Treffer: Abstract P1-04-25: Fast and accurate detection of tumor-immune cell interactions with deep learning

Background: Our previous studies demonstrated liquid biopsy-derived circulating tumor cells (CTCs) not only serve as predictive biomarkers of therapy response and overall survival of breast cancer patients, but also drive new metastasis, especially by multicellular CTC clusters with enriched stemness and up to 50 times higher metastatic potential than single CTCs (Cancer Discovery, 2019 and 2023). However, CTC detection remains challenging, and its characterization is limited due to the rare frequencies in blood cells and labor-intensive manual validations of CTC images analyzed by CellSearch and a few other methods. Methods: Utilizing FDA-approved CTC CellSearch platform and advanced single-cell sequencing technologies for CTC and immune cell profiling, we have analyzed over 2,000 blood biospecimens longitudinally collected from patients with advanced breast cancer (N=445), including CTC images and CTC interactions with immune cells. Harnessing the power of machine learning and deep learning algorithms for CTC and immune cell image analyses, this study investigates the significance of CTCs, immune cells, and CTC clusters in predicting patient prognosis. Results: By analyzing CellSearch raw images obtained from blood of breast cancer patients, we developed a python-based CTCpose analysis platform for cell feature analyses in connection with clinical relevance of various cell populations and intercellular communications. Our analysis automatically identified CTCs, immune cells, and CTC clusters differentiated by homogeneity or heterogeneity in cellular composition. By utilizing artificial intelligence (AI) algorithms, we extracted cellular features and classifications to identify different cell types and cell clusters at an accuray and sensitivity over 97%. Specifically, we found expression patterns of biomarkers such as cytokeratin, CD45, and DAPI to discern the spatial distribution and intensity within samples. Conclusions: The study has explored the potential prognostic values of CTCs, immune cells, and CTC immune cell interactions by correlating their presence, abundance, and spatial information, with clinical outcomes. These outcomes may include patient survival, disease progression, and treatment response. By combining multidimensional data derived from cell morphology, biomarker expression, and spatial relationships, we aim to develop predictive models capable of stratifying patients into unique risk groups. Citation Format: Huiping Liu, Joshua Squires, Youbin Zhang, Yuanfei Sun, Kaiyu Liu, Andrew Hoffmann, David Scholten, Leonidas C Platanias, Yuan Luo, Carson Stringer, Massimo Cristofanilli, William J. Gradishar. Fast and accurate detection of tumor-immune cell interactions with deep learning [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr P1-04-25.