Abstract A004: A minimal gene panel approach for cost-effective identification of tumor-reactive T cells

Abstract Personalized T cell therapies have shown remarkable efficacy in refractory cancers by utilizing a patient's own tumor-reactive T cells to selectively eliminate malignant cells. This precision enables treatment of disseminated disease and metastases often inaccessible to conventional therapies, while minimizing off-target toxicity, with some patients achieving complete and durable responses. The key to developing these therapies lies in accurately identifying and isolating the rare tumor-reactive T cells from patient samples. Machine learning tools such as predicTCR can accurately identify tumor-reactive T cells on the basis of their expression of 100 diagnostic transcripts, identified through the use of SHAP eXplainable AI. We hypothesized that targeted sequencing of only this minimal gene set would maintain classification accuracy while reducing sequencing costs, data processing requirements, and turnaround time. We tested this hypothesis by implementing Nanopore's targeted long-read sequencing platform combined with a custom gene-enrichment strategy focused on these key predictive markers. This approach enabled selective capture and sequencing of the informative gene subset while excluding non-essential transcripts, reducing the sequencing burden by over 98%. We then developed and trained a machine-learning classifier optimized specifically for this streamlined dataset. The targeted sequencing model maintained high performance, achieving >95% accuracy relative to our full-transcriptome predicTCR model, while offering substantial improvements in both cost-effectiveness and computational efficiency. These results demonstrate that targeted sequencing of a minimal, rationally-selected gene panel can effectively identify tumor-reactive T cells without compromising classification accuracy. Taken together, this approach addresses critical barriers to clinical implementation of personalized T cell therapies by substantially reducing the financial, and technical burdens associated with single-cell transcriptomic sequencing. By making tumor-reactive T cell identification more accessible and cost-effective, this methodology has the potential to accelerate the development and broaden the availability of next-generation cancer immunotherapies for patients with limited treatment options. Citation Format: Marie-Therese Neuhoff, Chin Leng Tan, Michael Platten, Edward Wilhelm Green. A minimal gene panel approach for cost-effective identification of tumor-reactive T cells [abstract]. In: Proceedings of the AACR Immuno-Oncology Conference (AACR IO): Discovery and Innovation in Cancer Immunology: Revolutionizing Treatment through Immunotherapy; 2026 Feb 18-21; Los Angeles, CA. Philadelphia (PA): AACR; Cancer Immunol Res 2026;14(2 Suppl):Abstract nr A004.