL. Some suggestions for measuring predictive performance. J
Pharmacokinet Biopharm. 1981;9:503-12.
22. Kim JS, Aviles DH, Silverstein DM, Leblanc PL, Matti Vehaskari V. Effect of age,
ethnicity, and glucocorticoid use on tacrolimus pharmacokinetics in pediatric renal
transplant patients. Pediatr Transplant. 2005;9:162-9.
24. Staatz CE, Goodman LK, Tett SE. Effect of CYP3A and ABCB1 single nucleotide
polymorphisms on the pharmacokinetics and pharmacodynamics of calcineurin inhibitors:
Part I. Clin Pharmacokinet. 2010;49:141-75.
25. Jacobson PA, Johnson CE, West NJ, Foster JA. Stability of tacrolimus in an
extemporaneously compounded oral liquid. Am J Health Syst Pharm. 1997;54:178-80.
26. U.S. Department of Health Services, Food and drug Administration. Guidance for
Industry Population Pharmacokinetics. February 1999 [updated 31 january 2012];
Available from:
http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidan
ces/UCM072137.pdf.
137
27. David OJ, Johnston A. Limited sampling strategies for estimating cyclosporin area
under the concentration-time curve: review of current algorithms. Ther Drug Monit.
2001;23:100-14.
28. Hoyer PF. Therapeutic drug monitoring of cyclosporin A: should we use the area
under the concentration-time curve and forget trough levels? Pediatr Transplant. 2000;4:2-
5.
29. van der Meer AF, Marcus MA, Touw DJ, Proost JH, Neef C. Optimal sampling
strategy development methodology using maximum a posteriori Bayesian estimation. Ther
Drug Monit. 2011;33:133-46.
138
7 Figure Captions
Figure 1. Whole blood tacrolimus concentrations versus time after last dose in pediatric
liver transplant recipients. LOESS is a locally weighted scatterplot smoothing line.
Figure 3. Simulation of tacrolimus concentrations for typical subjects with low bias (3A,
3B) or high bias (3C, 3D). Full circles represent observed concentrations, full lines
represent prediction using all available concentrations, and dashed lines represent
prediction with 3 sampling times (Ctrough-C0.5-C2) and MAP-BE.
139
8 Tables
Table 1. Mean Population Pharmacokinetic Parameters for Tacrolimus in Pediatric Liver
Transplant Recipients
BSV CL BSV V
Ref n Pharmacokinetic parameters
(%) (%)
a 0.75 –0.00032 T (–0.057 BILI
(12) 18 CL/F= 10.4*(WT/70) *e( )*e )*(1–0.079*ALT) (l/h) 24.3 NR
V= 39.1*[1+4.57*(BSA–0.49)] (l)
(11) a 20 33.5 33.0
BILI <200 μmol/l: F= 0.197*(1+0.0887*WT–11.4) (%)
F= 19% (fixed)
CL/F= (0.134*1.8iFLAG+0.0181*2hFLAG*XPOD)*8.6*(WT/8.6)0.341*e(-0.0358*AST/53)
ALT = alanine aminotransferase; AST = aspartate aminotransferase; BILI = bilirubin concentration; BSA =
body surface area; BSV = between-subject variability; CL = clearance; CL/F = apparent oral clearance; F =
bioavailability; HF = hepatic function; n = number of patients; NR = not reported; POD = postoperative day;
T = time after initiation of treatment; V = volume of distribution; WT = body weight.
a
Tacrolimus whole blood concentrations were determined using the microparticle enzyme immunoassay
(MEIA) by use of an IMx analyzer
b
Tacrolimus blood concentrations were measured using a validated high-performance liquid chromatography
tandem mass–spectrometry assay specific for the parent drug
c
Tacrolimus blood concentrations were measured using the EMIT 2000 assay (Siemens Healthcare) and a
validated high-performance liquid chromatography tandem mass –spectrometry assay specific for the parent
drug
140
d
if POD was <21, then XPOD = POD; otherwise, XPOD = 21; if the donor was a CYP3A5 *1 allele carrier,
then hFLAG = 1; otherwise, 0; and if the intestinal MDR1 mRNA level was >0.22 amol/μg total RNA, then
iFLAG = 1; otherwise, 0.
141
Table 4. Parameter Estimates for the Final Model with Bootstrap Validation
Bootstrapb
Parameter Estimates
Parameter Bias
(RSE%)a Median 95% CI
(%)
Pharmacokinetic Parameterc
CL/F (l/h) 12.1 (10.1) 12.08 10.1, 14.9 -0.17
V1/F (l) 31.3 (42.8) 30.87 12.3, 70.3 -1.37
Q2/F (l/h) 30.7 (29.3) 28.9 13.1, 53.1 -5.86
V2/F (l) 290 Fix
Ka (1/h) 0.342 (33.3) 0.342 0.142, 0.656 0.00
Tlag (h) 0.433 (4.2) 0.4325 0.383, 0.456 -0.12
Between-Subject Variability (BSV)
BSV CL/F (%) 55.6 (9.6) 54.39 43.82, 65.12 -4.27
BSV V1/F (%) 126.1 (18) 120.50 69.50, 165.23 -8.68
BSV Q2/F (%) 84.0 (21.3) 81.0 44.61, 141.77 -7.07
Residual Variability
Residual Proportional
20.3 (12.1) 20.21 15.3, 24.9 -0.44
Error (%)
CL/F= apparent oral clearance; F= bioavailability; Ka= absorption rate; Q 2/F= apparent
distribution clearance; Tlag= lag time; V 1/F= apparent central volume of distribution;
V2/F= apparent peripheral volume of distribution.
a
Relative standard error calculated as the standard error of parameter estimate/parameter
estimate x 100%.
b
Median of 939 successful bootstrap samples from the 1000 runs with prediction intervals
(CI) calculated as the 2.5th and 97.5th percentiles.
c
Body weight was included in all pharmacokinetic parameters as an allometric fixed term
d
V2/F was fixed to a value estimated from a previous run in order to stabilize the model
144
MAP-BE a MRAb
Optimal Sampling Times (h)
ME% (95% CI) RMSE% (95% CI) ME% (95% CI) RMSE% (95% CI)
C 0-C 0.5-C 2 -2.57 (-30.94 , 17.43) 12.49 (0.71 , 31.16) -- --
C 0-C 1-C 4 -1.93 (-14.90 , 15.34) 8.44 (0.60 , 17.59) -4.98 (-8.37 , -1.59) 8.29 (3.29 , 11.28)
C 0-C 0.5-C 2-C 4 -1.93 (-14.66 , 10.94) 7.37 (1.02 , 15.85) -6.15 (-9.14 , -3.16) 8.48 (4.25 , 11.22)
C 0-C 1-C 2-C 4 -2.78 (-14.38 , 10.91) 7.42 (0.14 , 15.04) -5.07 (-8.34 , -1.81) 8.15 (3.22 , 11.07)
CI= 95% confidence interval; MAP-BE= maximum a posteriori Bayesian estimators; ME%= relative mean
prediction error; MRA= multiple regression analysis; RMSE%= relative root mean squared prediction error.
a
Validation in the present study was carried out by the jackknife technique
b
Validation in the Delaloye et al. study was carried out with an independent set of patients (validation group)
(14)
145
9 Figures
Figure 1
146
Figure 2
147
Figure 3
(
(
A B
s
s
n
n
t i o
t i o
5
5
2
2
t r a
t r a
n
n
c e
c e
0
0
n
n
2
2
o
o
C
C
d
d
5
5
r v e
r v e
1
1
s e
s e
b
b
0
0
O
O
1
1
0 2 4 6 8 10 12 0 2 4 6 8 10 12
T im e A f t e r L a s t D o s e ( h ) T im e A f t e r L a s t D o s e ( h )
l )
l )
/ m
/ m
g
g
( n
( n
C D
s
s
n
n
t i o
t i o
5
2
t r a
t r a
5
2
n
n
c e
c e
0
n
2
0
o
o
2
C
C
d
5
r v e
r v e
5
1
1
s e
s e
b
0
O
O
0
1
1
0 2 4 6 8 10 12 0 2 4 6 8 10 12
T im e A f t e r L a s t D o s e ( h ) T im e A f t e r L a s t D o s e ( h )
148
CHAPITRE V
149
DISCUSSION ET CONCLUSION
Les enfants ont toujours été des « orphelins thérapeutiques ». De nombreux médicaments
n’ont pas fait l’objet d’étude chez les enfants, leur utilisation n’est pas officiellement
autorisée, ce qui crée un déficit lorsque le besoin s’y présente chez cette population.
Plusieurs catastrophes au sein du groupe d’âge pédiatrique en témoignent.[71-73, 197] Les
recommandations posologiques pour la plupart des médicaments utilisés en pédiatrie sont
généralement établies en milligramme/kilogramme de poids corporel jusqu'à une dose
adulte maximale. Bien que la posologie établie en milligramme/mètre carré de surface
corporelle puisse être préférable, l'expérience clinique indique que les erreurs de mesure de
la taille (notamment chez les petits enfants et les nourrissons) et les erreurs de calcul de la
surface corporelle à partir du poids et de la taille sont courantes. L’adaptation de posologie
en pédiatrie à partir des adultes ne prend pas en compte les particularités pharmacologiques
qui caractérisent l’enfant tout au long de son développement. L’extrapolation de
l’utilisation des médicaments à partir des données chez l’adulte a provoqué des morbidités
significatives qui auraient pu être minimisées en effectuant la recherche chez les
enfants.[71, 198, 199] Cependant, l’évolution considérable de la science et de la
réglementation sur la recherche des médicaments en pédiatrie[200] et dans la modélisation
pharmacocinétique[201] ont permis d’améliorer la pharmacothérapie chez les enfants.
Les travaux effectués dans le cadre de cette thèse avaient pour objectif général
d’appliquer la pharmacocinétique de population en pédiatrie pour explorer les
caractéristiques propres à cette population, de décrire la variabilité pharmacocinétique des
médicaments utilisés chez les enfants immunosupprimés, en vue de l’individualisation du
traitement en termes de choix de régime posologique. Conjointement à ces études, des
outils permettant d’estimer l’exposition individuelle en limitant le nombre de prélèvements
et optimiser les temps de ces derniers ont été développés.
l’utilisation de formules incluant la surface corporelle peut être considérée difficile pour les
professionnels de la santé malgré la signification physiologique de ce paramètre et son
importance à expliquer la variabilité dans la pharmacocinétique du busulfan. Un point à
souligner dans les résultats de cet article est l’utilisation de doses plus élevées par BSA
chez les enfants que chez les adultes. Ces résultats confirment que l’extrapolation par
kilogramme ou par mètre carré à partir des doses adultes ne fonctionne pas et qu’il faut
déterminer des doses spécifiques pour les enfants.
Dans notre deuxième article, les résultats finaux n’incluaient aucune covariable de
mesure anthropométrique sur les paramètres PK. Durant le processus du développement du
modèle, le poids corporel était la covariable la plus significative sur les paramètres PK
comme attendu en pédiatrie. Cependant, le nombre de patients était limité dans cette étude
(n = 14) et les valeurs extrêmes de poids influençaient énormément les résultats et
masquaient l’effet réel. Alors que l’inclusion du poids améliorait la description des profils
cinétiques de certains patients, il détériorait celles d’autres profils. Le nombre restreint de
patients représente la limitation principale de cette étude surtout pour l’étude de la
pharmacocinétique d’un médicament administré par plusieurs voies et connu pour avoir une
grande variabilité interindividuelle. Les autres limitations de cette étude étaient le manque
de collecte de covariables comme le niveau d’inflammation, la dysfonction hépatique et le
génotypage qui auraient pu aider à mieux expliquer la variabilité.
L’utilisation du poids corporel a été choisie dans notre troisième article pour
prendre en compte l’effet de la croissance en pédiatrie. Le poids corporel a été inclus de
façon allométrique sur tous les paramètres pharmacocinétiques, de même que dans les
autres articles. Certains auteurs proposent d’estimer les exposants allométriques [202, 203]
et ceci a été appliqué dans notre premier article. Cependant, l’estimation de l’exposant
allométrique à partir de données humaines se heurte à des problèmes en raison de la
fourchette de poids relativement étroite et l'erreur d'estimation impliquée dans un tel
paramètre non linéaire. En conséquence, cet exposant a été fixé dans notre troisième article
comme proposé par plusieurs auteurs.[93, 94, 204] En effet, il est plus sûr d’utiliser les
152
paramètres fixes développés à partir de données plus larges dans les modèles allométriques.
L’ajustement allométrique représente une approche mécanistique qui repose sur des bases
théoriques et empiriques solides.[21, 93, 205, 206] L’utilisation de l'exposant fixé permet
de séparer les effets d’autres covariables de l'effet de la mesure anthropométrique.
L’ajustement allométrique permet également la comparaison des estimés en pédiatrie avec
ceux des adultes.
tacrolimus suggèrent que cette enzyme ne jouerait aucun rôle dans le métabolisme du
tacrolimus in vivo.[52] La pharmacogénétique permet d’étudier la variabilité
interindividuelle de la séquence de l’ADN génomique responsable d'une variabilité de la
réponse à certains médicaments. L’intérêt du génotypage dans la modélisation et pour le
suivi thérapeutique n’est pas démontré. Des études sur le polymorphisme du GST ont
montré une corrélation avec la variabilité dans la pharmacocinétique du busulfan.[139, 208]
L’effet du polymorphisme génétique sur la pharmacocinétique du tacrolimus a été rapporté
dans plusieurs études. Cependant, les cibles thérapeutiques du tacrolimus sont atteintes en
utilisant le suivi thérapeutique chez tous les patients quel que soit leur génotype. Il existe un
chevauchement important des valeurs des concentrations du tacrolimus pour les patients
des différents génotypes, et la connaissance de ce génotype ne suffit pas à déterminer la
dose optimale pour chaque patient. En transplantation rénale, malgré l'inclusion du
génotypage du CYP3A5 dans le modèle PK du tacrolimus, la variabilité interindividuelle
reste élevée.[209] Dans la transplantation hépatique, il faut considérer l’influence du
génotype du donneur en plus de celui du receveur. D’un point de vue clinique, le
polymorphisme génétique n’est pas nécessairement associé à une différence de phénotype.
Des exemples de cas cliniques observés au CHU Ste-Justine incluent le cas d’un nouveau-
né qui était métaboliseur lent du voriconazole avec des signes de toxicité et est devenu
métaboliseur ultra rapide après deux jours avec des concentrations indétectables. D’autres
exemples sont les nouveau-nés métaboliseurs ultra rapides mais qui à cause de leur âge se
présentent avec un phénotype de métaboliseurs lents. Les études pharmacogénétiques
constitueront sans doute dans un avenir proche un préalable à la commercialisation d’un
grand nombre de médicaments, au même titre que les études de toxicité ou de
pharmacocinétique. Ceci sera à l’origine de l’émergence d’une thérapeutique sur mesure
pour chaque patient, lui assurant un traitement efficace, moins toxique et à moindre coût.
Les modèles physiologiques représentent un autre type de modèles pour analyser les
données présentées dans cette thèse. Ces modèles physiologiques auraient pu inclure une
composante physiologique pour expliquer la maturation de la fonction rénale, l’ontogénie
156
des organes et des systèmes enzymatiques. Cependant, les données sur l’ontogénie des
transporteurs et des tubules rénaux ne sont pas disponibles, et beaucoup d’incertitude existe
dans ces modèles qui introduiraient plus de variabilité.
Le défi majeur des études en pédiatrie est la grande variabilité dans les profils
pharmacocinétiques due à la maturation, la croissance mais aussi à la forme galénique des
médicaments administrés. Les suspensions n’existant pas pour certains médicaments, sont
préparées par l’hôpital ou la pharmacie locale. Les biodisponibilités de ces suspensions ne
sont jamais évaluées et ceci introduit une variabilité énorme dans les études. Une solution
pourrait être le contrôle de la biodisponibilité et son introduction dans les études
pharmacocinétiques. Un autre défi qui découle des variabilités élevées est le nombre de
patients à inclure dans les études. Malgré que la POPPK soit une méthode robuste capable
de déterminer les paramètres PK à partir d’un nombre limité de patients, une certaine
158
Enfin, les travaux de cette thèse peuvent être vus comme un pas vers le
développement d’approches prédictives de la thérapie individualisée. Les outils
pharmacocinétiques développés s’inscriraient à part entière dans une démarche visant à
diminuer le taux d'échec thérapeutique et l’incidence des effets indésirables ou toxiques
chez les enfants immunosupprimés et ayant bénéficié d’une transplantation.
159
Bibliographie
1. Kang JS, Lee MH. Overview of therapeutic drug monitoring. The Korean journal of
internal medicine. 2009;24:1-10.
2. Kuehl P, Zhang J, Lin Y, Lamba J, Assem M, Schuetz J, Watkins PB, Daly A,
Wrighton SA, Hall SD, Maurel P, Relling M, Brimer C, Yasuda K, Venkataramanan R,
Strom S, Thummel K, Boguski MS, Schuetz E. Sequence diversity in CYP3A promoters
and characterization of the genetic basis of polymorphic CYP3A5 expression. Nat Genet.
2001;27:383-91.
3. Beaulieu P, Lambert C. Précis de pharmacologie: Du fondamental à la clinique.
Montréal LpdlUd, editor2010.
4. Rane A, Schwab M, Seyberth H. Pediatric clinical pharmacology: Springer; 2011.
5. Kearns GL, Abdel-Rahman SM, Alander SW, Blowey DL, Leeder JS, Kauffman
RE. Developmental pharmacology--drug disposition, action, and therapy in infants and
children. N Engl J Med. 2003;349:1157-67.
6. Novak E, Allen PJ. Prescribing medications in pediatrics: concerns regarding FDA
approval and pharmacokinetics. Pediatr Nurs. 2007;33:64-70.
7. Venkataramanan R, Swaminathan A, Prasad T, Jain A, Zuckerman S, Warty V,
McMichael J, Lever J, Burckart G, Starzl T. Clinical pharmacokinetics of tacrolimus. Clin
Pharmacokinet. 1995;29:404-30.
8. International conference on harmonisation of technical requirements for registration
of pharmaceuticals for human use. Clinical investigation of medicinal products in the
pediatric population E11. 2000; Available from:
http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Efficacy/E11/Ste
p4/E11_Guideline.pdf.
9. Strolin Benedetti M, Whomsley R, Baltes EL. Differences in absorption,
distribution, metabolism and excretion of xenobiotics between the paediatric and adult
populations. Expert opinion on drug metabolism & toxicology. 2005;1:447-71.
10. European Medicines Agency. Clinical Investigation of Medicinal Products in the
Paediatric Population. 2001; Available from:
http://www.emea.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/W
C500002926.pdf.
11. Bartelink IH, Rademaker CM, Schobben AF, van den Anker JN. Guidelines on
paediatric dosing on the basis of developmental physiology and pharmacokinetic
considerations. Clin Pharmacokinet. 2006;45:1077-97.
12. Hunt A, Joel S, Dick G, Goldman A. Population pharmacokinetics of oral morphine
and its glucuronides in children receiving morphine as immediate-release liquid or
sustained-release tablets for cancer pain. J Pediatr. 1999;135:47-55.
160
13. Bakshi SS, Britto P, Capparelli E, Mofenson L, Fowler MG, Rasheed S, Schoenfeld
D, Zimmer B, Frank Y, Yogev R, Jimenez E, Salgo M, Boone G, Pahwa SG. Evaluation of
pharmacokinetics, safety, tolerance, and activity of combination of zalcitabine and
zidovudine in stable, zidovudine-treated pediatric patients with human immunodeficiency
virus infection. AIDS Clinical Trials Group Protocol 190 Team. The Journal of infectious
diseases. 1997;175:1039-50.
14. Murthy BV, Pandya KS, Booker PD, Murray A, Lintz W, Terlinden R.
Pharmacokinetics of tramadol in children after i.v. or caudal epidural administration. Br J
Anaesth. 2000;84:346-9.
15. Komori M, Nishio K, Kitada M, Shiramatsu K, Muroya K, Soma M, Nagashima K,
Kamataki T. Fetus-specific expression of a form of cytochrome P-450 in human livers.
Biochemistry. 1990;29:4430-3.
16. Holford NH. A size standard for pharmacokinetics. Clin Pharmacokinet.
1996;30:329-32.
17. Anderson BJ, Holford NH. Mechanism-based concepts of size and maturity in
pharmacokinetics. Annual review of pharmacology and toxicology. 2008;48:303-32.
18. Mahmood I. Prediction of drug clearance in children from adults: a comparison of
several allometric methods. Br J Clin Pharmacol. 2006;61:545-57.
19. Knibbe CA, Krekels EH, van den Anker JN, DeJongh J, Santen GW, van Dijk M,
Simons SH, van Lingen RA, Jacqz-Aigrain EM, Danhof M, Tibboel D. Morphine
glucuronidation in preterm neonates, infants and children younger than 3 years. Clin
Pharmacokinet. 2009;48:371-85.
20. Peeters MY, Allegaert K, Blusse van Oud-Alblas HJ, Cella M, Tibboel D, Danhof
M, Knibbe CA. Prediction of propofol clearance in children from an allometric model
developed in rats, children and adults versus a 0.75 fixed-exponent allometric model. Clin
Pharmacokinet. 2010;49:269-75.
21. Anderson BJ, Meakin GH. Scaling for size: some implications for paediatric
anaesthesia dosing. Paediatr Anaesth. 2002;12:205-19.
22. Tod M, Jullien V, Pons G. Facilitation of drug evaluation in children by population
methods and modelling. Clin Pharmacokinet. 2008;47:231-43.
23. Holford N. Dosing in children. Clin Pharmacol Ther. 2010;87:367-70.
24. Alcorn J, McNamara PJ. Pharmacokinetics in the newborn. Adv Drug Deliv Rev.
2003;55:667-86.
25. Commare CE, Tappenden KA. Development of the infant intestine: implications for
nutrition support. Nutrition in clinical practice : official publication of the American
Society for Parenteral and Enteral Nutrition. 2007;22:159-73.
161
and blood cell binding, and the hepatocellular enzymatic activity on hepatic drug clearance.
Journal of pharmacokinetics and biopharmaceutics. 1977;5:625-53.
40. Yamazaki M, Suzuki H, Sugiyama Y. Recent advances in carrier-mediated hepatic
uptake and biliary excretion of xenobiotics. Pharm Res. 1996;13:497-513.
41. Balistreri WF. Immaturity of hepatic excretory function and the ontogeny of bile
acid metabolism. J Pediatr Gastroenterol Nutr. 1983;2 Suppl 1:S207-14.
42. Cresteil T, Beaune P, Kremers P, Celier C, Guengerich FP, Leroux JP.
Immunoquantification of epoxide hydrolase and cytochrome P-450 isozymes in fetal and
adult human liver microsomes. European journal of biochemistry / FEBS. 1985;151:345-
50.
43. Cresteil T. Onset of xenobiotic metabolism in children: toxicological implications.
Food additives and contaminants. 1998;15 Suppl:45-51.
44. Sonnier M, Cresteil T. Delayed ontogenesis of CYP1A2 in the human liver.
European journal of biochemistry / FEBS. 1998;251:893-8.
45. Ratanasavanh D, Beaune P, Morel F, Flinois JP, Guengerich FP, Guillouzo A.
Intralobular distribution and quantitation of cytochrome P-450 enzymes in human liver as a
function of age. Hepatology. 1991;13:1142-51.
46. Tateishi T, Nakura H, Asoh M, Watanabe M, Tanaka M, Kumai T, Takashima S,
Imaoka S, Funae Y, Yabusaki Y, Kamataki T, Kobayashi S. A comparison of hepatic
cytochrome P450 protein expression between infancy and postinfancy. Life Sci.
1997;61:2567-74.
47. Treluyer JM, Cheron G, Sonnier M, Cresteil T. Cytochrome P-450 expression in
sudden infant death syndrome. Biochem Pharmacol. 1996;52:497-504.
48. Treluyer JM, Gueret G, Cheron G, Sonnier M, Cresteil T. Developmental
expression of CYP2C and CYP2C-dependent activities in the human liver: in-vivo/in-vitro
correlation and inducibility. Pharmacogenetics. 1997;7:441-52.
49. Daly AK, Brockmoller J, Broly F, Eichelbaum M, Evans WE, Gonzalez FJ, Huang
JD, Idle JR, Ingelman-Sundberg M, Ishizaki T, Jacqz-Aigrain E, Meyer UA, Nebert DW,
Steen VM, Wolf CR, Zanger UM. Nomenclature for human CYP2D6 alleles.
Pharmacogenetics. 1996;6:193-201.
50. Lacroix D, Sonnier M, Moncion A, Cheron G, Cresteil T. Expression of CYP3A in
the human liver--evidence that the shift between CYP3A7 and CYP3A4 occurs
immediately after birth. European journal of biochemistry / FEBS. 1997;247:625-34.
51. Yang M, Zhu XR, Park PC, Titt U, Mohan R, Virshup G, Clayton JE, Dong L.
Comprehensive analysis of proton range uncertainties related to patient stopping-power-
ratio estimation using the stoichiometric calibration. Physics in medicine and biology.
2012;57:4095-115.
163
52. Kamdem LK, Streit F, Zanger UM, Brockmoller J, Oellerich M, Armstrong VW,
Wojnowski L. Contribution of CYP3A5 to the in vitro hepatic clearance of tacrolimus. Clin
Chem. 2005;51:1374-81.
53. Wrighton SA, Brian WR, Sari MA, Iwasaki M, Guengerich FP, Raucy JL, Molowa
DT, Vandenbranden M. Studies on the expression and metabolic capabilities of human liver
cytochrome P450IIIA5 (HLp3). Mol Pharmacol. 1990;38:207-13.
54. Schuetz JD, Beach DL, Guzelian PS. Selective expression of cytochrome P450
CYP3A mRNAs in embryonic and adult human liver. Pharmacogenetics. 1994;4:11-20.
55. van den Berg H, van den Anker JN, Beijnen JH. Cytostatic drugs in infants: a
review on pharmacokinetic data in infants. Cancer treatment reviews. 2012;38:3-26.
56. Furlan V, Perello L, Jacquemin E, Debray D, Taburet AM. Interactions between
FK506 and rifampicin or erythromycin in pediatric liver recipients. Transplantation.
1995;59:1217-8.
57. Gabrielsson J, Weiner D. Pharmacokinetic and pharmacodynamic data Analysis:
Concepts and Applications2000.
58. Barrett JS, Fossler MJ, Cadieu KD, Gastonguay MR. Pharmacometrics: a
multidisciplinary field to facilitate critical thinking in drug development and translational
research settings. J Clin Pharmacol. 2008;48:632-49.
59. Mould DR, Upton RN. Basic Concepts in Population Modeling, Simulation, and
Model-Based Drug Development. CPT: Pharmacometrics & Systems Pharmacology
2012;1.
60. Gieschke R, Steimer JL. Pharmacometrics: modelling and simulation tools to
improve decision making in clinical drug development. Eur J Drug Metab Pharmacokinet.
2000;25:49-58.
61. Grasela TH, Dement CW, Kolterman OG, Fineman MS, Grasela DM, Honig P,
Antal EJ, Bjornsson TD, Loh E. Pharmacometrics and the transition to model-based
development. Clin Pharmacol Ther. 2007;82:137-42.
62. Sheiner LB. Learning versus confirming in clinical drug development. Clin
Pharmacol Ther. 1997;61:275-91.
63. Sheiner LB, Steimer JL. Pharmacokinetic/pharmacodynamic modeling in drug
development. Annual review of pharmacology and toxicology. 2000;40:67-95.
64. Lee JY, Garnett CE, Gobburu JV, Bhattaram VA, Brar S, Earp JC, Jadhav PR,
Krudys K, Lesko LJ, Li F, Liu J, Madabushi R, Marathe A, Mehrotra N, Tornoe C, Wang
Y, Zhu H. Impact of pharmacometric analyses on new drug approval and labelling
decisions: a review of 198 submissions between 2000 and 2008. Clin Pharmacokinet.
2011;50:627-35.
164
65. Bhattaram VA, Bonapace C, Chilukuri DM, Duan JZ, Garnett C, Gobburu JV, Jang
SH, Kenna L, Lesko LJ, Madabushi R, Men Y, Powell JR, Qiu W, Ramchandani RP,
Tornoe CW, Wang Y, Zheng JJ. Impact of pharmacometric reviews on new drug approval
and labeling decisions--a survey of 31 new drug applications submitted between 2005 and
2006. Clin Pharmacol Ther. 2007;81:213-21.
66. Powell JR, Gobburu JV. Pharmacometrics at FDA: evolution and impact on
decisions. Clin Pharmacol Ther. 2007;82:97-102.
67. Barrett JS. The role of quantitative pharmacology in an academic translational
research environment. The AAPS journal. 2008;10:9-14.
68. Holford N, Karlsson MO. Time for quantitative clinical pharmacology: a proposal
for a pharmacometrics curriculum. Clin Pharmacol Ther. 2007;82:103-5.
69. Gabrielsson J, Weiner D. Pharmacokinetic & Pharmacodynamic Data Analysis:
Concepts and Applications. 4 ed2006.
70. Sheiner LB, Rosenberg B, Melmon KL. Modelling of individual pharmacokinetics
for computer-aided drug dosage. Computers and biomedical research, an international
journal. 1972;5:411-59.
71. Berde CB. Convulsions associated with pediatric regional anesthesia. Anesthesia
and analgesia. 1992;75:164-6.
72. Burns LE, Hodgman JE, Cass AB. Fatal circulatory collapse in premature infants
receiving chloramphenicol. N Engl J Med. 1959;261:1318-21.
73. Kauffman RE. Fentanyl, fads, and folly: who will adopt the therapeutic orphans? J
Pediatr. 1991;119:588-9.
74. Chatelut E, Boddy AV, Peng B, Rubie H, Lavit M, Dezeuze A, Pearson AD, Roche
H, Robert A, Newell DR, Canal P. Population pharmacokinetics of carboplatin in children.
Clin Pharmacol Ther. 1996;59:436-43.
75. Collart L, Blaschke TF, Boucher F, Prober CG. Potential of population
pharmacokinetics to reduce the frequency of blood sampling required for estimating kinetic
parameters in neonates. Developmental pharmacology and therapeutics. 1992;18:71-80.
76. Crome P, Flanagan RJ. Pharmacokinetic studies in elderly people. Are they
necessary? Clin Pharmacokinet. 1994;26:243-7.
77. Anderson BJ, Allegaert K, Holford NH. Population clinical pharmacology of
children: general principles. Eur J Pediatr. 2006;165:741-6.
78. Ette EI, Williams PJ. Population pharmacokinetics I: background, concepts, and
models. Ann Pharmacother. 2004;38:1702-6.
79. Beal SL, Sheiner LB. NONMEM user’s guides. NONMEM Project Group. San
Francisco: University of California - 1992. 1992.
165
106. Xie HG, Kim RB, Wood AJ, Stein CM. Molecular basis of ethnic differences in
drug disposition and response. Annual review of pharmacology and toxicology.
2001;41:815-50.
107. Wilkinson GR. Drug metabolism and variability among patients in drug response. N
Engl J Med. 2005;352:2211-21.
108. Hengstler JG, Arand M, Herrero ME, Oesch F. Polymorphisms of N-
acetyltransferases, glutathione S-transferases, microsomal epoxide hydrolase and
sulfotransferases: influence on cancer susceptibility. Recent results in cancer research
Fortschritte der Krebsforschung Progres dans les recherches sur le cancer. 1998;154:47-85.
109. Cotton SC, Sharp L, Little J, Brockton N. Glutathione S-transferase polymorphisms
and colorectal cancer: a HuGE review. Am J Epidemiol. 2000;151:7-32.
110. Coles BF, Chen G, Kadlubar FF, Radominska-Pandya A. Interindividual variation
and organ-specific patterns of glutathione S-transferase alpha, mu, and pi expression in
gastrointestinal tract mucosa of normal individuals. Arch Biochem Biophys. 2002;403:270-
6.
111. Komiya Y, Tsukino H, Nakao H, Kuroda Y, Imai H, Katoh T. Human glutathione
S-transferase A1, T1, M1, and P1 polymorphisms and susceptibility to prostate cancer in
the Japanese population. Journal of cancer research and clinical oncology. 2005;131:238-
42.
112. Holmstrom G, Borgstrom B, Calissendorff B. Cataract in children after bone
marrow transplantation: relation to conditioning regimen. Acta ophthalmologica
Scandinavica. 2002;80:211-5.
113. Santos GW. The development of busulfan/cyclophosphamide preparative regimens.
Semin Oncol. 1993;20:12-6; quiz 7.
114. Santos GW, Tutschka PJ, Brookmeyer R, Saral R, Beschorner WE, Bias WB,
Braine HG, Burns WH, Elfenbein GJ, Kaizer H, et al. Marrow transplantation for acute
nonlymphocytic leukemia after treatment with busulfan and cyclophosphamide. N Engl J
Med. 1983;309:1347-53.
115. DeLeve LD, Wang X. Role of oxidative stress and glutathione in busulfan toxicity
in cultured murine hepatocytes. Pharmacology. 2000;60:143-54.
116. Hassan M, Ehrsson H, Ljungman P. Aspects concerning busulfan pharmacokinetics
and bioavailability. Leuk Lymphoma. 1996;22:395-407.
117. Hassan M, Ljungman P, Bolme P, Ringden O, Syruckova Z, Bekassy A, Stary J,
Wallin I, Kallberg N. Busulfan bioavailability. Blood. 1994;84:2144-50.
118. Vassal G, Gouyette A, Hartmann O, Pico JL, Lemerle J. Pharmacokinetics of high-
dose busulfan in children. Cancer Chemother Pharmacol. 1989;24:386-90.
168
119. Tran HT, Madden T, Petropoulos D, Worth LL, Felix EA, Sprigg-Saenz HA,
Choroszy M, Danielson M, Przepiorka D, Chan KW. Individualizing high-dose oral
busulfan: prospective dose adjustment in a pediatric population undergoing allogeneic stem
cell transplantation for advanced hematologic malignancies. Bone Marrow Transplant.
2000;26:463-70.
120. Buggia I, Locatelli F, Regazzi MB, Zecca M. Busulfan. Ann Pharmacother.
1994;28:1055-62.
121. Hassan M, Ehrsson H, Smedmyr B, Totterman T, Wallin I, Oberg G, Simonsson B.
Cerebrospinal fluid and plasma concentrations of busulfan during high-dose therapy. Bone
Marrow Transplant. 1989;4:113-4.
122. McCune JS, Gibbs JP, Slattery JT. Plasma concentration monitoring of busulfan:
does it improve clinical outcome? Clin Pharmacokinet. 2000;39:155-65.
123. Czerwinski M, Gibbs JP, Slattery JT. Busulfan conjugation by glutathione S-
transferases alpha, mu, and pi. Drug Metab Dispos. 1996;24:1015-9.
124. Bredschneider M, Klein K, Murdter TE, Marx C, Eichelbaum M, Nussler AK,
Neuhaus P, Zanger UM, Schwab M. Genetic polymorphisms of glutathione S-transferase
A1, the major glutathione S-transferase in human liver: consequences for enzyme
expression and busulfan conjugation. Clin Pharmacol Ther. 2002;71:479-87.
125. Zwaveling J, Press RR, Bredius RG, van Derstraaten TR, den Hartigh J, Bartelink
IH, Boelens JJ, Guchelaar HJ. Glutathione S-transferase polymorphisms are not associated
with population pharmacokinetic parameters of busulfan in pediatric patients. Ther Drug
Monit. 2008;30:504-10.
126. Johnson L, Orchard PJ, Baker KS, Brundage R, Cao Q, Wang X, Langer E, Farag-
El Maasah S, Ross JA, Remmel R, Jacobson PA. Glutathione S-transferase A1 genetic
variants reduce busulfan clearance in children undergoing hematopoietic cell
transplantation. J Clin Pharmacol. 2008;48:1052-62.
127. Kusama M, Kubota T, Matsukura Y, Matsuno K, Ogawa S, Kanda Y, Iga T.
Influence of glutathione S-transferase A1 polymorphism on the pharmacokinetics of
busulfan. Clinica chimica acta; international journal of clinical chemistry. 2006;368:93-8.
128. Srivastava A, Poonkuzhali B, Shaji RV, George B, Mathews V, Chandy M,
Krishnamoorthy R. Glutathione S-transferase M1 polymorphism: a risk factor for hepatic
venoocclusive disease in bone marrow transplantation. Blood. 2004;104:1574-7.
129. Vassal G, Challine D, Koscielny S, Hartmann O, Deroussent A, Boland I, Valteau-
Couanet D, Lemerle J, Levi F, Gouyette A. Chronopharmacology of high-dose busulfan in
children. Cancer Res. 1993;53:1534-7.
130. Poonkuzhali B, Chandy M, Srivastava A, Dennison D, Krishnamoorthy R.
Glutathione S-transferase activity influences busulfan pharmacokinetics in patients with
169
beta thalassemia major undergoing bone marrow transplantation. Drug Metab Dispos.
2001;29:264-7.
131. Fitzsimmons WE, Ghalie R, Kaizer H. Anticonvulsants and busulfan. Annals of
internal medicine. 1990;112:552-3.
132. Trame MN, Bergstrand M, Karlsson MO, Boos J, Hempel G. Population
pharmacokinetics of busulfan in children: increased evidence for body surface area and
allometric body weight dosing of busulfan in children. Clin Cancer Res. 2011;17:6867-77.
133. Nguyen L. Integration of modelling and simulation into the development of
intravenous busulfan in paediatrics: an industrial experience. Fundam Clin Pharmacol.
2008;22:599-604.
134. Booth BP, Rahman A, Dagher R, Griebel D, Lennon S, Fuller D, Sahajwalla C,
Mehta M, Gobburu JV. Population pharmacokinetic-based dosing of intravenous busulfan
in pediatric patients. J Clin Pharmacol. 2007;47:101-11.
135. Nakamura H, Sato T, Okada K, Miura G, Ariyoshi N, Nakazawa K, Kitada M.
Population pharmacokinetics of oral busulfan in young Japanese children before
hematopoietic stem cell transplantation. Ther Drug Monit. 2008;30:75-83.
136. Schiltmeyer B, Klingebiel T, Schwab M, Murdter TE, Ritter CA, Jenke A, Ehninger
G, Gruhn B, Wurthwein G, Boos J, Hempel G. Population pharmacokinetics of oral
busulfan in children. Cancer Chemother Pharmacol. 2003;52:209-16.
137. Nguyen L, Fuller D, Lennon S, Leger F, Puozzo C. I.V. busulfan in pediatrics: a
novel dosing to improve safety/efficacy for hematopoietic progenitor cell transplantation
recipients. Bone Marrow Transplant. 2004;33:979-87.
138. Vassal G, Michel G, Esperou H, Gentet JC, Valteau-Couanet D, Doz F, Mechinaud
F, Galambrun C, Neven B, Zouabi H, Nguyen L, Puozzo C. Prospective validation of a
novel IV busulfan fixed dosing for paediatric patients to improve therapeutic AUC
targeting without drug monitoring. Cancer Chemother Pharmacol. 2008;61:113-23.
139. Ansari M, Lauzon-Joset JF, Vachon MF, Duval M, Theoret Y, Champagne MA,
Krajinovic M. Influence of GST gene polymorphisms on busulfan pharmacokinetics in
children. Bone Marrow Transplant. 2010;45:261-7.
140. Sanati H, Belanger P, Fratti R, Ghannoum M. A new triazole, voriconazole (UK-
109,496), blocks sterol biosynthesis in Candida albicans and Candida krusei. Antimicrob
Agents Chemother. 1997;41:2492-6.
141. Ghannoum MA, Kuhn DM. Voriconazole -- better chances for patients with
invasive mycoses. European journal of medical research. 2002;7:242-56.
142. Jeu L, Piacenti FJ, Lyakhovetskiy AG, Fung HB. Voriconazole. Clin Ther.
2003;25:1321-81.
170
143. Purkins L, Wood N, Greenhalgh K, Allen MJ, Oliver SD. Voriconazole, a novel
wide-spectrum triazole: oral pharmacokinetics and safety. Br J Clin Pharmacol. 2003;56
Suppl 1:10-6.
144. Purkins L, Wood N, Greenhalgh K, Eve MD, Oliver SD, Nichols D. The
pharmacokinetics and safety of intravenous voriconazole - a novel wide-spectrum
antifungal agent. Br J Clin Pharmacol. 2003;56 Suppl 1:2-9.
145. Purkins L, Wood N, Ghahramani P, Greenhalgh K, Allen MJ, Kleinermans D.
Pharmacokinetics and safety of voriconazole following intravenous- to oral-dose escalation
regimens. Antimicrob Agents Chemother. 2002;46:2546-53.
146. Lazarus HM, Blumer JL, Yanovich S, Schlamm H, Romero A. Safety and
pharmacokinetics of oral voriconazole in patients at risk of fungal infection: a dose
escalation study. J Clin Pharmacol. 2002;42:395-402.
147. Pascual A, Calandra T, Bolay S, Buclin T, Bille J, Marchetti O. Voriconazole
therapeutic drug monitoring in patients with invasive mycoses improves efficacy and safety
outcomes. Clin Infect Dis. 2008;46:201-11.
148. Pascual A, Nieth V, Calandra T, Bille J, Bolay S, Decosterd LA, Buclin T,
Majcherczyk PA, Sanglard D, Marchetti O. Variability of voriconazole plasma levels
measured by new high-performance liquid chromatography and bioassay methods.
Antimicrob Agents Chemother. 2007;51:137-43.
149. Scott LJ, Simpson D. Voriconazole : a review of its use in the management of
invasive fungal infections. Drugs. 2007;67:269-98.
150. Denning DW, Ribaud P, Milpied N, Caillot D, Herbrecht R, Thiel E, Haas A,
Ruhnke M, Lode H. Efficacy and safety of voriconazole in the treatment of acute invasive
aspergillosis. Clin Infect Dis. 2002;34:563-71.
151. Theuretzbacher U, Ihle F, Derendorf H. Pharmacokinetic/pharmacodynamic profile
of voriconazole. Clin Pharmacokinet. 2006;45:649-63.
152. Spencer CM, Goa KL, Gillis JC. Tacrolimus. An update of its pharmacology and
clinical efficacy in the management of organ transplantation. Drugs. 1997;54:925-75.
153. Robatel C, Rusca M, Padoin C, Marchetti O, Liaudet L, Buclin T. Disposition of
voriconazole during continuous veno-venous haemodiafiltration (CVVHDF) in a single
patient. J Antimicrob Chemother. 2004;54:269-70.
154. Roffey SJ, Cole S, Comby P, Gibson D, Jezequel SG, Nedderman AN, Smith DA,
Walker DK, Wood N. The disposition of voriconazole in mouse, rat, rabbit, guinea pig,
dog, and human. Drug Metab Dispos. 2003;31:731-41.
155. Karlsson MO, Lutsar I, Milligan PA. Population pharmacokinetic analysis of
voriconazole plasma concentration data from pediatric studies. Antimicrob Agents
Chemother. 2009;53:935-44.
171
156. Walsh TJ, Karlsson MO, Driscoll T, Arguedas AG, Adamson P, Saez-Llorens X,
Vora AJ, Arrieta AC, Blumer J, Lutsar I, Milligan P, Wood N. Pharmacokinetics and safety
of intravenous voriconazole in children after single- or multiple-dose administration.
Antimicrob Agents Chemother. 2004;48:2166-72.
157. Kino T, Hatanaka H, Hashimoto M, Nishiyama M, Goto T, Okuhara M, Kohsaka
M, Aoki H, Imanaka H. FK-506, a novel immunosuppressant isolated from a Streptomyces.
I. Fermentation, isolation, and physico-chemical and biological characteristics. J Antibiot
(Tokyo). 1987;40:1249-55.
158. Kino T, Hatanaka H, Miyata S, Inamura N, Nishiyama M, Yajima T, Goto T,
Okuhara M, Kohsaka M, Aoki H, et al. FK-506, a novel immunosuppressant isolated from
a Streptomyces. II. Immunosuppressive effect of FK-506 in vitro. J Antibiot (Tokyo).
1987;40:1256-65.
159. Liu J, Farmer JD, Jr., Lane WS, Friedman J, Weissman I, Schreiber SL. Calcineurin
is a common target of cyclophilin-cyclosporin A and FKBP-FK506 complexes. Cell.
1991;66:807-15.
160. Ellis D. Clinical use of tacrolimus (FK-506) in infants and children with renal
transplants. Pediatr Nephrol. 1995;9:487-94.
161. McDiarmid SV. The use of tacrolimus in pediatric liver transplantation. J Pediatr
Gastroenterol Nutr. 1998;26:90-102.
162. Reding R, Wallemacq PE, Lamy ME, Rahier J, Sempoux C, Debande B, Jamart J,
Barker A, Sokal E, De Ville de Goyet J, et al. Conversion from cyclosporine to FK506 for
salvage of immunocompromised pediatric liver allografts. Efficacy, toxicity, and dose
regimen in 23 children. Transplantation. 1994;57:93-100.
163. Swenson JM, Fricker FJ, Armitage JM. Immunosuppression switch in pediatric
heart transplant recipients: cyclosporine to FK 506. J Am Coll Cardiol. 1995;25:1183-8.
164. Reyes J, Mazariegos GV, Bond GM, Green M, Dvorchik I, Kosmach-Park B, Abu-
Elmagd K. Pediatric intestinal transplantation: historical notes, principles and controversies.
Pediatr Transplant. 2002;6:193-207.
165. Jindal RM, Dubernard JM. Towards a specific immunosuppression for pancreas and
islet grafts. Clin Transplant. 2000;14:242-5.
166. Treede H, Klepetko W, Reichenspurner H, Zuckermann A, Meiser B, Birsan T,
Wisser W, Reichert B. Tacrolimus versus cyclosporine after lung transplantation: a
prospective, open, randomized two-center trial comparing two different
immunosuppressive protocols. J Heart Lung Transplant. 2001;20:511-7.
167. Reyes J, Jain A, Mazariegos G, Kashyap R, Green M, Iurlano K, Fung J. Long-term
results after conversion from cyclosporine to tacrolimus in pediatric liver transplantation
for acute and chronic rejection. Transplantation. 2000;69:2573-80.
172
168. Cao S, Cox KL, Berquist W, Hayashi M, Concepcion W, Hammes GB, Ojogho OK,
So SK, Frerker M, Castillo RO, Monge H, Esquivel CO. Long-term outcomes in pediatric
liver recipients: comparison between cyclosporin A and tacrolimus. Pediatr Transplant.
1999;3:22-6.
169. Busuttil RW, Holt CD. Tacrolimus is superior to cyclosporine in liver
transplantation. Transplant Proc. 1998;30:2174-8.
170. Shapiro R. Tacrolimus in pediatric renal transplantation: a review. Pediatr
Transplant. 1998;2:270-6.
171. Yanik G, Levine JE, Ratanatharathorn V, Dunn R, Ferrara J, Hutchinson RJ.
Tacrolimus (FK506) and methotrexate as prophylaxis for acute graft-versus-host disease in
pediatric allogeneic stem cell transplantation. Bone Marrow Transplant. 2000;26:161-7.
172. Filler G, Grygas R, Mai I, Stolpe HJ, Greiner C, Bauer S, Ehrich JH.
Pharmacokinetics of tacrolimus (FK 506) in children and adolescents with renal transplants.
Nephrol Dial Transplant. 1997;12:1668-71.
173. Floren LC, Bekersky I, Benet LZ, Mekki Q, Dressler D, Lee JW, Roberts JP, Hebert
MF. Tacrolimus oral bioavailability doubles with coadministration of ketoconazole. Clin
Pharmacol Ther. 1997;62:41-9.
174. Lo A, Burckart GJ. P-glycoprotein and drug therapy in organ transplantation. J Clin
Pharmacol. 1999;39:995-1005.
175. Wallemacq PE, Furlan V, Moller A, Schafer A, Stadler P, Firdaous I, Taburet AM,
Reding R, Clement De Clety S, De Ville De Goyet J, Sokal E, Lykavieris L, Van Leeuw V,
Bernard O, Otte JB, Undre NA. Pharmacokinetics of tacrolimus (FK506) in paediatric liver
transplant recipients. Eur J Drug Metab Pharmacokinet. 1998;23:367-70.
176. Wallemacq PE, Verbeeck RK. Comparative clinical pharmacokinetics of tacrolimus
in paediatric and adult patients. Clin Pharmacokinet. 2001;40:283-95.
177. Beysens AJ, Wijnen RM, Beuman GH, van der Heyden J, Kootstra G, van As H.
FK 506: monitoring in plasma or in whole blood? Transplant Proc. 1991;23:2745-7.
178. Moreno M, Manzanares C, Castellano F, Medina E, Urruzuno P, Camarena C,
Manzanares J, Jara P. Monitoring of tacrolimus as rescue therapy in pediatric liver
transplantation. Ther Drug Monit. 1998;20:376-9.
179. Undre NA, van Hooff J, Christiaans M, Vanrenterghem Y, Donck J, Heeman U,
Kohnle M, Zanker B, Land W, Morales JM, Andres A, Schafer A, Stevenson P. Low
systemic exposure to tacrolimus correlates with acute rejection. Transplant Proc.
1999;31:296-8.
180. Jain AB, Fung JJ, Tzakis AG, Venkataramanan R, Abu-Elmagd K, Alessiani M,
Reyes J, Irish W, Warty V, Mehta S, et al. Comparative study of cyclosporine and FK 506
dosage requirements in adult and pediatric orthotopic liver transplant patients. Transplant
Proc. 1991;23:2763-6.
173
194. Staatz CE, Taylor PJ, Lynch SV, Willis C, Charles BG, Tett SE. Population
pharmacokinetics of tacrolimus in children who receive cut-down or full liver transplants.
Transplantation. 2001;72:1056-61.
195. Wallin JE, Bergstrand M, Wilczek HE, Nydert PS, Karlsson MO, Staatz CE.
Population pharmacokinetics of tacrolimus in pediatric liver transplantation: early
posttransplantation clearance. Ther Drug Monit. 2011;33:663-72.
196. Fukudo M, Yano I, Masuda S, Goto M, Uesugi M, Katsura T, Ogura Y, Oike F,
Takada Y, Egawa H, Uemoto S, Inui K. Population pharmacokinetic and pharmacogenomic
analysis of tacrolimus in pediatric living-donor liver transplant recipients. Clin Pharmacol
Ther. 2006;80:331-45.
197. Taussig HB. A study of the German outbreak of phocomelia. The thalidomide
syndrome. JAMA. 1962;180:1106-14.
198. Borowitz DS, Grand RJ, Durie PR. Use of pancreatic enzyme supplements for
patients with cystic fibrosis in the context of fibrosing colonopathy. Consensus Committee.
J Pediatr. 1995;127:681-4.
199. Olkkola KT, Aranko K, Luurila H, Hiller A, Saarnivaara L, Himberg JJ, Neuvonen
PJ. A potentially hazardous interaction between erythromycin and midazolam. Clin
Pharmacol Ther. 1993;53:298-305.
200. Saint Raymond A, Brasseur D. Development of medicines for children in Europe:
ethical implications. Paediatric respiratory reviews. 2005;6:45-51.
201. Pillai GC, Mentre F, Steimer JL. Non-linear mixed effects modeling - from
methodology and software development to driving implementation in drug development
science. J Pharmacokinet Pharmacodyn. 2005;32:161-83.
202. Schuttler J, Ihmsen H. Population pharmacokinetics of propofol: a multicenter
study. Anesthesiology. 2000;92:727-38.
203. Yukawa E, Satou M, Nonaka T, Yukawa M, Ohdo S, Higuchi S, Kuroda T, Goto Y.
Pharmacoepidemiologic investigation of clonazepam relative clearance by mixed-effect
modeling using routine clinical pharmacokinetic data in Japanese patients. J Clin
Pharmacol. 2002;42:81-8.
204. Capparelli EV, Englund JA, Connor JD, Spector SA, McKinney RE, Palumbo P,
Baker CJ. Population pharmacokinetics and pharmacodynamics of zidovudine in HIV-
infected infants and children. J Clin Pharmacol. 2003;43:133-40.
205. West GB, Brown JH, Enquist BJ. A general model for the origin of allometric
scaling laws in biology. Science. 1997;276:122-6.
206. West GB, Brown JH, Enquist BJ. The fourth dimension of life: fractal geometry and
allometric scaling of organisms. Science. 1999;284:1677-9.
175
207. Sim SC, Edwards RJ, Boobis AR, Ingelman-Sundberg M. CYP3A7 protein
expression is high in a fraction of adult human livers and partially associated with the
CYP3A7*1C allele. Pharmacogenet Genomics. 2005;15:625-31.
208. Ansari M, Krajinovic M. Can the pharmacogenetics of GST gene polymorphisms
predict the dose of busulfan in pediatric hematopoietic stem cell transplantation?
Pharmacogenomics. 2009;10:1729-32.
209. Zhao W, Fakhoury M, Baudouin V, Storme T, Maisin A, Deschenes G, Jacqz-
Aigrain E. Population pharmacokinetics and pharmacogenetics of once daily prolonged-
release formulation of tacrolimus in pediatric and adolescent kidney transplant recipients.
Eur J Clin Pharmacol. 2012.
210. Reed MD. Optimal sampling theory: An overview of its application to
pharmacokinetic studies in infants and children. Pediatrics. 1999;104:627-32.
211. Delaloye JR, Kassir N, Lapeyraque AL, Alvarez F, Lallier M, Beaunoyer M, Labbe
L, Theoret Y, Litalien C. Limited sampling strategies for monitoring tacrolimus in pediatric
liver transplant recipients. Ther Drug Monit. 2011;33:380-6.
176
Annexe I
177
CL, Vc et Ka
run1cptpropdiagka CL, Vc, Ka 1273.403a 71.750 0.589 16.7 90.9 66.0 -- 0.328
non-corrélés
CL, Vc corrélés,
run1cptpropcorCLVka CL, Vc, Ka et Ka non- 1151.610 -50.043 0.790 12.6 158 65.5 -- 0.219
corrélés
CL, Vc, Ka, CL et Vc
run1cptpropcorCLVlag 1180.803b -20.850 1.05 12.7 193 65.7 -- 0.236
LAG corrélés
CL et Vc
CL, Vc, Ka,
run1cptpropcorCLVkalag corrélés; et Ka 1151.610b -50.043 0.790 12.6 158 65.5 -- 0.219
LAG
non-corrélés
CL et Vc
CL, Vc, Ka,
run1cptpropcorCLVdiagkalag corrélés; et Ka, 1151.610b -50.043 0.790 12.6 158 65.5 -- 0.219
LAG
lag non-corrélés
CL, Vc
CL, Vc, Ka, b
run2cptpropcorka correlated; and 1137.859 -12.677 0.491 12.9 102 64.9 60.9 -- 0.216
VP(290), Q
Ka
CL, Vc, Ka,
run2cptpropdiagL CL and Vc
VP(290), Q, 1115.605 -34.931 0.171 12.6 19.5 65.0 156 -- 0.202
AG uncorrelated
LAG
CL, Vc, Ka,
run2cptpropcorLA CL and Vc
VP(290), Q, 1111.198 -39.338 0.177 12.7 20.1 64.7 149 -- 0.203
G correlated
LAG
CL, Vc, Ka, CL and Vc
run2cptpropcorLA
VP(290), Q, correlated, KA 1101.870 -48.666 1.33 12.9 147 65.3 56.3 -- 0.194
GdiagKA
LAG uncorrelated
CL, Vc, Ka, CL and Vc
run2cptpropcorLA
VP(290), Q, correlated, Q 1100.268 -50.268 0.316 12.9 32.3 64.9 139 -- 0.193
GdiagQ
LAG uncorrelated
AGdiagLAG VP(290), Q, LAG LAG uncorrelated 1108.458 -42.078 0.621 12.0 75.4 80.9 161 -- 0.189
AGdiagQKA VP(290), Q, LAG Q, KA uncorrelated 1097.646 -52.890 0.901 13.0 104 65.5 42.5 -- 0.190
OVCL VP(290), Q, LAG Q uncorrelated 1135.567 -14.969 0.335 12.1 29.8 55.4 122 -- 0.204
-2 0 1 2 10 30 50 14 18 22 0 50 150
1.0
ETA1
0.0
-1.0
0 1 2
ETA2
r = 0.31
p v alue 0.092
-2
AGE
5 10
r = -0.23 r = -0.11
p v alue 0.23 p v alue 0.58
0
50
WT
30
HT
120
r = -0.23 r = -0.17 r = 0.97 r = 0.96
p v alue 0.22 p v alue 0.38 p v alue <0.001 p v alue <0.001
60
BMI
22
p v alue 0.097 p v alue 0.87 p v alue 0.057 p v alue 0.0036 p v alue 0.14
14
1.4
BSA
0.8
p v alue 0.19 p v alue 0.43 p v alue <0.001 p v alue <0.001 p v alue <0.001 p v alue 0.026
0.2
150
POSTTDATE
p v alue 0.12 p v alue 0.57 p v alue <0.001 p v alue <0.001 p v alue <0.001 p v alue 0.17 p v alue <0.001
BMI= Indice de masse corporelle; BSA= Surface corporelle; ETA1= Effet aléatoire sur la
clairance; ETA2= Effet aléatoire sur le volume de distribution; HT= Taille; WT= Poids.
181
1.0
ETA1
0.0
-1.0
ETA2
0 1 2
r = 0.31
p v alue 0.092
-2
AST
60
r = -0.18 r = -0.14
40
p v alue 0.34 p v alue 0.45
20
ALT
120
400
GGT
200
r = 0.14 r = -0.17 r = 0.33 r = 0.56
p v alue 0.45 p v alue 0.36 p v alue 0.073 p v alue 0.0012
0
PA
300
25
BILI
15
r = -0.038 r = -0.27 r = -0.057 r = 0.31 r = 0.075 r = -0.34
p v alue 0.84 p v alue 0.15 p v alue 0.77 p v alue 0.091 p v alue 0.69 p v alue 0.069
5
ALB
40
1.0
ETA1
0.0
-1.0
ETA2
2
1
r = 0.31
-2 -1 0
p v alue 0.092
CREAT
0.6
r = -0.14 r = 0.018
p v alue 0.46 p v alue 0.92
0.2
150 250 350
CRCL
r = -0.043 r = -0.093 r = -0.62
p v alue 0.82 p v alue 0.63 p v alue <0.001
50
HB
140
r = -0.17 r = -0.14 r = 0.23 r = -0.1
90 110
p v alue 0.38 p v alue 0.45 p v alue 0.23 p v alue 0.6
HCT
0.35
SEX SEX
1.5
1.5
Individual Random Effect of Central Volume of Distribution
1.0
1.0
Individual Random Effect of Clearance
0.5
0.5
0.0
0.0
-0.5
-0.5
-1.0
-1.0
-1.5
-1.5
) ) ) )
13 17 13 17
e (n= (n= e (n= (n=
l e l e
ma Ma
l ma Ma
l
Fe Fe
p-value = 0.7173 p-value = 0.4206
Individual Random Effect of Clearance
-2 -1 0 1 2
No
(
n=
10
)
p-value = 0.349
Steroids
Yes(
n=
20
)
Individual Random Effect of Central Volume of Distribution
-2 -1 0 1 2
No
(
n=
10
)
Y
p-value = 0.6853
Steroids
es(
n=
20
)
184
185
2
Individual Random Effect of Central Volume of Distribution
Individual Random Effect of Clearance
1
1
0
0
-1
-1
-2
-2
) )
20 0) 20 0)
( n= =1 ( n= =1
own ll (n own ll (n
t- d F u t- d F u
Cu Cu
p-value = 0.1653 p-value = 0.1765
Individual Random Effect of Clearance
-2 -1 0 1 2
No
(
n=
25
)
p-value = 0.4334
Yes(
n=
5
)
Hepatic Impairment
No
(
n=
25
)
p-value = 0.3747
Y es(
n=
5 )
Hepatic Impairment
186
Individual Random Effect of Clearance
-2 -1 0 1 2
No
(
n=
22
)
p-value = 0.7076
Yes(
n=
CYP3A4 Inhibitors
8
)
Individual Random Effect of Central Volume of Distribution
-2 -1 0 1 2
No
(
n=
22
)
p-value = 0.0231
Y es(
n=
8 )
CYP3A4 Inhibitors
187
Individual Random Effect of Clearance
-2 -1 0 1 2
<
0.3
3(
n=
7)
p-value = 0.713
>=
Hematocrit
0.3
3(
n=
23
)
Individual Random Effect of Central Volume of Distribution
-2 -1 0 1 2
<
0.3
3(
n=
7)
>=
p-value = 0.8288
0.3
Hematocrit
3(
n=
23
)
188
Individual Random Effect of Clearance
-2 -1 0 1 2
15
1-5
00
(n=
3)
53
-15
0
GGT
(n=
13
)
<=
52
(n=
14
)
Individual Random Effect of Central Volume of Distribution
-2 -1 0 1 2
15
1-5
00
(n=
3)
53
-15
0 (n=
GGT
13
)
<=
52
(n=
14
)
189
Individual Random Effect of Clearance
-2 -1 0 1 2
45
-2
50
(n=
) 9
AST
p-value = 0.6536
<4
5 (n=
21
)
Individual Random Effect of Central Volume of Distribution
-2 -1 0 1 2
45
-2
50
(n=
) 9
AST
p-value = 0.9795
<4
5 (n=
21
)
190
Individual Random Effect of Clearance
-2 -1 0 1 2
<4
5 (n=
15
)
ALT
p-value = 0.8501
>=
45
(n=
15
)
Individual Random Effect of Central Volume of Distribution
-2 -1 0 1 2
<4
5 (n=
15
)
ALT
p-value = 0.003
>=
45
(n=
15
)
191
Individual Random Effect of Clearance
-2 -1 0 1 2
<=
32 (n=
7)
p-value = 0.3926
Albumin
>3
2 (n=
23
)
Individual Random Effect of Central Volume of Distribution
-2 -1 0 1 2
<=
32 (n=
7)
p-value = 0.3168
Albumin
>3
2 (n=
23
)
192
193
2
Individual Random Effect of Central Volume of Distribution
Individual Random Effect of Clearance
1
1
0
0
-1
-1
-2
-2
8 ) ) 8 ) )
22 22
(n= ( n= (n= ( n=
ys s ys s
da ay da ay
28 8d 28 8d
<= >2 <= >2
p-value = 0.3439 p-value = 0.9583
194
2
Individual Random Effect of Central Volume of Distribution
Individual Random Effect of Clearance
1
1
0
0
-1
-1
-2
-2
) )
13 1 7) 13 1 7)
(n= n= (n= n=
nt h
nt h( nt h
nt h(
2 mo mo 2 mo mo
<= >2 <= >2
p-value = 0.1645 p-value = 0.3016
195
2
Individual Random Effect of Central Volume of Distribution
Individual Random Effect of Clearance
1
1
0
0
-1
-1
-2
-2
) )
16 1 4) 16 1 4)
(n= n= (n= n=
nt h
nt h( nt h
nt h(
3 mo mo 3 mo mo
<= >3 <= >3
p-value = 0.2197 p-value = 0.7592
196
Annexe II
197
Corresponding author:
Dr. Catherine Litalien
Division of Pediatric Critical Care, Department of Pediatrics
Centre Hospitalier Universitaire Sainte-Justine
3175 chemin de la Côte Sainte-Catherine
Montréal (Québec), Canada H3T 1C5
Tel: (514) 345-4931 extension 2571
FAX: (514) 345-7731
Abstract
Objective: To develop and validate limited sampling strategies (LSSs) for tacrolimus in
paediatric liver transplant recipients.
Methods: Thirty six 12-hour pharmacokinetic profiles from 28 paediatric liver transplant
recipients (0.4 to 18.5 years) were collected. Tacrolimus concentrations were measured by
immunoassay and area under the curve (AUC0-12) by trapezoidal rule. LSSs consisting of 1,
2, 3, or 4 concentration-time points were developed using multiple regression analysis.
Eight promising models (2 per category) were selected based on the following criteria: r 2 ≥
0.90, inclusion of trough concentration (C0), and time points within 4 hours post-dose. The
predictive performance of these LSSs was evaluated in an independent set of data by
measuring the mean prediction error (%ME) and the root mean squared prediction error
(%RMSE).
Results: Five models including 2 to 4 time points predicted AUC 0-12 with a ± 15% error
limit. Bias (%ME) and precision (%RMSE) of LSS involving C 0, C1, and C4 (AUCpredicted =
9.30 + 3.69*C0 + 2.19*C1 + 4.69*C4) were -4.98% and 8.29%, respectively. Among single
time point LSSs, model using C0 had a poor correlation with AUC0-12 (r2 = 0.53) while the
one with C4 had the highest correlation with tacrolimus exposure (r 2 = 0.84).
1 Introduction
Tacrolimus (Prograf®; Fujisawa Healthcare Inc.) is the most prescribed
immunosuppressive agent in paediatric solid organ transplant recipients; the proportion of
paediatric patients receiving tacrolimus increased from 14.5% in 1997 to 63.2% in 2006. 1
As with cyclosporine, tacrolimus inhibits calcineurin and blocks the transcription of
cytokines that drive the proliferative T-cell response, particularly interleukin-2.
Therapeutic drug monitoring has become a standard of care for tacrolimus dosing
optimization because of its significant inter- and intraindividual pharmacokinetic (PK)
variability and its narrow therapeutic index. Although whole blood trough concentration
(C0) is the current method used to guide dose individualisation. However many paediatric
studies have shown that the relationship between tacrolimus C 0 and area under the
concentration-time curve (AUC) is highly variable, with coefficients of determination (r 2)
2-13
ranging from 0.30 to 0.88. Additionally, controversies remain about the relationship
between C0 and clinical outcomes. 14, 15
In liver transplantation, while a number of studies
have shown a significant correlation between C0 and tacrolimus nephro- and neurotoxicity,
reports failed to demonstrate a linear relationship between tacrolimus C 0 and the
16-18
development of graft rejection.
The aim of this study was to develop and validate LSSs for tacrolimus in paediatric
liver transplant recipients.
202
2 Methods
Patients and Study Design
Whenever available in the medical chart, the following data were recorded for each
patient: demographic parameters, type of liver transplantation, underlying diagnosis, time
post-transplantation at which the PK profile was performed, tacrolimus dosing regimen and
formulation, concomitant immunosuppression, blood chemistry (liver and renal function
tests, haemoglobin, hematocrit, and albumin), and presence of clinically relevant CYP3A4
and/or P-glycoprotein inducer or inhibitor.
203
Pharmacokinetic data were randomly split into two equal groups: a training group
and a validation group. In the training group, a multiple regression analysis was used to
determine the relationship between observed tacrolimus AUC 0-12 (dependent variable) and
the concentrations at various time points (independent variable or predictor). This was
carried out using the best-subset-regression in conjunction with the stepwise forward
selection technique. This method consists of starting with an equation with no predictors,
trying them out one by one sequentially, and calculating the coefficient of determination for
each equation (r2). Multiple linear regression models were developed in which 1, 2, 3, or 4
concentration-time points were used as predictors. All possible equations were derived and
the 2 best regression models using 1, 2, 3, and 4 predictors were identified based on the
following criteria: r2 ≥ 0.90, inclusion of C0, and time points within 4 hours post tacrolimus
administration.
204
1 N
ME = ¦ (Pred - Obs)
N i=1
1 N
MSE = ¦
N i=1
(Pr ed-Obs) 2
RMSE = MSE
1 N § Pred - Obs ·
ME (%) = ¦ ¨ Obs ¸¹ u100
N i=1 ©
2
1 N § Pred - Obs ·
RMSE (%) = ¦ ¨ Obs ¸¹ u100
N i =1 ©
where Pred is the predicted value of AUC 0-12, Obs is the AUC0-12 observed value of, MSE is
the mean squared prediction error, and N is the number of patients. A 95 % confidence
interval (CI) was calculated for each parameter. In addition, the predictive performance of
the LSSs was tested by the method of Bland and Altman. 31 Finally, an approach based on
CI was used to compare the predictive performance of the different regression models. 30
This consisted of computing the difference in MSE’s (ΔMSE) and ME’s (ΔME) between
the two models compared and calculating their CI. If the CI did not include zero, the model
with the smaller MSE or ME was considered significantly more precise or less biased,
respectively. On the other hand, when the CI included zero, the MSE or ME difference was
considered not significant. Clinical and PK data are presented as mean ± SD (range).
Comparisons between patient characteristics and PK parameters of the training and
validation groups were performed using the unpaired t-test for normally distributed data
205
and the Wilcoxon Rank-Sum test for skewed data. Statistical significance was defined at P
value d 0.05. All statistical analyses were performed using S-plus 8.1, SAS 9.2 (SAS
Institue Inc., Cary, NC, USA), and GraphPad Prism 5.0 (GraphPad Software, San Diego,
CA, USA).
206
3 Results
A total of 42 full PK profiles obtained from 31 liver transplant recipients were
available for this study. Thirty six profiles from 28 patients (15M/13F) aged between 0.4
and 18.5 years were included. Six profiles from 3 patients were excluded because either
they were not obtained under steady-state conditions (n=3), or they were obtained while
patients were receiving tacrolimus three times a day. The indications for whole liver (10
patients) or cutdown liver (18 patients) transplantation were: biliary atresia (n=11),
tyrosinemia (n=7), North American Indian childhood cirrhosis (n=2), fulminant hepatitis
(n=2), Alagille syndrome (n=2), histiocytosis (n=2), sclerosing cholangitis (n=1), and auto-
immune hepatitis (n=1).
LSS development
Two hundred fifty five regression equations were developed to predict tacrolimus
AUC0-12 using a maximum of 4 different concentration-time points. Among these
equations, 5 were identified as best regression models based on the predefined selection
criteria (models 1 to 5, Table 3). For comparison purposes, 4 equations that did not meet
these criteria (models 6 to 9) are also presented. Among the 3 single concentration-time
point equations shown, those involving C0 and C4 had the poorest (r2=0.53) and the highest
(r2=0.84) correlation with AUC0-12, respectively.
LSS validation
The %ME (bias) was also less than 15%, although all models tended to
underestimate tacrolimus AUC0-12. The relative performance of the different LSSs is shown
in Table 5. For three out of the four regression models using 3 or 4 concentration-time
points (models 1 to 3), precision and bias were significantly better than the model using 2
point equation (model 5). The fourth model using 3 time points (model 4) was significantly
more precise than model 5, but there was no difference in bias between the two models.
Finally, no significant difference was observed between the predictive capacity of LSSs
involving 3 and 4 concentration-time points
Figures 2A and 2B depict the linear correlation of observed AUC 0-12 and AUC0-12
predicted by model 3 (C0-C1-C4) and model 5 (C0-C4), respectively. Figures 2C and 2D
show the Bland and Altman analysis of the paired samples for both models, with the
average of predicted and observed AUC 0-12 ranging from 83.6 to 305.0 ng*h/mL. In
accordance with %ME, the predicted AUC 0-12 was lower than the observed AUC0-12, with a
bias of -5.4% for model 3 (Figure 2C) and -10.5% for model 5 (Figure 2D).
208
4 Discussion
This is the first study on the development and validation of LSSs for the prediction
of tacrolimus AUC0-12 in paediatric liver transplant recipients. Five LSSs using 2 to 4
concentration-time points obtained within 4 hours after tacrolimus dosing and including C 0
have been developed and provide an accurate and convenient method to predict tacrolimus
AUC0-12 in this population. All 5 models predicted tacrolimus AUC 0-12 with a ± 15%
prediction error limit; this is a clinically acceptable range because it represents the
deviation from the observed AUC that usually initiates dosage adjustment. 29, 32, 33
Among single concentration-time point LSSs, regression model using C0 had a poor
correlation with AUC0-12 (r2 = 0.53) while the one with C4 had the highest correlation with
tacrolimus exposure (r2 = 0.84). This is in accordance with other adult and paediatric
studies that have shown C0 to be a poor predictor of tacrolimus exposure in solid organ
transplantation.2, 3, 6, 8, 14, 34, 35
LSSs developed in this study involved a limited number of blood samples within a
short period of time after drug administration, originally suggested by Ting et al. 21 This
method is convenient both for the patient and the clinician, and is financially acceptable for
the institution. The inclusion of predose measurement in the equations was a deliberate
choice. Indeed, C0 allows checking for compliance, helps identifying patients with high
tacrolimus clearance, and is a routinely used marker by clinicians. 29, 36
In paediatric organ transplantation, data published regarding LSSs are scarce. One
report involving 14 liver recipients has suggested a 3 concentration-time points LSS (C1-
C4-C8), but without evaluation of its predictive performance. 28 In renal transplant recipients,
2, 3
two studies have proposed LSSs, although with inconvenient sampling times for an
outpatient setting (C6) and no validation with an independent set of data. 21, 29, 30
Both
studies showed a weak correlation between observed AUC and C 0 (r2 = 0.36 and r2 = 0.56)
and among single concentration-time point, the best correlation was observed with C 4.
These observations are in agreement with our results.
209
This study has some limitations that deserve further comments. Ting et al. 21 recently
suggested that LSS should only be applied on transplant patient populations that are
comparable to the population used to develop the LSS. As such, the predictive power of the
210
LSSs reported in this study cannot be guaranteed for patient populations other than
paediatric liver transplant recipients. By opposition to Bayesian derived LSSs, LSSs
developed with a multiple regression approach are less flexible in the sampling time where
collection of samples at exact times is necessary. 41 Immunoassays may overestimate
tacrolimus concentrations compared to high-performance liquid chromatography methods,
due to cross reactivity with tacrolimus metabolites.15 Thus, caution should be exerted in the
application of the proposed LLSs if tacrolimus concentrations are measured with analytic
techniques different than the specific immunoassay used in this study.
Even though the evaluation of the relative performance of the 5 LSSs has shown
that 3 and 4 concentration-time points LSSs were statistically more precise and less biased
than the one using 2 time points, all equations can be used with confidence. Therefore, it is
the ultimate clinician’s choice to decide which model to use according to the desired level
of precision, balanced with the costs and feasibility in clinical practice.
211
5 Conclusion
Trough concentration is a poor predictor of tacrolimus AUC 0-12 in paediatric liver
transplant recipients. However, LSSs using 2 to 4 concentration-time points obtained within
4 hours after tacrolimus dosing have been developed and provide a reliable and convenient
method to predict tacrolimus exposure in this population. By facilitating the accurate
measurement of tacrolimus full AUC, the proposed LSSs represent an important step that
will allow the undertaking of prospective trials aiming to better define tacrolimus target
AUC in paediatric liver transplant recipients, and to determine whether AUC-guided
monitoring is superior to C0-based monitoring in terms of efficacy and safety.
212
6 Acknowledgements
We thank Carole Viau and Dr. Oliver Karam for their technical assistance. We also
thank the nursing staff of the Multispecialty Clinic at CHU Ste-Justine for their help with
sample collection. The time and effort of Professor Manuel Pascual (Chief, Centre de
Transplantation d'Organes, CHUV, Lausanne) in accomplishing this project are also
gratefully acknowledged.
213
7 Conflict of Interest/Disclosure
Financial support came from the Research Centre of Centre Hospitalier
Universitaire Sainte- Justine. N.K. received a studentship from the Faculty of graduate
studies of Université de Montréal. L.L. is recipient of a Rx & D/CIHR Health Research
Foundation Career Award in Health Sciences.
214
8 References
1. Smith JM, Stablein DM, Munoz R, et al. Contributions of the Transplant Registry: The
2006 Annual Report of the North American Pediatric Renal Trials and Collaborative
Studies (NAPRTCS). Pediatr Transplant. 2007;11:366-373.
4. Filler G, Grygas R, Mai I, et al. Pharmacokinetics of tacrolimus (FK 506) in children and
adolescents with renal transplants. Nephrol Dial Transplant. 1997;12:1668-1671.
6. Kim JS, Aviles DH, Silverstein DM, et al. Effect of age, ethnicity, and glucocorticoid
use on tacrolimus pharmacokinetics in pediatric renal transplant patients. Pediatr
Transplant. 2005;9:162-169.
13. Webb NJ, Stevenson PJ, Lewis MA, et al. Pharmacokinetics of tacrolimus in paediatric
renal transplant recipients. Transplant Proc. 2002;34:1948-1950.
14. Staatz CE, Tett SE. Clinical pharmacokinetics and pharmacodynamics of tacrolimus in
solid organ transplantation. Clin Pharmacokinet. 2004;43:623-653.
16. Backman L, Nicar M, Levy M, et al. FK506 trough levels in whole blood and plasma in
liver transplant recipients. Correlation with clinical events and side effects.
Transplantation. 1994;57:519-525.
17. Kershner RP, Fitzsimmons WE. Relationship of FK506 whole blood concentrations and
efficacy and toxicity after liver and kidney transplantation. Transplantation. 1996;62:920-
926.
19. Undre NA, van Hooff J, Christiaans M, et al. Low systemic exposure to tacrolimus
correlates with acute rejection. Transplant Proc. 1999;31:296-298.
21. Ting LS, Villeneuve E, Ensom MH. Beyond cyclosporine: a systematic review of
limited sampling strategies for other immunosuppressants. Ther Drug Monit. 2006;28:419-
430.
23. Pisitkun T, Eiam-Ong S, Chusil S, et al. The roles of C4 and AUC0-4 in monitoring of
tacrolimus in stable kidney transplant patients. Transplant Proc. 2002;34:3173-3175.
25. Scholten EM, Cremers SC, Schoemaker RC, et al. AUC-guided dosing of tacrolimus
prevents progressive systemic overexposure in renal transplant recipients. Kidney Int.
2005;67:2440-2447. \
26. Ting LS, Partovi N, Levy RD, et al. Limited sampling strategy for predicting area under
the concentration-time curve of mycophenolic acid in adult lung transplant recipients.
Pharmacotherapy. 2006;26:1232-1240.
29. David OJ, Johnston A. Limited sampling strategies for estimating cyclosporin area
under the concentration-time curve: review of current algorithms. Ther Drug Monit.
2001;23:100-114.
30. Sheiner LB, Beal SL. Some suggestions for measuring predictive performance. J
Pharmacokinet Biopharm. 1981;9:503-512.
217
31. Bland JM, Altman DG. Statistical methods for assessing agreement between two
methods of clinical measurement. Lancet. 1986;1:307-310.
32. Meier-Kriesche HU, Kaplan B, Brannan P, et al. A limited sampling strategy for the
estimation of eight-hour neoral areas under the curve in renal transplantation. Ther Drug
Monit. 1998;20:401-407.
33. Wong KM, Shek CC, Chau KF, et al. Abbreviated tacrolimus area-under-the-curve
monitoring for renal transplant recipients. Am J Kidney Dis. 2000;35:660-666.
34. Dansirikul C, Staatz CE, Duffull SB, et al. Sampling times for monitoring tacrolimus in
stable adult liver transplant recipients. Ther Drug Monit. 2004;26:593-599.
35. Langers P, Press RR, den Hartigh J, et al. Flexible limited sampling model for
monitoring tacrolimus in stable patients having undergone liver transplantation with
samples 4 to 6 hours after dosing is superior to trough concentration. Ther Drug Monit.
2008;30:456-461.
36. Hoyer PF. Therapeutic drug monitoring of cyclosporin A: should we use the area under
the concentration-time curve and forget trough levels? Pediatr Transplant. 2000;4:2-5.
37. Balbontin FG, Kiberd B, Squires J, et al. Tacrolimus monitoring by simplified sparse
sampling under the concentration time curve. Transplant Proc. 2003;35:2445-2448.
38. Jorgensen K, Povlsen J, Madsen S, et al. C2 (2-h) levels are not superior to trough
levels as estimates of the area under the curve in tacrolimus-treated renal-transplant
patients. Nephrol Dial Transplant. 2002;17:1487-1490.
40. Miura M, Satoh S, Niioka T, et al. Limited sampling strategy for simultaneous
estimation of the area under the concentration-time curve of tacrolimus and mycophenolic
acid in adult renal transplant recipients. Ther Drug Monit. 2008;30:52- 59.
9 Figure Legends
Figure 1. Tacrolimus concentrations as a function of time plotted for the 38
pharmacokinetic profiles (training group, solid lines; validation group, dashed gray lines).
Figure 2. Correlation between the observed AUC 0-12 (ObsAUC0-12) and the predicted AUC0-
12 (PredAUC0-12) of tacrolimus calculated with the equations of model 3 (A) involving 3
concentration-time points (C0-C1-C4) and model 5 (B) involving 2 concentration-time
points (C0-C4), respectively. Bland and Altman analysis testing agreement between
tacrolimus ObsAUC0-12 and PredAUC0-12 calculated with the equations of model 3 (C)
involving 3 concentration-time-points (C0-C1-C4) and model 5 (D) involving 2
concentration-time points (C0-C4), respectively.
220
10 Tables
Table 1. Summary of patient characteristics on the day of the pharmacokinetic profile
(results expressed as number or mean ± SD or median (range))
Variable Training group Validation group P value
(18 profiles) (18 profiles)
Age (years) 12.0 (0.4–16.9) 2.9 (0.6–18.5) 0.20
Weight (kg) 34.7 (4.5–61.3) 17.3 (6.2–59.1) 0.58
Time after transplantation 24.1 (0.5–166.2) 4.2 (0.5–185.5) 0.46
(months)
Tacrolimus dose 0.11 (0.05–0.54) 0.19 (0.05–0.42) 0.44
(mg/kg/d)1
Tacrolimus formulations2
Suspension (5 mg/mL) 7 11
Capsule 11 6
Concomitant
immunosuppressive agents
None 1 0
Corticosteroids 7 9
Corticosteroids + MMF 0 6
MMF 10 3
AST (U/L) 31 (16–304) 47 (24–148) 0.06
ALT (U/L) 45 ± 27 (17–125) 93 ± 69 (14–249) 0.01
GGT (U/L) 51 (8–261)3 98 (13–886) 0.02
AP (U/L) 168 (49–448) 130 (66–567)3 0.53
Bilirubine (μmol/L) 10 (4–25)3 11 (5–110) 0.84
Hb (g/dL) 120 ± 21 (83–151) 120 ± 16 (84–143) 0.92
3
Hct (%) 0.34 ± 0.054 (0.25– 0.36 ± 0.04 (0.26–0.43) 0.48
0.44)4
221
Table 3. Regression equations for predicting tacrolimus AUC 0-12 in paediatric liver
transplant recipients with the associated coefficient of determination (r2)
Model Time-points Regression equation r2
1 C0, C1, C2, C4 AUCp = 9.15 + 3.65*C0 + 1.81*C1 + 0.51*C2 + 4.55*C4 0.99
2 C0, C0.5, C2, C4 AUCp = 9.51 + 3.52*C0 + 1.27*C0.5 + 1.52*C2 + 4.29*C4 0.99
3 C0, C1, C4 AUCp = 9.30 + 3.69*C0 + 2.19*C1 + 4.69*C4 0.99
4 C0, C1.5, C4 AUCp = 9.14 + 4.11*C0 + 1.69*C1.5 + 4.67*C4 0.98
5 C0, C4 AUCp = 17.93 + 5.79*C0 + 4.71*C4 0.94
6 C0, C2 AUCp = 18.56 + 6.97*C0 + 4.05*C2 0.71
7 C4 AUCp = 45.19 + 5.87*C4 0.84
8 C2 AUCp = 44.33 + 6.01*C2 0.58
9 C0 AUCp = 36.24 + 11.68*C0 0.53
AUCp, predicted area under the concentration-time curve; r2, coefficient of determination
224
2 C0, C0.5, C2, C4 12.58 (8.50–15.63) -8.74 (-13.37– -4.11) 8.48 (4.25–11.22) -6.15 (-9.14– -3.16)
3 C0, C1, C4 12.03 (6.59–15.68) -7.91 (-12.55– -3.28) 8.29 (3.29–11.28) -4.98 (-8.37– -1.59)
4 C0, C1.5, C4 15.54 (9.08–20.01) -10.68 (-16.46– -4.91)9.35 (5.60–11.98) -6.75 (-10.06– -3.44)
5 C0, C4 23.89 (14.25–30.65) -18.06 (-26.06– -10.05)11.77 (8.65–14.22) -9.80 (-13.14– -6.46)
CI: confidence interval; ME, mean prediction error; RMSE, root mean squared prediction
error
225
11 Figures
Figure 1
227
Figure 2A
228
Figure 2B
229
Figure 2C
230
Figure 2D
231
Annexe III
This permission applies only to copyrighted material that the Massachusetts Medical Society owns, and not to copyrighted
text or illustrations from other sources.
All content reproduced from copyrighted material owned by the Massachusetts Medical Society (MMS) remains the full and
exclusive copyrighted property of the MMS. The right to grant to a third party is reserved solely by the MMS.
Copyrighted MMS content may not be used in any manner that implies endorsement, sponsorship, or promotion of any
entity, product or service by the MMS or its publications. The MMS cannot authorize use of authors’ names on promotional
materials; such approval must be obtained directly from authors.
The New England Journal of Medicine (and its logo design) are registered trademarks of the Massachusetts Medical
Society. We do not grant permission for our logo, cover, or brand identity to be used in materials produced by other
organizations. NEJM does not issue grants of permission for blanket use of its material. Non-exclusive grants are issued for
identified content to be used in a specific manner. We do provide worldwide rights.
MODIFICATIONS/ADAPTATIONS
Grants of permission are issued for the material to be used as originally published by MMS. MMS does not approve
adaptations or modifications.
Formatting and stylistic changes and any explanatory material or figure legends used by the requestor must accurately
reflect the material as originally published in the New England Journal of Medicine.
*This grant covers the right to use the material in print and electronic formats. Figures/Tables that contain text, may be
translated.
233
Annexe IV