Molecular characterisation of treatment resistance in acute myeloid leukaemia

Paediatric acute myeloid leukaemia (pAML) is genetically distinct disease from adult AML, even though both are classified under the same disease and often group together in classification systems. However, the literature clearly shows that they are different diseases with unique characteristics. There are specific genetic aberrations found only in pAML that contribute to poorer outcomes for patients. While standard of care (SOC) treatments are effective for about 60-70% of children, unfortunately, 30-40% will relapse and fail to respond to further treatment. For these patients, the most successful treatment option is a haematopoietic stem cell transplant, which carries its own set of complex risks and challenges. Furthermore, current clinical trials in pAML often enrol patients who have relapsed or not responded to SOC therapy, meaning they have been exposed to previous treatments. Treatment selection is typically based on data from successful adult clinical trials or preclinical studies using drug-naïve cells, which may not accurately reflect the needs of these relapsed patients. In my thesis, we aimed to address a fundamental gap in research; identifying targeted therapies for relapse or refractory pAML by using biologically relevant cell models. This approach ensures we are effectively “hitting the target” in the quest of a personalised medicine approach. As a proof of principle, I developed a pAML cell model that is resistant to standard-of-care therapies and provide a detailed proteomic characterisation of this resistant model. This enabled us to identify suitable targets for personalised therapies that could potentially improve treatment outcomes for these patients. To start, I first reviewed the current known mechanisms of resistance in pAML and highlighting the targeted treatments currently being tested in clinical trials. This comprehensive review identified potential opportunities for targeted therapy, providing a strong foundation for my studies. Key highlights in this body of work included the development and characterisation of a robust, clinically relevant resistant model (R-FLT3-ITD), which has been validated in vivo. Interestingly, we found that resistant pAML cells downregulate a significant number of proteins compared to their sensitive counterpart. However, two pathways, CLEAR signalling and Hippo signalling were significantly activated in resistance. We further analysed our resistant model in compared to cells treated with an acute dose of cytarabine. This analysis revealed a shift in cell signalling pathways. In the acute setting, key pathways such as FLT3-ITD, oxidative phosphorylation (OXPHOS) and the tricarboxylic acid (TCA) cycle were significantly down regulated. We hypothesise that this down regulation acts as a survival mechanism, temporarily suppressing proliferation pathways to allow the cells to metabolically adapt to their environment before reactivating these pathways. We identify a number of key pathways, such as DNA damage repair regulated by ATM and adaption to oxidative stress through NRF2, along with ATF4 transcription factor promoting cell survival in response to ongoing standard-of-care chemotherapy. We validated these critical proteins using multiple orthogonal methods, including open-access databases and, where possible, patients samples, to confirm their clinical relevance. My characterisation of resistance by proteomics, lead us to investigate the oncogene B-cell lymphoma 6 (BCL6). High BCL6 expression is intrinsically linked to mitochondrial function and resistance disease within the literature, however we discovered a decrease in BCL6 protein expression in our resistance cells, which was also confirmed using a BCL6 chromatin immunoprecipitation and sequencing (ChIP-seq), showing a significant decrease in BCL6 protein residing at the promoter of genes in resistant cells. Encouraged by my supervisor, I reviewed the role of BCL6 in paediatric cancers (Chapter 4), which highlighted key gaps in the understanding of BCL6 in paediatric cancer. Our results suggested, BCL6 acts as a negative regulator for CLEAR signalling genes in the de novo setting, that is lost in resistance, however further investigation are still needed. Finally, my research established that key paths of resistance in pAML converge on the mitochondria and its remarkable ability to adapt to metabolic stress. We identified that mitochondrial proteins are significantly upregulated in our resistance model. We attempt to target mitochondrial proteins IDH1 and IDH2 with current small molecule inhibitors; Ivosidenib, GSK864 and Enasidenib along with electron transport chain (ETC), complex 1 inhibitor: IACS 010759, however, these studies failed to show a benefit. Therefore, we employed novel small molecular therapies; TR107 and ONC206, which target the mitochondrial protease, CLPP. TR107 and ONC206 lead to increased activation of CLPP, and degradation of the resistance promoting mitochondrial protein IDH3A. We validated TR107s effectiveness in vivo, where we observed improvements in median overall survival of mice engrafted with our pAML resistance model. With the aim to improve survival further and find combination therapies to use in combination with TR107, we turned to metabolomics studies, and demonstrated that 2-hydroxyglutarate (2-HG), a known oncometabolite, is significantly upregulated in resistance. 2-HG inhibits key epigenetic regulators, to which we show that TR107 is able to potentially increase histone posttranslational modification including H3K27 trimethylation (H3K27me3). Finally, using matched diagnostic and relapsed AML patient samples, we provide preliminary correlative evidence supporting the increased levels of IDH3A and H3K27me3 observed in our resistant pAML model through proteomics and immunoblot experiments.