关键词: T1期结直肠癌, 淋巴结转移, 预测模型, 危险因素

Abstract: With advancements in endoscopic techniques, an increasing number of T1 colorectal cancers (CRCs) can be treated with endoscopic resection. However, the presence of lymph node metastasis (LNM) remains a critical determinant for the necessity of additional radical surgery. Current guidelines recommend risk stratification based on histopathological features, yet these criteria lack standardization and exhibit limited predictive accuracy. Beyond established pathological markers, emerging evidence highlights the prognostic value of horizontal invasion width, perineural invasion, and immune microenvironment characteristics. In predictive modeling, conventional statistical approaches using clinicopathological features show moderate utility but require refinement. Molecular biomarkers, such as miRNA signatures, DNA methylation profiles, and proteomic patterns, demonstrate superior potential, though clinical adoption is hindered by cost constraints. Artificial intelligence (AI)-driven models, which minimize subjective bias and enable automated analysis, significantly enhance predictive performance by integrating histopathological imaging with multimodal data. In conclusion, optimal LNM risk assessment in T1 CRC necessitates a combination of traditional pathology and novel biomarkers. AI and multi-omics approaches represent promising avenues for precision stratification. Future efforts should focus on optimizing model generalizability and clinical applicability to guide personalized therapeutic decision-making.

Key words: T1 colorectal cancer, lymph node metastasis, prediction model, risk factors ,