P412: TRANSCRIPTOMIC CLASSIFICATION, RISK STRATIFICATION AND THERAPY SELECTION IN AML

HemaSphere, vol. 7, pp. e495610f

Abstract

Background: Subtyping of acute myeloid leukaemia (AML) by the World Health Organization (WHO) and International Consensus Classification (ICC) has made significant progress but is currently genetics-based. Studies have shown that transcriptomics can improve AML stratification, but researchers have yet to perform a large-scale analysis of AML transcriptomics. Aims: Perform a systematic unsupervised analysis of transcriptional AML data to improve classification, risk stratification and therapy selection in AML patients. Methods: We integrated and harmonised mRNAseq data from BEAT, TARGET, TCGA and LEUCEGENE with our in-house LUMC dataset (n=1337) and defined transcriptional AML clusters. In addition, we acquired corresponding data on genetics, survival and ex-vivo drug response. We used this data to update AML classes for our samples to the WHO and ICC 2022 standards and tested for differences in mutation status, outcome and drug sensitivity between the found transcriptional subtypes. Results: We defined 19 transcriptomic clusters, and in addition to known genetic subtypes, the clusters improved AML classification and discovered new subtypes (Figure 1). We found only KMT2A-MLLT1 and KMT2A-MLLT10 to share a gene expression profile with KMT2A-MLLT3, improving the subtyping of KMT2A-rearranged AML. Patients with a CEBPA bZIP mutation-like gene expression had the same favourable outcome, extending the CEBPA mutated subtype. We found four NPM1 and nine AML with myelodysplasia-related changes (AML-MRC) subtypes, further stratifying 65% of the AML patiens in our datatset. Besides (co-)mutation status, maturation stage and overall survival, we demonstrated significant differences in ex-vivo drug response between the new NPM1 and AML-MRC subtypes.Summary/Conclusion: Gene expression profiling gives an unprecedented overview of the landscape of AML and identifies new clinically relevant AML subtypes. We show the importance of transcriptomics for routine AML diagnostics and highlight its potential to benefit AML patients greatly. Keywords: Drug sensitivity, WHO classification, Acute myeloid leukemia, Gene expression profile

Authors 8

1. Jeppe F. Severens Aachen

   Leiden University Medical Center

   Affiliation as printed

   LUMC, Leiden, The Netherlands

2. Onur Karakaslar Aachen

   Leiden University Medical Center

   Affiliation as printed

   LUMC, Leiden, The Netherlands

3. Elena Sánchez‐López Aachen

   Leiden University Medical Center

   Affiliation as printed

   LUMC, Leiden, The Netherlands

4. Redmar R. van den Berg Aachen

   Leiden University Medical Center

   Affiliation as printed

   LUMC, Leiden, The Netherlands

5. Hendrik Veelken Aachen

   Leiden University Medical Center

   Affiliation as printed

   LUMC, Leiden, The Netherlands

6. Marcel J. T. Reinders Aachen

   Leiden University Medical Center

   Affiliation as printed

   LUMC, Leiden, The Netherlands

7. Marieke Griffioen Aachen

   Leiden University Medical Center

   Affiliation as printed

   LUMC, Leiden, The Netherlands

8. Erik B. van den Akker Aachen

   Leiden University Medical Center

   Affiliation as printed

   LUMC, Leiden, The Netherlands