Machine Learning-Based Identification of a Glycolysis- Related Prognostic Signature and CENPA in Endometrial Cancer
Year: 2026
Type: preprint
Abstract: Background Endometrial cancer (EC) is strongly associated with metabolic disorders, while glycolytic reprogramming contributes to tumor progression and adverse outcomes. This study aimed to identify glycolysis-related prognostic genes and develop a machine learning-based signature for EC risk stratification. Methods Transcriptomic and clinical data were obtained from The Cancer Genome Atlas. Glycolysis-related genes were collected from KEGG, MSigDB, and Reactome. Differential expression and univariate Cox analyses were used to identify prognostic candidates. Multiple machine learning algorithms and combinations were evaluated to construct a prognostic model. Functional enrichment, immune infiltration, somatic mutation, tumor mutational burden, and drug sensitivity analyses were performed. Single-cell RNA sequencing datasets GSE179640 and GSE173682 were analyzed to characterize CENPA expression across cell types and its clinical relevance. Results Among 5,990 differentially expressed genes, 99 overlapped with 290 glycolysis-related genes, and 16 were associated with overall survival. Among 117 model combinations, random survival forest plus generalized boosted model generated a seven-gene signature comprising CLDN9, CENPA, GPI, TFF3, FBP1, SPAG4, and AURKA. The 1-, 3-, and 5-year AUCs were 0.88, 0.81, and 0.84 in the training cohort and 0.73, 0.77, and 0.76 in the internal validation cohort, respectively. High-risk patients had poorer survival and distinct pathway, immune, and genomic features. Single-cell analysis showed heterogeneous CENPA expression, while elevated CENPA was associated with advanced stage and poorer survival. Conclusions The glycolysis-related signature stratified EC patients by survival risk and was associated with distinct biological features. CENPA emerged as a candidate prognostic biomarker requiring independent external validation and experimental investigation.
Source: Research Square
Authors: Hang Yu, Ze Liu, Yanjun Liu, Xiushuang Zheng, Liqun Wang
Language: English
Subfield: Obstetrics and Gynecology
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7226010717
