Modelling of Chemotherapy-induced Febrile Neutropenia using the Predicted Degree and Duration of Myelosuppression
E.K. Hansson(1), M. Sandström(2), H. Lindman(3), L.E. Friberg(1)
(1)Department of Pharmaceutical Biosciences, Division of Pharmacokinetics and Drug Therapy, Uppsala University, Sweden; (2)AstraZeneca R&D, Södertälje, Sweden (3)Department of Oncology, Radiology and Clinical Immunology, Uppsala University Hospital, Uppsala, Sweden
Objectives: Chemotherapy-induced neutropenic episodes are associated with the risk of developing the life threatening condition febrile neutropenia (FN) (fever ≥ 38.3 °C, neutrophil count ≤ 0.5 x 109/L). The aim of this study was to develop a pharmacokinetic-pharmacodynamic model describing the time course of neutrophils in breast cancer patients either treated with epirubicin and docetaxel (ET), or 5-fluorouracil, epirubicin, and cyclophoshamide (FEC) and to characterize the relationship between the predicted myelosuppression profile and FN. In a first step we investigate if, as proposed by Minami [1], a line (the duration of neutropenia) or the area of the neutrophil-time course below a certain value, is better related to the risk of developing FN than a single observation point (the neutrophil nadir value).
Methods: The analysis included 1053 neutrophil observations and 11 episodes of FN in 40 patients treated with ET and 1171 neutrophil observations and 6 episodes of FN in 60 patients treated with FEC. Individual concentration-time profiles of the drugs were predicted from previous models [2, 3] and the time course of myelosuppression was described using a semi-mechanistic model [4]. The PRIOR option in NONMEM VI was used to estimate separate potency parameters for the co-administered drugs. Each of the model-predicted myelosuppression variables (nadir, neutrophil baseline, duration or area of grade 3 or 4 neutropenia) were related to the FN data in a logistic regression model.
Results: The myelosuppression model could well characterize the neutrophil-time course following ET and FEC treatment and resulted in similar system-related parameter estimates as for other drugs [4]. The fit improved when using a fixed neutrophil half-life of 7 hours and Box-Cox transforming the data with a factor of 0.2. Of the investigated variables nadir and duration of grade 4 neutropenia were best related to FN.
Conclusions: Both the model-predicted nadir value and neutropenia duration were related to the development of febrile neutropenia. To be able to a priori predict which patients who are at risk of developing FN, information on previous treatment and patient characteristics will be evaluated, as well as other types of models, e.g. time to event models.
References:
[1] Minami, H., J Clin Oncol, 2005. 23: p. 405-406.
[2] Sandström, M., et al., Cancer Chemother Pharmacol, 2006. 58: p. 143-156.
[3] Sandström, M., et al., J Clin Oncol, 2005. 23: p. 413-21.
[4] Friberg, L., et al., J Clin Oncol, 2002. 20: p. 4173-4721.
