J Antimicrob Chemother

doi:10.1093/jac/dky192

Model of population pharmacokinetics of cidofovir in

immunocompromised children with cytomegalovirus and
adenovirus infection
Nadège Neant1,2†, Roman Klifa3†, Naı̈m Bouazza1,2, Despina Moshous3,4, Benedicte Neven3,4,
Marianne Leruez-Ville5–7, Stephane Blanche2,3, Jean-Marc Treluyer1,2,8,9, Deborah Hirt2,8‡ and Pierre Frange3,5,10*‡

1
Unité de Recherche Clinique, Assistance Publique – Hôpitaux de Paris (AP-HP), Hôpital Tarnier, Paris, France; 2EA7323, Université Paris
Descartes, Sorbonne Paris Cité, Paris, France; 3Unité d’Immunologie, Hématologie et Rhumatologie pédiatriques, AP-HP, Hôpital univer-
sitaire Necker – Enfants malades, Paris, France; 4INSERM UMR1163, Institut Imagine, Université Paris Descartes, Sorbonne Paris Cité,
Paris, France; 5Laboratoire de Microbiologie clinique, AP-HP, Hôpital Universitaire Necker – Enfants malades, Paris, France; 6EA7328,
Université Paris Descartes, Sorbonne Paris Cité, Paris, France; 7Centre National de Référence Herpes Virus, Laboratoire associé, Paris,
France; 8Service de Pharmacologie clinique, AP-HP, Groupe hospitalier Paris Centre, Hôpital Cochin, Paris, France; 9CIC-0901 INSERM,
Cochin-Necker, Paris, France; 10EA 7327, Université Paris Descartes, Sorbonne Paris Cité, Paris, France

*Corresponding author. Tel: !33 1 44 49 49 61; Fax: !33 1 44 49 49 60; E-mail: [email protected]
†These authors contributed equally to the work.
‡These authors contributed equally to the work.

Received 9 February 2018; returned 21 March 2018; revised 20 April 2018; accepted 22 April 2018

Objectives: To describe cidofovir pharmacokinetics and assess the link between concentration and safety/effi-
cacy in children.
Patients and methods: An observational study was conducted in 13 immunocompromised children receiving
cidofovir for adenovirus and/or cytomegalovirus infection. A population pharmacokinetic model was built and
AUC0–24 was derived for each patient. Virological success was defined as a decrease of the viraemia by
1 log10 copies/mL within 15 days of cidofovir initiation. The association between AUC0–24 and virological success
was assessed using a Wilcoxon test. An AUC0–24 cut-off value was determined using a Fisher’s exact test.
Results: Overall, 86 blood samples were analysed. A two-compartment model with first-order absorption and
elimination best described the cidofovir data. Virological success (VS) was reached in 6/8 children with adeno-
virus viraemia and in 1/4 children with cytomegalovirus viraemia. Patients with VS displayed a non-significant
higher median AUC0–24 compared with patients with virological failure: 48.6 (range 8.9–72.6) versus 19.1 (6.9–
22.7) mgh/L. Adenovirus-viraemic patients with an AUC0–24 value below 19.1 mgh/L had a higher probability of
treatment failure (P " 0.03). Aviraemic children with stool and/or nasopharyngeal adenovirus carriage cleared
the viral carriage within a month of cidofovir initiation. During treatment, 1/13 children developed a tubulopathy
but none of them had an increase in creatininaemia.
Conclusions: Cidofovir appears safe and reasonably well tolerated and seemed to have efficacy in a subset of
patients with adenovirus/cytomegalovirus infection. Therapeutic drug monitoring may be useful in children
receiving cidofovir and, in the case of adenovirus infection, targeting an AUC0–24 above 19.1 mgh/L could be
associated with higher probability of virological success.

Introduction and many adenovirus serotypes.2 However, its government ap-

proval is limited in France to the treatment of cytomegalovirus ret-
Adenovirus and cytomegalovirus infections are major causes of initis in HIV-infected adults; European Conference on Infections in
morbidity in immunocompromised patients (e.g. pneumonia, en- Leukaemia guidelines recommend its use as a first-line paediatric
cephalitis, ileitis, colitis, hepatitis and multi-organ failure) and can drug for the treatment of adenovirus-related disease and/or infec-
lead to a mortality risk of up to 60%.1 tions and as part of second-line antiviral regimens to treat cyto-
Cidofovir is an inhibitor of the viral DNA polymerase, with broad megalovirus infections after allogeneic HSCT.3 Cidofovir has been
in vitro efficacy against many viruses, including cytomegalovirus used in these indications for many years, despite the absence of

C The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved.
V
For permissions, please email: [email protected].
1 of 8
Downloaded from https://academic.oup.com/jac/advance-article-abstract/doi/10.1093/jac/dky192/5026318
by Miami University user
on 07 June 2018
 Neant et al.

R
randomized controlled trials demonstrating its efficacy and the CMV R-GENEV kit (bioMérieux, Marcy l’Étoile, France) and the ADENOVIRUS
R
absence of consensus on dosage or scheduling of this drug in these R-GENEV kit (bioMérieux), respectively.
indications. Moreover, the frequency and severity of cidofovir-
related adverse events, particularly nephrotoxicity, in immuno- Cidofovir administration and plasma drug determination
compromised children remain unclear. Patients received intravenous cidofovir (5 mg/kg of bodyweight) as 1 h infu-
Cidofovir is phosphorylated to its active form by intracellular sions at day 1, day 8 and then every 2 weeks (if continued). In order to de-
kinases. After administration intravenously, cidofovir is predomin- crease the risk of nephrotoxicity, children received concomitantly (i) oral
antly eliminated by renal excretion via glomerular filtration and ac- probenecid (2 g/1.73 m2, given 3 h prior to the cidofovir infusion, and two
tive tubular secretion. Between 80% and 100% of the dose is additional 1 g/1.73 m2 doses at 2 h and 8 h after the infusion, respectively);
excreted unchanged in the urine within 24 h. The terminal elimin- (ii) intravenous prehydration (normal saline infusion of 1 L/1.73 m2/h prior
ation t1=2 of cidofovir in plasma is short (2 h). However, the t1=2 of to cidofovir infusion); and (iii) intravenous posthydration (normal saline in-
intracellular phosphorylated cidofovir is longer (17–65 h).4,5 fusion of 1 L/1.73 m2 over 3 h after cidofovir infusion).
The pharmacokinetics of cidofovir in children have only been Blood samples were collected on multiple occasions (usually between
one and six times) at various times post-dose ranging from 1 to 13 h after
investigated in one previous study6 and no model of population
the end of the infusion. Mostly, three samples per occasion were obtained
pharmacokinetics has been reported to date in this population. at 1, 3 and 6 h after drug intake. Blood samples were drawn at least 1 week
Population-based modelling, in contrast with conventional kinetic after treatment initiation and during the treatment. The plasma cidofovir
studies, allows concentrations to be analysed simultaneously concentrations were determined using a validated HPLC method with fluor-
within a cohort. It enables a direct estimate (single-step process) escence detection, as previously reported by Momper et al.7 The linearity of
of the typical pharmacokinetics parameters (fixed effects) and the calibration curve of our method was evaluated over the range of 0.1–
their variability (random effects) for the observed concentrations. 10 mg/L with a good correlation coefficient (r2"0.9985). The lower limit of
It also detects the covariates that can explain inter-individual vari- quantification was set at 0.1 mg/L. The precision and accuracy values
ability of the pharmacokinetic parameters. From this model, ex- ranged from 5.6% to 9.7% and 0.5% to 6.4%, respectively.
posure/efficacy and exposure/toxicity relationships can be studied.
As far as we are aware, our study is the first to establish a model Population pharmacokinetic analysis
of population pharmacokinetics of cidofovir in immunocomprom- The population pharmacokinetics analysis was performed using a non-
ised children in order to guide cidofovir use in paediatric severe linear mixed-effect modelling approach with NOMMEM software version
adenovirus and cytomegalovirus infections. VII.8 One- and two-compartment models were investigated as structural
models to describe cidofovir pharmacokinetics using the ‘first-order condi-
tional estimation (FOCE) with interaction’ method. Several error models
Patients and methods (additive, proportional and combined) were tested to describe the residual
Ethics variability. The inter-individual variability and inter-occasion variability were
assumed to be log-normally distributed. Moreover, weight-based allomet-
This research was conducted in accordance with the Declaration of Helsinki ric scaling was tested to standardize the pharmacokinetics parameters.9
and national and institutional standards. Parents/guardians provided For the CL parameter, the allometric model was:
informed consent for the anonymous use of their children’s clinical and bio-
logical data for biomedical research at the time of data collection.  0:75
WTi
CLi ¼ hCL 
70
Patients and data collection
A retrospective observational study was conducted on immunocomprom- in which CLi is clearance in the ith individual with body weight WTi and hCL is
ised children admitted to the paediatric Immunology-Haematology unit of the typical value of the clearance in a standardized adult with a body
Necker Hospital (Paris, France) who were receiving cidofovir for adenovirus weight of 70 kg. The minimal value of the objective function was used to
and/or cytomegalovirus infections between 2014 and 2016. Plasma sam- discriminate between hierarchical models (P " 0.05, v2 distribution, 1 de-
ples were collected within the context of routine drug monitoring, which gree of freedom).
was performed to look for potential major underexposure or overexposure
to cidofovir. Because of the absence of such results, and the absence of con-
Covariate model
sensus on therapeutic drug monitoring of this drug in paediatrics, no dose
adjustment was performed in the children. Collection of samples was pro- Firstly, the association between available physiological covariates at base-
posed at 1, 3 and 6 h after drug intake. For each patient, we retrospectively line (sex, age, body weight, creatinine clearance and probenecid dose) and
reviewed the demographic, clinical, therapeutic and routine biological data pharmacokinetics parameters were investigated graphically. The empirical
(including cytomegalovirus viraemia, adenovirus viral load in blood and Bayes estimates of the individuals’ pharmacokinetics parameters obtained
stool, whole-blood count, lymphocyte T count, creatininaemia, kalaemia, from the base model were plotted against each covariate. They were then
calcaemia, phosphoraemia, glycosuria and b2-microglobulinuria). tested using a stepwise covariate model approach. The relationship be-
Weekly monitoring of adenovirus PCR and cytomegalovirus PCR in whole tween continuous covariates and pharmacokinetics parameters was mod-
blood was performed in HSCT recipients with at least one risk factor for life- elled using linear, exponential and power functions with the covariates
threatening viral infection [T cell depletion, unrelated donor, cord blood centred at their median values. The influence of categorical covariates was
transplantation, grade III or IV graft-versus-host disease (GVHD) and se- tested according to the equation:
vere lymphopenia], as recommended by the current European guidelines
2011.3 Similar weekly virological monitoring was performed in our unit in P ¼ h1  ½1 þ h2  cov
children with severe inherited immune deficiencies that confer high suscep-
tibility to severe viral infections. Adenovirus and cytomegalovirus viral loads where h1 represents the typical value of pharmacokinetics parameter
in whole blood, stools or nasopharyngeal samples were measured with the P with covariate value equal to 0 and h2 represents is the estimated

2 of 8
Downloaded from https://academic.oup.com/jac/advance-article-abstract/doi/10.1093/jac/dky192/5026318
by Miami University user
on 07 June 2018
 Paediatric model of cidofovir population pharmacokinetics JAC
influential factor in the pharmacokinetics parameter with covariate value CLCR on CL was estimated at 0.133 with poor precision and was not
equal to 1. The decision to include the covariate in the model was based on statistically significant (DOFV " 0). Covariate analysis showed that
a combination of statistical significance, mechanistic plausibility and the Vc could be influenced by body weight (DOFV " #5, with a reduc-
clinical relevance of the relationship.10
tion of 23% in inter-individual variability of Vc). This relationship
was modelled by a power function; the exponent for the effect
Model evaluation was estimated at 0.88. However, the confidence interval of the
The performance of the final model was evaluated on both graphical and estimated influential factor for the body weight was wide and
statistical criteria: objective function value, precision of parameter estimation ranged close to zero. Consequently, the simplest model was
[relative standard error (RSE) ,30%] and goodness-of-fit plots (population chosen and none of the covariates were retained. The pharmaco-
predictions versus the observed cidofovir plasma concentrations, conditional kinetics parameter estimates from the final model and their 95%
weighted residuals versus population predictions and time, and individual CIs are summarized in Table 2.
predictions versus cidofovir plasma concentrations). The 95% CI of the
The performance of the final model fits is represented by
parameters was assessed by the bootstrap method; 1000 bootstrap samples
were generated by resampling of the original data set. To examine the
goodness-of-fit plots in Figure 1. Figure 1(a and b) show agree-
agreement between the observed data and prediction intervals derived from ment between observed and model-predicted plasma cidofovir
the simulated data, a prediction-corrected visual predictive check (pcVPC) concentrations. Conditional weighted residual concentrations ver-
was performed by simulating 1000 replicates based on the final model. sus population-predicted concentrations (Figure 1c) and condition-
al weighted residuals concentrations versus time (Figure 1d)
Concentrations/effect relationships showed acceptable model performance: 97.6% of the values were
within +2 SD.11,12
AUC0–24 for the first dose was obtained from the estimated individual
The pcVPCs showed good overlap of the observed data and
pharmacokinetics parameters for each patient. In viraemic patients, viro-
logical success was defined as a decrease in the viraemia of at least
prediction concentration profiles (Figure 2).13
1 log10 copies/mL on day 15 after the first cidofovir infusion. For patients
with adenovirus and cytomegalovirus co-infections, virological success was Virological follow-up
defined as a decrease of 1 log10 copies/mL of both viremias.
Among eight children with adenovirus viraemia, six (75.0%)
A Wilcoxon test was performed on the overall population in order to as-
sess the association between cidofovir AUC0–24 and virological success.
reached virological success on day 15 (Figures S3–S5, S7, S8,
A Fisher’s exact test was then used to test the association between an S11–S13). The median duration to obtain undetectable viraemia
AUC0–24 cut-off value and efficacy outcome in the subgroup of adenovirus- was 11.3 days (range 5–30 days). None of these children had late
viraemic children. viral rebound. Of the two patients (patients #8 and #12) with
adenovirus viraemia and initial virological failure, one child (patient
Results #8) reached late undetectable viraemia and the remaining one
(patient #12) remained viraemic under the treatment. Four chil-
Demographic data dren had asymptomatic stool and/or nasopharyngeal adenovirus
Sixteen virological events occurred in 13 patients: eight adeno- carriage without viraemia at the time of cidofovir initiation. None
virus infections/diseases, four cases of asymptomatic adenovirus of them developed adenovirus viraemia during cidofovir treat-
carriage without viraemia and four cytomegalovirus infections/ ment. All four of these children cleared viral carriage within the
diseases. The baseline characteristics of the patients are sum- month following the first cidofovir infusion (Figures S2, S6, S9, S10).
marized in Table 1 and Table S1 (available as Supplementary Four children had cytomegalovirus viraemia at the time of cido-
data at JAC Online). All of them suffered from immune deficien- fovir initiation (Figures S1, S2, S7 and S13). Two of them had con-
cies and all but one had received HSCT. Their median age was comitant adenovirus viraemia and a third presented with
4 years (range 1–12 years). The median body weight was 14 kg concomitant adenovirus asymptomatic stool carriage. Among chil-
(IQR 10–20.3 kg). Overall, 86 results of cidofovir plasma levels dren with cytomegalovirus viraemia, one (25%) reached virological
were collected for the pharmacokinetics analysis. On average, success (patient #2). In this patient, undetectable viraemia was
two occasions per patient were available. obtained during cidofovir treatment. Among the three cytomegalo-
virus-viraemic patients with virological failure, one (33.3%) reached
late viraemia suppression on cidofovir treatment (patient #13).
Cidofovir population pharmacokinetics model
Two children had concomitant (patient #7) or consecutive
A two-compartment model with first-order absorption and elimin- (patient #13) cytomegalovirus and adenovirus viraemia. Both of
ation adequately described the data. The parameters of this model them reached virological success for the adenovirus infection but
were: elimination clearance (CL), intercompartmental clearance cidofovir failed to control the cytomegalovirus viraemia. The third
(Q), and central (Vc) and peripheral (Vp) volumes of distribution. co-infected child (patient #2) successfully controlled cytomegalo-
The residual variability was modelled using a proportional error virus viraemia and cleared adenovirus stool carriage on cidofovir.
model. The inter-individual variability of Q and Vp were not
retained owing to the poor precision of the estimate. A model
including the inter-occasion variability was unstable and the Association between drug concentrations and
weight-based allometric scaling did not improve the fit [change in viral replication
objective function value (DOFV) " 0]. In patients with adenovirus and/or cytomegalovirus viraemia at
For the covariate analysis, a power function best described the the time of cidofovir initiation, the median AUC0–24 was 21 mgh/L
correlation between CL and CLCR. The exponent for the effect of (range 7–73 mgh/L). Patients with virological success displayed a

3 of 8
Downloaded from https://academic.oup.com/jac/advance-article-abstract/doi/10.1093/jac/dky192/5026318
by Miami University user
on 07 June 2018
 Table 1. Patient characteristics at baseline

on 07 June 2018
4 of 8
Lymphocyte
GVHD Time interval count at Other
Patient Ex vivo at the time between HSCT the time of No. of Duration of Reason concomitant

by Miami University user

(age, Underlying T cell Conditioning of first ADV CMV and first CDV CDV initiation CDV CDV treatment for CDV antiviral
Neant et al.

years) disease Donor depletion regimen before HSCT CDV infusion infection infection infusion (days) (cells/mm3) infusions (days) interruption drugs

1 (1) Wiskott–Aldrich NA NA NA NA no CMV disease (West NA 400 6 46 failure foscarnet !

syndrome syndrome) ganciclovira
IVIR " 5 log10
copies/mL
2b (3) combined MUD no busulfan ! digestive (grade IV), asymptomatic 145 1100 2 14 success foscarnet !
immune fludarabine ! skin, pancreatic viraemia ganciclovira
deficiency ATG and pulmonary IVIR " 3.9 log10
copies/mL
2b asymptomatic stool 145 1100 2 14 clearance of foscarnet !
carriage no viral carriage ganciclovir
viraemia
3 (5) familial MUD no fludarabine ! skin (grade I) ADV disease (poly- no 57 100 3 21 success none
haemophagocytic melphalan pnoea/fever/
lymphohistiocytosis ! alemtuzumab abdominal pain)
IVIR " 5.2 log10
copies/mL)
4 (11) CD40 ligand MUD no busulfan ! skin (grade III) ADV disease (fever/ no 16 100 2 14 success aciclovir
deficiency fludarabine ! diarrhoea)
ATG IVIR " 4.6 log10
copies/mL
5 (8) Shwachman MMUD yes busulfan ! no ADV infection (vomit- no 23 200 2 14 success aciclovir
disease (CD19 fludarabine ! ing/diarrhoea)
and TCR) thiotepa ! ATG IVIR " 3.6
log10 copies/mL
6 (4) MHC class II MFD no busulfan ! skin, hepatic, asymptomatic stool no 13 0 2 14 clearance of aciclovir
expression fludarabine ! digestive carriage no viral carriage
deficiency ATG (grade II) viraemia
7b (1) combined MMFD yes busulfan ! no asymptomatic 14 0 3 21 success foscarnet !

Downloaded from https://academic.oup.com/jac/advance-article-abstract/doi/10.1093/jac/dky192/5026318

immune (CD45RA) fludarabine ! viraemia ganciclovir
deficiency ATG IVIR " 4 log10
copies/mL
7b CMV disease 14 0 7 97 failure foscarnet !
(chorioretinitis) ganciclovira
IVIR " 5.3 log10
copies/mL
8 (1) MHC class II MUD no fludarabine ! no asymptomatic vir- no 14 0 3 21 failure none
expression cyclophosphamide aemia
deficiency IVIR " 2.8 log10
copies/mL
9 (6) GM-CSF-receptor MUD no busulfan ! skin (grade II), asymptomatic naso- no 77 400 4 46 clearance of aciclovir
deficiency fludarabine ! digestive pharyngeal and viral carriage
ATG (grade IV) stool carriage no
viraemia
10 (1) CD40 ligand MUD no busulfan ! no asymptomatic stool no 24 100 2 14 clearance of aciclovir
deficiency fludarabine ! carriage no viral carriage
ATG viraemia
11 (3) combined MUD yes busulfan ! skin and ADV disease no 22 3800 2 14 success aciclovir
immune (CD45RA) fludarabine ! digestive (diarrhoea)
deficiency thiotepa ! IVIR " 5.2 log10
alemtuzumab copies/mL
 Paediatric model of cidofovir population pharmacokinetics JAC
ganciclovir‡

CDV, cidofovir; ADV, adenovirus; CMV, cytomegalovirus; GM-CSF, granulocyte-macrophage colony-stimulating factor; NA, not applicable; MUD, matched unrelated donor; MMUD, mis-

In the case of concomitant (children #2 and #7) or successive (child #13) cytomegalovirus and adenovirus infections in the same child, the characteristics and the evolution of each
Table 2. Population pharmacokinetic parameters of cidofovir from the

ganciclovir
valaciclovir !

valaciclovir !
final model
none

Estimated Bootstrap
Parameters shrinkage (%) RSE (%) median (95% CI)

CL (L/h) 3.2 14.3 3.1 (2.2–4.2)

and toxicity

Vc (L) 3.8 19.2 3.7 (2.0–5.6)

matched unrelated donor; MFD, matched family donor; MMFD, mismatched family donor; ATG, anti-thymocyte globulins; IVIR, initial viraemia; TCR, T cell receptor.
Vp (L) 2 7.7 5 (1.8–18)
success
failure
death

Q (L) 1.3 29 1.4 (0.3–3.9)

IIV CL (%) 45 (0.1) 41.6 42 (15–61)
IIV Vc (%) 51 (22) 69.1 55 (10–123)
35

Residual error (%) 42 36 41 (27–56)

Q, intercompartmental clearance; IIV CL, inter-individual variability on

5

CL; IIV Vc, inter-individual variability on Vc.

0

200

100

higher median AUC0–24 compared with patients with failure: 48.6

(range 8.9–72.6) versus 19.1 (6.9–22.7) mgh/L but this did not
reach statistical significance (P " 0.22) (Figure 3). However, in the
subgroup of adenovirus-viraemic patients (n " 8), all children with
67

55

98

virological success had an AUC0–24 value above 19.1 mgh/L

whereas all children with failure displayed an AUC0–24 value below
These drugs were initiated prior to the initiation of cidofovir and failed to control the cytomegalovirus viraemia.
IVIR " 3.7 log10

this threshold (Fisher’s exact test, P " 0.03). Such correlation could
asymptomatic

copies/mL

not be estimated for cytomegalovirus infections owing to the small

viraemia

number of cytomegalovirus-viraemic patients.

ADV disease (respira- no

no

Evolution of immune deficiency during

tory and digestive)
IVIR " 4.6 log10

IVIR " 4.6 log10

cidofovir treatment
asymptomatic
copies/mL

copies/mL

Eight out of 13 children (61.5%) had an increase in T cell count dur-

viraemia

ing cidofovir treatment. Among the 10 cytomegalovirus and/or

adenovirus-viraemic patients, the lymphocyte count increased
no

during cidofovir treatment in 7 (70%) children, including 5 patients

with virological success. Among the three remaining children with
hepatic and
skin, digestive,

(grade IV)

a stable lymphocyte count during cidofovir treatment, two

ocular

reached virological success.

no

Safety of cidofovir
No increase in creatininaemia was observed during or after the
cidofovir treatment (Table S1). Three children previously had a
fludarabine !

fludarabine !

tubulopathy, which did not worsen during cidofovir treatment.

thiotepa !

One patient with low cidofovir AUC0–24 levels (18.8 mgh/L) devel-
busulfan !

busulfan !

oped a tubulopathy after two cidofovir infusions (patient #13). We

ATG

ATG

viral infection are shown in a separate row.

did not observe cidofovir-related haematotoxicity.

(CD45RA)

Discussion
yes

no

As far as we are aware, this study is the first retrospective popula-

MUD

MUD

tion pharmacokinetics study of cidofovir in children with data

obtained in routine clinical practice. Few published data describe
cidofovir pharmacokinetics in paediatric patients.6 Because cidofo-
erythropoietic

vir is often prescribed as a first-line agent for adenovirus infections

deficiency

porphyria
12 (7) CD40 ligand

and as a second-line agent for cytomegalovirus infections in

13b (1) congenital

paediatric recipients of HSCT, a better knowledge of cidofovir

pharmacokinetics is required. The most adequate model was a
two-compartment model as reported in the literature.6 Estimated
13b

typical values for pharmacokinetics parameters in the final model

a
b

5 of 8
Downloaded from https://academic.oup.com/jac/advance-article-abstract/doi/10.1093/jac/dky192/5026318
by Miami University user
on 07 June 2018
 Neant et al.

(a) Individual predictions versus concentrations (b) Population predictions versus concentrations

Observed concentration (mg/L) 30 30

Observed concentration (mg/L)

25 25

20 20

15 15

10 10

5 5

5 10 15 20 25 30 5 10 15 20 25 30
Individual predicted concentration (mg/L) Population predicted concentration (mg/L)

(c) Conditional weighted residuals versus (d) Conditional weighted residuals versus time
population predictions

2 2
CWRES

CWRES

0 0

–2 –2

0 5 10 15 20 25 0 5 10
Population predicted concentration (mg/L) Time (h)

Figure 1. Goodness-of-fit plots for the final model. (a) Observed concentrations versus individual predicted concentrations. (b) Observed concentra-
tions versus population predicted (PRED) concentrations. (c) Conditional weighted residuals (CWRES) versus PRED. (d) CWRES versus time after dose
(h). The bold black broken lines display the trend of the data, the black broken lines indicate the expected trends, and the black circles represent the
observed data.

were as follows: CL " 3.2 L/h, Vc " 3.8 L, Vp " 2 L, Q " 1.3 L/h and clinically relevant thus this covariate was not retained in the final
inter-individual variability of CL and Vc were 45% and 51% respect- model. In a previous paediatric study, no effect of covariates was
ively. Pharmacokinetics parameters were estimated with good found.6 However, the influence of probenecid administration on
precision (RSE , 30%). The estimated cidofovir t1=2 of 2 h was cidofovir pharmacokinetics has been described in a pharmacokin-
shorter than the previously described value (5.9 h) in children etics study in an adult population. Indeed, a high dose of probene-
using standard approaches (two-stage methods with a two- cid was significantly associated with both a lower CL and volume
compartment model).6 However, the estimated cidofovir terminal at steady-state.4 Owing to sparse data, the covariate analysis car-
t1=2 of 2 h (95% CI 1.7–2.3 h) was most similar to the value reported ried out was limited. It was difficult to fully explain the variability
in a pharmacokinetic study in adults (mean + SD 2.3+0.5 h).4,5 of cidofovir pharmacokinetics. Consequently, we cannot reach
The allometric model, with fixed power exponents of 1 and conclusions on the effect of weight on the pharmacokinetics
0.75 for volume and CL terms, respectively, is often used in popula- parameters. In this study, it was not possible to define dose
tion pharmacokinetics studies in paediatrics but in our model it did recommendations.
not improve the results.9 The effect of body weight on Vc was the In our study, the median AUC0–24 observed for the first dose was
only significant covariate; a higher bodyweight was associated 21 mgh/L (range 7–73 mgh/L). For adenovirus infection, all children
with a higher Vc. However, the effect size was judged to be not with virological success had an AUC0–24 threshold above 19.1 mgh/L,

6 of 8
Downloaded from https://academic.oup.com/jac/advance-article-abstract/doi/10.1093/jac/dky192/5026318
by Miami University user
on 07 June 2018
 Paediatric model of cidofovir population pharmacokinetics
Cidofovir plasma concentrations (mg/L) JAC
40

60
30

AUC (mg·h/L)
20
40

10

0 20
2 4 6 8 10 12
Time (h)

Figure 2. pcVPCs of the final model. The black circles represent the
observed data. The black broken lines indicate the 5th and 95th percen- Failure Success
tiles of the observed data and the black continuous line indicates the Figure 3. AUC0–24 box plot as a function of virological efficacy. The boun-
50th percentile of the observed data. daries of the box indicate the first (Q1) and third (Q3) quartiles, respect-
ively. The line inside the box plot represents the median. The whiskers
link the box with Q1!1.5 % IQR and Q3!1.5 % IQR.
consequently this value was chosen to be associated with efficacy.
This result must be interpreted carefully because (i) our analysis was
performed with the data obtained from a limited number of patients; viraemia during cidofovir treatment. These results are in line with
and (ii) most of the adenovirus-viraemic patients with virological suc- the suggestion by Lion et al.21,22 that systematic screening of
cess had a concomitant increase in their lymphocyte counts, which adenovirus intestinal excretion should be part of routine testing
could contribute to control of the viral infection. during the post-HSCT period in children, in order to discuss antiviral
Conventional pharmacokinetics measurements do not accur- treatment in patients with high adenovirus carriage with the aim
ately reflect the duration of action of cidofovir since the antiviral of preventing invasive infection. However, this indication is contro-
effect is dependent on the intracellular concentrations of the ac- versial because the presence of adenovirus in stool specimens is a
very common finding during the post-transplant period and most
tive phosphorylated metabolites within cells. As noted above,
of these patients remain asymptomatic and clear the virus without
these metabolites (whose intracellular concentrations were not
the use of antiviral treatment.23 Current recommendations for
evaluated in our study) have a long intracellular t1=2 (48 h for the
adenovirus screening and monitoring as a basis for preemptive
HPMPCp-choline adduct14) which may contribute to the prolonged
treatment in patients at high risk for adenovirus disease are still
antiviral action of cidofovir.15
relatively diverse and further studies are needed to provide reliable
We herein described two cases of late viral suppression in
data permitting the establishment of standardized approaches.
patients (one adenovirus- and one cytomegalovirus-viraemic child)
Some factors may disturb the interpretation of the correlation
with initial virological failure at day 15. We could thus discuss the
between pharmacokinetics and virological follow-up. These in-
relevance of the criteria defining virological success during cidofovir clude antiviral and immune-suppressive co-medications and con-
treatment. We used here the criteria used by Ganapthi et al. and current immune reconstitution. In two patients, virological success
other studies16–18 (viraemia decrease of at least 1 log10 copies/mL was observed without concomitant change of the absolute
at day 15). However, we cannot exclude that this strict criterion lymphocyte count (Supplementary data). These data contrast
may underestimate the rate of cidofovir-related virological suc- with some other studies concluding that cidofovir alone is unable
cess, especially in cases with high initial viraemia and profound to mediate complete resolution of adenovirus viraemia in the ab-
underlying immune deficiency. Indeed, owing to underlying im- sence of immune recovery.24
munosuppression and corticosteroid use, a modest quantitative in- Toxicity is the major concern in the utilization of cidofovir. In this
crease in viral loads (or antigenaemia) during the first 2 weeks of study, no increase in creatinaemia was observed in patients.
antiviral therapy may occur in some patients, as previously sug- Moreover, only one of them developed cidofovir-induced tubulop-
gested in the case of some cytomegalovirus-infected HSCT recipi- athy. These results are in line with the review published by Ganapathi
ents receiving ganciclovir preemptive treatment.19,20 Thus, we et al., in which no significant nephrotoxicity (defined by an elevation
suggest that, in HSCT recipients with profound immune deficiencies of 50% of the baseline creatininaemia) was described in 9/14 paedi-
and adenovirus and/or cytomegalovirus infections, less strict cut- atric studies. In the remaining five paediatric studies, in which
off values may be used to define refractory infections, in order not nephrotoxicity was observed in 9% to 43% of patients, the duration
to underestimate the rate of virological success. of cidofovir treatment was much longer than in our patients.17
In three non-viraemic patients with isolated adenovirus stool There is no approved antiviral agent for the treatment of
carriage after HSCT, cidofovir was initiated in order to prevent adenovirus infection. Brincidofovir may be a promising antiviral
adenovirus disease. None of these children developed adenovirus drug and has in vitro and in vivo activity against adenovirus.18

7 of 8
Downloaded from https://academic.oup.com/jac/advance-article-abstract/doi/10.1093/jac/dky192/5026318
by Miami University user
on 07 June 2018
 Neant et al.

Infusion of donor memory T lymphocytes directed specifically 6 Caruso Brown AE, Cohen MN, Tong S et al. Pharmacokinetics and safety
against immunodominant viral antigens is one of the therapeutic of intravenous cidofovir for life-threatening viral infections in pediatric
options in cytomegalovirus and adenovirus infections. Recent hematopoietic stem cell transplant recipients. Antimicrob Agents
Chemother 2015; 59: 3718–25.
studies showed that rapidly prepared cytomegalovirus- and/or
7 Momper JD, Zhang S, Randhawa PS et al. Determination of cidofovir in
adenovirus-specific T cells seem efficient in a subset of HSCT recipi-
human plasma after low dose drug administration using high-performance
ents with severe infections, as one-third of the patients showed a
liquid chromatography-tandem mass spectrometry. J Pharm Biomed Anal
complete virological response in parallel with specific T cell expan- 2010; 53: 1015–21.
sion even in the presence of significant corticotherapy. However, 8 Aho AV, Ullman JD. Foundations of Computer Science. Computer Science
the safety of this approach was not adequately assessed and Press, 1992. http://cds.cern.ch/record/240851.
larger studies are needed to assess the clinical and biological 9 Anderson BJ, Holford NHG. Mechanism-based concepts of size and matur-
parameters associated with virological failure and success.25 ity in pharmacokinetics. Annu Rev Pharmacol Toxicol 2008; 48: 303–32.
10. Bonate PL. Pharmacokinetic-Pharmacodynamic Modeling and Simulation.
Conclusions Berlin: Springer, 2011. http://www.springer.com/br/book/9781441994844.
Cidofovir appears safe and reasonably tolerated for the treatment 11 Karlsson MO, Savic RM. Diagnosing model diagnostics. Clin Pharmacol
Ther 2007; 82: 17–20.
of adenovirus and cytomegalovirus infections in paediatric allo-
12 Hooker AC, Staatz CE, Karlsson MO. Conditional weighted residuals
geneic HSCT and seemed to have antiviral efficacy in a subset of
(CWRES): a model diagnostic for the FOCE method. Pharm Res 2007; 24:
patients. There is no current recommendation for pharmacokinet-
2187–97.
ics surveillance in cidofovir treatment, but our retrospective study
13 Bergstrand M, Hooker AC, Wallin JE et al. Prediction-corrected visual
suggests that (i) therapeutic drug monitoring of cidofovir may be predictive checks for diagnosing nonlinear mixed-effects models. AAPS J
recommended in children; and (ii) in the case of adenovirus infec- 2011; 13: 143–51.
tion, targeting an AUC0–24 above 19 mgh/L is associated with 14 Connelly MC, Robbins BL, Fridland A. Mechanism of uptake of the
higher probability of virological success. Larger prospective studies phosphonate analog (S)-1-(3-hydroxy-2-phosphonylmethoxy-propyl)cyto-
are needed to confirm the correlation between cidofovir AUC0–24 sine (HPMPC) in vero cells. Biochem Pharmacol 1993; 46: 1053–7.
and cidofovir virological efficacy. 15 De Clercq E. Clinical potential of the acyclic nucleoside phosphonates
cidofovir, adefovir, and tenofovir in treatment of DNA virus and retrovirus
infections. Clin Microbiol Rev 2003; 16: 569–96.
16 Grimley MS, Chemaly RF, Englund JA et al. Brincidofovir for asymptomatic
Funding adenovirus viremia in pediatric and adult allogeneic hematopoietic cell trans-
This study was carried out as part of our routine work.
plant recipients: a randomized placebo-controlled Phase II trial. Biol Blood
Marrow Transplant 2017; 23: 512–21.
17 Ganapathi L, Arnold A, Jones S et al. Use of cidofovir in pediatric patients
Transparency declarations with adenovirus infection. F1000Res 2016; 5: 758.
P. F. is a member of the MSD France advisory board for letermovir, and in- 18 Hiwarkar P, Amrolia P, Sivaprakasam P et al. Brincidofovir is highly effica-
vestigator of the Phase 3 study SHP620–303 of maribavir (Shire cious in controlling adenoviremia in pediatric recipients of hematopoietic cell
ViroPharma, Incorporated). All other authors: none to declare. transplant. Blood 2017; 129: 2033–7.
19 Gerna G, Lilleri D, Zecca M et al. Rising antigenemia levels may be mis-
leading in pre-emptive therapy of human cytomegalovirus infection in
allogeneic hematopoietic stem cell transplant recipients. Haematologica
Supplementary data 2005; 90: 526–33.
Table S1 and Figures S1 to S13 are available as Supplementary data at
20 Park SY, Lee SO, Choi SH et al. Paradoxical rising cytomegalovirus antige-
JAC Online.
nemia during preemptive ganciclovir therapy in hematopoietic stem cell
transplant recipients: incidence, risk factors, and clinical outcomes. J Clin
Microbiol 2011; 49: 4179–84.
References 21 Lion T. Adenovirus infections in immunocompetent and immunocom-
1 Hierholzer JC. Adenoviruses in the immunocompromised host. Clin promised patients. Clin Microbiol Rev 2014; 27: 441–62.
Microbiol Rev 1992; 5: 262–74. 22 Lion T, Kosulin K, Landlinger C et al. Monitoring of adenovirus load in stool
2 Morfin F, Dupuis-Girod S, Mundweiler S et al. In vitro susceptibility of adeno- by real-time PCR permits early detection of impending invasive infection in
virus to antiviral drugs is species-dependent. Antivir Ther 2005; 10: 225–9. patients after allogeneic stem cell transplantation. Leukemia 2010; 24: 706–14.
3 Matthes-Martin S, Feuchtinger T, Shaw PJ et al. European guidelines for diag- 23 Baldwin A, Kingman H, Darville M et al. Outcome and clinical course of
nosis and treatment of adenovirus infection in leukemia and stem cell trans- 100 patients with adenovirus infection following bone marrow transplant-
plantation: summary of ECIL-4 (2011). Transpl Infect Dis 2012; 14: 555–63. ation. Bone Marrow Transplant 2000; 26: 1333–8.
4 Cundy KC, Petty BG, Flaherty J et al. Clinical pharmacokinetics of cidofovir 24 Heemskerk B, Lankester AC, van Vreeswijk T et al. Immune reconstitution
in human immunodeficiency virus-infected patients. Antimicrob Agents and clearance of human adenovirus viremia in pediatric stem-cell recipients.
Chemother 1995; 39: 1247–52. J Infect Dis 2005; 191: 520–30.
5 European Medicines Agency. Vistide: résumé des caractéristiques du pro- 25 Creidy R, Moshous D, Touzot F et al. Specific T cells for the treatment of
duit. http://www.ema.europa.eu/docs/fr_FR/document_library/EPAR_-_Produ cytomegalovirus and/or adenovirus in the context of hematopoietic stem cell
ct_Information/human/000121/WC500052072.pdf. transplantation. J Allergy Clin Immunol 2016; 138: 920–4.

8 of 8
Downloaded from https://academic.oup.com/jac/advance-article-abstract/doi/10.1093/jac/dky192/5026318
by Miami University user
on 07 June 2018