NextDose: A web-based Bayesian dose forecasting tool

Last updated 28 September 2021



Target Concentration

TCI for busulfan is usually used as part of conditioning for bone marrow transplant. The conditioning aims to remove existing host stem cells so that transplanted stem cells can be grafted and form new cells. Initial exposure-response studies of busulfan led to an average steady state target of 0.77 mg/L (Bolinger, Zangwill et al. 2000). This is equivalent to an AUC over a 6 h dosing interval of 1125 umol/L*min which is the target recommended in the FDA label (FDA 2015) for busulfan used for bone marrow conditioning.

Busulfan is an alkylating agent which is thought to bind irreversibly to DNA thereby causing cell damage and death. This irreversible action would explain why the cumulative exposure during treatment would be a predictor of cell kill and thus bone marrow ablation. Thus the exposure target for busulfan is the cumulative AUC over the standard treatment period of 4 days. The AUC with a constant rate infusion over 4 days approaching a steady state concentration of 0.77 mg/L would be 74 mg/L*h.

A subsequent analysis proposed an acceptable range for a 4 day cumulative AUC of 78 to 101 mg/L*h (Bartelink, Lalmohamed et al. 2016). The middle of that acceptable range is 89.5 mg/L*h (usually rounded up to 90 mg/L*h) which is the target exposure equivalent to 0.94 mg/L average steady state concentration.

A 2 day cumulative AUC of 16.8 mg/L (equivalent to 0.35 mg/L average steady state) has been proposed for bone marrow conditioning for Fanconi anemia (Mehta, Emoto et al. 2019).

A dosing strategy for busulfan is to aim for a steady state average target of 0.94 mg/L with the target type option Css avg mg/L then measure concentrations after the first busulfan dose (Figure 1).

The predictive performance of NextDose for doses required to reach a target cumulative AUC has been evaluated and shown to have acceptable bias and precision (Lawson, Paterson et al. 2021).

Both software tools utilising Bayesian methods provided acceptable relative bias and precision of cumulative exposure estimations under the tested sampling scenarios. Relative bias ranged from median RE of 0.114.6% using InsightRX and from 3.47.8% using NextDose. Precision ranged from median RMSE of 0.190.32 mghL−1 for InsightRX and 0.080.1 mghL−1 for NextDose.

The authors note that the bias of the Bayesian dosing methods is based on comparison with a non-compartmental (NCA) method for calculating AUC. The NCA methods are known to be biased (~10%) which means the Bayesian dose predictions are probably more reliable (Bartelink, Lalmohamed et al. 2016).

Figure 1

The time course of observed busulfan concentrations after a single dose are shown with NextDose predictions in Figure 2.

Figure 2

The proposed dose to achieve an average steady state target of 0.94 mg/L is 233 mg/day (Figure 3).

Figure 3

The NextDose proposed dose with a 24 h dosing interval can be multiplied by 4 to predict the total dose required. Subsequent measurements of busulfan can be used to refine the daily dose requirement and subsequent doses calculated from the predicted 4 day total dose minus those already given.

An alternative method is to use a cumulative AUC target type e.g. mg/L*h (cum AUC 4 days) with a target value such as 90 mg/L*h (Figure 4).

Figure 4

You may choose any dosing interval that is appropriate e.g. 24 h and NextDose will calculate the dose for that dosing interval which will achieve the cumulative AUC target of 90 mg/L*h if the dose is repeated until day 4 at the specified dosing interval.

If the time remaining before the end of the cumulative AUC interval is less than 1.25 x the dosing interval a single final dose will be proposed. Otherwise the remaining dose to achieve the cumulative target will be divided among the remaining dosing intervals in the remaining time.

An example of using the cumulative AUC target is shown in Figure 5. It shows the AUC as a % of the cumulative target AUC (22%). The AUC% value is estimated using between occasion variability. This is followed by the proposed dose of 389 mg given once a day and the predicted remaining % of target exposure after this first dose (77.4%).


Figure 5

Note that the proposed dose of 389 mg is based on the assumption that clearance will remain unchanged. The McCune model for busulfan includes a time varying change in clearance. In order to account for this it is necessary to repeat the NextDose calculation for each of the proposed doses. This is illustrated in the following sequence of figures.

Figure 6 Predicted time course and proposed dose following the second dose of 389 mg.

Figure 7 Predicted time course and proposed dose following the third dose of 338 mg.

Figure 8 Predicted time course and proposed dose following the fourth dose of 318 mg.

In this case the proposed doses are predicted to achieve within 1% of the target exposure after 4 daily doses. Note how the proposed dose decreased with time reflecting the decrease in busulfan clearance that has been reported over this time period (McCune, Bemer et al. 2014).

If further concentration profiles are obtained this may reveal additional changes in clearance and thus in exposure. For the purposes of the following example the observed concentrations are the same after each dose. Because the observed concentrations do not increase in proportion to the dose this is explained by an increase in clearance over time. This is confirmed in the table of empirical Bayes estimates of clearance at the bottom of each of the following figures. Note that typically busulfan clearance decreases with time but that does not mean this occurs in all patients.

Figure 9 Observed concentrations, predicted time course and proposed dose following the second dose of 389 mg.

Figure 10 Observed concentrations, predicted time course and proposed dose following the third dose of 416 mg.

Figure 11 Observed concentrations, predicted time course and proposed dose following the fourth dose of 498 mg.


After the fourth daily dose the predicted exposure is under target by 7.65% and an additional dose of 136 mg is proposed to achieve the target.

Figure 12 Predicted time course and proposed dose following the fourth dose of 625 mg.

Figure 12 shows that if this proposed dose (136 mg) was given then the predicted exposure would reach the target within 1%.


NextDose Model

The NextDose pharmacokinetic model is based on (McCune, Bemer et al. 2014). This is a large study including infants to elderly adults. It accounts for body size and composition using fat free mass, maturation of clearance in infants and a small decrease in clearance over 4 days of treatment.


Bartelink, I. H., A. Lalmohamed, E. M. van Reij, C. C. Dvorak, R. M. Savic, J. Zwaveling, R. G. Bredius, A. C. Egberts, M. Bierings, M. Kletzel, P. J. Shaw, C. E. Nath, G. Hempel, M. Ansari, M. Krajinovic, Y. Theoret, M. Duval, R. J. Keizer, H. Bittencourt, M. Hassan, T. Gungor, R. F. Wynn, P. Veys, G. D. Cuvelier, S. Marktel, R. Chiesa, M. J. Cowan, M. A. Slatter, M. K. Stricherz, C. Jennissen, J. R. Long-Boyle and J. J. Boelens (2016). "Association of busulfan exposure with survival and toxicity after haemopoietic cell transplantation in children and young adults: a multicentre, retrospective cohort analysis." Lancet Haematol 3(11): e526-e536.

Bolinger, A. M., A. B. Zangwill, J. T. Slattery, D. Glidden, K. DeSantes, L. Heyn, L. J. Risler, B. Bostrom and M. J. Cowan (2000). "An evaluation of engraftment, toxicity and busulfan concentration in children receiving bone marrow transplantation for leukemia or genetic disease." Bone Marrow Transplant 25(9): 925-930.

FDA (2015). "Busulfex Product Label"

Lawson, R., L. Paterson, C. J. Fraser and S. Hennig (2021). "Evaluation of two software using Bayesian methods for monitoring exposure and dosing once-daily intravenous busulfan in paediatric patients receiving haematopoietic stem cell transplantation." Cancer Chemotherapy and Pharmacology.

McCune, J. S., M. J. Bemer, J. S. Barrett, K. Scott Baker, A. S. Gamis and N. H. G. Holford (2014). "Busulfan in Infant to Adult Hematopoietic Cell Transplant Recipients: A Population Pharmacokinetic Model for Initial and Bayesian Dose Personalization." Clinical Cancer Research 20(3): 754-763.

Mehta, P. A., C. Emoto, T. Fukuda, B. Seyboth, A. Teusink-Cross, S. M. Davies, J. Wilhelm, K. Fuller, A. A. Vinks and F. Boulad (2019). "Busulfan Pharmacokinetics and Precision Dosing: Are Patients with Fanconi Anemia Different?" Biol Blood Marrow Transplant 25(12): 2416-2421.

Copyright All rights reserved | Developed by Sam Holford & Nick Holford 2012-2021