Abstract 6575: A novel AI-driven spatial genomics platform enables tumor enrichment and analysis of heterogeneity

Spatial genomics and digital pathology continue to advance at a rapid pace, generating large amounts of data. There is a critical unmet need for advanced tools to expedite spatial data generation, visualization, and analysis, to facilitate an improved understanding of health and disease, especially with the advancement of personalized cancer treatments. In this study we present a novel method for the extraction and analysis of genomic information from specific regions of interest (ROI) within a series of formalin-fixed paraffin embedded (FFPE) breast cancer adenocarcinoma tissue sections. An artificial intelligence-based tumor profiling algorithm first identified tumor, stroma, inflammatory, and necrotic regions within the tissue, and segmentation results were confirmed by a Board Certified Pathologist. Automated lysate extraction was then performed in the selected ROIs using ink-jet technology and novel ink chemistry. The process was performed directly on the tissue slide and the crude lysates generated were transferred to a standard collection tube for subsequent sequencing. DNA was extracted from the selected areas before being purified and analyzed using a multi-cancer panel. Library QC showed optimal coverage at an average of >95% uniformity. Sequencing results showed an average coverage of 95% at 250x across all amplicons of the panel. Variant calling was successfully performed. Sequencing data is presented to demonstrate the high coverage achieved using this workflow as well as the resulting tumor heterogeneity. This novel technology provides an efficient solution for the enrichment of tumor content with high sequencing coverage from FFPE tissue sections. Novel spatial data visualization and analysis techniques allowed a comprehensive assessment of tumor heterogeneity. Further studies on both DNA and RNA should be conducted to validate this approach for clinical molecular testing. Citation Format: Bidhan Chaudhuri, Katie Konigsfeld, Marisa Sanchez, Jason Smith, Casey Laris, Claire Weston, John Butler. A novel AI-driven spatial genomics platform enables tumor enrichment and analysis of heterogeneity [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6575.