Variability in the pharmacokinetics of intravenous busulphan given as a single daily dose to paediatric blood or marrow transplant recipients

Abstract

AIM

To examine inter- and intrapatient variability in the pharmacokinetics of intravenous (i.v.) busulphan given as a single daily dose to children with malignant (n = 19) and nonmalignant (n = 21) disease.

METHODS

Busulphan (120 mg m⁻², 130 mg m⁻² or 3.2 mg kg⁻¹) was administered over median 2.1 h. Blood samples (4–10) were collected after the first dose, busulphan concentrations were measured and pharmacokinetic parameters, including clearance (CL) and area under the concentration–time curve (AUC), were determined using the Kinetica software (Innaphase). Interpatient variability was assessed as percent coefficient of variation (% CV). Intrapatient variability was assessed by calculating percent differences between observed full dose AUC and AUC predicted from an initial 65 mg m⁻² dose in 13 children who had busulphan pharmacokinetic monitoring.

RESULTS

Clearance of i.v. busulphan in 40 children was 4.78 ± 2.93 l h⁻¹ (% CV 61%), 0.23 ± 0.08 l h⁻¹ kg⁻¹ (% CV 35%) and 5.79 ± 1.59 l h⁻¹ m⁻² (% CV 27%). Age correlated significantly (p < 0.001) with CL (l h⁻¹) and CL (l h⁻¹ kg⁻¹), but not with CL (l h⁻¹ m⁻²). AUC normalized to the 130 mg m⁻² dose ranged from 14.1 to 56.3 mg l⁻¹.h (% CV 37%) and also did not correlate with age. Interpatient variability in CL (l h⁻¹ m⁻²) was highest in six children with immune deficiencies (60%) and lowest in seven children with solid tumours (14%). Intrapatient variability was <13% for nine (of 13) children, but between 20 and 44% for four children.

CONCLUSIONS

There is considerable inter- and intrapatient variability in i.v. busulphan CL (l h⁻¹ m⁻²) and exposure that is unrelated to age, especially in children with immune deficiencies. These results suggest that monitoring of i.v. busulphan pharmacokinetics is required.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• The pharmacokinetics of oral busulphan given four times daily has been extensively studied.

• Large inter- and intravariability in oral busulphan exposure has led to attempts at pharmacokinetic monitoring.

• However, there have been limitations in the pharmacokinetic analysis due to inadequate characterization of the elimination phase in a 6-h dosing interval, due to late absorption in some patients.

• Intravenous (i.v.) busulphan is a relatively new administration method and there have been relatively few studies on the pharmacokinetics of i.v. busulphan, especially when given as a single daily dose.

WHAT THIS STUDY ADDS

• Inter- and intrapatient variability in i.v. busulphan pharmacokinetics is comparable to that previously observed with oral busulphan, suggesting that pharmacokinetic monitoring is advisable.

• Children with immune deficiencies, in particular, have widely variable exposure.

Introduction

Busulphan is an alkylating agent that, when combined with other drugs such as cyclophosphamide, effectively ablates the marrow prior to blood or marrow transplantation (BMT). It is a toxic drug with a relatively narrow therapeutic index. High exposure to busulphan has been linked to the occurrence of veno-occlusive disease of the liver (VOD) [1–3], a clinical syndrome of fluid retention, painful hepatomegaly, jaundice and/or unexplained weight gain, which can often lead to multiorgan failure and death [4]. Other complications of high-dose busulphan include interstitial pneumonia, haemorrhagic cystitis, cataracts, convulsions, alopecia and mucositis [5]. Low busulphan exposure, however, has been linked to higher rates of disease relapse and engraftment failure [6, 7].

Following administration of oral busulphan wide inter- and intrapatient variability in systemic exposure has been observed [8–10]. Possible reasons for this variability include erratic intestinal absorption (leading to variable bioavailability), variable hepatic metabolism, genetic polymorphisms in glutathione-S-transferase (an enzyme involved in the metabolism of busulphan), patient age, diagnosis and concomitant medication.

In an attempt to reduce this variability, some centres have performed pharmacokinetic monitoring of busulphan with subsequent dose adjustments [7, 11–13]. However, the most common dosing schedules used in many of these studies have been 1 mg kg⁻¹ (or 37.5 mg m⁻²) four times daily for 4 days (total dose 16 mg kg⁻¹), the four times daily regimen. Therapeutic drug monitoring of oral busulphan is difficult under these circumstances because it is often not possible to obtain reliable estimates of busulphan area under the concentration–time curve (AUC) in the 6-h dosing interval because of late absorption in some patients [10]. Dose-to-dose variability is also high for oral busulphan, so that even after dose adjustments are made, the target AUC levels are often not reached [14–17]. Excessive vomiting is an additional complication for therapeutic drug monitoring of oral busulphan [15].

Intravenous (i.v.) busulphan is a relatively new administration method that is reported to be safe when administered as a single daily dose [18]. When compared with the oral or i.v. four-times-daily regimens, single-daily-dose i.v. busulphan has comparable hepatic toxicity, engraftment and 100-day relapse rates and less severe acute graft-versus-host disease when combined with 120 mg kg⁻¹ cyclophosphamide [19].

It is hoped that use of i.v. busulphan will lead to reduced inter- and intrapatient variability in busulphan exposure and more reliable therapeutic drug monitoring results. This is a reasonable expectation, since changes in bioavailability and absorption will no longer be of concern and it will be possible to obtain accurate AUC estimates, even with the 6-h dosing interval. The aims of this study were therefore to (i) examine inter- and intrapatient variability in the pharmacokinetics of i.v. busulphan given as a single daily dose to children with malignant and nonmalignant diseases and (ii) compare these with previous results obtained with oral busulphan.

Methods

Study design

This study was a prospective, single-centre investigation of the pharmacokinetics of i.v. busulphan in children. It was approved by The Children's Hospital at Westmead's Ethics Committee and all parents provided informed consent. Initially the study was designed to be purely observational. However, it was later felt to be necessary to intervene and adjust the busulphan dose in a few patients with very high or low busulphan AUC values.

Patients

The study included 40 children aged between 0.2 and 17.7 years who underwent autologous (n = 7) or allogeneic (n = 33) BMT between 2003 and 2007 as part of their treatment for malignant (n = 19) or nonmalignant (n = 21) diseases. The characteristics of the children are summarized in Table 1. BMT conditioning was busulphan/melphalan (n = 11), busulphan/cyclophosphamide (n = 6), busulphan/cyclophosphamide/etoposide (n = 2), fludarabine/busulphan (n = 4), fludarabine/busulphan/cyclophosphamide (n = 12) and fludarabine/busulphan/melphalan (n = 5). Clobazam 0.25 mg kg⁻¹ day⁻¹ was administered in two divided doses before and during busulphan treatment as anticonvulsant prophylaxis. Unlike phenytoin [20], clobazam does not interfere with busulphan pharmacokinetics. Other medications given concomitantly with busulphan are listed in Table 2.

Table 1.

Clinical data of 40 children who received i.v. busulphan in preparation for BMT

Age (years)
Median (interquartile range)	3.2 (1.5–9.1)
Weight (kg)
Median (interquartile range)	14.0 (10.7–33.2)
Height (cm)
Median (interquartile range)	92.0 (78.9–133.3)
Surface area (m2)
Median (interquartile range)	0.60 (0.48–1.10)
GFR (ml min−1 1.73 m2)
Median (interquartile range)	133 (107–165)
Gender (no. children)
Male	27
Female	13
Diagnosis (no. children)
Malignant diseases
Leukaemia
Acute lymphoblastic leukaemia	2
Acute myeloid leukaemia	9
Chronic myeloid leukaemia	1
Solid tumours
Ewing sarcoma	4
Neuroblastoma	3
Genetic diseases
Adrenal leukodystrophy	3
Hurler's syndrome	1
Aspartyl glucosaminuria	1
Fanconi's anaemia	2
Wiskott–Aldrich syndrome	5
Phosphoglycerate kinase deficiency	1
Haemophagocytic lymphohistiocytosis	2
Severe immune deficiencies
Omenn syndrome	3
Severe combined immunodeficiency disorder	2
Congenital neutropenia	1

Omenn syndrome is an autosomal recessive form of severe combined immunodeficiency disorder, originally described by Omenn in 1965 [46].

Table 2.

Numbers of children in each disease group who were given certain medications concomitantly with busulphan

	Malignant disease		Nonmalignant disease
Medication	Leukaemia	Solid tumours	Genetic disease	Immune deficiency
Total children in group (n)	12	7	15	6
Fluconazole	12	7	15	6
Ursodeoxycholic acid	12	7	15	6
Clobazam	12	7	15	6
Ondansetron	12	7	15	6
Acyclovir	1	0	0	0
Valacyclovir	1	0	0	0
Ganciclovir	2	0	1	0
Gentamicin	1	0	1	1
Vancomycin	0	0	2	2
Ticarcillin/clavulanic acid	0	0	1	3
Flucloxacillin	0	0	1	0
Voriconazole	0	0	1	0
Liposomal amphotericin B	0	0	1	3
Meropenem	0	0	1	0
Co-trimoxazole	0	0	0	2
Roxithromycin	0	0	0	1
Ceftriaxone	0	0	1	0
Trimeprazine	0	0	2	0
Cetirizine hydrochloride	0	0	1	0
Hydrocortisone	1	0	3	0
Fludrocortisone	0	0	3	0
Methylprednisolone	0	0	0	1
Loratidine	0	0	1	0
Metclopramide	0	1	0	0
Ciclosporin	0	0	1	0

Drug administration and blood sampling

Busulphan (i.v. Busulphex; Orphan Australia Pty Ltd, Berwick, Australia) was administered intravenously over median 2.1 h on each of 4 days. Single daily doses were 120 mg m⁻² (n = 3), 130 mg m⁻² (n = 34) or 3.2 mg kg⁻¹ (n = 3). All patients had a double lumen central line, so that one lumen could be used for drug administration and one for blood sampling. To avoid contamination, 5 ml of blood was withdrawn prior to taking each sample. Lithium heparin blood samples were collected from each patient after the first busulphan dose to characterize the pharmacokinetic profile. In 20 patients the blood collection times were 0 min, 30 min, 1 h, 2 h, 4 h, 6 h, 8 h, 10 h and 12 h after the infusion end. In the remaining 20 children only four blood samples were collected at 0 min, 1 h, 2 h and 4 h after the infusion end, as suggested by the limited sampling strategy of Vaughan et al.[21]. Plasma fractions were separated by centrifugation at 1200 g for 10 min at 4°C (Beckman CS-15R; Beckman Instruments, Fullerton, CA, USA) and were stored at −40°C until analysis. Samples were analysed on the day after collection.

In 13 children, the first 130 mg m⁻² dose was administered as two 65 mg m⁻² doses over 2 days to allow for pharmacokinetic analysis and dose adjustments, if necessary. Blood samples were taken after the first half dose (65 mg m⁻²) for pharmacokinetic analysis. The AUC results were doubled to obtain a predicted AUC for the 130 mg m⁻² dose. If the predicted results were within a normal range of 16–37 mg l⁻¹ h⁻¹, then the child received 130 mg m⁻² busulphan for the remaining three full doses. If not, a dose was given that targeted specific AUC values. A target AUC of 29 mg l⁻¹ h⁻¹ was generally used, derived from our previous study on oral busulphan in acute leukaemia [22]. However, modifications to this target were required in some patients. Bloods for pharmacokinetic analysis were again collected after administration of the first full 130 mg m⁻² dose (n = 7) or the dose that targeted specific AUC values (n = 6).

Busulphan assay

Busulphan was measured in plasma samples using our previously published gas chromatographic assay [23] that had acceptable accuracy and precision. For concentrations ranging from 0.1 to 7 μg ml⁻¹, the between-day coefficient of variation of the assay (% CV) was <10% (n = 17) and the overall deviation from the true concentration was <9%. The limit of quantification was approximately 0.1 μg ml⁻¹, with a % CV of 10%. The limit of detection of the assay was approximately 0.03 μg ml⁻¹. The calibration curve was linear over the range 0.1–9 μg ml⁻¹ busulphan. None of the other drugs that the patients were receiving (Table 2) would form diiodobutane after reaction with potassium iodide, so the analysis was highly specific for busulphan.

Pharmacokinetic analysis

The pharmacokinetic parameters for each individual were determined using a one-compartment modelling procedure implemented by the computer software, Kinetica 2000 (Innaphase, Philadelphia, PA, USA). Estimates of clearance (CL), AUC, volume of distribution at steady state (V_ss), elimination rate constant (k_e), elimination half-life (t_1/2) and mean residence time (MRT) were generated. AUC was normalized to the 130 mg m⁻² dose by dividing the first dose AUC by the mg m⁻² dose and multiplying by 130.

Assessment of inter- and intrapatient variability in i.v. busulphan pharmacokinetics

Interpatient variability in i.v. busulphan pharmacokinetic parameters was assessed by calculating the % CV, given as the standard deviation divided by the mean, multiplied by 100. Percent CV was determined for the total group of children, as well as for children in specific disease groups.

Intrapatient variability in i.v. busulphan pharmacokinetics was assessed by calculating the percentage differences between observed full-dose i.v. busulphan AUC values and those predicted from an initial 65 mg m⁻² dose in the 13 children who had i.v. busulphan pharmacokinetic monitoring. In this analysis either the predicted AUC was twice the AUC obtained for the 65 mg m⁻² dose (a predicted AUC for the full 130 mg m⁻² dose) or, for those children whose predicted AUC results were outside the normal range (n = 6), the specific AUC that was targeted.

Results

In all patients, the decline in busulphan concentrations after the dose was log linear, fitting a single-compartment model. Four-sample limited sampling adequately described the concentration–time curve and allowed accurate determination of AUC in these children. There was wide variability in busulphan concentrations following a single dose in 31 children who received 130 mg m⁻² busulphan, either initially (n = 23) or after pharmacokinetic assessment (Figure 1). This variability was also reflected in the wide range of pharmacokinetic parameter estimates observed in the total group of 40 children (Table 3). AUC normalized to a 130 mg m⁻² dose ranged from a minimum of 14.1 mg l⁻¹ h⁻¹ to a maximum of 56.3 mg l⁻¹ h⁻¹, with a median value of 21.9 mg l⁻¹ h⁻¹ and an interquartile range of 18.7–27.0 mg l⁻¹ h⁻¹ (Figure 2).

Figure 1.

Variability in busulphan concentrations following a single 130 mg m⁻² intravenous dose in 31 children

Table 3.

Pharmacokinetic parameters of i.v. busulphan in 40 children

Parameter	Mean	SD	% CV	Median	Interquartile range	Minimum	Maximum
Age (years)	5.5	5.1	93	3.2	1.5–9.1	0.2	17.7
CL (l h−1)	4.78	2.93	61	4.05	2.85–6.52	0.67	13.81
CL (l h−1 kg−1)	0.23	0.08	35	0.22	0.18–0.29	0.10	0.42
CL (l h−1 m−2)	5.79	1.59	27	5.94	4.81–6.96	2.31	9.22
Vss (l)	16.08	12.54	78	11.37	7.60–21.51	2.5	67.51
Vss (l kg−1)	0.74	0.16	22	0.71	0.64–0.83	0.42	1.2
Vss (l h−1 m−2)	18.4	3.8	21	18.1	15.6–20.5	11.9	29.7
MRT	3.43	1.29	38	3.15	2.62–3.85	1.6	8.8
ke (h−1)	0.32	0.10	31	0.32	0.26–0.38	0.11	0.63
t1/2 (h)	2.38	0.90	38	2.18	1.81–2.67	1.11	6.1
AUC (mg l−1 h−1)*	24.7	9.2	37	21.9	18.7–27.0	14.1	56.3

^*AUC = AUC normalized to a 130 mg m⁻² busulphan dose. SD, standard deviation; % CV, % coefficient of variation.

Figure 2.

Distribution of busulphan AUC normalized to the 130 mg m⁻² dose in 40 children

Interpatient variability (% CV) in i.v. busulphan CL (l h⁻¹) in the total group was 61%, but was lower when normalized to weight (35%) or body surface area (27%). There was a strong significant positive correlation between patient age and CL (l h⁻¹) (r = 0.81, P < 0.001, Figure 3A). There was a significant negative correlation between age and weight-normalized CL (r = 0.54, P < 0.001, Figure 3B), but no significant relationship with CL normalized to body surface area (Figure 3C). Normalized AUC was associated with a % CV of 37% and did not correlate significantly with age.

Figure 3.

Influence of patient age on busulphan clearance, non-normalized (A) and when normalized with weight (B) and surface area (C)

Interpatient variability in i.v. busulphan clearance and AUC (normalized) for children in specific diagnosis subgroups are shown in Table 4 and Figure 4. Coefficients of variation were highest for children with immune deficiencies (60% for surface area–normalized clearance, 49% for normalized AUC) and lowest for children with solid tumours (14% for surface area–normalized clearance, 15% for normalized AUC). Children with solid tumours had minimal concomitant medication, whereas children with immune deficiencies were receiving antimicrobial and other medications at the time of busulphan (Table 2). The minimum and maximum normalized AUC were both obtained by children with immune deficiencies. The child who had the lowest normalized AUC (14.1 mg l⁻¹ h⁻¹) had Omenn syndrome and was administered methylprednisolone and vancomycin in addition to the common medications of clobazam, fluconazole, ursodeocycholic acid and ondansetron. The child who had the highest normalized AUC (56.3 mg l⁻¹ h⁻¹) also had Omenn syndrome and was administered liposomal amphotericin B, gentamicin, ticarcillin/clavulanic acid, vancomycin and roxithromycin in addition to the common medications.

Table 4.

Variability in i.v. busulphan clearance and AUC in children with different diagnoses. Data are mean ± standard deviation (% CV)

	Malignant disease	Nonmalignant disease
Parameter (units)	Leukaemia	Solid tumours	Genetic disease	Immune deficiency
Children (n)	12	7	15	6
Age (years)*	5.5 2.0–17.7	3.5 0.3–16.8	2.5 0.4–14.4	0.78 (0.2–3.8)
CL (l h−1)	5.97 ± 3.04 (51%)	5.50 ± 1.73 (31%)	4.64 ± 3.03 (65%)	1.87 ± 1.69 (90%)
CL (l h−1 kg−1)	0.22 ± 0.07 (32%)	0.23 ± 0.07 (30%)	0.22 ± 0.06 (27%)	0.26 ± 0.13 (50%)
CL (l h−1 m−2)	6.33 ± 1.60 (25%)	6.11 ± 0.86 (14%)	5.61 ± 0.98 (17%)	4.80 ± 2.88 (60%)
AUC (mg l−1 h−1)	22.0 ± 6.6 (30%)	21.7 ± 3.2 (15%)	24.1 ± 5.8 (24%)	35.2 ± 17.1 (49%)

^*Data are median (range). AUC = AUC normalized to a 130 mg m⁻² dose.

Figure 4.

Box plot comparing the distribution of AUC normalized to the 130 mg m⁻² dose in 40 children in different diagnosis groups

Busulphan AUC monitoring was performed in 13 patients. The 130 mg m⁻² dose was predicted to be adequate (within a normal range of 16–37 mg l⁻¹ h⁻¹) following an initial 65 mg m⁻² dose for seven children. In six children the AUC values for the 130 mg m⁻² dose were predicted to be outside the normal range, so doses were given that targeted specific AUC values: 29 mg l⁻¹ h⁻¹ for four children (diagnoses: acute myeloid leukaemia, congenital neutropenia, severe combined immune deficiency and aspartyl glucosaminuria), 32 mg l⁻¹ h⁻¹ for one child with Fanconi's anaemia undergoing T-cell-depleted matched-unrelated BMT who had a high risk of engraftment failure, and 20 mg l⁻¹ h⁻¹ for one child with Omenn syndrome, who would have required an extremely high dose (269 mg m⁻²) to target an AUC of 29 mg l⁻¹ h⁻¹.

The data from the 13 children who underwent pharmacokinetic monitoring was used to assess how well the AUC measured from an initial 65 mg m⁻² dose could predict the AUC of the full 130 mg m⁻² dose or the targeted dose to provide an assessment of intrapatient variability. The results are shown in Table 5. AUC targeting was successful in nine children in whom observed and predicted AUC values were within 13% of each other. In four children the target AUC was not achieved: observed and predicted AUC values deviated from each other by 20–44%. The children ranged in age from 0.24 to 7.0 years (three were <2.5 years old) and had diagnoses of congenital neutropenia, x-linked phosphoglycerate kinase deficiency, adrenal leukodystrophy and Wiskott–Aldrich syndrome.

Table 5.

Differences between observed busulphan AUC values and those predicted from an initial 65 mg m⁻² dose

Patient	Predicted AUC (mg l−1 h−1)	Observed AUC (mg l−1 h−1)	% Difference
1*	29	30.6	5.5
2†	22.1	24.8	12.2
3*	32	30.4	−5.0
4*	29	34.9	20.3
5†	21.8	28.4	30.3
6*	29	31.9	10
7*	29	29.7	2.4
8*	20	21.1	5.5
9†	23.4	23.6	0.9
10†	21.5	23.9	11.2
11†	19.4	27.9	43.8
12†	17.4	16.8	−3.4
13†	22.2	28.6	28.8

Predicted AUC = the AUC that was targeted following an initial 65 mg m⁻² dose. This value was twice the AUC obtained for the 65 mg m⁻² dose (a predicted AUC for the 130 mg m⁻² dose) for those patients whose predicted results were within the normal range of 16–37 mg l⁻¹ h⁻¹*, or a specific AUC for those patients whose results were outside the normal range†. Observed AUC = the AUC obtained after children received a full 130 mg m⁻² dose or a dose that targeted a specific AUC. %Difference = (Observed AUC − Predicted AUC)/Predicted AUC × 100.

Our results for i.v. busulphan clearance are compared with those obtained from previously published studies on i.v and oral busulphan in children in Table 6.

Table 6.

Comparison of oral and i.v. busulphan clearance in previously published studies in children

Study (% CV)	Children (n)	Age range (years)	CL (l h−1) Mean ± SD (% CV)	CL (l h−1 kg−1) Mean ± SD (% CV)	CL (l h−1 m−2) Mean ± SD
i.v. busulphan
This study*	40	0.2–17.7	4.78 ± 2.93 (61%)	0.23 ± 0.08 (35%)	5.79 ± 1.59 (27%)
Schecter et al. (2007) [25]	45	0.25–16.2		0.21 ± 0.07 (33%)	5.52 ± 3.12 (57%)
Oechtering et al. (2005) [27]	17	0.9–17.3		0.21 ± 0.05 (26%)
Tran et al. (2004) [16]	20	0.8–14.9		0.24 ± 0.07 (29%)
Oral busulphan
Shaw et al. (2004)*[22]	12	1.4–13.5		0.21 ± 0.07 (33%)
Bertholle-Bonnet et al. (2007) [24]	100	0.1–18.1	4.84 ± 2.86 (59%)
Vassal et al. (1992) [8]	25	2–14		0.27 ± 0.08 (31%)
Hassan et al. (1991) [26]	9	1–13		0.29 ± 0.13 (45%)

^*In these studies busulphan was administered as a single daily dose. In the remaining studies busulphan was administered four times daily. CL, clearance, which is divided by F (bioavailability) for oral busulphan.

Discussion

In this study we have shown that there is considerable interpatient variability in i.v. busulphan pharmacokinetics which is comparable to the variability observed for oral busulphan (Table 6). Interpatient variability (% CV) for CL (l h⁻¹) in our study on i.v. busulphan was 61%, which is similar to the % CV of 59% observed by Bertholle-Bonnet et al.[24] in their large study on oral busulphan that included 100 children. We obtained a % CV for CL (l h⁻¹ kg⁻¹) of 35%, which was closest to that obtained by Schecter et al.[25] and very similar to our previous study on oral busulphan in children with acute leukaemia [22] and the other presented studies on oral busulphan [8, 26]. Two studies on i.v. busulphan [16, 27] had lower % CV values for CL (l h⁻¹ kg⁻¹) of 26% and 29%. The study of Oechtering et al.[27] mostly included children with malignancies, whereas our study included equal numbers of children with malignant and nonmalignant diseases.

Children with immune deficiencies administered oral busulphan have been previously observed to have more variable exposure than children with other diseases, including haematological malignancies, metabolic diseases and haemoglobinopathies [24]. Similarly, we have observed that the interpatient variability in i.v. busulphan clearance and normalized AUC is highest for children with immune deficiencies. The reasons for this are unclear. These children were very young (age range 0.2–3.8 years), and it has been suggested that their age may contribute to this wide variability [24]. It is clear from Figure 3 that age contributed to the interpatient variability in CL (l h⁻¹) and CL (l h⁻¹ kg⁻¹), but not CL (l h⁻¹ m⁻²). The % CV for CL (l h⁻¹ m⁻²) was still higher for the children with immune deficiencies than for those with genetic disease, solid tumours or leukaemia (60% vs. 17%, 14% and 25%, respectively).

Another possible reason for the wide variability in i.v. busulphan clearance in children with immune deficiencies is concomitant medication. Being immune-suppressed, these children had multiple infections and were on antimicrobial medication, whereas the child who had the lowest normalized AUC also received methylprednisolone. These drugs may have affected busulphan clearance, but numbers were too low for statistical analysis and we will continue to collect information on concomitant medications. To date, drugs known to impact on busulphan clearance include phenytoin, itraconazole, ketobemidone and metronidazole [20, 28–31], but none of the children in this study were receiving these drugs (Table 2). It is interesting that all these drugs are either metabolized by, inhibit or induce enzymes in the cytochrome P450 family [29–32], even though busulphan is primarily eliminated after conjugation with glutathione in the liver [33, 34]. The mechanism of the interaction may relate to some of the busulphan glutathione metabolites being further metabolized by the cytochrome P450 system or, as previously suggested [35], it could relate to depletion of liver glutathione content or alterations in the activity of glutathione-S-transferase. Many drugs have inductive or inhibitory effects on hepatic enzymes [36], and further studies are required to identify all those that affect busulphan pharmacokinetics. In the meantime, we recommend the monitoring of busulphan pharmacokinetics in patients who also require drugs with inductive or inhibitory effects on hepatic enzymes. Of the co-administered drugs given to the children in this study, those that have been shown to induce or inhibit cytochrome P450 isozymes include amphotericin B, ciclosporin, methylprednisolone, hydrocortisone, fludrocortisone, cotrimoxazole and roxithromycin [36–39].

Busulphan pharmacokinetics may also be influenced by polymorphisms in the genes GSTA1, GSTM1 or GSTP1 that code for the glutathione transferases GSTα, GSTμ and GSTp, respectively. GSTα is the predominant isoform catalysing the conjugation of busulphan with glutathione, whereas GSTμ and GSTp contribute approximately 46% and 18% of the activity of GSTα, respectively [33, 34]. In a small study that included 12 patients, busulphan clearance was significantly lower for three patients who were genotyped as heterozygous variants (GSTA1*A/*B) compared with nine patients who were classified as wild-type (GSTA1*A/*A) [40]. In thalassaemia, the occurrence of VOD after busulphan-based conditioning was associated with the GSTM1-null genotype [41]. Further studies in larger populations are required for a better understanding of the influence of pharmacogenomics on busulphan pharmacokinetics and pharmacodynamics.

Intrapatient variability in busulphan AUC was found to be <13% for nine (of 13) children, but between 20 and 44% for four children. Although some previous researchers have observed that dose-to-dose variability of i.v. busulphan is minimal, reflecting linear pharmacokinetics within individuals [42, 43], there is a note of caution in one paper, which reported difficulties in achieving target AUC values in six children < 2 years old [44]. It is unclear why busulphan clearance changed with dose in these children. However, the influence of concomitant medication needs further investigation. In the meantime, careful evaluation of i.v. busulphan pharmacokinetics is required in children with nonmalignant disease, especially if they are <2 years of age or are on medication that is known to alter the activities of hepatic enzymes.

We have previously argued that, for many diseases, a target AUC range that is associated with good transplant outcome has not yet been adequately established for either oral or i.v. busulphan [45]. Therefore, it was our original intention to conduct an observational study designed to establish such a target range without any modifications of busulphan dose based on the pharmacokinetic results. However, after finding such high and low exposure levels in children with nonmalignant disease, we felt it prudent to dose adjust for those patients with very high and very low AUC values. The target AUC range that we used was derived from our previous study [22] where, after 150 mg m⁻² day⁻¹ oral busulphan, the AUCs achieved by 27 children were mean ± SD 29 ± 8 mg l⁻¹ h⁻¹, providing us with a normal range of 21–37 mg l⁻¹ h⁻¹, adjusted to 16–37 mg l⁻¹ h⁻¹ after we observed that the 130 mg m⁻² i.v. dose tended to provide lower AUCs compared with the 150 mg m⁻² oral dose. This target range is still undergoing revision and is based on children with acute leukaemia who received only cyclophosphamide in addition to busulphan.

We now routinely monitor first-dose busulphan AUC to allow for dose adjustments if necessary. Our procedure for pharmacokinetic monitoring involves administering a 65 mg m⁻² test dose on day 1 of busulphan, then a second 65 mg m⁻² dose on day 2 (while pharmacokinetic analysis is being performed), followed by 3 days of 130 mg m⁻² busulphan if the exposure is predicted to be acceptable (e.g. within the normal range), or, if not, 3 days of a dose targeting an AUC of 29 mg l⁻¹ h⁻¹ (the mean AUC for the oral 150 mg m⁻² dose in acute leukaemia) or another target within the normal range, depending on the clinical condition of the child.

Pharmacokinetic monitoring of i.v. busulphan is facilitated by limited sampling strategies. The monophasic, log-linear decline in busulphan concentrations following the end of infusion should permit determination of AUC from only two data points. Vaughan et al. found that AUCs determined with four or five postinfusion samples were in good agreement with AUCs determined using 11 samples [21]. We agree with their recommendation still to analyse at least four concentration–time points, since AUC results need to be available rapidly and questionable results cannot be repeated.

To conclude, there is considerable inter- and intrapatient variability in i.v. busulphan clearance and exposure which is comparable to that observed with oral busulphan. Children with immune deficiencies have higher variability than children with other diseases, including genetic disease, solid tumours and leukaemia. The reasons for this are unclear, but the influence of concomitant medication and pharmacogenomics on busulphan pharmacokinetics needs to be studied further. These results suggest that therapeutic drug monitoring of i.v. busulphan is required. Target AUC levels that are associated with a good outcome now need to be identified for uniform disease populations.

C.E.N. is supported by the Leukaemia Research Support Fund of The Children's Hospital Westmead, and by NH and MRC Project Grant 396702. We thank the patients and their families for taking part in the study and the nursing staff in the oncology unit for their care of the patients, including taking blood samples for measurement of busulphan concentrations.