Use of plasma cytotoxic activity to model cytotoxic pharmacodynamics of anticancer drugs

Abstract

We have developed a pharmacokinetic/pharmacodynamic approach that integrates the disposition, cytotoxic activity and interaction of anticancer drugs. Fundamental to this approach is the measurement of the cytotoxicity, against a "target" cell line, of patient plasma collected at different times after administration of the anticancer agent(s). To illustrate this approach, we have studied the plasma cytotoxic activity (PCA), against HL-60 cells, of plasma from 11 acute myeloblastic leukemic patients treated with daunorubicin (DNR). Plasma, obtained before and serially for 24 h after DNR treatment, was assayed by HPLC for DNR and daunorubicinol (DNRol), its active metabolite. The corresponding observed PCA values (PCAobs) against HL-60 cells were also measured with a flow-cytometric cell-survival assay that we had developed previously. The pharmacodynamics, i.e. PCA, were co-modeled (dual Hill equation with an interaction term to allow synergism or antagonism) with the pharmacokinetics. The integration of the PCA profile provided the area under the observed PCA versus time curve (AUCobs). For each patient, we also generated an "interaction panel", by adding known amounts of DNR and DNRol to his or her pretreatment plasma. The corresponding cytotoxicities were measured, and then applied to the pharmacodynamic model. This provided a standard surface from which the PCA of each sample obtained after therapy was predicted (PCAprd), on the basis of assayed concentrations of DNR and DNRol in that sample. For plasma samples obtained after treatment, the model simultaneously fit all three outputs, i.e. PCA and DNR/DNRol concentration, very well. We observed substantial interpatient variability in HL-60 growth rate in medium containing patient pretreatment plasma, in DNR activity in pretreatment plasma, and in the in vitro activity (PCA) of plasma obtained after DNR treatment. We also compared the AUCprd to the AUCobs for each patient, and we identified a subset of 4/11 acute myeloblastic leukemic patients who had developed much more PCA after DNR administration that could be explained by the measured concentrations of DNR and DNRol. This may be due to unidentified active metabolites or to factors produced in the plasma in response to the treatment. This pharmacokinetic/pharmacodynamic model is promising to describe pharmacodynamics and interactions of anticancer drugs in cancer patients.