Preview

PDF, English Download (4MB) | Terms of use

Abstract

Acute myeloid leukemia (AML) is one of the most aggressive hematopoietic malignancies and has been recognized as a heterogeneous disease due to a lack of unifying characteristics. It is driven by different genome aberrations, gene expression changes, and epigenomic dysregulations. Therefore a multi-omics approach is needed to unravel the complex biology of this disease. This thesis deals with the challenges of identifying driver events that account for differences in clinical phenotypes and responses to treatment. The work presented here investigates the driver events of AML and epigenetics drug response profiles. The thesis consists of three main projects. The first study identifies recurrent mutations in AML carrying t(8;16)(p11;p13), a rare abnormality. The second project is identifying prospective drivers of mutation- negative nkAML. The third project concentrates on epigenetic changes after AML drugs. t(8;16) AML is a rare and distinguishable clinicopathological entity. Some previous reports that rep- resented the characteristics of patients with this type of AML suggest that the t(8;16) translocation could be sufficient to induce hematopoietic cell transformation to AML without acquiring other genetic alterations. Therefore here I evaluate the frequently mutated genes and compare them with the most frequent mutated genes in AML in general and AML carrying t(8;16) translocation. FLT3 mutation was found in 3 patients of my cohort, a potential target for therapy with tyrosine kinase inhibitors. However, exciting finding was the mutations in EYS, KRTAP9-1, PSIP1, and SPTBN5 that were depicted earlier in AML. Elucidating different layers of aberrations in normal karyotype no-driver acute myeloid leukemia pro- vides better biology insight and may impact risk-group stratification and new potential driver events. Therefore, this study aimed to detect such anomalies in samples without known driver genetic abnor- malities using multi-omic molecular profiling. Samples were analyzed using RNA sequencing (n=43), whole genome sequencing (n=43), and EPIC DNA methylation array (n=42). In 33 of 43 patients, all three layers of data were available. I developed a pipeline looking for a driver in any layer of data by connecting the information of all layers of data and utilizing public genomic, transcriptomic, and clinical data available from TCGA. Genetic alterations of somatic cells can drive malignant clone formation and promote leukemogenesis. Therefore I first built a mutation prioritization workflow that checks each patient’s genomic mutation drivers. Here I use the information on the allele frequency of the specific mu- tation combining information from WGS and RNA sequencing data. Finally, I compared each mutation on a positional level with AML and other TCGA cancer cohorts to assess the causative genomic muta- tions. I found potential driver stopgain mutation in genes implicated in chromosome segregation during mitosis and some tumor suppressor genes. I found new stopgain mutations in cancer genes (NIPBL and NF1). Since fusions are increasingly acknowledged as oncology therapeutic targets, I investigated potential driver fusion events by evaluating high-confidence and in-frame cancer-related fusion findings. As a result, I found specific gene fusion patterns. Kinases activated by gene fusions define a meaningful class of oncogenes associated with hematopoietic malignancies. I identify several novel and recurrent fusions involving kinases that potentially play a role in leukemogenesis. I detected previously unreported fusions involving known cancer-related genes, such as PIM3- RAC2 and PROK2- EIF4E3. In addition, outliers, such as gene expression levels, can pinpoint potential pathogenic events. Therefore, combining my AML cohort with a healthy control group, I determined aberrant gene expression levels as possible pathogenic events using the deep learning method. Finally, I combined the data and looked for a com- parison to the methylation pattern of each patient. Overall, the analysis uncovered a rich landscape of potential drivers. In different data layers, I found an altered genomic and transcriptomic signature of different GTPases, which are known to be involved in many stages of tumorigenesis. My methods and results demonstrate the power of integrating multi-omics data to study complex driver alterations in AML and point to future directions of research that aim to bridge gaps in research and clinical applications. Furthermore, I provide in vitro evidence for antileukemic cooperativity and epigenetic activity between DAC and ATRA. I performed differential methylation analysis on CpG resolution and across genomic and transposable elements regions, enhancing the results’ statistical power and interpretabil- ity. I demonstrated that single-agent ATRA caused no global demethylation, nor did ATRA improve the demethylation mediated by DAC. In summary, combining multi-omics profiling is a powerful tool for studying dysregulated patterns in AML. Furthermore, multi-omics profiling performed on mutation- negative nkAML reveals several promising drivers. My findings not only go beyond augmenting my understanding of the heterogeneity landscape of AML but also may have immediate implications for new targeted therapy studies.