Development and application of methodologies and infrastructures for cancer genome analysis within Personalized Medicine

Abstract

[eng] Next-generation sequencing (NGS) has revolutionized biomedical sciences, especially in the area of cancer. It has nourished genomic research with extensive collections of sequenced genomes that are investigated to untangle the molecular bases of disease, as well as to identify potential targets for the design of new treatments. To exploit all this information, several initiatives have emerged worldwide, among which the Pan-Cancer project of the ICGC (International Cancer Genome Consortium) stands out. This project has jointly analyzed thousands of tumor genomes of different cancer types in order to elucidate the molecular bases of the origin and progression of cancer. To accomplish this task, new emerging technologies, including virtualization systems such as virtual machines or software containers, were used and had to be adapted to various computing centers. The portability of this system to the supercomputing infrastructure of the BSC (Barcelona Supercomputing Center) has been carried out during the first phase of the thesis. In parallel, other projects promote the application of genomics discoveries into the clinics. This is the case of MedPerCan, a national initiative to design a pilot project for the implementation of personalized medicine in oncology in Catalonia. In this context, we have centered our efforts on the methodological side, focusing on the detection and characterization of somatic variants in tumors. This step is a challenging action, due to the heterogeneity of the different methods, and an essential part, as it lays at the basis of all downstream analyses.

On top of the methodological section of the thesis, we got into the biological interpretation of the results to study the evolution of chronic lymphocytic leukemia (CLL) in a close collaboration with the group of Dr. Elías Campo from the Hospital Clínic/IDIBAPS. In the first study, we have focused on the Richter transformation (RT), a transformation of CLL into a high-grade lymphoma that

leads to a very poor prognosis and with unmet clinical needs. We found that RT has greater genomic, epigenomic and transcriptomic complexity than CLL. Its genome may reflect the imprint of therapies that the patients received prior to RT, indicating the presence of cells exposed to these mutagenic treatments which later expand giving rise to the clinical manifestation of the disease. Multiple NGS- based techniques, including whole-genome sequencing and single-cell DNA and RNA sequencing, among others, confirmed the pre-existence of cells with the RT characteristics years before their manifestation, up to the time of CLL diagnosis. The transcriptomic profile of RT is remarkably different from that of CLL. Of particular importance is the overexpression of the OXPHOS pathway, which could be used as a therapeutic vulnerability. Finally, in a second study, the analysis of a case of CLL in a young adult, based on whole genome and single-cell sequencing at different times of the disease, revealed that the founder clone of CLL did not present any somatic driver mutations and was characterized by germline variants in ATM, suggesting its role in the origin of the disease, and highlighting the possible contribution of germline variants or other non-genetic mechanisms in the initiation of CLL.