Identification and Validation of a LASSO-Based Diagnostic Signature for PCOS Endometrial Dysfunction Using Integrated Bioinformatics

Background Polycystic ovary syndrome (PCOS) is the most common endocrine disorder among reproductive-aged women, affecting 8–13 % of the population worldwide. It is defined by the 2003 Rotterdam criteria and is frequently accompanied by endometrial dysfunction, yet non-invasive molecular biomarkers for diagnosis remain scarce. This study aimed to identify a robust gene signature for PCOS endometrial dysfunction through comprehensive bioinformatic analyses. Methods Three public endometrial microarray datasets (GSE103465, GSE4888, GSE51901) were downloaded from the GEO database. Differential expression analysis was performed using limma (|log₂FC| > 1, FDR < 0.05). Functional enrichment analyses (GO and KEGG) were carried out using clusterProfiler. A Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression model was constructed to screen the optimal gene signature, and its diagnostic performance was evaluated by receiver operating characteristic (ROC) curves in both training and validation sets. Results A total of 200 differentially expressed genes (DEGs) were identified, mainly enriched in extracellular matrix remodeling, inflammatory response and angiogenesis pathways. A 50-gene LASSO signature was established, achieving an AUC of 0.816 in the training cohort and 0.766 in the independent validation cohort. Conclusions The LASSO-derived gene signature exhibits strong discriminatory power for PCOS endometrial dysfunction and may serve as a novel diagnostic resource for clinical translation.