PDF, English Restricted access: Repository staff only until 12 May 2024. Login+Download (26MB) | Terms of use

Abstract

The human body is made of trillions of cells that are the building blocks of all living things. Hematopoietic cells are a set of such building blocks. They are made in the bone marrow and carry out vital tasks, including fighting off infections, facilitating wound healing and carrying oxygen through the body. In health, hematopoietic cells are crucial for the maintenance of normal blood cell production which is necessary for a healthy life. In Acute Myeloid Leukemia (AML), errors in DNA repair result in the acquisition of genetic variations that drive malignant changes and growth of abnormal myeloid cells. These malignant cells fail to perform their vital functions and further hamper the production of normal blood cells.

AML is driven by a group of cells in the bone marrow with self-renewing capacity that give rise to a diverse progeny of abnormal myeloid cells. AML remains lethal due its complex and plastic cellular nature characterized by a high degree of intra-tumor heterogeneity. With the introduction of single-cell technologies, advancements in characterizing the genetic and non-genetic landscape of AML has improved. However, attempts to connect the different levels of heterogeneity, identify and target the disease-driving leukemic stem cells (LSCs) and assess the resulting functional outcomes are largely still lacking. Moreover, identifying the patients who best benefit from novel targeted therapies compared to standard cytotoxic therapies remains a challenge.

In this thesis, I first investigated the intra-patient heterogeneity of complex karyotype AML using an integrated single-cell multi-omics framework that combines structural variant discovery and nucleosome occupancy profiling (scNOVA) with concurrent immunophenotypic and transcriptomic profiling (CITE-seq). Using this framework, I revealed complex structural variant landscapes in single CK-AML cells, marked by ongoing karyotype evolution with frequent involvement of chromothripsis along with linear and circular breakage-fusion-bridge events mediating genomic remodeling. I further unveiled extensive cell-to-cell karyotype instability, exemplified by instable chromosome intermediates, like complex marker and ring chromosomes. Next, I characterized the intra-patient heterogeneity and revealed the existence of genetically distinct subclones with unique nucleosome occupancy, and transcriptomic and immunophenotypic features. By transplanting these cells into immunocompromised mice, I observed predominantly monoclonal expansion of subclones with high genomic complexity that were enriched for stemness-associated phenotypes, including high 17-gene stemness scores and expression of stem cell markers such as CD49F and CD90. Furthermore, I showed that these disease-driving LSCs showed resistance to standard chemotherapy ex vivo but could be targeted by BH3 mimetics. Finally, in an index patient, I showed that the patient-derived xenograft system recapitulated the subclone-specific evolution also during disease progression in the patient, offering a promising model to study relapse. Together, these data provide unique insights into the ongoing genetic and phenotypic complexity of CK-AML, highlight the clinical relevance of intra-patient heterogeneity in tumor evolution, and offer promising avenues to functionally explore and target the disease-driving LSCs.

Next, I explored the clinical relevance of disease-driving LSCs in AML by taking part in investigating how they can be used to predict response to a newly-approved targeted therapy comprising the BCL-2 inhibitor venetoclax in combination with azacytidine. By integrating transcriptomic, functional and clinical data we aimed to identify predictors of clinical response to this combination therapy. We revealed that while more differentiated monocytic cells had high MCL-1 expression and showed resistance to venetoclax and azacytidine, they consistently lacked disease-initiating potential and thus did not fuel leukemogenesis. In contrast, the cells with consistent LSC potential expressed high levels of BCL-2 and could be efficiently targeted ex vivo. We further showed that combining BCL-2, BCL-xL and MCL-1 protein expression ratios in these disease-driving LSCs, could be used to determine the clinical response to venetoclax and azacytidine. This flow cytometry-based “Mediators-of-Apoptosis-Combinatorial-Score” (MAC-Score) predicted initial response with a positive predictive-value of >97% and was associated with increased event-free survival. These data show that the combinatorial levels of BCL-2-family members in the disease-driving LSCs are a key determinate of response to venetoclax and azacytidine and that affordable techniques can be used to reliably predict response to this therapy.

In summary, I investigated different levels of intra-patient heterogeneity in CK-AML patient samples using an integrated single-cell multi-omics framework and explored the resulting functional outcomes. I also took part in predicting clinical response to the newly-approved therapy of venetoclax in combination with azacytidine by establishing a flow cytometry-based response score. Collectively the thesis emphasizes the importance of better identifying and characterizing the disease-driving LSCs to improve our understanding of AML as a dynamic disease entity and to offer effective ways to assess and target the disease-driving LSCs.