Multi-stage transcriptome analysis identifies hub genes and regulatory mechanisms driving cervical cancer progression

Cervical cancer (CC) ranks as the fourth most prevalent cancer in women worldwide. While biomarkers exist, previous studies are often limited by public dataset heterogeneity, incomplete spectrum coverage, or lack of clinical validation. To address these limitations, this study aims to profile fresh tissues spanning the full continuum to systematically identify hub genes with dysregulated expression during malignant progression.

This study collected cervical epithelial tissue from 10 patients each with low-grade squamous intraepithelial lesions (LSIL), high-grade squamous intraepithelial lesions (HSIL), and squamous cell carcinoma (SCC) for transcriptome sequencing. The differentially expressed genes (DEGs) analysis and Mfuzz clustering were employed to identify candidate genes showing monotonic expression trends. Candidate genes were prioritized using a protein-protein interaction (PPI) network with three centrality algorithms and validated in Gene Expression Omnibus to define hub genes. Subsequently, Gene Set Enrichment Analysis (GSEA), immune infiltration analysis, and regulatory network analyses were performed. Finally, hub gene expression was validated via reverse transcription quantitative polymerase chain reaction (RT-qPCR) in an independent clinical cohort.

A total of 1,654 DEGs (HSIL vs. LSIL) and 6,635 DEGs (SCC vs. HSIL) were identified. Mfuzz clustering and PPI analysis yielded nine candidates, with BUB1B, KIF14, and MELK finalized as hub genes via GEO validation. GSEA showed the three hub genes were significantly enriched in multiple shared pathways and exhibited stage-specific differences. Immune infiltration analysis showed no significant association between hub-gene expression and immune cell fractions. Regulatory network prediction implicated miRNAs (e.g., hsa-miR-192-5p, hsa-miR-215-5p, hsa-miR-193b-3p) and TFs (e.g., AF4, E2F1, FOXA1, KDM5B) as potential upstream regulators. The lncRNAs (e.g., KCNQ1OT1, LINC01089, XIST3) were linked to these miRNAs. Finally, RT-qPCR in independent clinical samples confirmed hub-gene expression changes consistent with the bioinformatics findings.

This study found the progressive upregulation of BUB1B, KIF14, and MELK across the LSIL-HSIL-SCC continuum, strengthening existing evidence of their association with cervical oncogenesis and highlighting their potential as candidate biomarkers for CC progression. However, the single-center design, limited sample size, and exploratory nature of the immune infiltration, regulatory network, and drug prediction analyses warrant cautious interpretation and underscore the need for further experimental validation.