In this open-label, single-center, non-randomized patients with AML (Acute Myeloid Leukemia) and receiving all induction chemotherapy and consolidation consisting of cytarabine under the care usual for this pathology, will be included. Each patient will be followed and for the development of toxicities, treatment response and progression-free survival. In addition to the usual care set out above, each patient will undergo a series of constitutional genetic investigations conducted by NGS on markers related to pharmacokinetics cytarabine. Another set of blood samples will also calculate, according to a Bayesian approach, individual pharmacokinetics of cytarabine and its metabolite, arabinosine-uracil.

This study should allow the correlation between pharmacogenetics and patient plasma exposure, that would eventually balance improved efficacy / toxicity of this molecule through a customization regimens, achieved so far on a empirical basis. If validation of our data, a dosage of therapeutic pre CDA could help in predicting pharmacodynamics of cytarabine individual dose adjustment, as is done for the 5-FU and DPD.

Background and Rationale: The development of personalized medicine in oncology has so far relied on the use of somatic biomarkers to inform the therapist about the choice of the molecule or molecules to be administered based on the genetic and molecular profile of each blood disease. In this project, we propose to extend the therapeutic individualization strategy targeting dosage domain. Today, cytarabine is one of two pillars of the treatment of leukemia Acute Myeloid Leukemia (AML) in combination with an anthracycline plus cytarabine ("3 + 7" therapeutic scheme) during the induction course and usually during monotherapy consolidation treatments. According to the treatment regimens and protocols, cytarabine is prescribed at a standard dose (SD = 100-200mg / m² / day), intermediate dose (ID = 1 to 1.5 g / 12H for all 3 days) or high dose (DH = 2 to 3 g / m² every 12H during 3 days). The choice between these different dose levels remains highly debated (Lowenberg et al. 2013) with a very narrow risk-benefit balance. Indeed, the various cooperative groups compared two dose regimens with factors between the minimum dose and maximum ranging from 1.7 times to 34 times between groups (Ex Australian arm group 1400 mg / m vs 48 000 mg / m cumulative dose cytarabine ). The magnitude of unmatched dose differentials illustrates the complexity of understanding and apprehension of this pivotal molecule for the treatment of AML. Despite the many studies that focused on the optimal dose of cytarabine ranging from 1400 mg / m² to 90 000 mg / m² (German group) (including currently French Intergroup of leukemia-ALFA FILO adult under the BIG1 protocol), none evaluated the relevance of the a priori individual dose adjustment depending on the pharmacogenetic patient data. In current practice, the doses are adapted a posteriori, and reduced empirically following the observed toxicity of occurrence (20% of patients) (Lowenberg et al. 2013). This adaptation a posteriori is a loss of opportunity for the patient. Similarly, under dosed patients for fear of toxicity is also another lost chance. Our hypothesis is that the optimal cytarabine dose depends not only on the characteristics of the patient's pathology (risk groups including cytogenetic data, biology, molecular), but also the patient's individual characteristics (genetic status of metabolic enzymes and carriers). A mathematical model of PK / PD kind could, based on early observations of circulating levels, be able to quickly predict the pharmacodynamic effect in each patient, allowing a rapid individualization of dosages. Such a tool could enable, in future, to propose dose adjustments early after initiation of treatment before the onset of toxicity, predicting that exposure levels of cytarabine correlate with the patient's clinical evolution.

Inclusion Criteria:

• Patient aged over 18 years old
• Patient with acute myeloid leukemia
• Patient treated with cytarabine
• Patient having signed an informed consent form
• Patient having signed an authorization to practice a constitutional genetic analysis
• Need for effective contraception in patients of childbearing age.
• Patient affiliated to a social security scheme

Exclusion Criteria

• Not obtaining free, informed and signed consent
• Patient participating in another biomedical research
• Pregnant patients