Extended Data Fig. 6: Quantitative and spatial immune characterization through immunohistochemistry and multispectral imaging.

a, Immunohistochemistry (IHC) quantification of the immune checkpoints CTLA4, IDO1, TIGIT and TIM3. Each of the tested markers was validated in SCC tissue (top). P values are derived from a non-parametric one-tailed Mann–Whitney U test used to validate increase in SCC compared to normal tissue. b, Comparison of PD-L1 densities between the stroma of normal tissue (stage 0) and SCC (stage 8), derived from multiplex immunohistochemistry. c, Clustering of normalized immunohistochemistry expression. d, A methodology for spatial analysis of multispectral imaging data. A whole slide is reconstructed from the individual images. On the basis of the tissue categorization, the images are masked to exclude the blank areas. Immune-cell densities are calculated as the number of cells per tissue area (m²). Spatial localization is analysed within the selected region of interest. e, Representative examples of CK⁻PD-L1⁺ in both SCC and severe dysplasia. Single-positive PD-L1 cells (CK⁻PD-L1⁺) were generally immune cells that were located in the stroma, with morphological similarities to infiltrating macrophages. f, We calculated the area between the theoretical and the empirical curve because deviations between the two can indicate clustering or segregation patterns (see Fig. 4b, bottom) to confirm that epithelial cells segregate from CD3 T cells in high-grade lesions, independently from the distance threshold of 25 μm.