Functional Data Analysis of Spatial Clustering Identifies Prognostic T Cell Patterns in Ovarian Cancer

Area of Science:

• Oncology
• Computational Biology
• Immunology

Background:

• Spatial proteomic imaging assesses immune cells in tumors.
• Current clustering methods rely on fixed radii, potentially missing patterns.

Purpose of the Study:

• To develop a novel framework for analyzing spatial clustering across multiple scales.
• To investigate the prognostic value of immune cell spatial organization beyond abundance.

Main Methods:

• Functional data analysis (FDA) and functional principal component analysis (FPCA) were used.
• Cox proportional hazards models incorporated spatial features and immune cell abundance.
• Analysis included multiplex immunofluorescence data from 773 ovarian cancer patients.

Main Results:

• Higher CD3+ and CD8+ T cell abundance correlated with improved survival.
• Spatial features, particularly for CD8+ T cells, independently predicted survival.
• Prognostic effects of immune infiltration depended on spatial clustering; high abundance with low clustering showed best outcomes.

Conclusions:

• Immune cell spatial organization offers prognostic information beyond cell counts.
• Diffuse immune infiltration may indicate more effective anti-tumor activity in ovarian cancer.
• FDA provides a flexible framework for analyzing spatial clustering and identifying prognostic features.