Machine learning-based identification of glycolytic-related genes highlights BPGM as a potential therapeutic target in endometriosis

Endometriosis is a prevalent gynecological disorder characterized by the ectopic growth of endometrial-like tissue, often leading to chronic pelvic pain, infertility, and reduced quality of life. Despite available treatments, therapeutic outcomes remain unsatisfactory with frequent recurrence, underscoring the need for reliable biomarkers and novel therapeutic targets. This study aimed to identify glycolysis-related genes involved in endometriosis using integrated bioinformatics and experimental validation, and explore their potential diagnostic and biological significance. Transcriptomic datasets from three studies in the Gene Expression Omnibus were analyzed using differential expression analysis and machine learning algorithms, including LASSO, Random Forest, and SVM-RFE. Cellular heterogeneity and gene expression patterns were further delineated using single-cell RNA sequencing. Key computational findings were validated through molecular and cellular experiments. Integrated bioinformatics analysis identified ALDH9A1, BPGM, and ALDH3A2 as key glycolysis-related candidate genes significantly dysregulated in endometriosis. Single-cell resolution revealed BPGM was predominantly upregulated in epithelial cells within ectopic lesions. Experimental validation confirmed elevated BPGM expression in endometriotic tissues. In 12Z cells, BPGM knockdown reduced proliferation, migration, and invasion, while also significantly decreasing lactate production, suggesting its involvement in glycolytic reprogramming. Our findings identify BPGM as a glycolysis-associated factor involved in the progression of endometriosis and suggest that it may serve as a potential biomarker and therapeutic candidate for this disease. Not applicable. This is a bioinformatics study with validation using archived human tissue sections, no prospective intervention or participant assignment was performed.