Characterizing the tumor microenvironment of a colorectal adenocarcinoma sample with the Orion spatial biology platform and HALO quantitative image analysis
This SITC 2024 poster from Indica Labs and RareCyte pairs single-round multiplex immunofluorescence and whole-slide imaging on the Orion platform with HALO AI image analysis, quantifying nearly two million cells across one high-grade colorectal adenocarcinoma section.
Presented by NeoGenomics.
Read or download to see:
- A 13-channel multiplex immunofluorescence panel was imaged in a single staining round on the Orion platform. A Hoechst nuclear stain and 12 antibody markers — Ki-67, CD3e, CD4, CD8a, CD20, CD68, CD163, FOXP3, PD-1, PD-L1, Pan-CK and Vimentin — were captured together in one imaging round, with tissue autofluorescence isolated as an additional channel for morphology.
- Nearly two million cells were segmented across the whole slide and quantified separately in tumor and stroma. HALO AI networks classified tumor, stroma, necrosis and glass, improved the nuclear and membrane segmentation, and read each marker’s density by compartment — Pan-CK-positive cells reached 9,626.82 cells/mm² in tumor against 328.15 in stroma, while CD3e-positive T cells ran the other way, 1,745.58 in stroma versus 164.94 in tumor.
- Five cell phenotypes were defined by marker combination and placed in space relative to one another. Activated tumor cells (Ki-67⁺Pan-CK⁺), cytotoxic T cells (CD8a⁺CD3e⁺), T-helper cells (CD4⁺CD3e⁺), tumor-associated macrophages (CD163⁺) and regulatory T cells (CD4⁺CD3e⁺FOXP3⁺) were quantified across tumor and stroma; HALO proximity analysis then placed cytotoxic T cells an average of 99 µm from activated tumor cells, alongside density-heatmap and infiltration analyses of the same populations.
