Abstract 775: Use of a novel deep learning-based virtual IF and IHC platform for instant multiplexing and quantification of CD4, CD8, PD-L1, FOXP3, and CK AE1/3 in non-small cell lung cancer

Background and Purpose: Non-small cell lung cancer (NSCLC) remains a leading cause of cancer mortality, with the tumor-immune microenvironment playing a pivotal role in disease progression and response to immunotherapy. Biomarkers such as CD4, CD8, PD-L1, FOXP3, and CK AE1/3 are critical for understanding T-cell biology and tumor-immune interactions. Traditional immunofluorescence (IF) and immunohistochemistry (IHC) is time-intensive, resource-heavy, limited in multiplexing capabilities, and requires separate tissue sections for each biomarker, consuming rare, precious tissue. We developed a deep learning-based virtual IF and IHC platform that requires only a single H&E-stained digital slide to virtualize and quantify multiple biomarkers, enabling instant spatial per-cell biomarker analysis while preserving tissue for future studies. Experimental Procedures: Digital NSCLC samples were analyzed using our platform to virtualize biomarkers CD4, CD8, PD-L1, FOXP3, and CK AE1/3. Virtual multiplexing of CK AE1/3 with immune-related biomarkers was performed to spatially map tumor-immune interactions. Automated quantitative assessment of biomarker-positive cells was completed instantly through our platform, and validation was conducted against traditional IHC methods. Results: Our platform virtualized all five biomarkers with per-cell fidelity using only a single H&E digital slide as input, eliminating the need for additional physical tissue sections. Virtual multiplexing enabled detailed spatial analysis of T-cell subsets (CD4+ helper and CD8+ cytotoxic T cells) alongside immune checkpoints such as PD-L1 and regulatory T-cell marker FOXP3. Corresponding biomarker positivity was automatically and instantly quantified as part of this analysis. These results revealed insights into immune evasion mechanisms, including immune suppression and T-cell exclusion zones within the tumor microenvironment. The approach delivered these results in seconds, with strong concordance to traditional IHC methods, thus saving time, costs, and tissue. Conclusions: Our virtual IF and IHC platform transforms biomarker analysis for NSCLC by preserving rare tissue, reducing costs, and accelerating analysis. By enabling precise spatial mapping and quantification of tumor-immune dynamics from a single H&E slide, the platform empowers researchers with rapid, actionable insights to advance immunotherapy and precision oncology strategies. Citation Format: Kenneth To, Digvijay Yadav, Rohan Kamra Lyons, George Hennen, Rafay Azhar, Louis Vaickus, Christopher Jackson. Use of a novel deep learning-based virtual IF and IHC platform for instant multiplexing and quantification of CD4, CD8, PD-L1, FOXP3, and CK AE1/3 in non-small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 775.