• Citer

Dissecting Tumor Heterogeneity in Triple-Negative Breast Cancer Through Spatial Transcriptomics

par Wang, Xiaoxiao
Président du jury Le Moine, Alain
Promoteur Sotiriou, Christos
Publication Non publié, 2026-10-01

Thèse de doctorat

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

Triple negative breast cancer (TNBC) is an aggressive breast cancer (BC) subtype associated with poor outcomes. Although the introduction of immunotherapy has significantly improved survival in the early-stage TNBC, a substantial proportion of patients still experience disease recurrence after neoadjuvant or adjuvant treatment, while others suffer short- and long-term toxicities with considerable human and financial burdens on healthcare systems. It is now widely recognized that TNBC is not a single disease entity but a collection of biologically distinct subtypes with substantial inter- and intratumoral heterogeneity. Yet, treatment decisions still rely largely on conventional clinicopathological markers, which fail to capture the full biological complexity of the disease. Consequently, ongoing research aims to identify more robust biomarkers to refine patient stratification and guide therapeutic decisions.Bulk transcriptomics analyses previously identified at least five TNBC molecular subtypes namely basal-like (BL), immunomodulatory (IM), mesenchymal (M), mesenchymal stem-like (MSL), and luminal androgen receptor (LAR)—each characterized by distinct transcriptomic, genomic, and tumor microenvironment profiles. However, this classification was derived from bulk RNA sequencing, which masks spatial intratumoral heterogeneity and oversimplifies the complexity of the tumor ecosystem, limiting their clinical applicability.This thesis aims to deepen our understanding of TNBC heterogeneity and identify novel biomarkers for treatment optimization using spatially resolved approach in a retrospective cohort of 92 upfront-resected early-stage TNBCs. To this end, we performed spatial transcriptomic (ST), a technology that enables two-dimensional mapping of gene expression across entire tumor sections and directly links molecular profiles to tissue architecture.For the first objective, we conducted a systematic histomorphological annotation of hematoxylin and eosin–stained slides. We demonstrated that distinct morphological patterns correlate strongly with bulk RNA-seq–derived TNBC molecular subtypes, suggesting that morphology-based classification could complement—or in some settings substitute—bulk transcriptomic profiling.For the second objective, we used spatial deconvolution to extract compartment-specific gene expression profiles from tumor and stromal regions to identify the specific features that characterize TNBC molecular subtypes. Applying molecular classification independently to each compartment revealed that the IM and MSL subtypes are predominantly stroma-driven, modulating global subtype assignments (BL vs. IM and M vs. MSL) with potential clinical implications. By contrast, LAR tumors remained stable across compartments, underscoring their distinct intrinsic transcriptional identity.For the third objective, we focused on tertiary lymphoid structures (TLS), key organized immune aggregates within the tumor microenvironment. Using ST, we derived a 30-gene TLS gene expression signature predictive of prognosis and immunotherapy response across TNBC, other BC subtypes, and several non-breast malignancies.For the fourth objective, we leveraged ST data to redefine TNBC beyond preexisting bulk classifications. Using a stepwise clustering pipeline, we identified nine recurrent patient-level spatial archetypes that capture geographical intratumoral heterogeneity and map distinct clinical behaviors. Several archetypes were associated with prognosis and highlighted therapeutically actionable pathways, providing a rationale for new treatment approaches tailored to the heterogeneous biology of TNBC.In conclusion, this work enhances our understanding of TNBC heterogeneity, challenges the long-standing view of TNBC as a uniform entity, and provides a biological foundation to involve biomarker-driven treatment strategies.