Carlos Fernandez, Ignacio Gonzalez, Rubin Lubomirov, Salvador Fudio
Pharma Mar S.A., Colmenar Viejo, Madrid, Spain
Objectives: Lurbinectedin is a new anticancer agent that inhibits activated transcription, induces DNA double-strand breaks generating apoptosis and modulates tumor microenvironment. Reversible myelosuppression (anemia, lymphopenia, leukopenia, neutropenia and thrombocytopenia) is the most frequent abnormality related to treatment with lurbinectedin. In this work, data from two phase I and three phase II trials were pooled to evaluate the time course of absolute neutrophil count in patients treated with lurbinectedin as single agent.
Methods: The dataset contained 2069 absolute neutrophil count observations from 156 patients, with lurbinectedin doses ranging from 0.02 to 6.9 mg/m2, given as 1-h i.v. infusion and at schedules of Day 1, and Days 1 and 8 in cycles of three weeks. The absolute neutrophil count dataset was merged with the pharmacokinetic dataset, with 2673 lurbinectedin plasma concentrations available from cycles 1 and 2 for those patients. The population PK/PD model for neutropenia was developed based on lurbinectedin pharmacokinetic model, which was linked to a transit model based on the proposed by Quartino [1]. Once the base model was achieved, several covariates where explored to identify the sources of variability. Datasets were produced with SAS Enterprise Guide v7.11, models were executed with NONMEM 7.3 and covariate analysis was performed by using stepwise covariate model from PsN v4.6.0. Additionally, the final model was validated through pcVPC and bootstrap analysis also from PsN.
Results: A neutropenia model with transit compartments and feedback effects on mean transit time and proliferating cell pool was suitable to describe the time‑course of absolute neutrophil count. The absolute neutrophil count dataset was Box-Cox transformed with lambda of 0.2465 which was the optimum value for the current dataset, while the pharmacokinetic dataset was log-transformed. Both residual variabilities were managed as additive errors with inter-subject variability. The observed baseline of absolute neutrophil count was used as covariate acknowledging the neutrophils residual variability following the B2 procedure [2]. The following covariates modified the neutropenia incidence: absolute neutrophil count at baseline, alpha-1-acid glycoprotein, presence of ascites, the use of Granulocyte-colony stimulating factors (G-CSF), prior use of anthracyclines, and the use of CYP3A inducers. The lurbinectedin drug effect was included into the model using a sigmoid Emax function governed by the parameters Emax, EC50 and a Hill exponent with a value of 2.16. The use of G-CSF doubled EC50 and almost cut in half the mean transit time which markedly reduced neutropenia.
Conclusions: This modeling exercise has characterized the relationship between lurbinectedin exposure and neutropenia incidence, identifying the sources of variability. The model shows that lurbinectedin-induced neutropenia can be managed with dose reductions and/or G-CSF, and is useful for supporting the design of new clinical trials.
References:
[1] Quartino et at. Invest New Drugs. 2012 Apr;30(2):833-45.
[2] Dansirikul, C., Silber, H.E. & Karlsson, M.O. J Pharmacokinet Pharmacodyn (2008) 35: 269.
Reference: PAGE 27 (2018) Abstr 8558 [www.page-meeting.org/?abstract=8558]
Poster: Drug/Disease Modelling - Oncology