Thesis defended on december 4, 2025 to obtain the degree of Doctor of the Université Grenoble Alpes

Abstract
Clear cell renal cell carcinoma (ccRCC), the most common kidney cancer subtype, is characterized by profound metabolic reprogramming and a tumor microenvironment (TME) potentially rich in immune cells. Although immune checkpoint inhibitors (ICIs) have transformed the therapeutic landscape of ccRCC, durable clinical responses remain limited, highlighting the urgent need to unravel the mechanisms of immune evasion and therapeutic resistance. In this thesis, we sought to identify novel biomarkers of immune resistance in ccRCC through network biology and patient-derived tumor organoid models. We firstly established a sample-specific gene co-expression network framework to identify network features predictive of patient response to nivolumab. Our results showed increased gene connectivity and stronger negative gene-gene associations are associated with clinical responses, and that incorporating these features into gene expression-based machine learning models improved the accuracy of predicting treatment benefit. We next applied single-cell gene regulatory networks to dissect cell-type–specific transcriptional programs underlying tumor progression. This approach identified regulatory interactions and transcription factors linked to tumor grade in both epithelial and CD8+ T cells, and highlighted potential regulatory interactions within the PD-1 pathway that may serve as candidate biomarkers of immunotherapy. Also, our results revealed cell-type-specific gene regulatory relationships may be associated with patient survival outcomes. We finally evaluated the transcriptional fidelity of patient-derived tumor organoids relative to their parental tumors. We observed that stress- and hypoxia-related transcriptional changes introduced from tumor dissociation, and the re-emergence of immune-related signatures during the formation of immune-infiltrated organoids. These support the utility of patient-derived organoids as robust preclinical platforms for immunotherapeutic studies. Together, these analyses demonstrate that metabolic reprogramming and immune infiltration, captured by network features, organoid modeling, and regulatory interactions, are interconnected processes that collectively drive immune evasion and therapeutic resistance in ccRCC. In conclusion, this work advances our understanding of the metabolic–immune axis in ccRCC, identifies candidate biomarkers with potential predictive values, and provides a conceptual framework for the development of precision immunotherapy strategies targeting both tumor metabolism and immune regulation. ​