Novel multiplexed imaging spatial analysis method development and application on colorectal cancer liver metastasis and pancreatic pathologies

Thesis

The microenvironment plays a decisive role in disease progression, therapeutic resistance, and immune evasion. While single-cell transcriptomics has revolutionised our understanding of cellular heterogeneity and molecular mechanisms, it requires tissue dissociation and therefore loses spatial context, limiting its ability to reveal how cellular interactions and tissue architecture shape disease.

A systematic framework to interrogate tumour ecosystems at single-cell resolution in situ, while capturing immune-stromal-epithelial interactions, extracellular matrix organization, and morphological evolution, remains lacking.

In this thesis, we developed and applied computational pipelines for multiplexed imaging of whole-slide tissue sections. These pipelines addressed key challenges in annotation, artifact removal, boundary analysis, extracellular matrix quantification, and pseudotime reconstruction of tissue evolution. The framework was applied to colorectal cancer liver metastasis (CRLM) and to three pancreatic pathologies including pancreatic ductal adenocarcinoma (PDAC), IgG4-related autoimmune pancreatitis (IgG4-AIP), and pancreatic neuroendocrine tumour (PNET), as well as normal pancreas, to map niche-level organization across normal, malignant, autoimmune, and indolent contexts.

In CRLM, spatial analysis revealed growth pattern-dependent differences: desmoplastic metastases exhibited proliferative immune and stromal niches, whereas replacement metastases were dominated by proliferating tumour cells and hepatocyte co-option. In pancreatic pathologies, multiplexed imaging highlighted how immune, stromal, and epithelial compartments assemble into distinct niches that diverge across PDAC, PNET, and IgG4-AIP, reflecting disease-specific tissue architectures and evolutionary trajectories. Quantitative tissue schematics and assembly rules uncovered substantial inter-sample heterogeneity, with limited overlap in spatial motifs across disease groups.

This work demonstrates that multiplexed imaging can resolve tissue architectures and immune-stromal dynamics not accessible to dissociated single-cell methods. By integrating morphology, spatial organization, and niche-level interactions, it provides a scalable framework for identifying spatial biomarkers and therapeutic vulnerabilities. These findings highlight the potential of multiplexed imaging, in concert with transcriptomic and proteomic approaches, to advance precision oncology and spatial systems biology.

Authors: Wu, S

• DOI: 10.5287/ora-6r29mnx6d
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: tumour evolution; multiplexed imaging; pancreatic cancer; cell-cell communication; fibrosis; colorectal cancer liver metastasis; tumour immunology; spatial biology; digital pathology
• Subjects: Bioinformatics; Pancreatic acinar cells--Cancer; Colon (Anatomy)--Cancer; Pathology; Cancer; Immune response