Halo imaging analysis software

The compositions of major immune cells in TLSs of ESCCs were examined on FFPE ESCC slides using mfIHC and the PANO Multiplex IHC kit (Panovue, BJ, China) with antibodies for CD20 (Abcam, Boston, MA), CD21 (Abcam), Ki67 (Abcam), CD4 (Abcam), CD8 (ZSGB‐BIO, Beijing, China), LAMP3 (Abcam) and 4ʹ‐6ʹ‐diamidino‐2‐phenylindole (DAPI) as described previously.⁷ The stained slides were scanned using a Vectra Polaris Automated Quantitative Pathology Imaging System and batch analysed with HALO imaging analysis software (Indica Labs, Corrales, NM). Each TLS was segmented into a B cell zone (CD20+ cell and CD21+ cell‐clustered region) and a T cell zone (CD4+ cell and CD8+ cell‐clustered region). Cells in TLSs were phenotyped as B cells (CD20+), FDCs (CD21+), proliferating cells (Ki67+), CD4+ T cells (CD4+), CD8+T cells (CD8+) and mature DCs (LAMP3+). The intensity for each marker was recorded. The intratumoural infiltration of CD8+ T cells, CD4+ Th cells, Treg cells, memory T cells, natural killer cells, DCs and macrophages were evaluated on ESCC tissue microarrays as previously described.⁷

Fourteen days after tumor implantation, 6 mg/kg of Texas Red-conjugated 3 kDa dextran (TRD, ThermoFisher) was injected intravenously through the tail vein. TRD was allowed to circulate for 20 min, animals were exsanguinated by intracardiac perfusion of PBS with subsequent perfusion of 4% PFA solution for brain tissue fixation. Extracted brains were embedded in paraffin and sectioned at 5 μm thickness. Formalin-fixed sections were then subjected to either TRD staining or immunohistochemical (IHC) staining. Whole slide images were obtained at high resolution, 20×g using an AT2 scanner (Aperio, Leica Biosystems, Buffalo Grove, IL). Double staining for CD31/Claudin5 were performed on the Leica Biosystems Bond RX autostainer using the Bond Polymer Refine Kit (Leica Biosystems DS9800), with omission of the Post Primary reagent, DAB and Hematoxylin. Slides were double stained for CD31 (Abcam ab28364, 1:100), and Claudin-5 (Invitrogen 35-2500, 1:50). Images were captured using the Aperio Scanscope FL whole slide scanner (Leica Biosystems) into whole slide digital images and OPAL Fluorophore 520 (AKOYA) as per previously published [25 (link)]. Image analysis was performed using Halo imaging analysis software (v3.3.2541.423, Indica Lab’s, Corrales, NM) using Halo algorithm (Area Quantification FL v2.3.4) which were used to quantify positive area of TRD, CD31, and Claudin5.

The pMLKL was classified by labeling index using the following formula: Labeling index (%) = number of pMLKL-positive cells × 100/number of total cells counted. Inflammatory/immune cell infiltration was initially determined by H&E staining. To quantify the tumor-infiltrating inflammatory/immune cells, three non-overlapping fields with high numbers of tumor-infiltrating immune cells (hot spots) were selected (Supplementary Fig. S4). The stained slides were scanned and digitally converted into virtual slides using NanoZoomer (Hamamatsu Photonics K.K., Japan). Inflammatory/immune cells were counted under high power magnification (× 400) by two of the investigators (TL and HU) using digital image analysis (Halo imaging analysis software; Indica Labs, Corrales, New Mexico, USA). For statistical analyses, low and high CD8 + T cells, FOXp3 T cells, and CD163 + macrophages were grouped using the median as a cutoff.