From cell counts to cellular interactions: Cu-Cyto and the co-localization index as a spatial framework for the tumor immune microenvironment of rectal cancer
Cohort study reveals stromal immune cell co-localization predicts relapse-free survival in rectal cancer, indicating spatial metrics improve prognostic assessment.
Key Points
To develop and evaluate a deep learning image cytometry platform and spatial co-localization metric that quantifies multi-cellular interactions within the rectal cancer tumor immune microenvironment.
Engineered Cu-Cyto, a deep learning cytometry system utilizing a bit-pattern kernel-filtering algorithm and an off-target labeling strategy to detect and classify approximately 20 cell types from standard immunohistochemistry whole-slide images.
Formulated the Co-Localization Index to convert classification probabilities and nuclear coordinates into quantitative spatial interaction scores between two or three cell types.
Evaluated the spatial distribution and prognostic value of CD103⁺CD8⁺ tissue-resident memory-like T cells in patients with rectal cancer treated with neoadjuvant chemoradiotherapy.
Stromal density of CD103⁺CD8⁺ T cells independently predicted relapse-free survival, whereas intratumoral density showed no independent prognostic association.
The Co-Localization Index captured tri-cellular spatial biology among CD103⁺CD8⁺ T cells, malignant tumor cells, and stromal elements that compartment-aware cell density alone could not resolve.