Background

The role of tumour innervation with respect cancer progression is an emerging field of study. One of the pervasive mechanisms associated with metastasis and poor prognosis in various cancer types is perineural invasion (PNI), by which tumour cells invade the space surrounding a nerve. It is hypothesized that PNI is a consequence of a complex multi-step process where peripheral nerves, malignant cells, and stromal cells cooperate to drive processes such as neurotropism, ECM remodelling and epithelial-mesenchymal transition (EMT).

Extensive transcriptomic profiling on PNI positive tumours reveals upregulation of neurotrophic factors, EMT signatures, matrix metalloproteinases (MMPs) and various cell adhesion factors, leading to the postulation that peripheral nerves within the TME undergo series of regeneration processes upon damage. Unravelling whether such processes create less-resistant paths for cancer cell migration is fundamental to establish the initiation mechanisms of PNI and its significance in tumour proliferation.

Yet, most published transcriptomics studies on PNI were limited by the lack of spatial information. Analysis at bulk and single cell level failed to locate the cell populations in an invaded peri-neural niche thus establish causal relationship between the aforementioned transcriptomic response and PNI. While emerging spatial transcriptomic technologies present opportunities for the research area, adequate analysis methods are yet to be solidified.

Objectives

The major objective of this summer project is to develop computational analysis methods for spatial transcriptomic analysis as the cornerstone of the study.

Methodology

To drive spatial-level characterization of cellular heterogeneity and interaction at the peri-neural niche, a computational pipeline with respect to spatial transcriptomics data generated by the CosMx® Spatial Molecular Imager (SMI) platform will be developed, covering the following aspect of analysis:

• Cell phenotyping and validation: Annotating cell types with results validated using two methods. 1) marker-based cell type identification, 2) mapping to publicly available reference scRNA atlas.
• Enrichment or neighbourhood analysis: Performing enrichment analysis by cell type using the relevant gene signatures or neighbourhood clustering and projection to single cell reference to characterize pathway activities and the heterogeneity of cell populations.
• Cell trajectory inference in spatial context: Measuring intercellular transfers, reconstructing tumour proliferation processes, and identify transient states, branching points etc.
• Cell-cell interaction in spatial context: combining ligand-receptors signal and cellular spatial information (e.g. distance and distribution) to reveal how malignant cells interact with other cell populations to gain invasion potentials.