Spatial Prostate Assessment and the Circulating Environment - discovering the lethal clone (SPACE-X)

Background Prostate cancer (PCa) diagnosis has advanced significantly, but we still lack certainty about who needs treatment.

PCa is inherently heterogeneous, as demonstrated by my team’s recent work, published in Nature, using spatially resolved transcriptomics and copy number status inference.

We unveiled complex clonal dynamics within the prostate, discovering that epithelial clones traverse histological and grade boundaries. This hinders accurate diagnosis, prognosis and treatment planning.

The next frontier is therefore to identify the specific clones responsible for cancer progression during the diagnostic process.

Metastasis serves as a reliable surrogate for lethality in prostate cancer, and so lymph node metastases constitute a potentially 'lethal clone’, given their progression to skeletal metastases. Aims As an Advanced Clinician Scientist Fellow, my research aims to: 1.

Investigate lethal clone behaviour within the prostate, identifying essential cellular neighbourhoods responsible for local advancement, invasion, and metastasis (WP1) 2.

Develop two windows into clonal lethality: SPACE-CTC, establishing the clonal status of CTCs, identifying their origin in localised PCa; and SPACE-Vision, repurposing open-source image recognition algorithms for morphological clone detection (WP2&3) 3. Validate a non-ST-dependent method of identifying lethal PCa (WP4).

Methods Multifaceted integration of spatial transcriptomics (ST), proteomics, and multiplex imaging to map spatial dynamics of clonal heterogeneity in primary PCa and lymph node metastases.

Data from second-generation Visium™ technology applied to selected archival multifocal prostatectomy cases with co-existent lymph node metastases.

Concurrently, I will lead a program to isolate and analyse CTCs from men with localised PCa using the Genesis CelSelect system. Single-cell RNA sequencing and clonal annotation used to correlate CTC profiles with primary tumour clones.

Model development of a machine learning-based image analysis tool integrating histopathological and ST data to automatically identify lethal clones from standard biopsy samples, leveraging convolutional neural networks (CNNs) and vision transformers (ViTs) to enhance the accuracy and robustness of clone detection.

Both methods will then be validated in retrospective (n=1034) and prospective (n=150) cohorts to assess their predictive value.

How the results of this research will be used: Overdiagnosis in PCa necessitates better identification of potential lethality.

If successful, this research could transform PCa decision-making, reducing harm for patients while targeting treatment effectively.

In summary, this grant application seeks funding to comprehensively study PCa heterogeneity, leveraging spatial transcriptomics, CTC analysis, and image recognition algorithms to improve diagnosis and treatment decision-making. This could have a profound impact on PCa care and aligns with the mission of Cancer Research UK.