Tissue microarrays (TMAs) were constructed [50 (link)]. For CD3, CD8 and FOXP3, antigen retrieval was performed, and deparaffinized tissue sections in EDTA Solution (pH 8.0) (Zymed, San Francisco, CA) were treated by a microwave for 30 sec in a pressure cooker. For CD45RO (the official gene symbol PTPRC), Citrate Buffer (pH 6.0) (Zymed) was used for antigen retrieval. Tissue sections were incubated with peroxidase block (5 min); protein block (20 min); and then, primary antibody against CD3 (rabbit polyclonal anti-human CD3 antibody, clone F7.2.38, 1:250 dilution, Dako Cytomation, Carpinteria, CA), CD8 (mouse monoclonal anti-human CD8 antibody, clone C8/144B, 1:100 dilution, Dako Cytomation), CD45RO (mouse monoclonal anti-human CD45RO antibody, clone UCHL1, 1:100 dilution, Dako Cytomation) or FOXP3 (mouse monoclonal anti-human FOXP3 antibody, clone 206D, 1:50 dilution; BioLegend, San Diego, CA) for 1 hr at room temperature. Next, we applied Envision System HRP-labeled polymer anti-rabbit (for CD3) or anti-mouse (for the other markers) (Dako Cytomation) for 30 min, diaminobenzidine (5 min) and hematoxylin counterstain (1 min).
After staining for each T-cell subset, TMA slides were scanned by an automated scanning microscope and Ariol image analysis system (Genetix, San Jose, CA) (Figure 1 ). The software was used to count the number of positive nuclei in each tissue core. We marked neoplastic epithelial areas, to exclude non-neoplastic areas (stroma, normal mucosa, and necrotic regions). We calculated the average density (cells/mm2) of each tumour-infiltrating T-cell subset. In addition, we analysed subset T-cell density in stromal areas as well as in whole TMA core (Supplementary Tables 1 -2 ).
Immunohistochemical evaluation of T-cell subsets in cancer tissue has been a challenge, and there has been no standardized method. Pre-analytical variables such as tissue processing may have considerable impact on antigenicity of each T-cell subset, which may be substantially influenced by a subtle difference in conditions of immunohistochemical procedure. We tried to minimize such measurement errors in a number of ways. We constructed TMA and performed the immunohistochemical procedure in a very similar condition for all specimens. We obtained two to four tumour tissue cores from each case, considering within-tumour heterogeneity.
After staining for each T-cell subset, TMA slides were scanned by an automated scanning microscope and Ariol image analysis system (Genetix, San Jose, CA) (
Immunohistochemical evaluation of T-cell subsets in cancer tissue has been a challenge, and there has been no standardized method. Pre-analytical variables such as tissue processing may have considerable impact on antigenicity of each T-cell subset, which may be substantially influenced by a subtle difference in conditions of immunohistochemical procedure. We tried to minimize such measurement errors in a number of ways. We constructed TMA and performed the immunohistochemical procedure in a very similar condition for all specimens. We obtained two to four tumour tissue cores from each case, considering within-tumour heterogeneity.