Fan Wu Haotian Cao Siqi Ren Jiaying Wu Xin Liu ^* Qunxing Li ^* Qiuping Xu ^* Jiali Chen Ruixin Wang ^* CHEN Suling Shijia Kuang ^* Binbin Xia ^* Yanyan Li ^* Liansheng Wang ^* Jintao Li ^* Bowen Li ^* Jinsong Li ^* Tianjun Lan ^*

• Sun Yat-sen Memorial Hospital, Guangzhou, China

Background: Substantial studies reveal that tertiary lymphoid structure (TLS) correlate with prognosis and immunotherapy response in various types of cancers. However, the predictive value of TLS, the specific immune cell subtype within TLS and their anti-tumor mechanisms remain unclear.Methods: Based on 23 TLS-related genes (TLSRGs), we utilized bioinformatics methods to construct a scoring system, named TLSscore. By integrating RNA and single-cell sequencing data, we assessed the utility of TLSscore in head and neck squamous cell carcinoma (HNSCC). Flow cytometric sorting was used to isolate specific T cells subtypes, in vivo and in vitro experiments were conducted to demonstrate its anti-tumor effects.Results: The TLSscore model was constructed and specific TLSscore-genes were found to consistently align with the spatial location of TLS. TLSscore has proven to be a robust predictive model for predicting survival prognosis, immune cell infiltration, somatic mutation and immunotherapy response. Notably, a specific PD1 + CXCL13 + CD8 + T cell subtype was identified within TLS. Both in vivo and in vitro experiments demonstrated that PD1 + CXCL13 + CD8 + T cell might represent a functional cell subtype exerting anti-tumor effects during the process of immune surveillance.Our study presents a predictive model for TLS, which can evaluate its presence and predicts survival prognosis and immunotherapy response in HNSSC patients. Additionally, we identify a specific subtype of T cells that might elucidate the mechanism of TLS function in antitumor activities. This T cell subtype holds the potential to be a prognostic marker and a target for adoptive cell therapy (ACT) in the future.