129 Enhancing cellular specificity of RNA-ISH spot detection using dual-channel staining and AI-based analysis methods

<h3>Background</h3> RNA staining techniques are rapidly gaining prominence within the field of cancer research, holding promise for novel diagnostic and therapeutic applications. However, the lack of robust validation of probes targeting newly discovered RNA molecules remains a critical bottleneck for clinical translation. Accurate identification of RNA within specific cells, particularly in cytoplasmic regions, presents a significant challenge due to difficulties in determining cell boundaries and ensuring probe specificity. <h3>Methods</h3> Molecular Instruments’ Dual Channel™ validation method addresses these challenges head-on by assessing both the specificity and sensitivity of HCR™ RNA-ISH. This comprehensive evaluation facilitates the translation of high-value RNA-ISH assays from research to clinical settings, ultimately enabling reliable RNA quantification. Traditionally, cell boundary delineation has been achieved with membrane stains. However, this approach introduces a layer of complexity during subsequent image analysis. Furthermore, conventional RNA-ISH methods depend on harsh proteases to allow for bulky amplification reagents to reach RNA targets in situ, which often compromise tissue morphology, thereby hindering the co-detection of membrane stains. HCR™ RNA-ISH offers a distinct advantage through its use of small, background suppressed DNA reagents that enable a truly protease-free workflow, which safeguards tissue and epitope integrity. This allows for co-detection with virtually any protein marker on any given tissue section, maximizing the utility of valuable patient-derived samples. The integration of Visiopharm’s Phenoplex software, a powerful deep-learning solution powered by AI along with co-detection of cellular membrane markers, further enhances the identification of cells and their boundaries. User-drawn annotations are used to train an algorithm for the precise recognition of cell boundaries based on membrane stains. This significantly improves cell boundary detection and therefore RNA spot localization, paving the way for more accurate spot counts and a deeper understanding of RNA expression within the complex tumor microenvironment. <h3>Results</h3> Preliminary data indicates that utilizing HCR™ RNA-ISH coupled with membrane segmentation markers and immune checkpoint markers like PD-L1, in both healthy and cancerous lung tissue, leads to demonstrably improved accuracy and reliability in RNA detection. <h3>Conclusions</h3> The innovative Dual Channel™ validation of HCR™ RNA-ISH represents a significant advancement in RNA staining techniques, ensuring both specificity and sensitivity without compromising tissue morphology. By leveraging Visiopharm’s Phenoplex deep-learning algorithms for precise cell boundary recognition, this method promises to enhance RNA quantification accuracy. These improvements are crucial for the clinical translation of RNA-ISH assays, potentially transforming cancer diagnostics and therapeutic monitoring by providing reliable RNA measurements within the tumor microenvironment. <h3>Ethics Approval</h3> All FFPE tissue samples were obtained from biobanks under IRB approvals.