Phenochart software

Two panels of antibodies were used to perform mIF on CRC TMAs (n=787). On panel 1, a fully automated mIF assay was developed on the Ventana Discovery Ultra autostainer platform (Roche Tissue Diagnostics, software version RUO Discovery Universal V21.00.0019). Staining was performed on 4 µm thick sections of previously constructed TMAs with the optimised antibodies (Table S1). A negative control slide was used on each staining run to rule out non-specific staining. Whole slide images were captured at 10x magnification using the PhenoImager HT multispectral slide scanner (Akoya Biosciences V1.0.13), TMA maps were applied using Phenochart software (Akoya Biosciences V1.1.0), and core images were captured at 20x magnification. Core images were spectrally unmixed using Inform software (Akoya Biosciences, software version 2.5.1).
mIF panel 2 was stained using an autostainer (Thermofisher) with optimised antibodies (Table S1). The slides were scanned by NanoZoomer S60 digital slide scanner (Hamamatsu, USA) with 20x magnification. TMA maps were applied for further analysis. Visiopharm (version 2021.02.5.10297), a digital precision pathology software, was used to perform the analysis. The percent positive cells of total cells detected for each marker were calculated (Figure S1).

Whole slide scans were imaged on the Vectra 3.0 Automated Quantitative Pathology Imaging System (Akoya) using the 20× objective. We randomly selected 12 multispectral image regions each slide using Phenochart software (Akoya Biosciences) at 40× magnification, which were then analyzed with inForm software (v2.4.10, Akoya) to unmix adjacent fluorochromes, subtract autofluorescence, segment the tissue into tumor and stroma regions, segment the cells into nuclear, cytoplasmic, and membrane compartments, and to phenotype the cells according to morphology and cell marker expression based on either their PD-L1 status or their expression of CD68, CD11c, and CK (CD68⁻CD11c⁻CK⁻ cells were defined as “other”). Independent projects were created to phenotype each cellular marker, then merged, consolidated, and analyzed in R Studio using the phenoptrReports plug-in (Akoya Biosciences) to quantify the total number of PD-L1⁺ cells that overlapped CD68/CD11c/CK/other cells. We also used phenoptrReports to quantify entire-cell PD-L1 expression in each phenotype category in all 23 samples. Use of human samples was approved by IRB of University of Colorado AMC (COMIRB Protocol 16-2436).

Liver tissues were processed for hematoxylin and eosin (H&E) staining and immunohistochemistry (IHC) staining for CK-19 and Ki67 at the Mouse Model Core at the VCU Massey Cancer Center (Richmond, VA, USA). Picro Sirius Red Staining was performed using the commercial Kit (Abcam, USA) with the paraffin-embedded tissue sections according to the manufacturer's instructions. Small intestine tissues were processed for H&E staining. Alcian blue staining was performed using the Alcian blue Stain Kit (Abcam, USA). Immunofluorescence staining of ZO-1 was performed with the paraffin-embedded tissue sections according to the manufacturer's instructions. All the stained slides were scanned using a Vectra Polaris Automated Quantitative Pathology Imaging System (Akoya Biosciences, MA, USA), and the images were captured using Phenochart software (Akoya Biosciences, MA, USA).