Construction of a prognostic model for endometrial cancer related to programmed cell death using WGCNA and machine learning algorithms
Background
Programmed cell death (PCD) refers to a regulated and active process of cellular demise, initiated by specific biological signals. PCD plays a crucial role in the development, progression, and drug resistance of uterine corpus endometrial carcinoma (UCEC), making the exploration of its relationship with UCEC prognosis highly clinically relevant.
Methods
Data from UCEC patients and control cohorts were obtained from The Cancer Genome Atlas (TCGA) database. Differentially expressed genes (DEGs) were identified and subsequently intersected with a PCD gene set to discern PCD-related differentially expressed genes (PCD-DEGs). To isolate core prognostic PCD-DEGs, methods including consistency clustering analysis, weighted gene co-expression network analysis (WGCNA), univariate Cox regression analysis, and five machine learning techniques for dimensionality reduction were utilized. Validation of three core prognostic PCD-DEGs was conducted using RT-qPCR, and these genes were used to develop a prognostic model. Additionally, an analysis of drug sensitivity was performed.
Results
Consistency clustering analysis revealed significant differences in prognosis and tumor microenvironment among subtypes, strongly associated with various immune subtypes. The three core prognostic PCD-DEGs identified—SRPX, NT5E, and ATP6V1C2—were instrumental in constructing the lasso prognostic model and nomogram. Receiver Operating Characteristic (ROC) curve analysis confirmed the model’s strong prognostic performance and clinical applicability. The high-risk group exhibited lower tumor mutation frequencies, a higher propensity for immune escape, reduced response to immune therapy, and potential benefits from potent chemotherapy drugs.
Conclusion
This study developed a prognostic model related to PCD for UCEC using comprehensive bioinformatics analyses. The model demonstrates robust predictive performance and holds significant potential for clinical application, thereby facilitating precise stratification and personalized treatment of UCEC patients.