An improved limited sampling method for individualised ... - Nature

Bone Marrow Transplantation, (1997) 20, 347–354  1997 Stockton Press All rights reserved 0268–3369/97 $12.00

An improved limited sampling method for individualised busulphan dosing in bone marrow transplantation in children DS Chattergoon1,2, EF Saunders1, J Klein2, S Calderwood1 , J Doyle1, MH Freedman1 and G Koren2 Divisions of 1Hematology/Oncology and 2Clinical Pharmacology, Hospital for Sick Children, and University of Toronto, Toronto, Canada

Summary: Busulphan (BU) pharmacokinetic (PK) studies in children undergoing bone marrow transplantation suggest that individual BU dosing may be necessary to optimise BU systemic exposure. Optimising BU systemic exposure may improve outcome and decrease toxicity in BMT. Because of practical limitations in obtaining blood from children and for financial reasons, a limited sampling method (LSM) is needed. However, such methods for BU have not been validated in children. In the present study, we individualized oral BU dosing in 10 children to target an area under the curve of BU (BU AUC) of 900–1400 mM/min based on BU AUC0–` calculated from nine serum BU concentrations performed after a BU test dose of 40 mg/m2. We validated a LSM using 3 BU concentrations to determine AUC. Six of nine patients studied (one patient non-evaluable), required their doses modified (3, lower; 3, higher). The mean percent dose change was 26.2% (range −33.3% to +45.3%). Our three sample LSM BU AUC0–` (1098 ± 344, mean ± 1 s.d.) correlated highly with our nine sample BU AUC0–` (1132 ± 389, Pearson r = 0.98, P = 0.0001) and was not significantly different by t-test (P = 0.3). The mean percentage difference between the three sample LSM AUCs and the nine sample AUCs in each of our patients was 7.5%, (range −10.99% to +9.4%). Trough levels correlated extremely well with AUC (r = 0.95, P = 0.0001). Individual BU dosing, based on AUC, is necessary in most children to achieve targeted levels of BU therapy. An LSM of three BU concentrations performed at 0.5 h, 1 h and 6 h post-BU test dose closely predicts the AUC calculated from nine sampling points. Keywords: children; busulphan; limited sampling method; individualised dosing; pharmacokinetics; bone marrow transplantation

for more than 20 years.1–5 Early studies in children, using the standard total adult busulphan dose of 16 mg/kg, reported decreased toxicity, increased graft rejection and increased neoplastic relapse in children compared to adults.2–11 Subsequent pharmacokinetic studies have confirmed that, at this dose, children achieve significantly lower busulphan systemic exposure as measured by the area under the curve (AUC) of busulphan vs time.12–20 Busulphan dosing, based on surface area, in children (busulphan 600 mg/m2) has been associated with BU AUCs that were similar to adults and with acceptable toxicity, but the wide inter-patient variability in AUCs persists.14–18 Age, disease, circadian rhythms, food, drugs and pretransplant liver abnormalities have been identified as causes of variability in busulphan disposition.20 Grochow19 reported an increased risk of hepatic venoocclusive disease (VOD) in patients whose AUC was greater than 1500 mm/min and a reduction in the risk of VOD in adults who received dose adjustment to target an AUC 900–1500 mm/min. Many authors have suggested that individual busulphan dosing based on therapeutic drug monitoring may be necessary to achieve maximal busulphan systemic exposure while minimising toxicity.18–21 To this end, we initiated a study of individualising busulphan dosing based on measured BU AUC, with an intended target range of 900 to 1400 mm/min. In most centers nine to 12 blood samples are drawn to derive BU AUC. This frequency of blood sampling, even if drawn from a central venous line, may create anxiety in both the children and parents. The procedure is timeconsuming for staff, and the multiple drug assays are expensive. Therefore, one of our objectives was to develop a limited sampling method for determining busulphan AUC which would yield results equivalent to the traditional ninesample model derived AUC. Patients and methods

Busulphan (BU), an alkylating agent, has been used in allogeneic and autologous bone marrow transplant conditioning regimens, for both malignant and non-malignant diseases, Correspondence: Dr G Koren, Division of Clinical Pharmacology, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8 Received 6 January 1997; accepted 11 April 1997

Patients Ten consecutive children undergoing autologous or allogeneic bone marrow transplantation at the Hospital for Sick Children, Toronto were enrolled in the study (Table 1). Ethics approval for this study was obtained from the institutional Research Ethics Board. Informed consent was obtained from the parents of all the children in the study and assent was obtained from all children 7 years or older.

Limited sampling for individualised Bu dosing DS Chattergoon et al

348

Table 1

Patient characteristics

Patient No.

Age (years)

Diagnosis

Donor

Conditioning

2 1.5 4.4 11.5 5.3 6.6 1.3 7.3 2.2

HS MLD ANLL-CR1 ANLL-CR1 ANLL-CR1 ALL-CR2 ANLL-CR1 ANLL-CR1 Osteo

MUD MUD AUTO ALLO AUTO ALLO AUTO AUTO MUD

BU/CY BU/CY BU/CY BU/CY BU/CY VP16/BU/CY BU/CY BU/CY BU/CY/TBI

1 2 3 5 6 7a 8 9 10

HS = Hurlers syndrome; MLD = metachromatic luekodystrophy; ANLL = acute nonlymphocytic leukemia; ALL = acute lymphoblastic leukemia; Osteo = osteopetrosis; BU = busulphan; CY = cyclophosphamide (50 mg/kg × four doses); TBI = total body irradiation (300 cGy, single dose); VP-16 = etoposide (60 mg/kg × one dose); CR1 = first complete remission; CR2 = second complete remission. Patient 4 was excluded from pharmacokinetic analysis because of vomiting. a See text: Bulsulphan administration.

Busulphan administration All patients received a ‘test/therapeutic’ dose of busulphan (40 mg/m2, to the nearest tablet; Burroughs-Wellcome, Montreal, Quebec; 2 mg/tablet) prior to their busulphan course. Body surface area was calculated using the formula of Haycock et al,22 based on weight and height. Patients were kept nil per os (NPO) for a minimum of 2 h before and 1 h after administration of the test dose. The test dose was given as crushed tablets in water, via a nasogastric tube, as fast as possible or over a maximum of 5 min. Patient 4 received the dose orally. Patient 7 received VP16/busulphan/cyclophosphamide conditioning. To avoid any as yet unknown interaction between VP-16 and busulphan, the busulphan test dose was performed prior to the VP-16 administration, and the patient received no busulphan for 48 h after the VP-16 allowing VP-16 to be eliminated from the body (t.b = 11.5 h). Most patients received the dose at 0600 h but two patients received their dose at 1300 h. All children received phenytoin, during the busulphan course for seizure prophylaxis. The modified busulTable 2

phan dose, based on calculations described below, was given q 6 h × 15 doses beginning at 0500 h, 36–48 h following the test dose. Antiemetic prophylaxis given was granisetron + dexamethasone (five patients), ondansetron + dexamethasone (two patients), metoclopramide + diphenhydramine + dexamethasone (three patients). Blood sampling Nine blood samples (2–3 ml heparinised blood) were drawn through the patient’s central venous line, via an INTERLINK tubing system, just before administering the busulphan test dose and 0.25, 0.5, 1, 1.5, 2, 3, 4 and 6 h postdose. Trough busulphan levels were the 6-h post-dose levels. Samples were sent immediately to the laboratory, on ice, and were centrifuged within 1.5 h. The plasma was frozen at −20°C until assayed, usually during the same or the next day. Busulphan assay Plasma busulphan concentrations were determined using gas chromatography with electron capture detection (GCECD) with 1,5-pentanediol dimethanesulfonate as the internal standard (a modification of a previously described method).23,24 The extraction process was as follows: 250 ml of patient’s plasma and 10 ml internal standard (25 mg/ml) were mixed with 750 ml sodium iodide 4 M. The mixture was vortexed for 30 s and 400 ml hexane was added. The mixture was vortexed again, and incubated for 60 min at 70°C. The sample was cooled to room temperature and centrifuged. One microliter of the hexane layer was injected into a gas chromatograph with an electrochemical detector. The detection limit of the assay was 0.1 mmol. Commercial plasma was spiked with known concentrations of busulphan (0.1, 0.2, 0.5, 1.0, 2.0, 5.0, 10.0 mmol) and processed as above, simultaneously with the patient’s plasma. The results were analysed by linear regression to obtain an equation for the standard curve. This equation was used to calculate the busulphan concentrations from the patient’s samples. The interday coefficient of variation for low and high quality assurance samples (n = 44) respectively were

BU AUC achieved with test dose and recommended dose adjustments

Patient No.

Test dosea (mg/kg)

AUCKinfit mm/min

% dose change

Adjusted BU dose: mg/m 2/dose

Adjusted BU dose: mg/kg/dose

1 2 3 5 6 7 8 9 10 Median Range

1.59 1.77 1.55 1.14 1.61 1.44 1.83 1.30 1.71 1.59 1.14–1.83

1036.3 1238.1 1475.9 1090.9 916.2 598.8 1041.1 1941 852.9 1041 599–1941

0 −10.1 −14.3 0 +21.4 +45.3 0 −33.3 +33.3

39.3 34.6 32.9 40.0 47.2 55.2 39.13 27.18 55.17 39.3 27.2–55.2

1.59 1.59 1.33 1.14 1.95 2.03 1.83 0.95 2.29 1.59 0.95–2.29

a

Test dose was based on 40 mg/m2 /dose. Mean % dose change in six patients: ±26% (range, −33 to +45).

−33–+45


as follows: low QA, 3.6%; high QA, 2.7%. The intraday coefficient of variation for low and high quality assurance samples (n = 5) respectively were as follows: low QA, 4.6%; high QA, 1.5% (Dr Slattery, Fred Hutchinson Cancer Research Center, Seattle, WA, USA, personal communication).

Comparison of limited sampling methods Using the same methods described below, four LSMs were compared to the standard AUCkinfit by using data from selected time points in our patients in the formulae developed by Vassal et al16 (study in children), Schuler et al21 (study in adults) and our three-sample method (above).

Pharmacokinetics

Statistical analyses

The area under the curve of busulphan to infinity (AUC) was calculated using two methods. (1) Manually using the trapezoidal rule until the last data point and estimation of the AUC from the last data point to time infinity based on the elimination rate constant (Ke) obtained from the log BU concentration vs time curve;25 and (2) computer fitting of the busulphan concentration-time data to a one compartmental model with first order absorption and elimination kinetics, using a nonlinear regression analysis program (KINFIT).26 Results were expressed as mm/min.

The limited sampling AUCs were compared to the ninesample Kinfit-derived AUC (regarded as our gold standard) using the Pearson correlation co-efficient27 and the two tailed paired t-test.28 To assess the limited sampling methods, the percentage difference (% diff) between the ‘LSM-derived AUC0–` (limited AUC) and the ‘Kinfit-derived AUC0–` (AUCkinfit) was calculated for each patient. The difference, whether positive or negative, was used to calculate the mean percent difference (mean % diff) between the limited AUCs and the AUCkinfit. A post hoc power analysis, for a two-tailed test, was performed to determine the power of this study to detect a difference of 10% of the mean percentage difference between the AUCkinfit and our three-sample LSM AUC. AUC0–` was correlated with trough levels using least square regression analysis.

Calculation of the modified dose The adjusted BU dose was calculated as follows (rounded to the nearest tablet): Adjusted dose =

target AUC × test dose. test dose AUC

The intended acceptable target BU AUC range was 900– 1400 mm/min. Calculations were based on achieving a target AUC of 1300 mm/min. Some patients’ doses were calculated to achieve a lower target, due to concerns of toxicity (see Discussion for the rationale of the different targets used). Development of the limited sampling method Limited sampling AUCs were calculated using a combination of the trapezoidal rule (for concentrations to the penultimate measured BU concentration, Cx, where x is the time at the second to last measured busulphan concentration data point for the particular data set) and logarithmic rule (from the penultimate measured BU concentration to infinity using the formula: Cx/Ke). This was done using twosample, three-sample and four-sample time points as shown below. The limited AUCs were compared as described under statistical analysis. The formulae used were as follows: (1) Two-sample (1, 6 h): 30C1h + 300C1h/(Ln C1h Ln C6h) (2) Three-sample (1, 1.5, 6 h): 45C1h + 15C1.5h 270C1.5h/(Ln C1.5h − Ln C6h) (3) Four-samples: (a) (1, 1.5, 2, 6 h): 45C1h + 30C1.5h + 15C2h 270C2h/(Ln C2h − Ln C6h) (b) (1, 1.5, 3, 6 h): 45C1h + 60C1.5h + 45C3h 180C3h/(Ln C3h − Ln C6h) (c) (1, 1.5, 4, 6 h): 45C1h + 30C1.5h + 75C4h 120C4h/(Ln C4h − Ln C6h)

− +

Clinical monitoring The patients were monitored for hepatic VOD and mucositis according to previously published criteria. 8 The clinical criteria used for the diagnosis of VOD were the presence of two or more of the following occurring within 21 days post-transplantation in the absence of other causes of liver disease: bilirubin .2 mg/dl, unexplained weight gain 5% of baseline weight or presence of ascites, and hepatomegaly or right upper quadrant tenderness. Complete blood counts were monitored daily with differential counts every alternate day. Bilirubin levels were monitored twice weekly or as clinically indicated.

Results Pharmacokinetics The data from patient 4 was excluded from analysis because of vomiting of the BU test dose. The natural logarithm of the measured busulphan concentrations vs time, for the other nine patients, is shown in Figure 1. Eight of nine patients had an observed tmax (time to observed maximal concentration) at either 60 or 90 min. The busulphan concentrations obtained fitted well to the one compartmental, first order pharmacokinetic model used.

+ + +

Busulphan AUC The computer derived BU AUC0–` (AUCkinfit), for each patient is shown in Table 2. The mean (± s.d.) AUC obtained with the test dose of 40 mg/m2 was 1132 (±389)

349


Table 3

Comparison of limited AUCs vs AUCkinfit (this study)

Calculated AUCs mmol/min/l

Mean AUC (± s.d.)

Pearson r

t-test P value

Mean % difference

Range %

AUC0–` Kinfit

1132 (±389)

—

—

—

—

AUC0–` manual

1151 (±372)

0.98 (P = 0.0001)

0.41

3.3

−3 to 13

Limited AUC 1, 6 h

1065 (±363)

0.95 (P = 0.0001)

0.15

8.4

−30 to 9

Limited AUC 1, 1.5, 6 h

1098 (±344)

0.98 (P = 0.0001)

0.29

7.5

−11 to 9

Limited AUC 1, 1.5, 2, 6 h

1096 (±332)

0.97 (P = 0.0001)

0.30

7.4

−10 to 9


1110 (±349) 1103 (±304)

0.98 (P = 0.0001) 0.95 (P = 0.0001)

0.46

7.1

−9 to 12

0.53

8.9

−15 to 14


2

In(X) of BU concentrations (µmol/l)

Development of the limited sampling method Figure 2 shows the linear regression plot of the three-sample AUC vs AUCkinfit in which the correlation (r = 0.976, P = 0.001) was excellent. The limited AUCs derived from the two-sample, three-sample and the various four-sample limited sampling methods outlined in Table 3 were highly significantly correlated with, and were not statistically different from AUCkinfit. However, when the percentage difference between the limited AUC and AUCkinfit was calculated for each individual patient, the range of the percentage difference varied according to the limited sampling method used. The three-sample limited AUC and the four-sample limited AUC using either the BU concentration at 120 min or 180 min resulted in AUCs that most closely approximated the AUCkinfit with a mean absolute percentage difference of 7.5, 7.4 and 7.07% respectively. The range of the percentage differences were also quite similar and close to ±10% of our standard, AUCkinfit.

1.5

1

0.5

0

–0.5

–1 0

50

100

150

200

250

300

350

400

2000

Time (mins)

mm/min. The mean (± s.d.) AUC0–` using the trapezoidal rule was 1151 (±372) mm/min. Busulphan dose adjustment Six of nine evaluable patients needed dose adjustments. The actual targeted BU AUC and the recommended dose changes are shown in Table 2. In the six patients whose doses were adjusted, the mean dose change (calculated from the absolute value of the dose change, whether positive or negative) was 26.2% (range −33.3% to +45.3%). Patient 2 received a dose adjustment based on a manual calculation of AUC prior to obtaining the KINFIT program. The Kinfit-derived AUC is given in Table 2.

y = 0.862× + 121.508 Correlation co-efficient, r = 0.976, P = 0.0001

1800

Figure 1 Linear plot of natural logarithm of measured BU concentrations vs time of nine patients.

AUC three-sample

350

1600

1400

1200

1000

800

600 400

600

800

1000

1200

1400

1600

1800

AUCkinfit Figure 2

Linear regression: AUC three-sample vs AUCkinfit.

2000


2000

The power analysis, for a two-tailed test comparing the means of two related groups and to detect a difference of 10% in the mean percent difference between the means of the AUCkinfit and our three-sample LSM AUC yielded a Zb value of 9.2. This corresponds to a b of ,0.005 (Zb0.005 = 2.58). Therefore, this study has a power of .99% to detect at the 5% significance level a difference in the mean percent difference between our standard AUCkinfit and our three-sample LSM AUC of 10%.

1800

1600

1400

AUC µM/min

351

Results of statistical power analysis

1200

1000

Correlation between AUC0–` and busulphan trough concentrations (test dose)

800

The AUCkinfit of the test dose was significantly correlated (r = 0.945, P = 0.0001) with the corresponding trough BU concentration level as shown in Figure 4.

600

400 0

1

2

3

4

5

6

7

8

9

10

Toxicity

Patient No.

Six patients had mucositis (two mild and four severe) requiring morphine, two required morphine doses .50 mg/kg/h. Patient 7 developed liver disease of uncertain etiology. Other toxic manifestations included: pulmonary haemorrhage, one; mild hemorrhagic cystitis, two; and leukoencephalopathy, one. We encountered no VOD during this study despite the markedly higher total BU doses used (15.5 to 36 mg/kg) compared to our previous standard dose (16 mg/kg).

AUCkinfit µ M/min (this study) AUC 1, 1.5, 6 h µM/min (this study) AUC 0.5, 6 h µM/min (Ref. 16) AUC 1,4 h µM/min (Ref. 21) AUC 1, 2, 4 h µM/min (Ref. 21)

Figure 3 AUC in each patient by four limited sampling methods and AUCkinfit.

Comparison of published limited sampling methods

Patient outcome

Figure 3 and Table 4 show the results of applying our busulphan concentration time data to the formulae reported by Vassal et al16 (study in children) and Schuler et al21 (study in adults). It also shows the percentage differences of the AUCs calculated using these methods to the AUCkinfit.

Patient 1 had autologous marrow recovery. The dose for this patient was not adjusted because of possible toxicity. Five patients (two allogeneic and three matched unrelated donor) are fully engrafted. The medians (ranges) of the time to the following outcome measures were: reticulo-

Table 4

Four limited sampling methods compared to AUCkinfit

Methods

Sampling times (h)

Mean AUC ± s.d.

Mean % difference

Range, %

Pearson: r, (P)

t-test P

AUC0–` Kinfit

nine samples

1132 (±389)

Present study

1, 1.5, 6

1098 (±344)

7.52 (±2.7)

−11 to 9.4

0.98 (0.0001)

0.29

Vassal et al

0.5, 6

1132 (±437)

9.58 (±9.3)

−26 to 16

0.95 (0.0001)

0.99

Schuler et al

1, 4

931 (±156)

17.55 (±10.2)

−40 to 22

0.86 (0.0001)

0.05

Schuler et al

1, 2, 4

884 (±170)

18.97 (±12.9)

−43 to 12

0.8 (0.0001)

0.03

Limited sampling methods: formulae: Present study 45C 1h + 15C1.5h + 207C1.5h/(Ln C1.5h − Ln C6h) AUC = 122 + 0.97C 0.5h + 13.94 C6h Vassal et al 16 Schuler et al 26 AUCLSM 1, 4 = 782 + 1.42C 1h + 3.74C4h AUCLSM 1, 2, 4 = 289 + 1.16C1h + 1.06C2h + 3.16C 4h


352

2000

y = 802.895× + 235.32, r2 = 0.895

1800

AUCkinfit µM/min

1600

1400

1200

1000

800

600

400 0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Trough BU level µ mol/l Figure 4

Linear regression of AUCkinfit vs trough BU concentrations.

cyte count of 10 × 109/l: 20 days (range: 14–56); absolute granulocyte count of .0.5 × 109/l: 18 days (range: 10–52); and platelet count of .50 × 109/l: 29 days (range: 21–68). Three of four patients (3, 6 and 9) with autotransplants for AML failed to achieve a platelet count of 50 × 109/l by 2.5 to 9 months post-transplant. Of these, patients 6 and 9 require ongoing platelet and red cell transfusions. Discussion Busulphan pharmacokinetics in children are significantly different from those in adults. There is wide inter-patient variability of busulphan kinetics in children. Recent evidence supports the hypothesis that toxicity and outcome in children may depend on the BU AUC achieved.29 Many authors have therefore recommended busulphan therapeutic drug monitoring and individualised busulphan dose adjustment to optimise BU systemic exposure while minimising regimen-related toxicity. Our study shows clinically significant busulphan dose adjustments (−33.3% to +45.3%) are necessary in a large proportion of children (.66%) despite surface area based dosing. This finding suggests that individualised dose adjustment is necessary in children receiving high dose busulphan therapy prior to bone marrow transplantation. Schuler et al21 recently suggested that busulphan monitoring is not warranted except for research purposes. Their study population included only two patients 18 years of age or younger. This conclusion may be appropriate only in adults. The standard nine to 12 sample pharmacokinetic study used to derive BU AUC, is the most accurate. Disadvantages of performing a full pharmacokinetic study include the following: more blood sampling is required; it is more time-consuming for staff and it is much more expensive

than only three samples. Limited sampling methods may be almost as accurate and overcome these disadvantages. Trough BU levels have not been evaluated. The coefficient of variation of many tests regarded as being accurate is in the order of ±10%. Therefore, a limited sampling method with an accuracy of ±10% is reasonable. We compared our LSM AUCs with our AUCkinfit and LSMs from the literature using our data. Our three-sample and four-sample methods resulted in LSM AUCs that were highly significantly correlated and not statistically different from our standard (AUCkinfit). However, our LSM AUCs had a narrower range and a smaller mean percentage difference when compared to the other methods described below. The LSM AUCs obtained using the method of Vassal et al16 were highly correlated and not statistically different from our standard. However, the range of differences between the LSM AUCs and the standard was much wider and with a higher mean percentage difference than our three-sample method. The LSM AUCs using both methods of Schuler et al21 were less well correlated, statistically different from the standard AUCkinfit (P < 0.05), and had a much wider range of differences from the standard. Their patient population was almost all adult and therefore their methods may not be applicable in children. Also, Schuler’s calculations were based on AUC0–6h at first dose, whereas we calculated AUC0–`. The AUC0–6h calculated using our busulphan concentration–time data (data not shown), was still significantly different from that calculated by the Schuler et al21 method. Hassan et al30 recently suggested a three-sample LSM for children at 1 h, 3 h and 6 h. Our data applied to this linear regression equation resulted in a mean BU AUC of 1097 mm/min which is not significantly different (P = 0.54) from our AUCkinfit. The mean percentage difference was 10.9%, which is higher than that of our three-sample method (7.5%) and that of Vassal et al16 (9.58%). The range of differences between the LSM AUCs and the standard was −20.3% to 12.4%. The differences in the AUCs obtained may be due to variations in the method of drug administration, and differences in the times to maximal concentrations among the studies. It must be noted that Vassal et al16 and Hassan et al30 use different units of measurement (concentration, ng/ml; AUC, ng/h/ml). The regression equations reported by these groups are based on these units and therefore one cannot apply concentrations in mm directly to these equations. To obtain an accurate AUC from these equations a correction factor (×60) must be used to obtain an AUC in mm/min. This is based on the following conversions: 1 mm = 246.3 ng/ml and 1 mm/min = 4.1 ng/h/ml. Our results suggest that either a three-sample or a foursample method can be used. Both methods yield similar results. There is no significant advantage to using the foursample method. There will be less blood sampling and other associated costs by using the three-sample method. Therefore, we now use our three-sample limited sampling method to test patients BU AUC before dose adjustment, and after dose adjustment to assess the effectiveness of dose adjustment. It may also be used to carry out further dose adjustments. The AUCkinfit was highly correlated with the trough


busulphan concentrations achieved. Further study is necessary to test whether trough busulphan levels are a reliable predictor of BU AUC. Trough levels could become very useful for monitoring BU systemic exposure and perhaps for further dose adjustment during the course of busulphan therapy. Eight of nine of our patients had observed times to maximal concentration (tmax ) at either 60 or 90 min. Previously reported tmax values varied widely ranging from 30 min to 180 min30 and as long as 255 min.12 It is possible that the formulation of the busulphan tablets and their dissolution characteristics may be a factor. Ehninger et al31 recently reported studies in a canine model using a solution of busulphan in dimethyl sulphoxide (DMSO). They suggested that the problems of physico-chemical dispersion of the tablets is a source of inter-patient and intra-patient variability. The tmax after oral dosing ranged from 30 to 90 min and AUCs achieved were ,10% different from that obtained by intravenous administration of the same dose of this solution. Our findings of a fairly consistent tmax in our patients, after administration of crushed tablets in water, may provide human data to support this theory. DMSO may not be necessary. Perhaps, busulphan packaged in powdered form in a gel capsule may help to reduce this variability in busulphan absorption but any variability in gel solubility may complicate this approach. Hassan et al32 reported the variability of busulphan bioavailability (bioavailability, F = 0.22 to 1.20) in children. This may be a major factor in the variability in observed BU AUC. The optimal BU AUC exposure required for efficacy in children is not known and may be different for different diseases. We have chosen to target an AUC (1300 mm/min) between the mean adult level (1268 mm/min19) and the level reportedly associated with an increased risk of serious toxicity. An upper limit of AUC of 1400 mm/min was used as a margin of safety to reduce the risk of VOD assuming a ±10% laboratory error/variability. For patients in whom there were concerns about the high busulphan dose, the dose was calculated to target an AUC of 1100 mm/min, based on data where the mean AUC achieved in children with a dose of 40 mg/m2 was 1105 mm/min.18 Failing this, the minimum AUC targeted was 900 mm/min, the lowest value of our ‘acceptable range’. We have partly validated this therapeutic window in children, but more clinical studies are needed. There is evidence that higher busulphan exposure yields better results in unrelated transplants.29 As further data are acquired, it is likely that analysis will suggest that different levels of busulphan exposure may be efficacious for different diseases. By adjusting doses based on measured AUC, the inter-patient variability in BU AUC achieved will be markedly reduced and the toxicity data from this approach will be helpful in guiding therapy in the future. An increased risk of VOD was reported in patients achieving BU AUC .1500 mm/min.19 We thought it prudent to keep the BU exposure level below 1400 mm/min for an extra margin of safety. This was the reason for decreasing doses in three of six of our patients who had dose adjustment. No patient in this series developed VOD with the current doses ranging from 0.91 to 2.3 mg/kg/dose

(total BU doses ranging from 15.5 to 36 mg/kg). This includes those patients whose target AUC was near 1300 mm/min. Previously, our standard busulphan doses were 1 mg/kg/dose for patients .3 years old and 1.25 mg/kg/dose for children ,3 years old. Our patients now receive busulphan to target an AUC of 1300 mm/min. Yeager et al18 reported VOD in one of seven patients and Slattery et al29 reported VOD in five of 28 patients (children ,18 years of age). These patients received fixed dose busulphan, 27.7 mg/kg and 14 to 17 mg/kg total dose, respectively. Recently, Slattery et al29 reported an increased risk of grade 3–4 toxicity (including VOD) in adults and children receiving an average steady state busulphan concentration .900 ng/ml (equivalent to BU AUC of approximately 1320 mm/min). Data were also presented on two children which suggests VOD may develop at BU levels much less than 900 ng/ml. Cyclophosphamide, etoposide, total body irradiation and other drugs may cause VOD independently or contribute to its development. Previous liver disease or previous liver irradiation may also be factors. Further study taking these factors into account is necessary to determine whether targeting BU AUC will result in decreasing VOD in children. The incidence of mucositis in this study (67%) is comparable to that reported in previous studies (57%,18 64%30). The other toxicities reported are more likely related to irradiation and cyclophosphamide. Patients with mucopolysaccharide storage disease have problems with graft rejection and autologous marrow recovery, especially in unrelated transplants.33 Patient 1 may be representative of this. He had Hurler’s syndrome and received an unrelated donor transplant. He did not have T cell depletion or any other predisposing factors to graft rejection. This patient was our first study patient. We would have targeted a higher AUC if we had previous experience using the much larger doses that would have been required. Subsequently, Slattery et al29 reported that higher levels of busulphan exposure were associated with better engraftment rates in unrelated donor transplants. Perhaps the risk of graft failure in storage diseases with unrelated donor transplants requires maximal target BU AUC. We have changed our protocol in patients with storage diseases to busulphan, cyclophosphamide 60 mg/kg × two doses and total body irradiation 300 cGy × one dose. We do not use antithymocyte globulin or T cell depletion with this protocol. In those patients who engrafted, the time to engraftment is comparable to other published data.18 We think that the engraftment problems in three patients were related to aggressive chemotherapy of their acute nonlymphocytic leukemia prior to marrow harvest and probably not related to BU-CY conditioning. Long-term outcome data, especially relapse in malignancy, is not available. Further follow-up is necessary. Our data strongly suggest that individualised busulphan dosing is necessary in most children to achieve maximal targeted systemic exposure. Toxicity and engraftment rates are acceptable within the targeted range of busulphan AUC. The proposed three-sample limited sampling method gives an accurate estimate of busulphan AUC with improved results compared to previous reported methods. This

353


354

method is easier to perform, reduces blood sampling and saves money and time. We are currently prospectively assessing the accuracy of the AUC achieved after dose adjustment vs the targeted AUC using this limited sampling method.

14 15

Acknowledgements

16

This work was supported by grants from the Physicians’ Services Incorporated Foundation and Paediatric Consultants, the Hospital for Sick Children. Dr Chattergoon is a Terry Fox Physician-Scientist Research Fellow. We also thank the nursing staff of the Bone Marrow Transplant Unit and Dr Shinya Ito from the Division of Clinical Pharmacology for their assistance and support during this study.

17 18

19

References 1 Porcellini A, Manna A, Moretti L et al. Busulphan and cyclophosphamide as conditioning regimen for autologous bone marrow transplantation in acute lymphoblastic leukemia. Bone Marrow Transplant 1989; 4: 331–333. 2 Lucarelli G, Galimberti M, Delfini C et al. Marrow transplantation for thalassemia following busulphan and cyclophosphamide. Lancet 1985; 1: 1355–1357. 3 Blazar BR, Ramsay NKC, Kersey JH et al. Pretransplant conditioning with busulphan (myeleran) and cyclophosphamide for nonmalignant diseases. Transplantation 1985; 39: 597– 620. 4 Shaw PJ, Hugh-Jones K, Hobbs JR et al. Busulphan and cyclophosphamide cause little early toxicity during displacement bone marrow transplantation in 50 children. Bone Marrow Transplant 1986; 1: 193–200. 5 Hobbs JR, Shaw PJ, Hugh-Jones K et al. Engraftment rates related to busulphan and cyclophosphamide dosages for displacement bone marrow transplants in 50 children. Bone Marrow Transplant 1986; 1: 201–208. 6 Brugieres L, Hartmann O, Benhamou E et al. Veno-occlusive disease of the liver following high-dose chemotherapy and autologous bone marrow transplantation in children with solid tumours: incidence, clinical course, and outcome. Bone Marrow Transplant 1988; 3: 53–58. 7 Jones RJ, Lee KSK, Beschorner WE et al. Veno-occlusive disease of the liver following bone marrow transplantation. Transplantation 1987; 44: 778–783. 8 McDonald GB, Sharma P, Matthews DE et al. Veno-occlusive disease of the liver after bone marrow transplantation: diagnosis, incidence, and predisposing factors. Hepatology 1984; 4: 116–122. 9 McDonald GB, Sharma P, Matthews DE et al. The clinical course of 53 patients with veno-occlusive disease of the liver after bone marrow transplantation. Transplantation 1985; 39: 602–608. 10 Rollins BJ. Hepatic veno-occlusive disease. Am J Med 1986; 81: 297–306. 11 Vassal G, Hartmann O, Benhamou E. Busulphan and venoocclusive disease of the liver. Ann Intern Med 1990; 112: 881. 12 Vassal G, Gouyette A, Hartmann O et al. Pharmacokinetics of high-dose busulfan in children. Cancer Chemother Pharmacol 1989; 24: 386–390. 13 Grochow LB, Jones RJ, Brundrett RB et al. Pharmacokinetics

20 21 22

23 24 25 26 27 28 29

30

31 32 33

of busulfan: correlation with veno-occlusive disease in patients undergoing bone marrow transplantation. Cancer Chemother Pharmacol 1989; 25: 55–61. Grochow LB, Krivit W, Whitley CB, Blazar B. Busulfan disposition in children. Blood 1990; 75: 1723–1727. Hassan M, Oberg G, Bekassy AN et al. Pharmacokinetics of high-dose busulphan in relation to age and chronopharmacology. Cancer Chemother Pharmacol 1991; 28: 130–134. Vassal G, Deroussent A, Challine D et al. Is 600 mg/m2 the appropriate dose of busulfan in children undergoing bone marrow transplantation? Blood 1992; 79: 2475–2479. Regazzi MB, Locatelli F, Buggia I et al. Disposition of highdose busulfan in pediatric patients undergoing bone marrow transplantation. Clin Pharmacol Ther 1992; 53: 45–52. Yeager AM, Wagner JE Jr, Graham ML et al. Optimisation of busulphan dosage in children undergoing bone marrow transplantation: a pharmacokinetic study of dose escalation. Blood 1992; 80: 2425–2428. Grochow LB. Busulfan disposition: the role of therapeutic drug monitoring in bone marrow transplantation induction regimens. Semin Oncol 1993; 20 (Suppl. 4): 18–25. Vassal G, Fischer A, Challine D et al. Busulphan disposition below the age of three: alteration in children with lysosomal storage disease. Blood 1993; 82: 1030–1034. Schuler U, Schroer S, Ku¨hnle A et al. Busulphan pharmacokinetics in bone marrow transplant patients: is drug monitoring warranted? Bone Marrow Transplant 1994; 14: 759–765. Haycock GB, Schwartz GJ, Wisotsky DH. Geometric method for measuring body surface area: a height–weight formula validated in infants, children and adults. J Paediatr 1978; 93: 62–66. Embree L, Burns RB, Heggie JR et al. Gas-chromatographic analysis of busulphan for therapeutic drug monitoring. Cancer Chemother Pharmacol 1993; 32: 137–142. Ehrsson H, Hassan M. Determination of busulphan in plasma by GC-MS with selected ion monitoring. J Pharmaceut Sci 1983; 72: 1203–1205. Gibaldi M, Perrier D. Pharmacokinetics, Drugs and the Pharmaceutical Sciences, 2nd edn. Marcel Dekker: New York, 1982. D’Argenio DZ, Schumitzky A. A program package for simulation and parameter estimation in pharmacokinetic systems. Comp Prog Biomed 1979; 9: 115–134. Dawson-Saunders B, Trapp RG (eds). Summarizing data. Basic and Clinical Biostatistics, 2nd edn. Appleton and Lange: Norwalk, CT, USA, 1994, pp 52–53. Dawson-Saunders B, Trapp RG (eds). Estimating and comparing means. Basic and Clinical Biostatistics, 2nd edn. Appleton and Lange: Norwalk, CT, USA, 1994, pp 107–110. Slattery JT, Sanders JE, Buckner CD et al. Graft rejection and toxicity following bone marrow transplantation in relation to busulfan pharmacokinetics. Bone Marrow Transplant 1995; 16: 31–42. Hassan M, Fasth A, Gerritsen B et al. Busulphan kinetics and limited sampling model in children with leukemia and inherited disorders. Bone Marrow Transplant 1996; 18: 843–850. Ehninger G, Schuler U, Renner U et al. Use of a water-soluble busulfan formulation. Pharmacokinetic studies in a canine model. Blood 1995; 85: 3247–3249. Hassan M, Ljungman P, Bolme P et al. Busulphan bioavailability. Blood 1994; 84: 2144–2150. Krivit W, Henslee-Downey J, Klemperer M et al. Survival in Hurler’s disease following bone marrow transplantation in 84 patients. Bone Marrow Transplant 1995; 15 (Suppl. 1): 182–185.


355