NextDose: A web-based Bayesian dose forecasting tool

Last updated 9 August 2020


Methotrexate is the pioneering example of the use of target concentration intervention (TCI) to substantially improve clinical outcome. In 1998 Evans and his team of pharmacists and medical doctors revealed the results of a clinical trial of TCI showing the 5 year survival of children with acute lymphoblastic leukaemia (ALL) was extended from 66% to 76% simply by individualizing the dose to achieve a target exposure of methotrexate (Evans, Relling et al. 1998). This improvement in outcome is arguably bigger than any new medicine tested since then for ALL.

Target Concentration

The acceptable range for methotrexate used by Evans et al. was an AUC of 580 to 950 micromol/L*h. AUC was effectively integrated from 0 to infinity because doses were administered weekly. The target AUC was 800 micromol/L*h. Methotrexate clearance was estimated using concentrations measured at 1 and 6 h after the dose with a 2 compartment Bayesian model. The individual clearance was then used to predict future doses. The individualized dosing approach was not strictly TCI because it had a TDM element. If the estimated AUC was within the acceptable range then the dose was not changed. An individualized dose was only used if the estimated AUC was outside the acceptable range.

NextDose Model

The pharmacokinetic model was developed in a group of children administered high dose methotrexate for the treatment of a variety of cancers (Hirankarn, Holford et al. 2012). Theory based allometry was used to develop a two compartment model with first-order elimination after intravenous infusions of methotrexate of known duration. Total body weight was the best predictor of clearance and central and peripheral volumes. Intercompartmental clearance was more closely associated with fat free mass. Maturation of clearance was not identifiable because the majority of children were older than 2 years. Although methotrexate is known to be renally eliminated the use of creatinine clearance to predict renal function accounted for only 20% of total clearance and explained 3% of the population parameter variability of clearance. This is explicable by the narrow distribution of renal function in this group of children. Both between subject and between occasion variability were important to describe unpredictable differences in PK. In addition, it was essential to account for the covariance of the random effects in order to accurately estimate between occasion variability of clearance (Holford, Hirankarn et al. 2012).

Copyright All rights reserved | Developed by Sam Holford & Nick Holford 2012-2020




Evans, W. E., M. V. Relling, J. H. Rodman, W. R. Crom, J. M. Boyett and C. H. Pui (1998). "Conventional compared with individualized chemotherapy for childhood acute lymphoblastic leukemia." N Engl J Med 338(8): 499-505.

Hirankarn, S., N. H. G. Holford, E. Dombrowsky, D. Patel and J. S. Barrett (2012). "Pharmacokinetics of high-dose methotrexate in children with cancer: A mechanism-based evaluation of clearance prediction." PAGANZ

Holford, N. H. G., S. Hirankarn, E. Dombrowsky, D. Patel and J. S. Barrett (2012). "What is the between cycle variability in methotrexate clearance?" PAGE 21 (Abstr 2474): [].