Proteogenomic profiling to identify therapeutic targets for acute myeloid leukaemia

Acute myeloid leukaemia (AML) is the most common and fatal acute leukaemia, with a 5-year survival rate of 25%. AML is a heterogenous cancer, with multiple recurring mutations identified, driving activation of complex signalling cascades. Conventional single-agent therapies targeting AML-associated mutations provide only short-term therapeutic benefit. Adding proteomic information to genomic platforms – "proteogenomics" – holds promise to identify druggable pathways. We have expanded this concept to incorporate phosphoproteomic profiling and ex vivo drug screening, enabling identification of activated signalling pathways in individual patients.

Aim: To perform multi-omic and drug sensitivity profiling of primary AML patient samples, to identify novel molecular targets for improved therapies.

Ex vivo drug sensitivity profiling has been performed on 71 patients. Genomic profiling was performed using the SOPHiA Genetics Myeloid Panel. Transcriptomic and proteomic analysis was performed on bone marrow mononuclear cells from 24 AML patients at diagnosis, and 5 healthy control samples. Paired-end 150bp sequencing was performed on poly-A enriched RNA using an Illumina NextSeq500 Quantitative shotgun proteomics was performed using TMT16 labelling and mass spectrometry analysis. Analysis was performed using Perseus, MoCluster, and Kinase Enrichment Analysis. Results were compared to clinical genomic information.

We identified both known and novel associations between drug sensitivities and mutation subgroups. Consistent with previous reports, NPM1 mutations were associated with higher sensitivity to venetoclax. Samples with NPM1 mutation were also significantly more sensitive to Aurora kinase/CDK inhibitor JNJ7706621 compared to NPM1 wildtype. NPM1 has documented roles in cell cycle regulation, and these results suggest that mutant NPM1 may sensitise to cell cycle targeted therapies. Multi-omic analyses of 23 intermediate/high risk patients, 9 of which had no actionable mutations, identified further drug targets. Integrated multi-omic clustering defined 4 molecular subtypes of AML, each with distinct therapeutic vulnerabilities; including DNA repair and metabolic pathways. Proof-of-principle follow-up using in vitro, ex vivo and in vivo preclinical models shows that pharmacological inhibition of DNA-protein kinase combined with standard AML therapy is a novel approach for some AML subtypes including FLT3-mutant AML. Validation of this approach was exemplified by individual case studies where our profiling platform correlated with clinical outcome. The results of these and further case studies will be presented.

This study presents a unique pharmaco-phospho-proteogenomic characterisation of AML, providing unprecedented insight into the biology of AML. This is a powerful approach for functional characterisation and therapeutic target identification in AML, both for discovery and precision medicine approaches.