This is a multicentre study on biobanked bone marrow and blood samples of AML patients, conducted by Exscientia GmbH. The study aims to compare the drug response measured 'ex vivo', this means outside of the body, in the samples and the documented outcome of the respective patient's clinical treatment. To measure this Ex Vivo Drug Response (EVDR), Exscientia will use it's AI-( Artificial Intelligence)-based precision medicine platform. In this platform, the cells of each sample are split and distributed in a number of small vials, to which different approved or experimental AML drugs are added. The cells are left with the drugs for a certain period of time (no culturing or expansion is done). After that, the cells are stained (coloured) by using specific dyes and the rates of dead cancer cells in each of these small vials is determined via automated microscopy. The EVDR shows how well the drugs killed the cancer cells in the sample. Taking clinical data into account, which is information on e.g. the patients health status or genetic markers, the EVDR could reveal which patients might especially benefit from the treatment.

If a reproducible correlation between the EVDR and the patient's clinical treatment outcome is found, the scFDS platform could be used in the future to improve treatment selection for AML patients.

The study will include biobanked samples from newly diagnosed patients, treated with cytarabine + daunorubicin (classical 7+3 or CPX-351) or venetoclax + azacitidine and after favourable results in an interim analysis, biobanked samples from R/R AML FLT3 mutant patients, treated with Gilteritinib might be included.

Key procedures include:

• Viable tumour tissues (i.e. bone marrow or blood) taken prior to therapy are provided by biobanks to Exscientia's central lab (or delegated central laboratory),
• Ex vivo drug response against commonly given standard of care drugs is evaluated in viable tumour tissues, Exscientia-owned drug candidates might be included in the assay for pre-clinical testing.
• Clinical patient data are collected,
• Relationship of EVDR to clinical response is evaluated.

Primary key hypothesis: Ex vivo drug response (EVDR) is significantly associated with Complete Response (CR).

Secondary key hypothesis: EVDR predicts achieving CR with 80% sensitivity and specificity.

The outcome of this observational clinical study will have broad implications both for the clinical routine, preclinical drug development, and translational cancer research. If a robust correlation between drug response measured ex vivo in tumour samples and clinical outcome can be identified, this will pave the way for:

• the use of functional drug testing as a tool for personalised treatment decision making in the clinical routine, in particular where classical molecular precision medicine approaches fail to prioritise effective therapies, and
• the use of human tumour samples as clinically relevant model systems for preclinical development of new drugs and translational cancer research that can potentially overcome the limited clinical relevance of mouse and other animal models.

Inclusion Criteria:

1. Patients aged 18 years or older.

2. Signed informed consent form that permits use of sample in the proposed study (including permission for genetic analysis).

3. Sample of a newly diagnosed patient (sample at time point diagnosis); after interim analysis the population may include samples from patients with FLT3 mutated relapsed/refractory AML.

4. The sample may not be older than 5/ 10 years (depending on location)

5. Confirmed diagnosed AML according to WHO or ICC criteria; after interim analysis the population may include FLT3 mutated R/R AML according to ELN 2022 criteria after interim analysis.

6. Patient received one of the following therapies after sampling for which response data is available:

   1. 7 + 3 (with or without additional GO or TKIs)
   2. CPX-351
   3. Venetoclax and AZA in combination or alone
   4. If R/R AML, FLT3 mutated: treated with Gilteritinib

7. Characteristics of sample taken prior to therapy as specified in the study protocol.

8. Availability of complete dataset as specified in the study protocol.

Exclusion Criteria:

1. Known or suspected HIV or active Hepatitis B and/or C infection or active COVID-19 infection (if information not available, samples can still be included) at time of sample collection.
2. Known active infection of bone marrow.
3. Known pregnancy.
4. Received systemic anticancer treatment or radiotherapy within 4 weeks of sampling (pre-treatment of hydroxyurea and/or low dose cytarabine allowed).
5. Patient is diagnosed with Acute Promyelocytic Leukaemia (APL).
6. Patients with treatment for any other oncologic neoplasm at time of sample collection.
7. Patients for whom CR could not be assessed (e.g. death before re-staging).
8. Inclusion of samples from the same patient at both diagnosis and relapse is not permitted. In such cases, only the diagnostic sample is to be used.

This study is planning to include biobanked samples of the following patient populations:

1. Newly diagnosed fit AML patients (excl. APL), defined as treated with Cytarabine + Daunorubicin (classical 7+3 or CPX-351) independent or combination with GO or a TKI (e.g., Midostaurin).
2. Newly diagnosed unfit AML patients, defined as treated with Venetoclax and Azacitidine (alone or in combination).

Upon favourable results in an interim analysis, R/R AML FLT3 mutant patients, treated with Gilteritinib, for which we expect different covariates exist, might also be included.