Characterization of G2/M checkpoint classifier for personalized treatment in uterine corpus endometrial carcinoma.

Abstract

Background: Uterine Corpus Endometrial Carcinoma (UCEC) is a highly heterogeneous tumor, and limitations in current diagnostic methods, along with treatment resistance in some patients, pose significant challenges for managing UCEC. The excessive activation of G2/M checkpoint genes is a crucial factor affecting malignancy prognosis and promoting treatment resistance.

Methods: Gene expression profiles and clinical feature data mainly came from the TCGA-UCEC cohort. Unsupervised clustering was performed to construct G2/M checkpoint (G2MC) subtypes. The differences in biological and clinical features of different subtypes were compared through survival analysis, clinical characteristics, immune infiltration, tumor mutation burden, and drug sensitivity analysis. Ultimately, an artificial neural network (ANN) and machine learning were employed to develop the G2MC subtypes classifier.

Results: We constructed a classifier based on the overall activity of the G2/M checkpoint signaling pathway to identify patients with different risks and treatment responses, and attempted to explore potential therapeutic targets. The results showed that two G2MC subtypes have completely different G2/M checkpoint-related gene expression profiles. Compared with the subtype C2, the subtype C1 exhibited higher G2MC scores and was associated with faster disease progression, higher clinical staging, poorer pathological types, and lower therapy responsiveness of cisplatin, radiotherapy and immunotherapy. Experiments targeting the feature gene KIF23 revealed its crucial role in reducing HEC-1A sensitivity to cisplatin and radiotherapy.

Conclusion: In summary, our study developed a classifier for identifying G2MC subtypes, and this finding holds promise for advancing precision treatment strategies for UCEC.