Ida Netterberg, Lena E. Friberg, Elisabet I. Nielsen, Mats O. Karlsson
Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: Myelosuppression, as evaluable by absolute neutrophil count (ANC), is the dose-limiting toxicity for many chemotherapeutic agents. Today, monitoring of ANC between dosing occasions is, for practical reasons, very limited and consequences include (i) patients experiencing life-threatening conditions possible to counter-act with rescue therapy, (ii) patients returning to the oncological clinic for a new course, only to be told that their ANC is too low to start the next cycle, and (iii) cautious dosing in view of the limited possibility to monitor myelosuppression. Technological advances may make daily ANC monitoring in home-labs possible in the near future. This study aim to investigate how such data, together with model-based analysis and prediction, could improve therapy.
Methods: ANC for 601 patients administered a 1-hour infusion of 75 or 100 mg/m2 docetaxel [1] were simulated at baseline and thereafter once daily from day 3 to 21, given the model by Friberg et al. [2] and parameter estimates according to the analysis by Kloft et al. [3]. The ANC at each day during the cycle was predicted in scenarios of varying sampling duration and frequency. Characteristics of the myelosuppression time-course, i.e. time to baseline (TBA), time to nadir (TNADIR) and ANC value at nadir (VNADIR), were computed using individual parameters estimated from simulated data and compared to the respective characteristic computed from the true individual profile. The imprecision in the predictions were expressed as root mean squared error (RMSE). Also, the ANC was forecasted in the different scenarios and evaluated as a relative estimated error (REE). The analysis was carried out using NONMEM 7.2 [4].
Results: Sampling ANC longer than approximately 15 and 12 days for TBA/TNADIR and VNADIR respectively, did not improve the imprecision. With denser sampling schedules the imprecision decreased. The RMSE of VNADIR decreased from 0.439 (805%) to 0.303 (346%) 109 cells/L with an ANC measurement only at baseline and data at baseline and daily from day 3 to 6, respectively. Around nadir, i.e. day 7 for docetaxel, a one day forecast of the ANC resulted in a larger bias than a one day forecast both earlier and later in the cycle.
Conclusions: Increased number of measurements of ANC together with model predictions could improve therapy with respect to patient safety, e.g. predicting nadir, and convenience for the patient and the clinic, i.e. schedule the start of the next cycle.
References:
[1] Bruno R, Vivler N, Vergninol JC, De Phillips SL, Montay G and Sheiner LB: A population pharmacokinetic model for docetaxel (Taxotere): Model building and validation. J Pharmacokinetics Biopharm. 1996, 24(2):153-172
[2] Friberg LE, Henningsson A, Maas H, Nguyen L, and Karlsson MO: Model of Chemotherapy-Induced Myelosuppression With Parameter Consistency Across Drugs. J Clin Oncol. 2002, 20(24):4713-4721
[3] Kloft C, Wallin J, Henningsson A, Chatelut E, Karlsson MO: Population Pharmacokinetic-Pharmacodynamic Model for Neutropenia with Patient Subgroup Identification: Comparison across Anticancer Drugs. Clin Cancer Res. 2006, 12(18):5481-5490
[4] Beal S, Sheiner L, Boeckmann A, Bauer R: NONMEM Users Guides. Ellicott City. Icon Development Solutions; 2009., 1989-2009
Reference: PAGE 23 () Abstr 3231 [www.page-meeting.org/?abstract=3231]
Poster: Drug/Disease modeling - Oncology