A Nomogram Model Integrating Transcriptomics and Clinical Features for Predicting the Risk of Cervical Intraepithelial Lesion Progression

Objectives: LSIL patients may progress to HSIL due to delayed screening, increasing the risk of invasive cancer. Identifying progression mechanisms and risk factors is crucial for early intervention.
Methods: Transcriptomic data from the GEO database were analyzed for differentially expressed genes (DEGs). GO and KEGG enrichment analyses identified biological functions and pathways, while a molecular interaction network (STRING) screened core genes. Immunohistochemistry on patient samples validated key proteins associated with progression. A prediction model using Cox regression and a random forest algorithm was developed based on clinical data, with performance assessed by ROC and decision curve analysis (DCA).
Results: We identified 221 upregulated and 161 downregulated DEGs. Key progression-related genes included CDKN2A, CALML5, GINS2, KCNH1, MCM2, MKI67, and ESR1. Immunohistochemical validation in 155 patients confirmed that Eag1, p16INK4a, and Ki-67 correlated with poor outcomes. Colposcopic lesion involvement across cervical quadrants was protective, while bacterial vaginosis was a risk factor. The prediction model demonstrated high accuracy (AUC: 0.914/0.753 for 1-year and 0.941/0.769 for 2-year progression). DCA confirmed clinical benefits.
Conclusion: This study identifies key biomarkers and clinical factors influencing cervical lesion progression and provides a predictive model for early intervention and personalized management.