Construction of a Predictive Modelbased on Tertiary Lymphoid Structures on the Survival of Patients With Gastrointestinal Malignant Tumors.

To investigate the prognostic significance of tumor-associated tertiary lymphoid structures (TLS) in patients with gastrointestinal malignancies and to develop a survival prediction model based on TLS. One hundred and seventy six patients undergoing surgical resection for gastric or colorectal cancer were included. Hematoxylin-eosin staining and immunohistochemistry were used to assess the density and spatial distribution of TLS within tumor tissues. Patients were categorized into TLS-rich group (RG), TLS-poor group (PG), and TLS-negative group (NG). Clinical variables and TLS characteristics were integrated, and Cox proportional hazards regression, along with random forest algorithms, were employed to construct a prognostic prediction model. The median overall survival in the RG group (30 months) was significantly longer than in the PG group (24 months) and the NG group (18 months) (p < 0.05). Patients exhibiting TLS both intratumorally and peritumorally, secondary TLS, TLS within metastatic lesions, or TLS containing germinal center (GC) B cells derived a greater survival benefit. The random forest model demonstrated excellent predictive performance, with a C-index of 0.81 in the training set and 0.80 in the external validation set. The presence and density of TLS serve as independent prognostic biomarkers in patients with gastrointestinal malignancies. The TLS-based prediction model provides a robust tool to support individualized clinical decision-making. Clinical Trial Number: Not applicable.