Abstract PD4-01: PD4-01 Spatial transcriptomics reveals a substantial heterogeneity in TNBC tumor and stroma compartments with potential clinical implications

Abstract

Abstract Background Triple negative breast cancer (TNBC) is a heterogeneous disease characterized by at least five molecular subtypes, namely basal like (BL), immunomodulatory (IM), luminal androgen receptor (LAR), mesenchymal (M) and mesenchymal stem like (MSL), associated with distinct gene expression, genomic and tumor microenvironment (TME) profiles. Recent technological advances allow to investigate intratumor geographic heterogeneity ignored by bulk tumor analyses. Here, we deployed spatial transcriptomics (ST) to interrogate tumor and stroma compartments heterogeneity and assess its association with clinical outcome. Methods Spatial transcriptomics (Visium® Spatial Gene Expression, 10X Genomics) was performed on a retrospective series of 94 case-control TNBC samples matched for known clinic-pathological parameters with available long term outcome. Detailed morphological annotations spanning 11 histomorphological categories were performed by a breast dedicated pathologist assisted by the automated QuPath digital pathology software. Bioinformatic analyses were performed using in house pipelines. Results We investigated the distribution of each morphological category across the five TNBC molecular subtypes. We found that LAR, M and MSL, even though they had less tumor cells, had more patches of very small size (p< 0.0001). On the other hand, BL and IM had few large and dense patches. Stroma had an opposite distribution compared to tumor, with LAR and MSL being enriched with stroma (p< 00001). Tumor infiltrating lymphocytes were specific of the IM (p< 0.0001) and to a lesser extent the BL subtype. Normal structures, like fat tissue (p< 0.008), lactiferous ducts (p< 0.03) and vessels (p< 0.02), were more present in MSL and LAR. The differences between the molecular subtypes are mirrored at the level of their cell composition and tumor organization, suggesting the possibility to assess TNBC molecular subtypes from imaging data alone. At the gene expression level, spatial deconvolution analyses revealed the co-existence of tumor and stroma compartments from different TNBC subtypes within a tumor sample of a given subtype as defined by bulk tumor analysis. Interestingly, these different tumor-stroma combinations were associated with prognosis. For example, M tumors associated with MSL stroma seem to have a better prognosis than M tumors with an M stroma (p=0.001). Furthermore, spatial resolution of the gene expression identified 418 individual clusters (median 4 clusters per sample) associated with specific molecular and cellular features highlighting a substantial intra-patient heterogeneity. These clusters were further grouped into 11 ecotypes associated with distinct hallmarks and pathways, including EMT, angiogenesis, DNA repair and immune profiles revealing an important inter-patient heterogeneity beyond TNBC classification. Interestingly, 2 ecotypes were identified within the IM subtype associated with distinct clinical outcome, with ecotype 6 characterized by high EMT, mesenchymal stroma and worse prognosis (p = 0,021). Conclusion To our knowledge, this is the largest study demonstrating the substantial intra- and inter-patient heterogeneity characterizing TNBC at an unprecedented level, with differences both in tumor and stroma composition as well as spatial organization and clinical outcome. Our results hightlight the need to consider TNBC heterogeneity for patient care and future clinical development including immunotherapy. Citation Format: Xiaoxiao Wang, David Venet, Frédéric Lifrange, Denis Larsimont, Mattia Rediti, Linnea Stenbeck, David Gacquer, Floriane Dupont, Ghizlane Rouas, Matteo Serra, Joakim Lundeberg, Françoise Rothé, Christos Sotiriou. PD4-01 Spatial transcriptomics reveals a substantial heterogeneity in TNBC tumor and stroma compartments with potential clinical implications [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr PD4-01.