1213 Development and multiplexing of virtual immuno-oncology biomarkers for enhanced research and discovery
Year: 2024
Type: conference-abstract
Abstract: Background The complexity of the tumor microenvironment in immuno-oncology (IO) research necessitates advanced tools for precise biomarker analysis. Our innovative virtual immunohistochemistry (IHC) pipeline has been adapted to develop a comprehensive panel of IO biomarkers, including CD206, CD4, CD8, FOXP3, PD-L1, and a broad spectrum cytokeratin (CK AE1/3). This platform provides detailed per-cell phenotyping based on annotations directly from H&E slides, allowing users to bypass physical IHC or immunofluorescence (IF). This approach significantly reduces time and cost compared to traditional methods. Methods We utilized a previously validated method for annotating hematoxylin and eosin slides, integrating both IHC and IF for initial biomarker labeling to create a robust reference dataset. H&E slides were first scanned using a whole slide imager, then de-stained, and subjected to IHC or IF before re-scanning. The resulting IHC/IF images were used to annotate the hematoxylin and eosin images at the per-cell level. These annotations were then used to train an artificial intelligence model, which performs per-cell predictions directly on H&E slides. Our platform processes multiple virtual biomarkers simultaneously, maintaining the spatial context of tissue architecture and ensuring accurate representation of the tumor microenvironment. This enables detailed analysis of cellular interactions and spatial relationships, providing crucial insights for IO research. Results Initial qualitative assessments indicate that our platform effectively visualizes and differentiates various immune cell types and their states within the tumor microenvironment. The per-cell prediction capability allows for nuanced analysis of the immune landscape, identifying potential biomarkers for therapeutic targeting, and enhancing the understanding of immune cell infiltration and distribution in tumors. By eliminating the need for physical IHC or IF, our platform accelerates the research process, offering faster and more cost-effective solutions. Conclusions Our virtual IHC platform, capable of per-cell predictions directly from H&E slides, represents a significant advancement in IO research. This technology provides a powerful tool for exploring tumor-immune dynamics, aiding in the discovery of novel biomarkers and the development of new therapeutic strategies. The efficiency and precision of our platform promise to accelerate research and improve clinical outcomes, saving time and resources. By leveraging the detailed and comprehensive analysis capabilities of our virtual IHC pipeline, researchers can gain deeper insights into the tumor microenvironment. This advancement underscores the critical role of innovative technologies in driving forward cancer research and treatment, making the process faster, more efficient, and cost-effective. Acknowledgements We would like to thank the Pathology Translational Research Shared Resource and Scott Palisoul at Dartmouth-Hitchcock Medical Center for their assistance with the wet lab and whole slide imaging. Ethics Approval The study received IRB approval from the A*STAR IRB: 2021-188 ‘Multiple Tumor Multiplex IHC Staining’. Consent No sensitive or identifiable information is included in the study.
Authors: Christopher Jackson, Digvijay Yadav, Felicia Wee, Jeffrey Lim, Willa WY Yim, Kenneth To, Joe Yeong
Language: English
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4404065308