Population PK Modelling of Treosulfan in Paediatric Allogeneic Transplant Patients
Theo Reijmers (1), Claudia Hemmelmann (2), Karl-Walter Sykora (3), Ajay Vora (4), Jochen Kehne (2), Ann-Kristin Möller (2), Joachim Baumgart (2), Jeroen Elassaiss-Schaap (1,5)
(1) Venn Life Sciences ED BV, Breda, The Netherlands; (2) medac GmbH, Wedel, Germany; (3) Hanover Medical University, Hanover, Germany; (4) Great Ormond Street Hospital for Children NHS Foundation Trust, London, UK; (5) PD-value BV, Houten, The Netherlands
Objectives: Treosulfan, a bifunctional alkylating prodrug, is currently being developed by medac GmbH as a component of conditioning treatment prior to haematopoietic stem cell transplantation (HSCT) in adults and children. An initial population PK model was developed to select dosages for new trials in paediatric patients from 1 month to 18 years of age. During interim analyses this model was updated with new paediatric PK-data to contribute (in addition to clinical safety and efficacy data) to decision finding for potential modification of dose recommendations for the ongoing medac-sponsored trials MC-FludT.16/NM (EudraCT-Number: 2013-005508-33) and MC-FludT.17/M (EudraCT-Number: 2013-005508-33). In addition to the data covering different age groups the model also needed to handle data that was measured with different bioanalytical methods.
Methods: The initial population PK model for treosulfan was developed with NONMEM using treosulfan PK-data from 7 previously conducted clinical trials consisting of 93 adults and 23 children (age range: 0.4 to 17 years) [1]. To recommend doses in paediatric patients, the potential influence of certain covariates was investigated. Next, this model was updated on a regular basis using PK-data from the new paediatric allogeneic HSCT trials MC‑FludT.16/NM (17 children; recruitment ongoing) and MC‑FludT.17/M (59 children; recruitment closed). For modelling different subpopulations (adults vs. children) both a dichotomization approach and use of a Bayesian prior were evaluated [2].
Results: The initial population PK model for treosulfan consisted of 2 compartments with first order distribution and elimination processes. Covariate analysis revealed that BSA was the only relevant covariate for clearance (allometric coefficient CL-BSA = 1.29) and volumes of distribution (allometric coefficients V1-BSA = 2.05 & V2-BSA = 1.24). The model provided an adequate fit to the data and model diagnostics revealed no relevant bias. The dose recommendation for treosulfan was 10 g/m2 (BSA<=0.5m2), 12 g/m2 (BSA between 0.5 m2 and 1.0 m2), and 14 g/m2 (BSA>1.0 m2). The functional relationship between clearance and BSA within the estimated population PK model was applied to derive dose recommendations for the new paediatric trials. The update of the initial PK model contained a shift parameter (on model prediction parameter F in $ERROR) that allowed modelling of data from the new paediatric trials measured with a different validated bioanalytical method. Also, an additional covariate‑model parameter relationship between BSA and intercompartmental clearance was identified. Both the dichotomization and Bayesian approach gave similar results. Ultimately the approach closest to the initial model, the dichotomization approach, was chosen. All model parameters could be estimated with adequate precision (RSE < 25% & 45% for fixed and random effect parameters). Bootstrap results were in good agreement with results directly obtained from NONMEM. VPCs of the updated model showed a certain bias for the largest BSA paediatrics group (BSA > 1 m2). Concentrations between 2 and 3 hr after start of infusion are within the prediction range but almost all concentrations are above the median of the simulations for this group. According to the model-based updated dosing scheme, for children with a BSA of 0.4, 0.5, 0.9 and 1.0 m2 a limited increase in dose is considered based on these interim data.
Conclusions: The population PK model for treosulfan is robust and does accurately predict exposure in children. Inclusion of interim PK-data from newly included paediatric patients resulted in a significant update of the model. From a population PK modelling perspective, a slight refined dosing for some patients was recommended.
References:
[1] van den Berg PJ, Ruppert M, Sykora K-W, Beier R, Beelen DW, Hilger RA, Scheulen ME, Wachowiak J, Nemecek ER, Spaans E, Lioznov M, Pichlmeier U, Wullf B, Baumgart J. A preliminary population pharmacokinetic model for dose selection of Treosulfan used in conditioning treatment prior to haematopoietic stem cell transplantation (HSCT) in children. Physicians Poster Sessions, EBMT 2014; PH-P543.
[2] Cella M, Gorter de Vries F, Burger D, Danhof M, Della Pasqua O. A model-based approach to dose selection in early pediatric development. Clinical Pharmacology & Therapeutics, 87 (2010); 294-302.
