Biology and bioinformatics oriented projects
Our goal is to identify biomarkers that can serve either as therapeutic targets or as clinical decision tools to guide the selection of optimal treatment strategies. These biomedical-oriented projects also highlighted the need for more advanced mathematical and computational approaches tailored to high-dimensional tabular data—an essential aspect of developing precision medicine strategies. For more details, see the Mathematics and computer science -oriented projects section.
1. Breast Cancer
Breast cancer remains the leading cause of cancer-related death among women, primarily due to treatment resistance in metastatic cases. While some patients are diagnosed with de novo metastatic breast cancer, the majority of metastatic cases arise from progression of the initial disease following treatment failure.
Heterogeneity in components forming tumor microenvironment (TME) is now considered as one of the major causes of treatment resistance. Indeed, TME is inextricably involved in cancer progression - from carcinoma in situ to local invasion and dissemination - and variation in the expression of several of its components have been associated with breast cancer recurrence. Given the strong inter-patient variability in TME composition, there is a growing need for personalized treatment approaches, which in turn require the identification of novel biomarkers to better characterize tumor biology.
In our first project - PredAlgoBC, funded by the European Commission Horizon 2020 program /Marie Skłodowska-Curie Action - we focused on identifying biomarkers associated with response to hormone therapy in patients with early-stage breast cancer. We developed supervised machine learning models using integrated transcriptomic datasets selected from public repositories. Using supervised machine learning models applied to integrated transcriptomic datasets from public repositories, we discovered that neuronal progenitor markers and perineural invasion were associated with poor outcomes in patients treated with hormone therapy (Basseville et al., Cancer Research Communications, 2022) . The study also resulted in the development of two nervous system-related gene expression which demonstrated strong predictive and prognostic performance across multiple cohorts. These findings underscore the potential of neuronal components as novel biomarkers and therapeutic targets in breast cancer.
We are now expanding this research through the INTRICATE project - funded by La Ligue Contre Le Cancer and Perseverance - which investigates the biological interactions between tumoral nerves, cancer cells, and other components of the TME using spatial transcriptomics.
2. Malignant peritoneal mesothelioma (MPM)
MPM is a rare and invasive tumor formerly considered a lethal neoplasm. Although the overall prognosis remains sometimes poor, outcomes have significantly improved with the combined use of cytoreductive surgery and intraperitoneal chemotherapy, reaching an overall survival of 34-100 months. Several factors make the disease difficult to treat, including late diagnosis due to its rarity and nonspecific symptoms, tumor aggressiveness, and limited treatment options, particularly for patients unsuitable for surgery. Moreover, MPM can be confused with other peritoneal tumors, especially its benign form (BPM) or metastases from ovarian serous carcinoma (OSC), leading to misdiagnosis that can worsen the prognosis. Studies indicate a lack of biomarkers for reliable histopathological distinction between the diseases.
Moreover, despite recent therapeutic advances, comprehensive molecular characterization of MPM remains limited, hindering the development of targeted therapy and immunotherapy. In contrast, pleural mesothelioma has benefited from earlier molecular profiling, enabling new strategies like FDA-approved immunotherapy. Although current MPM treatments are largely extrapolated from pleural mesothelioma, caution is needed due to molecular and immunological differences between the two. However, existing studies on MPM molecular profiling involve small cohorts and mainly focus on genetic alterations rather than expression. This highlights a significant research gap that must be addressed to support therapeutic transfer, as expression markers are also essential for the development of new therapeutic strategies.
In this context, we initiated a collaborative study ICO/HCL to explore the proteomic landscape of MPM in comparison with BPM and OSC (thanks to AMARAPE funding). Our goals are: (1) to identify biomarkers that help pathologists distinguish MPM from OSC and BPM, and (2) to profile MPM to gain insights into tumor behavior and vulnerabilities.
