A precision dosing application for patients treated with docetaxel and G-CSF

Hurry, C.; Villette, C.; Mistry, H.; Millen, J.; Chassagnole, C.

Purpose Produce a precision dosing application for clinicians to control neutropenia in patients treated with both docetaxel and G-CSF. Context Chemotherapy induced neutropenia (CIN) poses serious harm to patients due to the heightened risk of severe infection. Accordingly, chemotherapy dose is assessed at the beginning of each cycle. The therapeutic window of chemotherapy is determined from population studies, with an individual's dose often scaled by their body surface area. This leads to a large number of patients being over- or under- dosed. We previously developed an application [1] which uses weekly neutrophil counts from the first cycle of docetaxel treatment to predict the level of neutropoenia in subsequent cycles for a given dose of docetaxel. However, in the original app the administration of G-CSF, used as a prophylactic treatment for neutropenia, was not considered. G-CSF administration lacks standardisation and the COVID-19 pandemic has created a highly risk averse environment to infections, raising the prospects that a clinician will administer G-CSF. MethodsAiming to predict the level of neutropenia after the first cycle we draw from models in the literature of CIN for G-CSF treated patients [2,3,4]. We adapt the G-CSF induced feedback effect on neutrophil development in these models to account for features specific to docetaxel patients;notably the rebound in neutrophils above baseline seen after nadir [5], and the apparent short maturation time of neutrophils [4]. Using data in the public domain from the comparator arm of a phase III clinical trial for metastatic hormone-resistant prostate cancer (clinical trial number NCT00617669), we identify 27 patients treated with docetaxel and G-CSF with weekly blood tests in the first and second cycle. We calibrate our model by minimising a Bayesian objective function to establish a model that predicts the neutrophil count in the second docetaxel cycle from the first. We then simulate the effect of different combinations of docetaxel, G-CSF, and the timing of G-CSF administration. Conclusions Current progress includes a model which captures the rebound above baseline in neutrophil count seen in docetaxel patients. We will show that our model captures the effect of filgrastim and pegfilgrastim on neutrophil counts, and can determine how a docetaxel patient may best benefit from G-CSF treatment and/or a change in dose of docetaxel.