Itziar Irurzun Arana (1), Marcella Petrone (1), Claire Smith (2), Simon Smith (2), Bienvenu Loembé (2), Alan Lau (2), Emma Dean (2).
(1) Clinical Pharmacology and Quantitative Pharmacology, BioPharmaceuticals R&D, AstraZeneca R&D, Cambridge, UK. (2) AstraZeneca R&D, Cambridge, UK.
Objectives: Ceralasertib (AZD6738) is an inhibitor of the ATR protein kinase, the apical kinase in the replication stress response DNA-damage induced checkpoint pathway [1], currently being tested in patients with solid malignancies as orally daily single (OD) or twice a day (BID) administration. The main dose limiting toxicity of ceralasertib is thrombocytopenia. In order to support the dose and schedule decision for ceralasertib monotherapy RP2D, and analyse the safety and tolerability of the compound, we utilised emerging phase I clinical data to model the relationship between exposure and platelet/neutrophil cell counts and calculate the probability of developing severe (≥grade 3) thrombocytopenia/neutropenia.
Methods: The population PK (popPK) model was built using emerging data from 230 patients either dosed with ceralasertib monotherapy (continuous doses from 20 to 240mg OD or BID), or in combination with carboplatin or the PD-L1 inhibitor durvalumab 1500mg Q4W (intermittent doses up to 320mg) as these drugs did not show any impact on the PK of ceralasertib. The inclusion of a nonlinear behaviour into the model was explored in order to account for the less than dose proportional increase in exposure observed in the dose range from 20 to 320 mg.
For the exposure-safety model, preliminary data from 51 subjects who received ceralasertib + durvalumab in a phase I combination study was used for model building, assuming that the myelosuppressive effects were driven by ceralasertib only based on the mechanism of action. Individual PK profiles were derived using the previously developed ceralasertib popPK model as matched PK and platelet/neutrophil observations were generally not available. Friberg model [2] was used as the structural PD model to describe drug-induced myelosuppression where the individual PK profiles affected the proliferation of the haematological cells.
Results: A popPK and a PK-safety models for ceralasertib were developed describing ceralasertib-induced thrombocytopenia and neutropenia. The popPK model consisted of a two-compartment model with linear absorption and elimination and a relative bioavailability parameter that decreased linearly with increasing ceralasertib dose. The PK-safety model demonstrated that individual PK profiles affected the proliferation of the cells by a linear inhibitory function.
Model and simulation results showed that a two-week off treatment period is needed for platelet/ neutrophil recovery. A shorter off treatment period would lead to an incomplete recovery of the platelets /neutrophils for some patients which would provoke treatment delays and more grade 3 and 4 adverse events at the following cycles. For the dose of 240 mg BID ceralasertib for 14 days, the probability of having ≥grade 3 neutropenia and thrombocytopenia in a cycle was <20%. For the dose of 160 mg BID this probability decreased to a <15%.
Conclusions: These modelling and simulation results suggest that doses of 160mg BID or 240 mg BID ceralasertib for 14 days are safe and tolerated as the probability of developing severe treatment related haematological toxicities is low (<20%) and platelets/neutrophil counts recover after two weeks off treatment period. This modelling and simulation exercise is expected to be refined on a regular basis, with the integration of further clinical trial data.
References:
[1] Foote, K.M., et al. J Med Chem. 61:9889-907 (2018).
[2] Friberg, L. E., et al. J. Clin. Oncol. (2002).
Reference: PAGE 29 (2021) Abstr 9798 [www.page-meeting.org/?abstract=9798]
Poster: Drug/Disease Modelling - Oncology