Multi-omics analysis of relapsed acute myeloid leukemia

The prognosis for patients suffering from acute myeloid leukemia (AML) remains unsatisfactory and survival is often measured in months. Although the majority of patients achieve complete remission after aggressive treatment, most of them relapse within a few years. Those patients that relapse frequently show accelerated disease progression and therapy resistance and represent the major clinical challenge in AML oncology. The advent of massive parallel sequencing launched the detailed understanding of the molecular basis of AML leukemogenesis, however, studies focused on relapse and primary resistant (R/PR) AML remain sparse. This thesis explores the spectrum of molecular alterations present in R/PR AML, using a multi-omics analysis approach on sequential primary patient specimens from 48 adult and 25 pediatric R/PR AML patients. In Paper I we applied genome wide next generation sequencing to investigate genomic alterations in adult and pediatric R/PR AML. We identified recurrent alterations affecting MGA, ARID1A and H3F3A, specific for adult R/PR AML cases. In addition, we reported previously unappreciated internal tandem duplications in UBTF, solely found in pediatric cases. In Paper II we showed an association between a pro-inflammatory signature and AML relapse, utilizing transcriptome wide RNA sequencing. Further, through a novel machine learning based analysis we were able to depict gene interactive networks and predictive features in AML relapse. In Paper III we performed DNA methylation analysis to further understand transcriptional changes during disease progression that could not be explained by genomic alterations. We identified hypermethylation of promoter associated sites of RNF180 and DSC3 at relapse and in treatment resistant AML samples, respectively, as well as concordant downregulated transcription of these genes. In Paper IV we were able to confirm some of the above mentioned alterations at the proteomic level by exploiting high resolution mass spectrometry data. In addition, we showed higher levels of mitochondrial related proteins at AML relapse. In summary, molecular associations identified in this thesis, together with AML-specific neoantigens discovered via a proteogenomic approach in Paper IV, predict novel therapeutic targets and/or help to further optimize current treatment schemes. We envision that knowledge gathered through our studies will shed further light on the molecular characteristics underlying disease progression, thus contributing to prolong AML patient survival