Prediction of Area under the Cyclosporine Concentration
Versus Time Curve in Children Undergoing Hematopoietic
Stem Cell Transplantation
L. Lee Dupuis
1
,2
,3
,4
,*
, Winnie Seto
1
,4
, Oliver Teuffel
2
, Paul Gibson
7
,
Kirk R. Schultz
8
, John D. Doyle
2
,5
, Adam Gassas
2
,5
, R. Maarten Egeler
2
,3
,5
,
Lillian Sung
2
,3
,6
, Tal Schechter
2
,3
,6
1Department of Pharmacy, SickKids, Toronto, Ontario, Canada 2Division of Haematology/Oncology, SickKids, Toronto, Ontario, Canada 3Research Institute, SickKids, Toronto, Ontario, Canada
4Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada 5Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada
6Department of Paediatrics, University of Toronto, Toronto, Ontario, Canada
7Pediatric Haematology/Oncology, Children’s Hospital, London Health Sciences Centre, London, Ontario, Canada 8Division of Pediatric Hematology/Oncology/BMT, BC Children’s Hospital, University of British Columbia, Vancouver, British
Columbia, Canada Article history: Received 31 August 2012 Accepted 28 October 2012 Key Words: Cyclosporine Pediatrics Pharmacokinetics Area under the curve Acute graft versus host disease
a b s t r a c t
This prospective study aimed to validate a previously developedfirst-dose limited sampling strategy (LSS) to predict the area under the cyclosporine concentration-versus-time curve (AUC) and to develop and then validate an LSS to predict cyclosporine AUC at steady state. This two-center Canadian study included children (ages .4 to 17.2 years) undergoing myeloablative allogeneic hematopoietic stem cell transplantation receiving cyclosporine for acute graft-versus-host disease prophylaxis. There were three cohorts, each incorporating 24 AUC determinations:first-dose LSS validation, steady-state LSS development, and steady-state LSS vali-dation. Patients contributing data to either of the development cohorts were excluded from the corre-sponding validation group. Cyclosporine was given every 12 hours as a 2-hour infusion. Cyclosporine AUC was determined after administration of thefirst cyclosporine dose (8 samples) and then once weekly (9 samples) until engraftment. Steady-state LSSs were developed using stepwise multiple linear regression. An LSS was considered to provide an acceptable estimate of AUC if the lower limit of the 95% confidence limit (CL) of the intraclass coefficient was .8 or higher and both bias and precision were 15% or less. Fifty-three children age .4 to 18 years participated. Cyclosporine concentrations drawn up to 4 hours from the start of the infusion correlated most strongly with AUC. The previously developedfirst-dose LSSs and three steady-state LSSs met criteria for acceptability. The intraclass coefficients of the three-pointfirst-dose LSS validation cohort, three-point steady-state LSS development cohort, and three-three-point steady-state LSS validation cohort were .974 (95% CL: .941 to .988), .984 (95% CL: .965 to .993), and .993 (95% CL: .984 to .997), respectively. The three-point first-dose (2, 6, and 8 hours) and steady-state (2, 2.5, and 8 hours) LSSs are valid measures of cyclosporine AUC after intravenous administration over 2 hours. Their use in a prospective evaluation of the relationship between cyclosporine AUC and hematopoietic stem cell transplantation clinical outcomes in children is suggested.
Ó2013 American Society for Blood and Marrow Transplantation.
INTRODUCTION
Acute graft-versus-host-disease (aGVHD) is the primary
cause of treatment-related mortality after hematopoietic
stem cell transplantation (HSCT) and has a great impact on
morbidity and quality of life
[1]
. Calcineurin inhibitors
(cyclosporine or tacrolimus) in combination with
short-course methotrexate or corticosteroids are most commonly
used for aGVHD prevention. Because inter- and
intra-individual variability in cyclosporine disposition is high,
the whole blood trough concentration (C
0) is routinely used
to individualize cyclosporine doses in HSCT patients.
However, evidence to support a relationship between C
0values and aGVHD prevention is limited and highly variable.
For example, reported target C
0values range from 85 to
500 ng/mL
[2-6]
. As a result, there is no consensus regarding
the optimal target C
0range, and targets vary widely among
institutions
[7]
.
Prevention of acute graft rejection correlates more
strongly with area under the cyclosporine
concentration-versus-time curve (AUC) than C
0in both adult and pediatric
solid organ transplant patients
[8,9]
. Cyclosporine dose
adjustment based on AUC rather than C
0has consequently
become the standard of care. This approach has only just
begun to be evaluated in HSCT
[10-19]
. Classical
determina-tion of AUC entails a large total blood sample volume, high
workload to obtain the samples, high assay costs, and
potentially a high level of patient inconvenience or distress.
Blood volume is especially a concern in pediatrics. A valid
limited sampling strategy (LSS) would facilitate large-scale
evaluations of the relationship between cyclosporine AUC
and clinical outcomes such as aGVHD prophylaxis in
pediatrics.
We described the relationship between AUC and whole
blood cyclosporine concentrations at various single time
American Society for Blood
ASBMT
and Marrow TransplantationFinancial disclosure:See Acknowledgments on page 422.
*Correspondence and reprint requests: L. Lee Dupuis, MScPhm, Department of Pharmacy, SickKids, 555 University Ave., Toronto, Ontario, Canada, M5M 1X3.
E-mail address:lee.dupuis@sickkids.ca(L.L. Dupuis).
1083-8791/$esee front matterÓ2013 American Society for Blood and Marrow Transplantation. http://dx.doi.org/10.1016/j.bbmt.2012.10.031
points within a dosing interval after the
fi
rst intravenous
(i.v.) cyclosporine dose given as a 2-hour infusion in 24
children undergoing HSCT
[20]
. Strong correlations between
AUC and each of the concentrations obtained within the
fi
rst
4 hours after the start of the infusion (2, 2.5, 3, and 4 hours)
(Spearman rho coef
fi
cient
¼
.717 to .868) were observed
but only a moderate relationship (Spearman rho
coef-fi
cient
¼
.442 to .553) between AUC and the concentrations
obtained at the remaining time points. The correlation
between C
0and AUC was moderate (Spearman rho
coef
fi
cient
¼
.457). Using these data, we developed an LSS for
the prediction of AUC achieved in children after
adminis-tration of the
fi
rst i.v. cyclosporine dose. An LSS for prediction
of cyclosporine AUC after steady-state cyclosporine i.v.
administration has not yet been developed.
The primary purpose of this prospective study was to
validate the LSS previously developed to estimate
cyclo-sporine AUC after the
fi
rst i.v. cyclosporine dose and to
develop and validate an LSS for use once steady state was
achieved in children undergoing HSCT. These LSSs will be
used in future studies of the relationship between HSCT
clinical outcomes and cyclosporine AUC and cyclosporine
concentrations at time points other than C
0.
METHODS
This study was approved by the Research Ethics Boards of both SickKids and BC Children’s Hospital. Patients or guardians provided informed consent and/or assent.
Study Population
Patients less than 18 years of age who were undergoing allogeneic myeloablative HSCT at either SickKids or Women’s and Children’s Hospital of British Columbia and who were scheduled to receive cyclosporine for aGVHD prophylaxis were eligible to participate. Children who received voriconazole or posaconazole within 14 days of thefirst cyclosporine dose, or at any time until engraftment, were excluded. Patients receivingfl uco-nazole were included in this study becausefluconazole does not alter the disposition of i.v. cyclosporine[21,22].
Cyclosporine Administration, Blood Sampling, and Dose Adjustment
Cyclosporine was administered as a 120-minute (15 minutes) i.v. infusion (including the lineflush) every 12 hours starting on day3,2, or1 (where the day of HSCT is day 0). One lumen of a central venous access device was used consistently for cyclosporine dose administration. Cyclo-sporine dose adjustments were made by the clinical team caring for the patient using the blood samples obtained for clinical care in accordance with the institutional target C0concentrations (100 to 200 ng/mL).
All samples for cyclosporine concentration determination were ob-tained from a lumen of a central venous access device not used for cyclo-sporine administration. After thefirst i.v. cyclosporine dose, blood samples were obtained for determination of cyclosporine concentration at 2 (end of the infusion), 2.5, 3, 4, 6, 8, 10, and 12 hours after the start of the 2-hour infusion. Samples drawn within 10 minutes of the ideal sampling time were acceptable.
After day 0, blood samples were obtained for determination of cyclo-sporine concentration after the morning dose weekly on a standard day of the week (eg, every Monday) until engraftment (absolute neutrophil count>.5109/L on 2 consecutive measurements separated by more than 1
day) occurred. In total, 9 blood samples per week were obtained: just before the morning cyclosporine dose and at 2 (end of the infusion), 2.5, 3, 4, 6, 8, 10, and 12 hours after the start of the 2-hour infusion. Blood samples were obtained weekly for no longer than 7 weeks after transplantation.
Cyclosporine Assay
Whole blood cyclosporine concentrations were determined in a single laboratory using tandem mass spectrometry. The interday coefficient of variation of this assay is 3.6% to 4.2% (Warren Walsh, Resource Technologist, Department of Laboratory Medicine, SickKids, personal communication, October 27, 2009) Study samples were frozen at20C until assayed.
Data Analysis
Statistical analysis was performed using SAS Enterprise Guide (version 4.3, SAS Institute, Cary, NC) or SPSS (version 17, SPSS Inc., Chicago, IL). APvalue of .05 or less was considered to indicate statistical significance.
Pharmacokinetics
The cyclosporine AUC achieved over the 12-hour dosing interval was calculated using all data points (AUC-12) for all patients using non-compartmental methods (Pharsight Phoenix WinNonLin 6.0.0.1648, St Louis, MO). The relationship between AUC and whole blood cyclosporine concentrations at individual time points was assessed by the Spearman rho correlation coefficient.
First-Dose LSS Validation
The previously developedfirst-dose LSS[20]was validated using the cyclosporine concentrations obtained from thefirst 24 patients enrolled for whom 8 properly timed cyclosporine concentrations were available. Each patient’s AUC was calculated using the previously developed LSSs. Agree-ment between the AUC-12 and the AUC calculated using each LSS was assessed using a single-measure, one-way, random effect, intraclass coeffi -cient (ICC)[23,24]. Agreement was considered to be acceptable if the lower limit of the 95% confidence limit of the ICC was .8 or higher. The extent of agreement between AUC-12 and predicted AUC was also assessed using Bland-Altman analysis. The limits of agreement (mean % difference2 standard deviations, where % difference¼100[(AUC-12eAUC-LSS)/(mean of
AUC-12 and AUC-LSS)]) were calculated. It was not possible to determine the clinically significant limits of agreement a priori for cyclosporine because the clinical significance of AUC-12 with respect to HSCT outcomes is not yet established. Bias (mean prediction error) and precision (root mean squared error) of 15% or less were considered to be acceptable[25]. In summary, an LSS was considered to provide an acceptable estimate of AUC-12 if the lower limit of the 95% confidence limit of the ICC was .8 or higher and both bias and precision were 15% or less. The association between patient age and percent bias of the AUC estimated by the 3-point LSS was undertaken using linear regression analysis.
Steady-State LSS Development and Validation
Steady state was defined as having been achieved once the patient had received at least 4 consecutive, identical cyclosporine doses. Thefirst 24 patients enrolled for whom all 9 cyclosporine concentrations were available at steady state constituted the development group; the second group of 24 patients enrolled for whom 9 properly timed cyclosporine concentrations were available at steady state constituted the validation group. None of the patients in the validation group was included in the development group.
LSSs were developed at steady state using stepwise multiple linear regression analysis [25]. LSSs involving 5 or fewer data points were preferred for model parsimony and clinical utility. Adjusted r2values greater
than .9 were considered acceptable. Agreement between AUC-12 and the AUC calculated using each steady-state LSS was assessed using ICC and Bland-Altman analysis as described above, as were the criteria for accep-tance of a steady-state LSS as an estimate of AUC-12. The association between patient age and percent bias of the AUC estimated by the 3-point LSSs was undertaken using linear regression analysis.
RESULTS
From June 2008 to March 2012, 104 patients were
screened for study eligibility. Thirty did not meet study entry
criteria, 12 declined to participate, the attending physician of
5 patients asked that they not be approached, and 3 were
missed due to logistical problems. Fifty-four patients were
enrolled; 1 later withdrew before sampling at steady state
occurred. Of the patients who enrolled, blood sampling
deviated from protocol after administration of the
fi
rst dose
in 14 patients. Furthermore, blood sampling deviated from
protocol or cyclosporine dosing never achieved steady state
in 3 and 2 patients beyond the
fi
rst cyclosporine dose,
respectively. Data from 53 children are reported here.
Demographic data relating to the study patients are
pre-sented in
Table 1
. It is notable that mycophenolate mofetil
was more commonly used in the steady-state validation
cohort (7 of 24 patients) compared with the other cohorts
(0 or 1 of 24 patients). Fourteen patients contributed data to
both the
fi
rst-dose LSS validation group and the steady-state
LSS development group, whereas 7 contributed data to both
the
fi
rst-dose LSS validation group and the steady-state LSS
validation group.
First-Dose LSS Validation
Table 2
illustrates that whole blood cyclosporine
concentrations drawn within the
fi
rst 4 hours from the start
of the infusion correlated strongly with AUC-12 (Spearman
rho coef
fi
cient
¼
.66 to .91) Correlations with other time
points were moderate.
The four previously developed
fi
rst-dose LSS met the
criteria for acceptability (
Table 3
). That is, adjusted r
2was
>
.9, ICC indicated good agreement, and mean bias and
precision were
<
15%. In addition, Bland-Altman analysis
indicated good agreement between the AUC-12 and
pre-dicted AUC values (
Table 3
,
Figure 1
A). No signi
fi
cant
asso-ciation between patient age and percent bias of the AUC
estimates was observed (adjusted r
2¼
.0332). AUC-12 and
AUC values calculated using the LSSs are presented in
Table 4
.
Steady-State LSS Development and Validation
The mean day of sampling relative to day 0 was
day
þ
18.0
8.7 and day
þ
17.6
7.4 in the steady-state LSS
development and validation cohorts, respectively. Whole
blood cyclosporine concentrations drawn from 2 to 4 hours
from the start of the 2-hour infusion correlated most strongly
with AUC-12 (
Table 2
; Spearman rho coef
fi
cient
¼
.80609 to
.87913 [validation cohort] and .85409 to .95238
[develop-ment cohort]). Correlations with other time points were also
signi
fi
cant in both cohorts. However, although the
correla-tions were weak to moderate at time points beyond 4 hours
from the start of the 2-hour infusion in the development
cohort, correlations with AUC-12 were strong at all time
points in the validation cohort.
Three steady-state LSSs were developed that met criteria
for acceptability: the three-, four-, and
fi
ve-point LSSs
(
Table 3
,
Figure 1
B). That is, ICC indicated good agreement
between the AUC-12 value and that predicted by the three-,
four-, and
fi
ve-point LSSs, adjusted r
2was
>
.9, and both
mean bias and precision were
<
15%. The Bland-Altman
analysis also indicated good agreement. When evaluated in
the validation cohort, the 3 acceptable steady-state LSSs also
met criteria for acceptability and demonstrated good
agree-ment with AUC-12 values (
Table 3
,
Figure 1
C). No signi
fi
cant
association between patient age and percent bias of the AUC
estimates
was
observed
in
either
cohort
(adjusted
r
2¼
.0187 [development cohort] and
.0249 [validation
cohort]). AUC-12 and AUC values calculated using the LSSs
are presented in
Table 4
.
DISCUSSION
Using
complete
pharmacokinetic
pro
fi
les
collected
prospectively, we developed and validated LSSs for
predict-ing AUC of cyclosporine after administration as a 2-hour i.v.
infusion to children undergoing HSCT both after the
fi
rst
dose and at steady state. Speci
fi
cally, we found that a 3-point
LSS is a valid method for estimating cyclosporine AUC after
both the initial i.v. cyclosporine dose and at steady state. This
information will facilitate assessment of cyclosporine AUC by
clinicians and the assessment of the relationship, if any,
between cyclosporine AUC and HSCT outcomes in future
studies.
Cyclosporine dose adjustment using the trough whole
blood cyclosporine concentration (C
0) remains the standard
in HSCT, although the target concentration range is highly
variable among HSCT centers as stated previously. Several
investigators have reported cyclosporine AUC in HSCT
patients after i.v. administration
[3,10-14,16-19,26,27]
. The
method used to calculate or estimate cyclosporine AUC was
not described in some of these reports
[13]
, and serum rather
than whole blood concentrations (the accepted standard
matrix) were reported in another
[10]
. LSSs have been
developed and validated for estimation of AUC-12 in HSCT
patients older than 15 years receiving oral cyclosporine
[17]
.
The proposed LSSs had minimal bias. Agreement between
AUC-12 and the AUC estimates generated by the LSSs was not
evaluated. Most pediatric patients receive i.v. cyclosporine
Table 1Description of patients in First Dose Development (N¼24) and Validation (N¼24) Cohorts and Steady-State Development (n¼24) and Validation (n¼24) Cohorts
Parameter First-Dose Development Cohort First-Dose Validation Cohort Steady-State Development Cohort Steady-State Validation Cohort Mean age, yr,SD (range) 8.84.81 (.5-16.9) 8.84.89 (.4-17.2) 10.25.06 (.4-17.2) 9.33.54 (3.3-17.2) Day of sampling (no. days after HSCT) d d 18.08.70 (10-39) 17.67.39 (4-32)
Mean cyclosporine dose
(mg/kg;SD) 1.5.05 2.1.57 2.31.28 2.51.57 (mg/m2;SD) NA 58.49.64 69.234.04 69.635.79 Number of patients Sex (male:female) 16:8 15:9 16:8 15:9 Underlying diagnosis ALL 8 5 4 7 AML 10 9 10 7 Other malignancies 2 2 3 3 Nonmalignancies 4 8 7 7 HSCT conditioning Cyclophosphamide-containing 21 24 24 22 Other 3 0 0 2 GVHD prophylaxis CSA alone 0 2 0 2 CSAþ Methotrexate 24 13 17 11 Mycophenolate mofetil 0 1 0 7 Corticosteroid 0 6 6 4 Methotrexateþcorticosteroid 0 2 1 0
immediately after HSCT and convert to oral dosing several
weeks later, usually after engraftment. Because achievement
of a minimum C
0has been reported to be most important
just before engraftment
[2]
, it may be most clinically relevant
to evaluate cyclosporine AUC when patients are receiving i.v.
cyclosporine.
The relationship between cyclosporine AUC after i.v.
administration and HSCT outcomes such as aGVHD is just
beginning to be explored
[11,13,17-19,26,27]
. Among these
are 2 pediatric studies. In the
fi
rst, Willemze et al.
retro-spectively estimated cyclosporine pharmacokinetic
param-eters including AUC in 91 children receiving i.v. cyclosporine
over 2 hours
[19]
. Cyclosporine concentrations were assayed
by either radioimmunoassay or
fl
uorescence polarization
immunoassay. No comment regarding the comparability of
the results of these assays was provided. AUC-12 was
esti-mated using Bayesian methods based primarily on C
0values
that the authors stated led to a potential error in AUC
esti-mations of approximately 20%. During the second week after
HSCT, AUC values greater than 3000
m
g/L/h were associated
with lower overall survival rates and higher risk of relapse,
most especially in children with acute myelogenous
leukemia. No association between AUC and the incidence of
acute GVHD was observed.
Gerard et al. reported the second evaluation of
cyclo-sporine AUC in children undergoing HSCT
[26]
. In this
retro-spective analysis, cyclosporine AUC in blood and other tissues
were estimated using Bayesian models. The sampling
schedule and the number of cyclosporine blood
concentra-tions available to inform the construction of the
pharmaco-kinetic model were not described. Children received
cyclosporine either as a continuous infusion (30 patients) or
as a 2-hour infusion every 12 hours (31 patients). Thirty-four
patients (56%) developed aGVHD. It was not clear if the
relationship between C
0and aGVHD occurrence was
evalu-ated. However, there was a signi
fi
cant relationship between
cyclosporine AUC at steady state (de
fi
ned as the day before
the occurrence of aGVHD or day 13 after HSCT) and aGVHD.
For patients with underlying malignancy, the authors
concluded that cyclosporine doses be adjusted to achieve an
AUC of 10 h
mg/L over 24 hours (10,000
m
g/L/h over 24 hours;
5000
m
g/L/h over 12 hours), which corresponds to a C
2and C
12of 1000 and 80
m
g/L, respectively. It was furthermore
sug-gested that an AUC of 16 h
mg/L over 24 hours (16,000
m
g/L/h
over 24 hours; 8000
m
g/L/h over 12 hours) corresponding
to a C
2and C
12of 1600 and 120
m
g/L, respectively, be targeted
for patients without underlying malignancy.
The
fi
ndings of these retrospective studies require
prospective validation before speci
fi
c AUC values can be
adopted as the standard of care in clinical practice. The
availability of LSSs for estimation of cyclosporine AUC, such
as presented here, that can be performed without the use of
computer software will facilitate this process.
We found that the correlation between all data points and
AUC-12 was remarkably stronger in the steady-state LSS
validation cohort than in the other 3 study cohorts. Another
Table 3Bias and Precision of Estimates of AUC Using Limited Sampling Strategies for i.v. Cyclosporine Involving Fewer Than Five Data Points and Extent of Agreement between Values of AUC Calculated Using All Data Points and Using the LSS
Sample Time (hours from start of 2-h infusion)
LSS Equation Adjusted r2 Mean Bias,
% (95% CI)
Precision (%) ICC (95% CI) Limits of Agreement (%) Validation offirst-dose LSS[20] 2, 6, and 8 1.429C2þ5.722C8þ2.741C6þ380.862 .965 2.1 (.1-4.3) 5.0 .974 (.941 to .988) 13.4-8.9 2, 2.5, 6, and 8 1.372C2þ5.449C8þ2.870C6þ.470C2.5þ 165.709 .985 2.5 (3.9-1.2) 3.2 .979 (.952-.991) 4.5-9.8 2, 2.5, 4, 6, and 8 1.229C2þ5.242C8þ.740C6þ.514C2.5þ 2.7C4þ23.740 .993 2.2 (2.9-1.4) 2.3 .994 (.985-.997) 1.7-6.1 Development of steady-state LSS 2, 2.5, and 8 1.22C2þ1.54C2.5þ8.26C8þ292 .9851 .34 (2-1) 3.5 .984 (.965-.993) 7.2-7.3 2, 2.5, 4, and 8 1.22C2þ.78C2.5þ2.55C4þ5.91C8þ331 .9920 .19 (1-1) 4.9 .994 (.986-.997) 4.6-4.3 2, 2.5, 3, 4, and 8 1.26C2þ1.65C3þ2.43C4þ4.80C8þ261 .9967 .13 (0-1) 1.2 .995 (.988-.998) 2.9-2.9 Validation of steady-state LSS 2, 2.5, and 8 1.22C2þ1.54C2.5þ8.26C8þ292 .9875 3.5 (2-5) 3.3 .993 (.984-.997) 9.3-3.3 2, 2.5, 4, and 8 1.22C2þ.78C2.5þ2.55C4þ5.91C8þ331 .9937 2.3 (1-3) 3.5 .996 (.991-.998) 7.3-2.7 2, 2.5, 3, 4, and 8 1.26C2þ1.65C3þ2.43C4þ4.80C8þ261 .9935 1.7 (1-3) 1.4 .998 (.996-.999) 7.1-3.9
CI indicates confidence interval.
An LSS was considered to provide an acceptable estimate of AUC-12 if the r2value was>.9, lower limit of the 95% confidence limit of the ICC was .8 or higher, and
both bias and precision were 15% or less.
Table 2
Spearman Rho Correlation Coefficients for AUC-12 and Whole Blood Cyclosporine Concentrations at Time Points after the Start of a 2-Hour Infusion Sampling
Time (h)
First-Dose LSSValidation Cohort Steady-State LSS Development Cohort Steady-State LSS Validation Cohort Spearman
Rho Coefficient
Significance (Two-Tailed)
Spearman Rho Coefficient
Significance (Two-Tailed)
Spearman Rho Coefficient
Significance (Two-Tailed) 0 NA NA .43845 .0321 .80409 <.0001 2 .91324 <.0001 .87913 <.0001 .85478 <.0001 2.5 .66014 <.0004 .83652 <.0001 .95238 <.0001 3 .71739 <.0001 .80609 <.0001 .91933 <.0001 4 .90039 <.0001 .83826 <.0001 .94609 <.0001 6 .82775 <.0001 .54502 .0059 .92345 <.0001 8 .74543 <.0001 .47097 .0202 .90522 <.0001 10 .65810* <.0006 .45605 .0251 .89841 <.0001 12 .68099* <.0007 .34181* .1021 .87671 <.0001 *n¼23.
notable distinction in this cohort was the increased use of
mycophenolate mofetil compared with the other 3 cohorts.
The potential in
fl
uence of mycophenolate mofetil on
cyclo-sporine pharmacokinetic disposition has been identi
fi
ed
previously
[28]
and deserves further exploration.
For the purpose of this study, steady state was de
fi
ned as
having been achieved when patients had received at least 4
identical cyclosporine doses. Judging by the discrepant
correlations between C
0and the concentration 12 hours after
the start of the infusion (C
12), true steady state likely was not
achieved in all patients. Indeed, the C
0and C
12differed
by
>
20% in 6 (25%) and 4 patients (17%) in the development
and validation
steady-state LSS
cohorts,
respectively.
Fluctuations in dose times, infusion duration and/or rate,
liver function, and extent of enterohepatic circulation may
preclude achievement of steady state in HSCT patients in the
fi
rst weeks after HSCT. Interpretation of our study
fi
ndings
may also be limited by differences between cohorts such as
differences in aGVHD prophylaxis, pharmacogenetic
poly-morphism
[29]
, and underlying diagnosis. Despite these
possible limitations, the proposed LSSs performed well in
estimating AUC-12 within each patient.
The LSSs developed and validated here permit con
fi
dent
estimation of AUC-12 by clinicians. It must be noted that
their use is limited to the population in which they were
developed. That is, they may be used to estimate
cyclo-sporine AUC in children undergoing HSCT age .4 to
<
18 years
of age and receiving cyclosporine as a 2-hour i.v. infusion.
Use of the 3-point
fi
rst-dose (2, 6, and 8 hours) and
steady-state (2, 2.5, and 8 hours) LSSs to prospectively evaluate
the relationship between cyclosporine AUC and HSCT
outcomes in children is suggested.
ACKNOWLEDGMENTS
We are thankful to the children and their families who
participated in this study; Warren Walsh and Daksha Jadav,
Department of Pediatric Laboratory Medicine, SickKids; the
Clinical Research Assistants at SickKids and Children &
Women
’
s Health Centre of British Columbia (Entela Zhuleku,
Kim Pagel, Alexandra Airhart, Calvin Liu, Sheliza Moledina,
and Colleen Fitzgerald); and the nursing and pharmacy staff
of the Blood and Bone Marrow Transplant units, SickKids,
and Children & Women
’
s Health Centre of British Columbia.
Financial disclosure:
This study was funded by the
Pedi-atric Oncology Group of Ontario and C
17Research Network.
L.S. is supported by a New Investigator Award from the
Canadian Institutes of Health Research.
Con
fl
ict of Interest Statement
: There are no con
fl
icts of
interest to report.
REFERENCES
1. Jacobsohn DA. Acute graft-versus-host disease in children. Bone Marrow Transplant. 2008;41:215-221.
2. Punnett A, Sung L, Price V, et al. Achievement of target cyclosporine concentrations as a predictor of severe acute GvHD in children receiving cyclosporine and methotrexate prophylaxis.Ther Drug Monit. 2007;29:750-757.
3. Martin P, Bleyzac N, Souillet G, et al. Clinical and pharmacological risk factors for acute graft-versus-host disease after paediatric bone marrow transplantation from matched-sibling or unrelated donors.
Bone Marrow Transplant. 2003;32:881-887.
Table 4
Mean and Predicted AUC-12 Value Estimates Using the LSSs for Estimation of Cyclosporine AUC Estimation after i.v. Administration over 2 Hours
Mean AUC-12 (SD)mg/L/h) 95% CI Validation offirst-dose LSS AUC-12 2509672 2226-2793 Three point 2546588 2297-2794 Four point 2431595 2180-2682 Five point* 2480671 2190-2770 Development of steady-state LSS AUC-12 48881316 4279-5496 Three point 48651305 4262-5468 Four point 49401289 4424-5455 Five point 49351290 4419-5451 Validation of steady-state LSS AUC-12 42001142 3746-4657 Three point 43241168 3856-4791 Four point 42821110 3837-4726 Five point 42551105 3813-4697 *n¼23.
Figure 1.Bland-Altman plots. Percent difference against mean for actual (AUC-12) and predicted area under the whole blood cyclosporine concentra-tion versus time curve during 12-hour dosing interval generated by LSS. Percent difference¼100[(AUC-12eAUC predicted by LSS)]/mean of AUC-12
and AUC predicted by LSS]. (A) First-dose 4-point LSS validation: % difference against mean AUC. (B) Steady-state 3-point LSS development: % difference against mean AUC. (C) Steady-state three-point LSS validation: % difference against mean AUC.
4. Kanda Y, Hyo R, Yamashita T, et al. Effect of blood cyclosporine concentration on the outcome of hematopoietic stem cell trans-plantation from an HLA-matched sibling donor.Am J Hematol. 2006;81: 838-844.
5. Ram R, Storer B, Mielcarek M, et al. Association between calcineurin inhibitor blood concentration and outcomes after allogeneic hemato-poietic cell transplantation. Biol Blood Marrow Transplant. 2012;18: 414-422.
6. Malard F, Szydlo R, Brissot E, et al. Impact of cyclosporine-A concen-tration on the incidence of severe acute graft-versus-host-disease after allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2010;16:28-34.
7. Ruutu T, Niederwieser D, Gratwohl A, Apperley JF. A survey of the prophylaxis and treatment of acute GVHD in Europe: a report of the European Group for Blood and Marrow, Transplantation (EBMT). Chronic Leukaemia Working Party of the EBMT.Bone Marrow Trans-plant. 1997;19:759-764.
8. Levy G, Thervet E, Lake J, Uchida K. Patient management by Neoral C2 monitoring: an international consensus statement. Transplantation. 2002;73:S12-S18.
9. Nashan B, Cole E, Levy G, Thervet E. Clinical validation studies of Neoral C2 monitoring: A review.Transplantation. 2002;73:S3-S11.
10. Schrauder A, Saleh S, Sykora KW, et al. Pharmacokinetic monitoring of intravenous cyclosporine A in pediatric stem-cell transplant recipients. The trough level is not enough.Pediatr Transplant. 2009;13:444-450. 11. Barkholt L, Remberger M, Bodegard H, et al. Cyclosporine A (CSA) 2-h
concentrations vary between patients without correlation to graft-versus-host disease after allogeneic haematopoietic stem cell trans-plantation.Bone Marrow Transplant. 2007;40:683-689.
12. Willemze AJ, Cremers SC, Schoemaker RC, et al. Ciclosporin kinetics in children after stem cell transplantation.Br J Clin Pharmacol. 2008;66: 539-545.
13. Hendriks MP, Blijlevens NM, Schattenberg AV, et al. Cyclosporine short infusion and C2 monitoring in haematopoietic stem cell transplant recipients.Bone Marrow Transplant. 2006;38:521-525.
14. Kong D, Shuttleworth P, Bailey M, Grigg A. CsA 2-h concentration correlates best with area under the concentration-time curve after allo-SCT compared with trough CsA.Bone Marrow Transplant. 2012;47:54-59. 15. Balci Y, Tavil B, Karabulut E, et al. Cyclosporine level at the second hour in pediatric hematopoietic stem cell transplant patients. Exp Clin Transplant. 2011;5:329-335.
16. Furukawa T, Kurasaki-Ida T, Masuko M, et al. Pharmacokinetic and pharmacodynamic analysis of cyclosporine A (CsA) tofind the best single time point for the monitoring and adjusting of CsA dose using twice-daily 3-h intravenous infusions in allogeneic hematopoietic stem cell transplantation.Int J Hematol. 2010;92:144-151.
17. Hadjibabaie M, Vazirian I, Iravani M, et al. Development and validation of limited sampling strategies for estimation of cyclosporine area under the concentration-time curve in hematopoietic stem cell transplant patients.Ther Drug Monit. 2011;33:673-680.
18. Inoue Y, Saito T, Ogawa K, et al. Pharmacokinetics of cyclosporine A at a high-peak concentration of twice-daily infusion and oral adminis-tration in allogeneic haematopoietic stem cell transplantation.J Clin Pharm Ther. 2011;36:518-524.
19. Willemze AJ, Press RR, Lankester AC, et al. CsA exposure is associated with acute GVHD and relapse in children after SCT. Bone Marrow Transplant. 2010;45:1056-1061.
20. Sibbald C, Seto W, Taylor T, et al. Determination of area under the whole blood concentration versus time curve afterfirst intravenous cyclosporine dose in children undergoing hematopoietic stem cell transplant: limited sampling strategies. Ther Drug Monit. 2008;30: 434-438.
21. Osowski CL, Dix SP, Lin LS, et al. Evaluation of the drug interaction between intravenous high-dose fluconazole and cyclosporine or tacrolimus in bone marrow transplant patients.Tranplantation. 1996; 61:1268-1272.
22. Canafax DM, Graves NM, Hilligoss DM, et al. Interaction between cyclosporine and fluconazole in renal allograft recipients. Trans-plantation. 1991;51:1014-1018.
23. Muller R, Buttner P. A critical discussion of intraclass correlation coefficients.Stat Med. 1994;13:2465-2476.
24. Lee J, Ong CN. Statistical evaluation of agreement between two methods for measuring a quantitative variable.Comput Biol Med. 1989; 19:61-70.
25. Ting LS, Partovi N, Levy RD, et al. Limited sampling strategy for pre-dicting area under the concentration-time curve of mycophenolic acid in adult lung transplant recipients. Pharmacotherapy. 2006;26: 1232-1240.
26. Gerard C, Bleyzac N, Girard P, et al. Links between cyclosporine exposure in tissues and graft-versus-host disease in pediatric bone marrow transplantation: analysis by a PBPK model.Pharm Res. 2011; 28:531-539.
27. Wilhelm A, de Graaf P, Veldkamp A, et al. Population pharmacokinetics of cyclosporin in haematopoietic allogeneic stem cell transplantation with emphasis on limited sampling strategy.Br J Clin Pharmacol. 2012; 73:553-563.
28. Pape L, Froede K, Strehlau J, et al. Alterations of cyclosporin A metab-olism induced by mycophenolate mofetil.Pediatr Transplant. 2003;7: 302-304.
29. Qui F, He X-J, Sun Y-X, et al. Influence of ABCB1, CYP3A4*18B and CYP3A5*3 polymorphisms on cyclosporine A pharmacokinetics in bone marrow transplant recipients.Pharmacol Rep. 2011;63:815-825.