Single-cell atlas of pancreatic cancer reveals microenvironment ecosystems associated with prognosis and treatment responses

Thesis

Due to its late-stage diagnosis and resistance to broad range of therapies, pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal solid malignancies. The tumour microenvironment (TME), comprising of diverse malignant, stromal, and immune cell states, exhibits significant heterogeneity which profoundly influences patient prognosis, metastatic potential, and treatment responses. Advances in single-cell RNA-sequencing (scRNA-seq) technologies have enabled studies to perform high-resolution characterization of the neoplastic and TME cellular landscape, overcoming previous limitations of bulk transcriptomic approaches. Yet, the limited number of donors, variable experimental techniques, and inconsistent cell type definitions prevent these studies from serving as a universal reference for this disease. Integrating multiple single-cell datasets to construct cellular atlases can address these limitations, enabling a holistic interrogation of the cellular landscape while capturing extensive inter-patient variability necessary for effective patient stratification.

Therefore, nine publicly available PDAC datasets were integrated to construct a comprehensive single-cell atlas. This single-cell atlas includes data from 126 individuals and a total of 565,584 cells, encompassing a range of disease states, and treatment conditions. We reveal a diverse repertoire of cell states, many of which are strongly prognostic with patient outcomes across independent cohorts, highlighting their potential as biomarkers for clinical prognosis. Importantly, we discover five robust cellular ecosystems (CEs) associated with distinct clinical outcomes, malignant cell states, and genomic alterations. Notably, one CE defined by immunosuppressive and proliferative cell states across multiple cell types significantly predicts response to immunotherapy yet exhibits marked resistant to chemotherapy across multiple cohorts. Additionally, the identified CEs show limited concordance with previously reported prognostic subtypes, indicating that our classification framework provides novel biological and clinical insights into the PDAC TME. Taken together, these CEs may facilitate effective patient stratification and may inform the development of novel therapeutic strategies.

Authors: Demetriou, C

• DOI: 10.5287/ora-nr4pvynr5
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: pancreatic cancer; tumour microenvironment; ecotypes; patient stratification
• Subjects: Bioinformatics; tumour microenvironment; Pancreatic acinar cells--Cancer