The LSIL were defined as lesions confined to the lower one-third of the cervical epithelium. Cytological features included mild disturbance of nuclear polarity, minimal mitotic activity, evenly distributed chromatin, and the presence of koilocytes characteristic of HPV infection. The HSIL involved two-thirds or more of the cervical epithelium. Cells showed significant loss of nuclear polarity, an increased nuclear-to-cytoplasmic ratio, and extension of atypical cells to the upper two-thirds or full thickness of the epithelium.

Eligible participants were women aged 20–50 years with a history of sexual activity, whose cervical histopathology met the diagnostic criteria for LSIL or HSIL, and who had not received vaginal or cervical medications in the previous month. Patients in the lesion group were required to have a confirmed diagnosis from the hospital’s pathology department. Control group participants were required to have negative HPV and ThinPrep cytology test (TCT) results. All participants were fully informed of the study objectives and procedures and provided written informed consent.

Exclusion criteria included: (1) pathological diagnoses from external institutions not reviewed or confirmed by the hospital, or cases with diagnostic discrepancies; (2) before treatment with loop electrosurgical excision procedure (LEEP) or conization for cervical lesions; (3) detection of abnormal glandular cells or histologically confirmed malignant lesions; (4) coexisting severe systemic illness requiring urgent intervention or advanced cardiac, hepatic, or renal dysfunction; (5) irregular vaginal bleeding during pregnancy, lactation, menstruation, or prolonged abnormal bleeding; and (6) poor compliance, incomplete clinical data, or withholding of relevant medical information.

This study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Heilongjiang University of Traditional Chinese Medicine (Approval No. HZYLLKY202300701). All procedures were conducted following the relevant ethical guidelines and regulations. Written informed consent was obtained from all participants before enrollment.

A search of the GEO database using the terms “cervical intraepithelial neoplasia,” “cervical HSIL,” and “cervical high-grade intraepithelial lesion” initially identified 588 eligible datasets. Following detailed screening, inspection, and data correction, two datasets, GSE63514 and GSE75132, were selected for comparative analysis. Both datasets were generated using high-throughput sequencing on the GPL570 platform, which has a detectable nucleotide (nt) length of 600. The Affymetrix Human Genome U133 Plus 2.0 Array was employed, providing comprehensive coverage of the U133 set of the human genome along with 6,500 additional genes and over 47,000 transcripts, ensuring high sensitivity and accuracy of detection (16).

The GSE63514 dataset comprises 128 sequencing samples, including tissue samples from 24 normal controls, 14 cases of CIN I, 22 cases of CIN II, 40 cases of CIN III, and 28 cases of cervical cancer. The GSE75132 dataset contains 41 samples, including 11 normal controls, 10 cases of persistent HPV16 infection, 4 cases of CIN II, 9 cases of CIN III, and 7 cases of cervical cancer.

The basic characteristics of the two datasets were analyzed as follows;

In the GSE75132 dataset, baseline alignment was first performed, followed by comparative analysis. A volcano plot ( Figure 1A ) and scatter plot ( Figure 1B ) were generated to visualize the differential gene expression. In Figure 1A , blue indicates downregulated genes and red indicates upregulated genes. In Figure 1B , green denotes downregulated genes, while red denotes upregulated genes. A total of 2,266 differentially expressed genes (DEGs) were identified, including 951 significantly upregulated genes and 1,315 significantly downregulated genes ( Figure 1C ). In Figure 1C , green represents downregulated genes, while red represents upregulated genes.

In the GSE63514 dataset, analysis of normal tissues versus high-grade lesion tissues revealed 2,208 DEGs ( Figures 1D, E ). Among these, 1,203 genes were significantly downregulated and 1,005 genes were significantly upregulated ( Figure 1F ). In Figure 1D , blue represents downregulated genes and red represents upregulated genes; in Figure 1E , green denotes downregulated genes and red denotes upregulated genes; in Figure 1F , green indicates downregulated genes and red indicates upregulated genes.

The DEGs from the two datasets were uploaded separately to the Venny tool (https://bioinfogp.cnb.csic.es/tools/venny/index.html) to generate a Venn diagram ( Figure 2 ). The GSE75132 dataset contained 2,266 DEGs, while the GSE63514 dataset contained 2,208 DEGs. A total of 371 DEGs were found to be shared between the two datasets. Further analysis of these 371 common DEGs revealed 221 downregulated genes and 150 upregulated genes.

The results from the STRING database analysis were downloaded and imported into Cytoscape software to construct a PPI network for the 371 differentially expressed genes. The resulting network comprised 288 nodes and 1,274 edges ( Figure 3A ). Using the CytoHubba plugin, the top 10 hub genes within the PPI network were identified ( Figure 3B ). These included vascular endothelial growth factor A (VEGFA), matrix metalloproteinase-9 (MMP9), hyaluronan receptor CD44 (CD44), cyclin D1 (CCND1), cyclin B1 (CCNB1), C-X-C motif chemokine ligand 8 (CXCL8), estrogen receptor 1 (ESR1), Toll-like receptor 2 (TLR2), C-X-C chemokine receptor 4 (CXCR4), and signal transducer and activator of transcription 1 (STAT1). The average degree value of the PPI network is 7.26 ( Figure 3C ), and the correlation between genes is good ( Figure 3D ).

Analysis of the differentially expressed genes identified 102 enriched biological processes (BP), 33 cellular components (CC), and 15 molecular functions (MF). The top 10 terms from each category were selected for GO visualization ( Figure 4 ). The results revealed several processes closely associated with CSIL.

Key biological processes included inflammatory response, cellular response to interferon-γ, immune response, signal transduction, positive regulation of protein phosphorylation, positive regulation of interleukin-6 production, apoptotic processes, and positive regulation of IκB kinase/NF-κB signaling. The enriched cellular components were primarily associated with cytoplasmic constituents and the extracellular environment. The dominant molecular functions were related to protein binding, enzyme binding, and interactions with associated cofactors and protein complexes.

Comparative analysis identified 29 signaling pathways associated with cervical HSIL lesions. The key pathways are illustrated in Figure 5 . Importantly enriched pathways included Kaposi sarcoma-associated herpesvirus infection, proteoglycans in cancer, general cancer pathways, interactions between viral proteins and cytokines or cytokine receptors, natural killer cell-mediated cytotoxicity, human immunodeficiency virus type 1 infection, human cytomegalovirus infection, the PI3K–Akt signaling pathway, Toll-like receptor signaling pathway, cytokine–cytokine receptor interaction, Ras signaling pathway, nucleotide metabolism, leukocyte transendothelial migration, DNA replication, and lipid and atherosclerosis-related processes.

GSEA enrichment analysis predicted three key genes closely associated with the disease: serine/threonine kinase 33 (STK33), ribosomal protein S14 (RPS14), and anaplastic lymphoma kinase (ALK). The expression levels of all three genes demonstrated an upward trend in the disease samples ( Figure 6A ).

GSEA identified the three most relevant biological processes for this disease: the TNF-α signaling pathway mediated by NF-κB, the interferon-γ response, and the immune response ( Figure 6B ).

Studies have demonstrated that STK33 is expressed across multiple tissues, with the strongest hybridization signals observed in the testes, fetal lungs, and heart, followed by the pituitary gland, kidneys, pancreas, thyroid gland, and uterus. Moderate signals are seen in the aorta, hematopoietic system, and digestive tract, whereas no detectable hybridization is observed in nervous system tissues (17). In tumor-related diseases, STK33 has been shown to suppress mitochondrial apoptosis by inhibiting BAD activity, enhancing tumor cell survival and proliferation. These findings suggest that STK33 may represent a potential therapeutic target for malignancies driven by oncogenic mutations (18).

Literature evidence indicates that RPS14 can disrupt the MDM2–p53 interaction, resulting in significant accumulation of p53. Elevated p53 levels can induce cell cycle arrest, promote DNA damage repair, and trigger cellular senescence and apoptosis (19). RPS14, a component of the 40S ribosomal subunit, is involved in erythroid differentiation and has been implicated in hematological disorders such as myelodysplastic syndrome and 5q syndrome (20). In colon cancer, overexpression of RPS14 has been reported to activate the PI3K/Akt signaling pathway, enhancing tumor cell survival (21).

Alterations in the ALK gene, including recombination, fusion, mutation, amplification, and alternative splicing, have been identified in various malignancies. ALK also mediates NF-κB signaling and promotes activation of the NLRP3 inflammasome in macrophages. These properties suggest that ALK-targeted interventions may represent a novel therapeutic strategy for NLRP3 inflammasome–mediated diseases (22, 23).

A total of 19 normal cervical tissues, 14 LSIL tissues, and 15 HSIL tissues were collected from outpatient and inpatient cases. Normal cervical tissues were obtained from patients undergoing total hysterectomy for non-neoplastic conditions at the First Affiliated Hospital of the Heilongjiang University of Chinese Medicine. All normal tissue donors tested negative for both HPV and TCT. LSIL and HSIL cervical tissues were confirmed by the hospital’s Department of Pathology and met the inclusion and exclusion criteria described earlier.

Comparison of age distribution among the groups revealed statistically significant differences (P< 0.01). In the control group, most patients were aged 40–50 years and primarily underwent total hysterectomy for benign conditions such as adenomyosis or multiple uterine fibroids. LSIL cases were more common in younger patients, predominantly within the 20–30 year and 30–40 year age groups. However, HSIL cases were mainly concentrated in the 30–40 year and 40–50 year age ranges ( Table 3 ).

This study found that the age of onset for LSIL was primarily concentrated in the 30–40-year age group, reflecting a younger onset pattern. However, HSIL onset was more common in the 40 to 50-year age group. Statistical analysis indicated a positive correlation between age and disease severity, which may be associated with the prolonged duration of viral infection.

The relationship between age and disease severity can be summarized as follows:

1. Variation in lesion severity across age groups.

Chang HK et al. reported that the peak onset age for LSIL is 25–29 years, for HSIL is 30–34 years, and for cervical cancer is 70–74 years. Lesion severity tends to increase with age (24). Younger patients generally demonstrate higher rates of lesion regression and complete remission, with lower rates of progression (25).

2. Differences in the natural regression rate of SIL by age.

HSIL represents a precancerous condition that can progress to cervical cancer. Meta-analysis has shown that the regression rate differs by lesion grade. For CIN 1, spontaneous regression occurs in approximately 40% of cases. Advanced lesions (CIN 2 or CIN 3) have a higher risk of progression, with CIN 2 showing a progression rate of about 10.28%. Age is negatively correlated with regression rate; older patients show lower spontaneous regression rates (26).

3. Association of age with abnormal HPV and TCT results.

The average latency between carcinogenic HPV infection and the development of cervical cancer is approximately 25–30 years. With advancing age, the prevalence of high-risk HPV (HR-HPV) positivity and abnormal TCT findings increases (27). HPV infection is most prevalent among younger and middle-aged women, particularly those aged 25–35 years, due to higher sexual activity. However, in older women, factors such as lower educational levels, reduced immune function, and hormonal changes contribute to a higher detection rate of SIL (28).

Analysis of baseline characteristics across the patient groups revealed no significant differences (P > 0.05) in parity, smoking history, alcohol consumption, vaginal microbiota diversity, or microbiota density between the LSIL, HSIL, and control groups. However, significant differences were observed in the age at first sexual intercourse and the number of sexual partners (P< 0.05 and P< 0.01, respectively).

Statistical evaluation ( Table 4 ) indicated that age, age at first sexual intercourse, and number of sexual partners were associated with the occurrence of SIL. In comparison, parity, smoking history, alcohol consumption, vaginal microbiota diversity, and microbiota density showed no significant association in this study. Comparative analysis further demonstrated that the lesion groups (LSIL and HSIL) were characterized by earlier sexual debut and a higher number of sexual partners compared with the control group. This study revealed that host factors (including age, vaginal microbiota composition, and immune status) and environmental factors (such as early sexual behavior, multiple sexual partners, and smoking tendency) jointly affect the activation level of the PI3K/Akt/NF - κ B signaling pathway. These findings emphasize the importance of host environment interaction in the development of cervical lesions. In particular, the IFN - γ response shown by GSEA analysis suggests a close relationship between this pathway and host immune response, providing a theoretical basis for the development of combination therapy strategies that combine immune regulation (such as enhancing antiviral response) and targeted pathway inhibition (such as using ALK or PI3K inhibitors), which may improve the therapeutic effect of cervical precancerous lesions.

Analysis of HPV genotyping and TCT examination results in the LSIL and HSIL groups revealed that, compared with the LSIL group, the HSIL group had a significantly higher detection rate of high-risk HPV types 16 and 18 (P< 0.05). The LSIL group, however, showed relatively higher detection rates of other high-risk HPV types.

In terms of cytology, most patients in the LSIL group had NILM results, whereas the majority of HSIL patients showed significant cytological abnormalities (P< 0.01) ( Table 5 ). Overall, HSIL cases demonstrated a higher prevalence of HPV16/18 infection and abnormal TCT findings compared with LSIL cases.

Experimental analysis revealed that, compared with the control group, the target genes ALK and RPS14 were significantly overexpressed in the HSIL group (P< 0.01). When compared with the LSIL group, HSIL tissues revealed substantially higher expression of ALK (P< 0.01) and moderately higher expression of RPS14 (P< 0.05).

For key signaling pathway components, expression of PI3K, AKT, NF-κB, and IκB-α was significantly elevated in HSIL tissues relative to the control group (P< 0.01). Compared with the LSIL tissues, the HSIL group showed significant upregulation of all four pathway markers (P< 0.01). These findings are illustrated in Figure 7 (mRNA expression levels of target genes), where expression of ALK, RPS14, PI3K, AKT, NF-κB, and IκB-α is significantly increased in HSIL, suggesting overactivation of the ALK, RPS14, and PI3K/AKT/NF-κB signaling axis. Numerical data are summarized in Table 6 .

qRT-PCR analysis directly compared the expression of PI3K and AKT in cervical lesions and normal cervical tissues. NF-κB expression showed a positive correlation with lesion severity, increasing from LSIL to HSIL. IκB-α expression was low in normal cervical tissues, slightly reduced in LSIL, and significantly elevated in HSIL. These results indicate that the inflammatory signaling pathway becomes activated during LSIL and is strongly upregulated in HSIL.

During the experiment, the STK33 gene showed no detectable Ct values, even after modifying primers and increasing sample concentration. As a result, its expression data were not presented. This absence of detection is likely attributable to the low abundance of STK33 in cervical tissue. While the RPS14 gene yielded measurable values, the Ct values were excessively high, likely due to low initial template concentration. This suggests that RPS14 expression in cervical tissue is relatively low.

In comparison, the ALK gene demonstrated a more stable expression profile in cervical tissue, with consistent and readily detectable Ct values, indicating higher expression levels compared to STK33 and RPS14.

Correlation analysis between age, vaginal microbiota diversity, microbiota density, key genes, and signaling pathway markers in the LSIL and HSIL groups ( Table 7 ) revealed that cervical lesion severity was significantly associated with patient age, vaginal microenvironment, ALK and RPS14 expression, and activation of the PI3K/AKT/NF-κB signaling pathway (P< 0.05).

Analysis of key genes (ALK, RPS14) and PI3K/AKT/NF-κB pathway targets between the control and HSIL groups demonstrated that ALK, RPS14, PI3K, AKT, NF-κB, and IκB-α possess high diagnostic value for this disease ( Figure 8 ; AUC > 0.9, indicating excellent accuracy). Comparative evaluation further showed that the diagnostic performance of ALK was superior to that of RPS14 ( Table 8 ).