Abstract Background: Azacitidine (Aza) is a hypomethylating agent (HMA) used to treat patients with acute myeloid leukaemia (AML) or myelodysplastic syndrome (MDS) who are not eligible for intensive chemotherapy. The standard Aza dose and schedule is 75 mg/m2 daily for seven days, administered in 28-day cycles until disease progression occurs or severe toxicities arise. Recent protocols have associated it with venetoclax. However, only half of patients respond, and almost all will eventually relapse. Azacitidine undergoes metabolic detoxification driven by cytidine deaminase (CDA) in the liver. However, CDA is encoded by a highly polymorphic gene, resulting in significant variability: some patients do not respond to treatment, while others experience toxic death. Fifty per cent of patients with acute myeloid leukaemia (AML) or myelodysplastic syndrome (MDS) have deficient CDA enzyme activity (1). This variability suggests the need to consider individualising doses and understanding the kinetics of Azacytidine. In this real-world study, we monitored azacitidine plasma concentrations in order to develop a mathematical model to predict its pharmacokinetics (PK). Method: Twenty-one adult patients (13 male/8 female), with a mean age of 78.7 years (range 59-89), were treated with azacitidine, either alone or in combination with venetoclax. Aza plasma concentrations were measured at the first cycle (C1) at multiple time points. Aza was analysed using a validated mass spectrometry method. PK parameters were derived using a population model with the SAEM algorithm within the Monolix® software. The influence of various factors, including CDA activity, on the pharmacokinetics (PK) of Azacytidine was investigated using a stepwise multivariate procedure. Results: A one-compartment population pharmacokinetic (popPK) model with zero-order absorption and linear elimination was developed. This model incorporated CDA status as an absorption covariate, as well as basal serum albumin and body surface area as distribution volume covariates, and sex and renal impairment as clearance covariates. Conclusions: This initial characterisation of a population model marks the beginning of the process of dose individualisation, which aims to optimise efficacy while minimising undesirable toxicities. Validating this model and determining toxicity parameters (concentration-dependent, exposure-dependent and time-dependent toxicity) and survival parameters will improve the management of these complex diseases with poor prognoses. (1): High incidence of CDA deficiency in patients with hematological malignancies: perspectives and therapeutic implications. Donnette M, Ciccolini J, Pissier C, Costello R, Duffaud F, Salas S, Farnault L, Tichadou A, Arcani R, Jarrot PA, Ouafik LH, Venton G, Fanciullino R. Ann Oncol. 2021 May;32(5):684-686. Citation Format: Quentin Gerbault, Loic Osanno, Joseph Ciccolini, Laure Farnault, Geoffroy Venton, Raphaelle Fanciullino. A mathematical model for fitting adjusting azacytidine in AML or SMD patients abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 1834.
Gerbault et al. (Fri,) studied this question.