Abstract:

The diagnosis and treatment of endometrial carcinoma are gradually shifting from traditional histomorphological assessment to molecular classification and comprehensive precision management. However, the early non-invasive identification, prediction of immunotherapy efficacy, and recurrence monitoring still lack stable and reliable biomarkers. In the early non-invasive screening phase, serum proteins, circulating tumor cells, metabolomics, and circulating nucleic acids (circulating tumor DNA and exosomal circular RNA) significantly enhance the detection capability for extremely low-abundance tumor signals. Post-diagnosis immune assessment relies on checkpoints such as programmed death-ligand 1 (PD-L1), the tumor microenvironment (tertiary lymphoid structures), molecular classification, and novel immune regulatory genes to define tumor immunogenicity and predict immunotherapy response rates. Confronting the high heterogeneity of tumors, the integration of artificial intelligence and multi-omics technologies has overcome traditional diagnostic bottlenecks. Multimodal artificial intelligence, by integrating clinical and pathological features, substantially optimizes risk prediction and personalized decision-making. In the future, the combination and dynamic evaluation of multiple biomarkers are inevitable trends for improving precision diagnostic and therapeutic efficacy. There is an urgent need for multicenter prospective studies to standardize detection protocols and accelerate clinical translation to improve prognosis.

Key words: Endometrial neoplasms, Carcinoma, Early detection of cancer, Biomarkers, Artificial intelligence