Resumen de Genomic instability as a predictive biomarker for the application of DNA-damaging therapies in gynecological cancer patients
The natural course of tumors matches the progressive accumulation of genomic alterations, triggering a cascade of events that results in genomic instability (GI). This phenomenon includes copy number alterations and constitutes a genomic hallmark that defines specific outcomes beyond histology and other molecular features of the tumor.
In the context of gynaecologic oncology research, GI has gained strength in the last years allowing the stratification of patients according to prognosis and response to certain DNA-damaging agents, such as platinum-based therapies and PARP inhibitors. Particularly in ovarian and endometrial cancers, it has been described a molecular subgroup characterized by high copy number alterations (CNA) related to good prognosis and better response to chemotherapy. This relationship highlights GI as a predictive and prognostic biomarker. Hence, a GI-based model translated into clinical practice would constitute a tool for optimizing clinical decision-making.
The era of personalised medicine arrived together with the coming of integrative studies, where results of high-throughput techniques are combined to obtain a comprehensive molecular landscape of the diseases, bringing a new paradigm to characterize the tumors beyond classical anatomic and histological characteristics.
This thesis proposes a global study of the phenomenon of GI as a prognostic and predictive biomarker of treatment response in gynaecological cancers, mainly focused on high-grade ovarian cancer and endometrial cancer. Through the development of an NGS-based strategy with the adaptation of available pipelines of analysis, we obtained GI profiles on formalin-fixed paraffin-embedded samples in a reliable, portable, and cost-effective approach, with the combination of Machine Learning tools to fit prognostic and predictive models based on the integration of omic data.
Based on that premise, we fit and validated, in well-characterized clinical cohorts, three single-source models and an integrative ensemble model (Scarface Score) that proved to be able to predict response to DNA-damaging agents in a clinical scenario of High-Grade Serous Ovarian Cancer. In addition, a mutational-based algorithm (12g algorithm) with prognostic impact was developed and validated for endometrial cancer patients. This algorithm achieved a GI-based stratification of patients. Finally, a panel of ovarian cancer cell lines was characterized at the response, genetic and genomic level, interrogating homologous recombination repair pathway status and its associated GI profiles, completing the molecular landscape, and establishing the basis and breeding ground of future preclinical and clinical studies.
The results reported in this Doctoral Thesis provide valuable clinical management tools in the accomplishment of a reliable tailored therapy. Additionally, future studies in different tumor types and drugs for implementation of the predictive model can be planned, using as a base the defined one but re-establishing new and specific cut-offs.
