Model-based optimization of G-CSF treatment during cytotoxic chemotherapy

Abstract

PurposeAlthough G-CSF is widely used to prevent or ameliorate leukopenia during cytotoxic chemotherapies, its optimal use is still under debate and depends on many therapy parameters such as dosing and timing of cytotoxic drugs and G-CSF, G-CSF pharmaceuticals used and individual risk factors of patients.MethodsWe integrate available biological knowledge and clinical data regarding cell kinetics of bone marrow granulopoiesis, the cytotoxic effects of chemotherapy and pharmacokinetics and pharmacodynamics of G-CSF applications (filgrastim or pegfilgrastim) into a comprehensive model. The model explains leukocyte time courses of more than 70 therapy scenarios comprising 10 different cytotoxic drugs. It is applied to develop optimized G-CSF schedules for a variety of clinical scenarios.ResultsClinical trial results showed validity of model predictions regarding alternative G-CSF schedules. We propose modifications of G-CSF treatment for the chemotherapies ‘BEACOPP escalated’ (Hodgkin’s disease), ‘ETC’ (breast cancer), and risk-adapted schedules for ‘CHOP-14’ (aggressive non-Hodgkin’s lymphoma in elderly patients).ConclusionsWe conclude that we established a model of human granulopoiesis under chemotherapy which allows predictions of yet untested G-CSF schedules, comparisons between them, and optimization of filgrastim and pegfilgrastim treatment. As a general rule of thumb, G-CSF treatment should not be started too early and patients could profit from filgrastim treatment continued until the end of the chemotherapy cycle.Electronic supplementary materialThe online version of this article (10.1007/s00432-017-2540-1) contains supplementary material, which is available to authorized users.

Clinical role of filgrastim in the management of patients at risk of prolonged severe neutropenia: An evidence‐based review

“…Some models have been proposed to optimise G‐CSF administration. Although pegfilgrastim, a PEGylated form of the growth factor, increases patient compliance, filgrastim appears to be more ductile, enabling its administration to be individually tailored to patients …”

“…The authors reported that, for the highest risk group, filgrastim should begin 6‐8 days after the start of therapy, cycle with at least eight injections. Timing is less important for medium‐ and low‐risk groups, with six and four injections, respectively, obtaining sufficient neutrophil recovery …”

Integration of mathematical model predictions into routine workflows to support clinical decision making in haematology

“…In this (still ongoing) study clinicians from different hospitals and cancer centres outside our consortium are involved as test users. The implemented models itself as well as the MAGPIE model server have already been tested and validated independently ( [1][2][3][4][5][6][7][8]12]). Although tested for model correctness and for general usability, the presented framework is still a prototype.…”

“…2 Schematic outline (screenshots) of framework components/features and information flows. Our prototype provides management of patient-identifying data (1) and corresponding medical data (2) complemented by an integrated graphical representation (3). Mathematical model predictions can be generated interactively for user-defined parameter settings (slider-based parameter selection) and visualized in comparison to the clinical data (4).…”

“…tumour aggressiveness, chemosensitivity, pharmacokinetics and -dynamics of anticancer drugs, risk factors), which are difficult to predict empirically. Hence, we sought to establish a number of disease and treatment models for haematological malignancies, such as high-grade Non-Hodgkin's lymphomas (NHL) ( [1][2][3][4]) and chronic myeloid leukaemia (CML) ( [5][6][7][8]). In addition to providing a better general understanding of the disease mechanisms and treatment effects, these models identify patient-specific parameterizations, which are essential to provide individually tailored predictions.…”