A Holistic Framework for Cervical Cancer Screening and Risk Stratification: Integrating Etiology, Morphology, Epigenetics, and Microecology

Abstract

Cervical cancer (CC) remains a major global health burden despite being highly preventable through early detection. Current screening strategies, predominantly high-risk human papillomavirus (HR-HPV) testing and liquid-based cytology (LBC), have limited specificity, leading to unnecessary colposcopic referrals and overtreatment. This narrative review proposes a conceptual holistic screening framework—termed HADm—that integrates four complementary pillars into a unified risk stratification system: (1) Etiology (HR-HPV genotyping), (2) Morphology (artificial intelligence-assisted LBC, Ai-LBC), (3) Epigenetics (DNA methylation [DNAm] biomarkers, with PAX1 methylation as an example), and (4) Microecology (genital pathogens and vaginal microbiota). Persistent HR-HPV infection is the primary etiological factor, while disease progression is modulated by host epigenetic alterations and the local microenvironment. DNAm markers may serve as molecular indicators of malignant transformation and have shown high specificity for cervical intraepithelial neoplasia grade 2 or worse (CIN2+) and grade 3 or worse (CIN3+) among human papillomavirus (HPV)-positive women in some studies. Ai-LBC may improve both sensitivity and standardization of morphological interpretation, whereas genital pathogen and microbiota may help explain persistent HPV infection and chronic inflammation. However, evidence for the combined HADm framework remains limited, and prospective multicenter validation is lacking. This framework represents a shift from single-analyte testing toward integrated risk assessment, but its incremental clinical value, cost-effectiveness, and feasibility in low- and middle-income countries require rigorous evaluation. If validated, HADm may support precise risk stratification, individualized follow-up, and more efficient resource allocation, aligning with the World Health Organization (WHO) global goal of accelerating CC elimination by 2030.

Introduction

Clinical Burden and Current Screening Landscape

Cervical cancer (CC) is one of the most important causes of cancer-related morbidity and mortality among women worldwide, despite being highly preventable and, if detected early, highly curable.¹ According to global estimates, approximately 660,000 new cases and 350,000 deaths occurred in 2022, with more than 80% of cases and deaths concentrated in low- and middle-income countries.² This disproportionate burden reflects persistent inequalities in access to vaccination, screening, and timely treatment.

The natural history of CC is well characterized. Persistent infection with high-risk human papillomavirus (HR-HPV) is the primary etiological factor, and HR-HPV DNA is detectable in virtually all cases of invasive CC.^3,4 However, most human papillomavirus (HPV) infections are transient and cleared by the host immune system within 1–2 years; only a minority persist and progress to cervical intraepithelial neoplasia (CIN) and invasive carcinoma.⁵ This distinction between transient infection and progressive disease is central to modern screening: the clinical challenge is not merely to detect HPV, but to identify which HPV-positive women harbor or will develop clinically significant lesions.

Current screening strategies rely predominantly on HR-HPV testing and liquid-based cytology (LBC), either alone or in combination.⁶ HR-HPV testing has high sensitivity (often exceeding 90%) for detecting cervical intraepithelial neoplasia grade 2 or worse (CIN2+) and provides a strong negative predictive value, allowing screening intervals to be extended safely.⁷ As a result, HPV testing has been adopted as the preferred primary screening method in many high-income countries and is recommended by the World Health Organization (WHO) as part of its global strategy to accelerate CC elimination.⁶ Cytology, by contrast, has lower sensitivity but higher specificity for immediate morphological risk and remains widely used, particularly where HPV testing infrastructure is limited.^8,9

Limitations of Current HR-HPV-Based Screening

Despite these advances, current screening approaches have important limitations.

First, HPV testing has suboptimal specificity. Approximately 10–15% of women test positive for HR-HPV, but only a small proportion will progress to cervical intraepithelial neoplasia grade 3 or worse (CIN3+). This leads to unnecessary colposcopy referrals, biopsies, and overtreatment, as well as patient anxiety and increased healthcare costs. Although HPV16/18 genotyping improves risk discrimination, many non-16/18 HR-HPV infections remain difficult to stratify.

Second, cytology is limited by subjectivity, inter-observer variability, and moderate reproducibility.^10,11 Even with LBC, sensitivity for CIN2+ is variable, and quality depends heavily on the availability of trained cytopathologists and cytotechnologists. In many low- and middle-income countries, workforce shortages make widespread, high-quality cytology-based screening difficult to sustain.

Third, current triage strategies—such as repeated HPV testing, cytology, or p16/Ki-67 dual staining—still leave a substantial proportion of HPV-positive women in an indeterminate risk category. This uncertainty contributes to inconsistent follow-up, delayed diagnosis, and unnecessary interventions.

Fourth, the vaginal microenvironment and co-infections are not routinely assessed in screening programs, despite growing evidence that genital pathogens and vaginal microbiota may influence HPV persistence and lesion progression.¹² However, most available evidence is observational; associations between specific microorganisms and cervical disease do not establish causality, and the clinical utility of microbiome-based risk prediction remains unproven.

Finally, patient perspectives are often underrepresented in screening research. HPV positivity is associated with anxiety, stigma, and concerns about partner transmission,¹³ which may affect follow-up adherence and quality of life. A more complex, multi-component screening pathway must therefore be evaluated not only for diagnostic performance but also for acceptability, accessibility, and psychological impact.

Evidence Gap and Rationale for Holistic Framework

These limitations highlight a persistent evidence gap: existing screening tools are effective at detecting infection and morphological abnormalities, but they do not adequately capture the biological heterogeneity that determines whether an HPV infection will regress, persist, or progress. HPV genotyping addresses part of this gap by identifying women at higher etiological risk.¹⁴ Morphological assessment, particularly with artificial intelligence-assisted liquid-based cytology (Ai-LBC), may improve standardization and sensitivity.^15,16 DNA methylation (DNAm) biomarkers may reflect epigenetic changes associated with malignant transformation and may help distinguish transient infection from progressive disease.¹⁷ Genital pathogen and vaginal microbiota profiling may clarify co-factors that promote persistent infection and chronic inflammation.¹²

Each of these components has been studied individually, but they are rarely integrated into a single risk stratification framework. The proposed holistic screening framework—termed HADm—aims to address this gap by combining four complementary pillars:

• (1)Etiology: HR-HPV genotyping;
• (2)Morphology: Ai-LBC;
• (3)Epigenetics: DNAm biomarkers, with PAX1 methylation (PAX1m) as a representative example;
• (4)Microecology: genital pathogens and vaginal microbiota.

The HADm framework is proposed as a complementary, risk-based triage strategy rather than a replacement for existing HPV-based screening. Its intended purpose is to improve discrimination between women who require immediate colposcopy and those who can be safely followed at longer intervals.

Established Evidence versus Proposed Framework

There is strong evidence that HR-HPV testing is sensitive for CIN2+ and provides excellent negative predictive value.¹⁸ There is moderate evidence that HPV16/18 genotyping improves risk stratification.¹⁹ There is growing evidence that DNAm markers, including PAX1m, ZNF582 Methylation (ZNF582m), JAM3 methylation (JAM3m), and EPB41L3 methylation (EPB41L3m), may improve specificity for CIN2+/CIN3+ among HPV-positive women, although performance estimates vary across populations and assays.^20–23 There is also increasing evidence that Ai-LBC can improve sensitivity and standardization compared with conventional cytology,¹⁶ but prospective, population-based validation remains limited. For microecology, evidence is mostly cross-sectional or observational, and causality has not been established.

By contrast, evidence supporting the combined HADm framework is currently lacking. No prospective, multicenter study has evaluated whether integrating all four pillars improves CIN2+/CIN3+ detection, reduces unnecessary colposcopy, or is cost-effective compared with standard triage. The framework should therefore be regarded as a conceptual proposal and a hypothesis-generating model, not as a validated clinical algorithm.

Scope, Terminology, and Feasibility Considerations

This article is a narrative review and conceptual proposal. It does not report original data, and no statistical analysis is applicable. Its objectives are to:

• (1)Summarize the biological and clinical rationale for each HADm pillar;
• (2)Critically appraise the strengths and limitations of available evidence;
• (3)Propose a practical screening, triage, and follow-up pathway;
• (4)Identify key challenges, including cost, infrastructure, standardization, and validation requirements;
• (5)Outline future directions, including urine-based liquid biopsy, point-of-care devices, and microbiome-based interventions.

Feasibility considerations are introduced early because they directly affect the clinical relevance of a four-component strategy. Integrating HPV genotyping, Ai-LBC, DNAm, and multiplex pathogen detection inevitably increases per-capita screening costs and requires advanced laboratory infrastructure, trained personnel, and quality control systems. These requirements may be difficult to meet in low- and middle-income countries (LMICs), where the burden of CC is highest. The framework therefore explicitly considers simplified algorithms—such as HPV genotyping plus PAX1m as a minimum core panel—and reserves Ai-LBC and microbiome analysis for referral centers or research settings.

Finally, the patient perspective is considered throughout. A more complex screening pathway must be acceptable to women, feasible for healthcare providers, and equitable across different settings. HPV positivity-related anxiety, follow-up adherence, and potential overtreatment are important outcomes that should be evaluated alongside diagnostic accuracy in future studies.

Etiology: HR-HPV Infection and Genotyping

HR-HPV as Primary Etiological Factor

Persistent HR-HPV infection is the principal etiological factor in cervical carcinogenesis, and HR-HPV DNA can be detected in nearly all cases of invasive CC.^3,4 More than 100 HPV subtypes infect humans.^24,25 Based on epidemiological and biological evidence, the International Agency for Research on Cancer (IARC) classifies 12 HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59) as carcinogenic to humans (Group 1), HPV68 as probably carcinogenic (Group 2A), and HPV26, 53, 66, 67, and 82 as possibly carcinogenic.

Most HPV infections are transient and cleared by the host immune system within 1–2 years; approximately 80% of infected women would regress automatically,⁵ and only <1% of infected people develop HPV-mediated cancers. HPV persistence lasting more than two years is associated with a significantly increased risk of developing high-grade CIN (CIN2/3). The oncogenic potential of HR-HPV is primarily attributable to the E6 and E7 oncoproteins.²⁶ E6 degrades p53 via the ubiquitin-proteasome pathway, inhibiting apoptosis and allowing DNA-damaged cells to evade immune elimination. E7 binds and inactivates retinoblastoma protein (Rb), leading to cell cycle dysregulation and uncontrolled proliferation. Viral DNA integration into the host genome may result in sustained E6/E7 expression and further facilitate malignant transformation.²⁷

Globally, HPV prevalence parallels CC incidence. In sub-Saharan Africa, HPV infection rates reach approximately 24%, and the region bears the highest global CC burden. In countries with organized screening programs, CC incidence has declined markedly.²⁸ In 2020, the WHO launched the Global Strategy to Accelerate the Elimination of Cervical Cancer, identifying HPV vaccination and regular screening as core measures.

HPV Genotyping Technologies and Clinical Performance

HR-HPV testing uses molecular methods to detect viral DNA or mRNA in cervical exfoliated cells, and detection principles are based mainly on nucleic acid hybridization and polymerase chain reaction (PCR).²⁹ Compared with cytology, HR-HPV testing has higher sensitivity (often exceeding 95%) but lower specificity, because most HPV infections are transient.

Main detection methods include:

• Nucleic acid hybridization: Hybrid Capture 2 (HC2)³⁰ was the first Food and Drug Administration (FDA)-approved HR-HPV test; it detects 13 HR-HPV types but cannot distinguish specific genotypes and has some cross-reactivity.

• PCR amplification: Real-time quantitative PCR platforms (eg, Roche Cobas 4800,³¹ Abbott RealTime) detect 14 HR-HPV types and provide separate reporting for HPV16/18. Multiplex PCR assays (eg, Euroarray, BD Onclarity³²) identify multiple HPV types and enable stratified reporting.

• Isothermal amplification: The Aptima³³ HPV assay detects HPV E6/E7 mRNA using transcription-mediated amplification (TMA), reflecting active oncogene expression rather than mere viral presence.

• Next-generation sequencing (NGS): NGS can detect all known HPV types and provide information on viral load and integration status,^34,35 but cost and technical complexity limit widespread use.

Assay results are not directly comparable because detection targets (L1 vs E6/E7), genotype coverage, and cut-off values differ. For example, HC2 and Cobas show approximately 85% concordance.³⁶

Clinical Utility and Limitations

HR-HPV testing is used in three main clinical scenarios:

• (1)Primary screening: The 2018 USPSTF (US Preventive Services Task Force) guideline recommends one of three strategies, including HR-HPV testing alone every 5 years, for women aged 30 to 65 years.³⁷ Based on ATHENA trial data, HPV primary screening increases sensitivity for CIN3+ by 29.3% compared with cytology (76.1% vs 47.8%) and has a negative predictive value exceeding 99.5%.¹⁸
• (2)Triage of cytological abnormalities: For women with atypical squamous cells of undetermined significance (ASC-US), HR-HPV testing using Hybrid Capture 2 (HC2) achieved a sensitivity of 96.3% for detecting CIN3 or worse, while referring 56.1% of women to colposcopy, compared with 100% referral under immediate colposcopy.³⁸
• (3)Post-treatment surveillance: Post-treatment HR-HPV testing predicts residual or recurrent CIN2+ with a pooled sensitivity of 91.0% (95% CI 82.3–95.5),³⁹ and an updated meta-analysis reports a sensitivity of 86.8%.⁴⁰

However, HPV testing has important limitations. Approximately 10–15% of women test positive, but the 5-year cumulative risk of progression to CIN3+ is 6.4% (95% prediction interval 0.0–13.5) in high-risk HPV-positive women with normal cytology.⁴¹ Undifferentiated HPV testing leads to unnecessary colposcopic biopsies, patient anxiety, and diminished quality of life. HPV16/18 genotyping improves risk discrimination, but non-16/18 HR-HPV infections remain difficult to stratify due to marked within-category risk heterogeneity.⁴² This is the principal rationale for adding methylation and morphological triage within the HADm framework.

Cost and Feasibility

HR-HPV testing has demonstrated favorable cost-effectiveness in organized screening programs, with systematic reviews consistently identifying primary HPV testing as the optimal or most cost-effective strategy compared with cytology-based screening.⁴³ In LMICs, however, coverage remains low because of infrastructure, workforce, and cost constraints. Self-sampling devices show acceptable concordance with clinician-collected samples and may expand coverage. Isothermal amplification and point-of-care platforms can reduce infrastructure requirements. In the HADm framework, HPV genotyping is the minimum core component and should be prioritized where resources are limited.

Morphology: AI-Assisted Liquid-Based Cytology

Cytomorphological Basis of Cervical Lesion Grading

CC development is a protracted process, and early lesions may regress. Cervical exfoliative cytology, including conventional Papanicolaou smear and LBC, is a cornerstone method for screening cervical precancerous lesions. Cytomorphological changes accompany lesion progression, including nuclear enlargement, hyperchromasia, an increased nuclear-to-cytoplasmic (N/C) ratio, and loss of cellular polarity.⁴⁴ In low-grade squamous intraepithelial lesions (LSIL), nuclei are mildly enlarged with slightly coarse chromatin, but cellular morphology remains relatively regular. In high-grade squamous intraepithelial lesions (HSIL), nuclei are markedly enlarged and hyperchromatic, the N/C ratio is significantly elevated, and multinucleation or abnormal mitotic figures may be present. In invasive CC, cellular polarity is completely lost, nuclear atypia is pronounced, and necrotic debris or inflammatory cells may be observed in the background.

Conventional cytology relies on pathologist interpretation and is limited by subjectivity, low efficiency, and a high false-negative rate. Manual interpretation is susceptible to fatigue and limited fields of view.

Ai-LBC Technical Workflow

Ai-LBC is an integrated artificial intelligence-assisted liquid-based cytology system spanning three functional layers. The hardware layer comprises automated sample processing instruments, high-resolution whole-slide imaging (WSI) scanners, and high-performance computing servers.⁴⁵ The algorithm layer employs deep learning architectures—predominantly convolutional neural networks for cell-level morphological feature extraction (eg, nuclear enlargement, hyperchromasia, increased N/C ratio) and increasingly Transformer-based models for contextual slide-level classification.^45,46 The application layer integrates with laboratory information systems (LIS) and hospital information systems (HIS) to embed AI-assisted reading into routine clinical workflows, and supports teleconsultation platforms for remote expert review and quality control.^45,47

The standardized workflow has five stages:

• (1)Cervical brush samples are preserved in cell buffer solution.
• (2)An automated slide preparation system produces monolayer cell slides using density gradient centrifugation and membrane adsorption.
• (3)A high-speed whole-slide scanner digitizes the entire slide at 20× or higher magnification.
• (4)Pre-trained deep neural networks perform multi-task analysis, including cell segmentation, morphological quantification, N/C ratio calculation, chromatin distribution assessment, and extraction of hundreds of characteristic parameters.
• (5)The system generates a structured report annotating suspicious regions according to the Bethesda System for pathologist review.

The entire process can be completed within 2 hours, whereas traditional methods require 24–48 hours.

Diagnostic Performance and Limitations

The latest third-generation Ai-LBC systems employ multimodal fusion learning frameworks, integrating visual features with textual or contextual information to enhance diagnostic accuracy and interpretability.⁴⁶ An improved Mask R-CNN model incorporating a PointRend module achieves high-precision segmentation of overlapping cervical cells, attaining a Dice similarity coefficient of 0.97 on the ISBI2014 cervical cytology dataset.⁴⁸ The pooled accuracy, sensitivity, and specificity of AI-assisted LBC were 90% (95% CI 85–94), 97% (95% CI 95–99), and 94% (95% CI 85–98), respectively.⁴⁹ Compared with conventional Papanicolaou smears, Ai-LBC increases the detection rate of CIN2+ lesions and exhibits comparable sensitivity to LBC with improved specificity.

However, most Ai-LBC studies are retrospective or cross-sectional, and prospective population-based validation is limited. Performance varies across scanners, staining methods, and populations. The “black box” nature of deep learning models raises interpretability concerns. AI remains an adjunct tool; final clinical decisions must integrate professional judgment and the overall condition of the patient.

Cost and Feasibility

Ai-LBC requires high-resolution scanners, stable electricity, computing infrastructure, and trained personnel. These requirements may be difficult to meet in LMICs. Cloud-based AI platforms can partially circumvent local infrastructure deficits, and portable scanners are increasingly available. In the HADm framework, Ai-LBC is proposed as a triage tool for HPV-positive women rather than a stand-alone primary screening test in all settings. Simplified algorithms may reserve Ai-LBC for referral centers.

Epigenetics: DNA Methylation Biomarkers

DNA Methylation as an Epigenetic Switch

Epigenetics refers to mechanisms that regulate gene expression without altering the DNA sequence, including DNAm, histone modification, and non-coding RNA regulation.⁵⁰ DNAm is the most common and best-studied epigenetic modification in cancer. DNAm participates in tumor progression by silencing tumor suppressor genes or activating oncogenes.⁵¹

DNAm detection has three potential advantages in CC screening:

• (1)High tissue specificity: Methylation levels of specific genes are low in normal tissues but elevated in precancerous lesions and invasive carcinoma, reducing false-positive interference.
• (2)Early warning capability: Methylation changes may occur earlier than morphological alterations, enabling identification of high-risk lesions even when cytology remains normal.
• (3)Prognostic assessment value: Methylation levels are positively correlated with lesion progression risk, potentially guiding clinical intervention timing.

PAX1m and Other Methylation Markers

Dozens of methylation-regulated genes associated with CC have been identified, including PAX1, ZNF582, DPP6, ZNF761, JAM3, FAM19A4, PCDHGB7, EPB41L3, CADM1, MAL, and GYPC.^20,21,52 These genes play important roles in cervical carcinogenesis, and their aberrant methylation levels correlate with CIN and CC. Methylation status may serve as a “molecular switch” for cervical malignant transformation.⁵³

PAX1 belongs to the PAX gene family and encodes a transcription factor involved in embryonic spinal and thymic development. In normal cervical tissue, PAX1 expression maintains the balance between epithelial cell differentiation and proliferation.⁵⁴ In CC, PAX1 is frequently downregulated due to hypermethylation of its promoter region, leading to cell cycle dysregulation and inhibition of apoptosis.⁵⁵ PAX1 expression suppresses phosphorylation of multiple kinases following stimulation by oncogenic growth factors such as epidermal growth factor (EGF) and IL-6. PAX1 also activates phosphatases including dual-specificity phosphatase 1/5/6 (DUSP1/5/6), and inhibits the EGF/ mitogen-activated protein kinase (MAPK) signaling pathway. PAX1 interacts with SET domain containing 1B (SET1B), enhancing histone H3K4 methylation and DNA demethylation of phosphatase-encoding genes.⁵⁶

Multiple studies have demonstrated that PAX1m levels progressively increase with lesion severity (normal → CIN1 → CIN2 → CIN3 → invasive carcinoma).^57–59 However, most PAX1m evidence comes from Chinese cohorts, and external validation in other populations remains limited. PAX1m should therefore be regarded as a representative example of DNAm biomarkers rather than the only candidate. Table 1 summarizes representative DNAm biomarkers, including ZNF582m,²¹ JAM3m,²² EPB41L3m,²³ FAM19A4m,⁶⁰ PCDHGB7 methylation (PCDHGB7m),⁶¹ and ZNF671 methylation (ZNF671m),⁶² and compares their reported performance and limitations.

Table 1.

Representative DNA Methylation Biomarkers for Cervical Cancer Triage

Biomarker	Gene Function / Pathway	Reported Sensitivity for CIN2+	Reported Specificity for CIN3+	Sample Type	Evidence Source and Setting	Key Limitations
PAX1m	Transcription factor; regulates epithelial differentiation and proliferation; activates phosphatases and inhibits EGF/MAPK signaling	Moderate to high; varies by assay and cut-off	High (often 85–90% in some studies)	Cervical scrapings; LBC specimens	Multiple studies, mainly Chinese cohorts; included in China’s Cervical Cancer Screening Guidelines II for HPV-positive triage	Cut-off standardization; external validation in non-Chinese populations limited; performance may vary by HPV genotype
ZNF582m	Zinc finger transcription factor; involved in transcriptional regulation	Moderate to high	High	Cervical scrapings; tissue	Case-control and cohort studies	Assay-dependent performance; fewer large prospective studies
JAM3m	Junctional adhesion molecule; cell-cell adhesion and signaling	Moderate	Moderate to high	Cervical scrapings	Case-control and cohort studies	Limited prospective population-based validation
EPB41L3m	Cytoskeletal adaptor protein; tumor suppressor candidate	Moderate	Moderate to high	Cervical scrapings	Cohort studies, including international collaborations	Performance varies with HPV status and lesion grade
FAM19A4m	Chemokine-like protein; immune regulation	Moderate	High	Cervical scrapings; self-samples	European cohorts; QIAsure commercial assay	Cost; availability; performance in LMICs less studied
Multi-gene panels (eg, QIAsure, GynTech)	Combination of methylation markers	Moderate to high	High	Cervical scrapings	Commercial assays; European and international cohorts	Higher cost; need for standardized reporting; limited head-to-head comparisons with PAX1m
PCDHGB7m	Protocadherin; cell adhesion	Moderate to high	High	Cervical scrapings	Prospective cohort study (CerMe)	Requires further multicenter validation
ZNF671m	Zinc finger protein; transcriptional regulation	Moderate to high	High	Cervical scrapings	Case-control and cohort studies	Limited external validation

Abbreviations: CIN, cervical intraepithelial neoplasia; EGF, epidermal growth factor; LBC, liquid-based cytology; LMICs, low- and middle-income countries; MAPK, mitogen-activated protein kinase.

Comparison with QIAsure, GynTech, and Other Assays

Several commercial methylation assays are available, including QIAsure (FAM19A4/miR124-2) and GynTech (multi-gene panel). These assays have shown high specificity for CIN3+ in European cohorts.^63,64 However, head-to-head comparisons with PAX1m are limited, and performance estimates vary across populations, sample types, and cut-off values. Table 1 provides a structured comparison.

Clinical Utility and Limitations

DNAm detection is used in two main clinical scenarios:

• (1)Triage of HPV-positive women: DNAm can distinguish transient HPV infection from precancerous lesions resulting from persistent infection. Studies have shown that DNAm has a specificity of 85–95% for CIN3+, significantly higher than cytology (approximately 75%).^22,59 A combined strategy of HPV + DNAm double positive → direct colposcopy; HPV positive but normal methylation → extended follow-up interval has been proposed.
• (2)Monitoring of precancerous lesion progression: Dynamic changes in methylation levels may inform conservative management decisions for CIN2 and post-treatment follow-up.⁶⁵ Failure of DNAm to decrease to very low levels after surgery may indicate residual disease or recurrence.

Limitations include assay-dependent cut-offs, limited external validation, population differences, and the need for standardized reporting. PAX1m testing has been incorporated into China’s Cervical Cancer Screening Guidelines II for triage of HPV-positive patients,⁶⁶ but its role in other health systems remains to be defined.

Cost and Feasibility

DNAm detection technologies (eg, quantitative methylation-specific PCR) are mature, relatively low-cost, and amenable to automation. However, bisulfite treatment, DNA extraction, and quality control require well-equipped molecular laboratories. In LMICs, simplified workflows and centralized testing may be necessary. Within the HADm framework, a minimum core panel of HPV genotyping plus PAX1m may be the most feasible approach, with broader multi-gene panels reserved for referral centers.

Microecology: Genital Pathogens and Vaginal Microbiota

Chlamydia Trachomatis, Ureaplasma Urealyticum, and Neisseria Gonorrhoeae

Chlamydia trachomatis (CT) is an intracellular prokaryotic microorganism that invades columnar epithelial cells at the squamocolumnar junction. It disrupts epithelial cells and induces inflammatory responses, leading to elevated pH in cervical and vaginal secretions, which may facilitate HPV invasion. CT infection approximately doubles the risk of HPV infection.⁶⁷ CT/HPV coinfection further suppresses Langerhans cell functions by reducing their antigen-presenting ability and density, and alters T-cell subsets, resulting in fewer CD4+ and CD8+ T cells and increased Treg infiltration, thereby impairing cell-mediated immunity.⁶⁸

Ureaplasma urealyticum (UU) is a commensal microorganism that can disrupt epithelial barriers and modulate immune responses, creating a chronic inflammatory environment that facilitates HPV persistence. UU infection is significantly associated with an increased risk of cervical cancer (OR = 1.81, 95% CI 1.31–2.51). HPV-positive individuals co-infected with genital mycoplasmas have a markedly elevated cervical cancer risk (OR = 3.13, 95% CI 2.04–4.79), demonstrating potential synergistic effects in tumor progression.⁶⁹ UU infection induces production of pro-inflammatory cytokines including tumor necrosis factor-alpha (TNF-α), interferon-gamma (IFN-γ), interleukin-6 (IL-6), and interleukin-8 (IL-8), contributing to cervical inflammation and epithelial tissue damage.⁷⁰ However, UU colonization is common in healthy women, and the distinction between colonization and clinically meaningful infection depends on bacterial load, species, and host immunity.

Neisseria gonorrhoeae (NG) invades and proliferates within the urogenital tract, causing urethritis and cervicitis. Epidemiological studies suggest that NG is an independent risk factor for cervical cancer.⁷¹ NG employs multiple immune evasion mechanisms, including secretion of nucleases that degrade neutrophil extracellular traps, inhibition of respiratory burst, activation of the NLR family pyrin domain containing 3 (NLRP3) inflammasome, modulation of macrophage differentiation, and binding of factor H to evade complement-mediated killing. NG promotes transforming growth factor-beta (TGF-β) production, suppressing T helper 1/T helper 2 (Th1/Th2)-mediated adaptive immune responses, and binds carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) on B and T cells to inhibit proliferation. These immunomodulatory effects may facilitate HPV persistence and contribute to cervical carcinogenesis.⁷²

Infections with CT, UU, NG, herpes simplex virus type 2 (HSV-2), Ureaplasma parvum, Mycoplasma hominis, and Mycoplasma genitalium are risk factors for HPV infection.⁷¹ Patients with CIN or cervical malignancies have significantly higher infection rates of CT and UU.^72,73 As shown in Table 2, HPV persistence and cervical lesion progression have been associated with alterations in the vaginal microbiota and with several genital pathogens.

Table 2.

Genital Pathogens and Vaginal Microbiota Associated with HPV Persistence and Cervical Lesion Progression

Microorganism / Community	Proposed Mechanism	Association with HPV Persistence or CIN	Detection Method	Potential Clinical Implication	Evidence Caution
Chlamydia trachomatis (CT)	Invades columnar epithelial cells; induces chronic inflammation; elevates vaginal pH; may impair Langerhans cell function and T-cell responses; encodes pORF5, which activates Nrf2/NQO1 and inhibits apoptosis	Approximately doubles the risk of HPV infection; associated with HPV persistence and CIN	Multiplex PCR; qPCR; mNGS	Screening and treatment in HPV-positive women; may reduce persistence risk	Association is consistent but causality is not established; confounding by sexual behavior and co-infections
Ureaplasma urealyticum (UU)	Produces proteases and urease; disrupts cell membrane stability; induces cytokine release (TNF-α, IFN-γ, IL-6, IL-8) and cervical inflammation	Associated with vaginal dysbiosis and increased risk of HSIL; independent factor for microenvironmental disturbance	Culture; multiplex PCR; qPCR	Consider treatment in symptomatic or persistent HPV infection; differentiate colonization from infection	Colonization is common; pathogenic role depends on load, species, and host immunity
Neisseria gonorrhoeae (NG)	Immune evasion via nuclease secretion, inhibition of respiratory burst, NLRP3 activation, TGF-β production, and CEACAM1-mediated T/B cell suppression	Associated with HPV infection and cervical inflammation; may facilitate persistence	Culture; multiplex PCR; qPCR	Test in sexually active women with risk factors; treat promptly	Evidence for direct effect on CC progression is limited
Herpes simplex virus type 2 (HSV-2)	Synergizes with HPV; may inhibit p53 and Rb pathways; induces cervical ulcers	Associated with increased HPV acquisition and persistence	qPCR; multiplex PCR	Consider testing in high-risk populations	Causality and incremental value in screening algorithms uncertain
Human immunodeficiency virus (HIV)	CD4+ T-cell depletion; impaired HPV immune surveillance	3- to 6-fold increased risk of CC	HIV serology; viral load	Prioritize HPV screening and early treatment	Well-established association; management pathways differ from general population
Gardnerella vaginalis	Disrupts cervical mucus barrier; contributes to bacterial vaginosis; promotes HPV infection and persistence	Increased abundance in HPV-infected women; associated with CIN	Nugent score; multiplex PCR; 16S rRNA sequencing; mNGS	Guide anti-infective therapy and vaginal microenvironment regulation	Association does not prove causality; co-occurrence with other anaerobes
Prevotella spp.	Anaerobic overgrowth; produces carcinogenic metabolites such as nitrosamines	Increased abundance in HPV-infected women; associated with dysbiosis	16S rRNA sequencing; mNGS	Potential marker of high-risk microenvironment	Evidence mainly cross-sectional
Lactobacillus crispatus and other Lactobacillus spp.	Maintain low vaginal pH (3.8–4.5); secrete H2O2 and antimicrobial peptides; inhibit pathogen colonization	Depletion associated with HPV persistence and CIN	16S rRNA sequencing; mNGS; qPCR	Probiotic supplementation; restoration of vaginal microbiota	Strain-specific effects; clinical trials limited
Bacterial vaginosis-associated community	Reduced Lactobacillus; overgrowth of anaerobes; chronic inflammation; carcinogenic metabolites	Associated with HPV persistence and cervical lesion progression	Nugent score; 16S rRNA sequencing; mNGS	Microbiome-targeted intervention; monitoring of treatment response	Standardization and causality remain unresolved

Abbreviations: CC, cervical cancer; CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus; HSIL, high-grade squamous intraepithelial lesion; mNGS, metagenomic next-generation sequencing; qPCR, quantitative polymerase chain reaction; TGF-β, transforming growth factor-beta; TNF-α, tumor necrosis factor-alpha.

Viral Co-Infections

HSV-2 may synergize with HPV to inhibit p53 and Rb pathways and induces cervical ulcers, increasing opportunities for HPV invasion. Human immunodeficiency virus (HIV) causes CD4+ T cell depletion, weakening HPV immune surveillance; HIV-infected individuals have a 3- to 6-fold increased risk of cervical cancer.⁷⁴

Bacterial Vaginosis and Lactobacillus Depletion

Bacterial vaginosis (BV) is characterized by a reduction in Lactobacillus species (eg, Lactobacillus crispatus, Lactobacillus gasseri) and overgrowth of anaerobic bacteria (eg, Gardnerella vaginalis, Prevotella spp.).⁷⁵ Lactobacillus crispatus maintains a low vaginal pH (3.8–4.5), inhibiting pathogen colonization, and secretes hydrogen peroxide (H₂O₂) and antimicrobial peptides.^75,76 Gardnerella vaginalis disrupts the cervical mucus barrier and alters cervicovaginal epithelial cell function through TLR2-dependent immune responses, potentially promoting HPV infection and persistence.⁷⁷ Prevotella spp. may produce carcinogenic metabolites such as nitrosamines. However, most evidence for microbiota-disease associations is cross-sectional or observational; causality has not been established, and microbiome-based risk prediction remains exploratory.

Detection Technologies

Traditional detection methods include culture (time-consuming, low sensitivity for anaerobes) and microscopic examination (Gram stain Nugent score for BV; wet mount for Trichomonas and fungi). Molecular methods include multiplex PCR (simultaneous detection of 10–20 pathogens with high sensitivity), 16S rRNA gene sequencing (comprehensive microbiota composition), metagenomic next-generation sequencing (mNGS; unbiased detection of all microorganisms and functional genes), and real-time quantitative PCR (qPCR) for specific pathogens. Rapid detection technologies include isothermal amplification (eg, loop-mediated isothermal amplification [LAMP]) and microfluidic chips.

Integration into Clinical Decision-Making

Incorporating microecology into the HADm framework requires careful consideration to avoid unnecessary complexity. We propose that microecology testing be used selectively:

• (1)In HPV-positive women with negative DNAm and negative Ai-LBC, microecology testing may help identify co-factors that promote persistence and inform anti-infective or probiotic intervention.
• (2)In women with persistent HPV infection, microecology profiling may guide targeted therapy and vaginal microenvironment regulation.
• (3)In research settings, microbiome analysis may help identify high-risk microbial combinations (eg, HPV + Gardnerella + low Lactobacillus) and generate hypotheses for future trials.

Microecology is not proposed as a stand-alone primary screening test. Its incremental clinical value and cost-effectiveness require prospective evaluation.

Cost and Feasibility

Multiplex PCR and qPCR are relatively affordable and can be integrated into existing molecular laboratories. 16S rRNA sequencing and mNGS are more expensive and require bioinformatics support, limiting their routine use in LMICs. Simplified panels targeting CT, UU, NG, and BV-associated bacteria may be more feasible. In the HADm framework, microecology testing is optional and should be prioritized for women with persistent HPV infection or those in research settings.

Proposed HADm Screening, Triage, and Follow-up Pathway

Rationale and Scope of the Proposed Pathway

The HADm framework is proposed as a conceptual, risk-based triage strategy that complements, rather than replaces, existing HPV-based screening. Its intended purpose is to improve discrimination between women who require immediate colposcopy and those who can be safely followed at longer intervals.

The pathway integrates four pillars:

• (1)Etiology: HR-HPV genotyping;
• (2)Morphology: Ai-LBC;
• (3)Epigenetics: DNAm biomarkers, with PAX1m as a representative example;
• (4)Microecology: genital pathogens and vaginal microbiota.

The pathway has not been prospectively validated. It is presented as a hypothesis-generating model to guide future comparative studies. Figure 1 summarizes the proposed screening, triage, and follow-up algorithm. Table 3 compares the four pillars in terms of targets, technologies, performance, clinical utility, limitations, and evidence level.

Figure 1.

Proposed Clinical Pathways of the Holistic Screening Framework (HADm).

Abbreviations: HPV, Human Papillomavirus; HR-HPV, High-Risk Human Papillomavirus; HPV16/18, Human papillomavirus genotypes 16 and/or 18; DNAm, DNA Methylation; AI-LBC, Artificial Intelligence-assisted Liquid-Based Cytology; ASC-US, Atypical Squamous Cells of Undetermined Significance; CIN1, Cervical Intraepithelial Neoplasia Grade 1; CIN2+, Cervical Intraepithelial Neoplasia Grade 2 or Worse.

Table 3.

Comparison of the Four HADm Pillars

Pillar	Main Target	Representative Technologies	Sensitivity for CIN2+/CIN3+	Specificity for CIN2+/CIN3+	Main Clinical Utility	Key Limitations	Evidence Level
Etiology	HR-HPV DNA or E6/E7 mRNA	HC2; Cobas 4800; Abbott RealTime; Aptima; BD Onclarity; NGS	High (often >90–95% for CIN2+)	Low to moderate (most HPV infections are transient)	Primary screening; identification of the causative agent; genotyping for HPV16/18 and other HR types	Low specificity; overdiagnosis; assay heterogeneity; cannot distinguish transient from persistent infection	Established
Morphology	Cytomorphological abnormalities	Conventional LBC; Ai-LBC (CNN/Transformer-based)	Moderate to high; Ai-LBC may improve sensitivity	Moderate to high; Ai-LBC may improve standardization and specificity	Lesion grading; triage of HPV-positive women; morphological confirmation	Subjective in manual reading; inter-observer variability; Ai-LBC requires digital scanners, algorithms, and validation	Established for LBC; emerging for Ai-LBC
Epigenetics	DNA methylation of tumor suppressor genes	qMSP; MSP; pyrosequencing; ddPCR; methylation arrays	Moderate to high for CIN3+; varies by marker and assay	High (often 85–90% for CIN3+ in some studies)	Triage of HPV-positive women; identification of progressive lesions; post-treatment surveillance	Assay-dependent cut-offs; limited external validation; population differences; PAX1m evidence mainly from Chinese cohorts	Emerging to moderate
Microecology	Genital pathogens and vaginal microbiota	Multiplex PCR; 16S rRNA sequencing; mNGS; qPCR; LAMP; microfluidic chips	Not established as a stand-alone screening test	Not established	Risk co-factor assessment; explanation of HPV persistence; guidance for anti-infective or probiotic intervention	Mostly observational evidence; causality not established; Ureaplasma may be colonizing rather than pathogenic; standardization lacking	Exploratory

Abbreviations: Ai-LBC, artificial intelligence-assisted liquid-based cytology; CIN, cervical intraepithelial neoplasia; CNN, convolutional neural network; ddPCR, digital droplet polymerase chain reaction; HC2, Hybrid Capture 2; HR-HPV, high-risk human papillomavirus; LAMP, loop-mediated isothermal amplification; LBC, liquid-based cytology; mNGS, metagenomic next-generation sequencing; MSP, methylation-specific polymerase chain reaction; NGS, next-generation sequencing; qMSP, quantitative methylation-specific polymerase chain reaction; qPCR, quantitative polymerase chain reaction.

Screening Strategies

Preferred Strategy

HR-HPV DNA testing is used as the primary screening tool. If HR-HPV is negative, repeat screening is performed after 5 years (or after 3 years if local guidelines recommend shorter intervals). If HR-HPV is positive, HPV genotyping is performed, followed by DNAm and/or Ai-LBC testing to assess lesion risk.

Alternative Strategy

HPV DNA combined with Ai-LBC is used as primary screening, with DNAm serving as the triage tool. Where HPV testing is unavailable, Ai-LBC may be used as the sole screening method; if the result is ASC-US/LSIL, supplementary DNAm or HR-HPV testing is performed.

In resource-limited settings, a simplified algorithm—HPV genotyping plus PAX1m as the minimum core panel—may be considered, with Ai-LBC and microecology reserved for referral centers or research settings.

Triage Management

Triage of HPV-positive women:

• HPV16/18 positive → direct referral to colposcopy;

• Non-16/18 HR-HPV positive → DNAm positive or cytology ≥ ASC-US → refer to colposcopy; DNAm negative and cytology negative → repeat HPV + DNAm at 12 months;

• HPV positive + genital pathogen positive → anti-infective therapy + repeat testing at 3–6 months.

Triage of cytological abnormalities:

• ASC-H (atypical squamous cells, cannot exclude HSIL) or worse → colposcopy;

• ASC-US/LSIL → HR-HPV positive or DNAm positive → refer to colposcopy; both negative → repeat HPV + DNAm at 6–12 months;

• Negative cytology but HR-HPV positive → repeat at 6–12 months; if DNAm positive → consider colposcopy.

These thresholds are proposed for evaluation and should not be regarded as fixed clinical recommendations until prospective validation is available.

Follow-up Surveillance

• (1)Low-risk population (HPV negative / Ai-LBC negative): routine screening, with HPV every 5 years or Ai-LBC every 3 years.
• (2)Intermediate-risk population (HPV positive but DNAm / Ai-LBC negative): repeat HPV + DNAm testing at 12 months. If HPV becomes negative, return to routine screening; if HPV remains positive but DNAm remains negative, appropriately extend the follow-up interval; if DNAm becomes positive, refer to colposcopy.
• (3)High-risk population (eg, CIN): For patients with CIN1/CIN2 prior to treatment who are DNAm negative, repeat testing at 6–12 months. If results normalize, the follow-up interval may be extended; if lesions persist or progress, corresponding treatment should be administered. At 6 months post-treatment for CIN, HPV + DNAm testing is performed. If both are negative, repeat testing at 12 months; if either is positive, colposcopy evaluation is performed to assess the risk of residual or recurrent disease.

Individualized Intervention

• (1)HPV persistent infection with abnormally elevated DNAm: Confirm lesions based on colposcopy biopsy results. For CIN2+, perform conization; for CIN1, conservative observation with immunomodulation may be considered.
• (2)Concurrent genital tract infection: Administer anti-infective therapy first, followed by regulation of the vaginal microenvironment to reduce the risk of persistent HPV infection. However, the benefit of routine anti-infective therapy in asymptomatic women remains unproven.
• (3)Dynamic methylation monitoring: Compare DNAm levels before and after treatment. A post-treatment decline may indicate effective treatment; persistent elevation post-treatment warrants vigilance for recurrence or progression. This monitoring strategy requires prospective evaluation.

Critical Appraisal and Challenges

While the HADm framework offers a conceptually appealing approach, its translation into practice faces substantial challenges. This section provides a balanced appraisal of the evidence and identifies key uncertainties.

Incremental Clinical Value and Cost-Complexity Trade-off

The central question is whether integrating four pillars provides sufficient incremental clinical benefit to justify the added complexity and cost compared with existing sequential triage strategies. Current evidence supports the individual performance of HPV testing, cytology, DNAm, and—to a lesser extent—microecology. However, no prospective study has evaluated whether the combined HADm algorithm improves CIN2+/CIN3+ detection, reduces unnecessary colposcopy, or is cost-effective compared with standard care. The framework should therefore be regarded as a conceptual proposal rather than a validated clinical pathway. Comparative studies are needed to determine whether a simplified algorithm (eg, HPV genotyping plus PAX1m) performs comparably to the full four-pillar model.

Cost-Effectiveness and Economic Viability

Integrating HPV genotyping, Ai-LBC, DNAm, and multiplex pathogen detection inevitably increases per-capita screening costs. Although the HADm framework may reduce unnecessary colposcopic referrals and overtreatment, upfront investment in equipment, reagents, and training is substantial. Comprehensive cost-effectiveness analyses of multi-dimensional screening remain scarce. While individual components have demonstrated favorable cost-effectiveness in high-income settings, the incremental cost-effectiveness ratio of the combined HADm approach requires rigorous evaluation. Prospective studies with embedded health economic analyses are urgently needed to define budget impact thresholds across different economic contexts.

Applicability in Low- and Middle-Income Countries (LMICs)

Paradoxically, regions with the highest CC burden—LMICs, where more than 80% of global cases occur—face the greatest implementation barriers. Infrastructure constraints include requirements for high-resolution scanners and stable electricity (Ai-LBC), well-equipped molecular laboratories (DNAm), and sophisticated thermal cyclers (multiplex PCR). Severe shortages of pathologists, cytotechnologists, and molecular biologists further compound these challenges.

Promising solutions are emerging. Self-sampling devices show acceptable concordance with clinician-collected samples for HPV and methylation testing. Isothermal amplification eliminates the need for expensive thermal cyclers. Cloud-based AI platforms can partially circumvent local infrastructure deficits. Portable, battery-powered PCR devices are increasingly available for decentralized testing. Implementation research should prioritize simplified HADm algorithms—for example, HPV genotyping plus PAX1m as the minimum core panel, with Ai-LBC and microbiome analysis reserved for referral centers.

Standardization Challenges in Data Integration

The HADm framework generates heterogeneous data—viral genotypes, cytomorphological parameters, methylation ratios, and microbial abundances—from distinct platforms with different units and cut-offs. No universal standards exist for HPV genotyping thresholds, methylation percentage cut-offs, or microbiome metrics. Divergent protocols for sample collection, preservation, DNA extraction, and analysis lead to poor inter-laboratory reproducibility. Moreover, while machine learning can integrate multi-dimensional data, the “black box” nature of deep learning models raises interpretability concerns.

Proposed solutions include developing international reference standards for each HADm component, establishing external quality assessment programs, applying explainable AI techniques to render risk scores interpretable, and creating web-based open-access risk calculators.

Clinical Validation and Prospective Evidence

Although individual HADm components have been validated in large cohorts, the integrated framework lacks prospective, multicenter validation. Critical unresolved questions include:

• What are the positive and negative predictive values of the HADm algorithm for CIN2+ and CIN3+ across diverse populations?

• Does HADm reduce interval cancers compared with current strategies?

• What is the optimal screening interval for each risk stratum?

• Can HADm safely replace colposcopy in selected low-risk groups?

• Is HADm acceptable to women and feasible for healthcare providers?

Large-scale, cluster-randomized controlled trials comparing HADm-guided management with standard-of-care are needed across different geographic settings, incorporating patient-reported outcomes and health-economic analyses.

Patient Perspective and Acceptability

A more complex screening pathway may increase anxiety, reduce follow-up adherence, and complicate communication. HPV positivity is already associated with stigma and concerns about partner transmission. Adding methylation and microbiome testing may heighten these concerns if results are not clearly explained. Conversely, a more precise risk assessment may reduce unnecessary interventions and reassure low-risk women. Acceptability, accessibility, and psychological impact should be evaluated alongside diagnostic accuracy in future studies. Shared decision-making and clear communication tools will be essential.

Future Perspectives

Urine-Based Liquid Biopsy

Urine testing offers a non-invasive, self-collectable alternative to cervical brushing. Emerging evidence demonstrates concordance with cervical specimens for HPV DNA and methylation detection.⁷⁸ Although sensitivity for pre-invasive CIN2/3 lesions remains suboptimal, ongoing refinements—including urine preservative optimization and pre-amplification strategies—hold promise for expanding screening coverage, particularly among under-screened populations. Urine-based testing could be integrated into the HADm framework as a self-sampling option, but its performance for methylation and microbiome assessment requires further validation.

Point-of-Care Devices

Microfluidics, Lab-on-a-Chip technology, and smartphone integration are converging toward integrated HPV + methylation cartridges. Paper-based colorimetric sensors and portable nanopore sequencers offer glimpses of accessible screening in resource-constrained environments. If validated, these technologies could enable decentralized HADm testing and reduce dependence on centralized laboratories. However, analytical performance, quality control, and data security must be rigorously evaluated.

Microbiome-Based Interventions

As understanding of the cervical-vaginal microbiome deepens, probiotic supplementation (eg, Lactobacillus crispatus strains), microbiome-targeted anti-inflammatory agents, and biomarker-guided antimicrobial therapy may emerge as novel preventive strategies. However, current evidence is mostly observational, and randomized trials are needed to determine whether microbiome modulation reduces HPV persistence or CIN progression. Within the HADm framework, microbiome testing could guide individualized intervention, but it should not be regarded as a stand-alone screening tool.

Health System and Implementation Research

Moving HADm from concept to practice will require:

• Collaborative consortiums to establish standardized protocols and reference materials;

• Public-private partnerships to drive low-cost diagnostic innovation;

• Health system strengthening, including workforce training and quality assurance;

• Patient engagement to ensure acceptability and equity;

• Implementation research to evaluate simplified algorithms in LMICs;

• Health-economic evaluation to inform reimbursement and resource allocation.

These efforts should be aligned with the WHO global strategy to accelerate CC elimination by 2030.

Conclusion

Current HR-HPV-based screening has high sensitivity but limited specificity, resulting in unnecessary colposcopy and overtreatment. The proposed HADm framework integrates etiology (HR-HPV genotyping), morphology (Ai-LBC), epigenetics (DNAm, eg, PAX1m), and microecology (genital pathogens and vaginal microbiota) into a multidimensional risk stratification pathway. Each individual component has supportive evidence, but the combined framework has not yet been validated prospectively. Its potential clinical value—including improved CIN2+/CIN3+ detection, reduced unnecessary colposcopy, and more efficient resource allocation—must be tested in large-scale, multicenter, cluster-randomized trials with embedded health-economic analyses and patient-reported outcomes. Simplified algorithms, such as HPV genotyping plus PAX1m as a minimum core panel, may improve feasibility in LMICs. Emerging technologies, including urine-based liquid biopsy and point-of-care microfluidic devices, could further expand screening coverage. With collaborative standardization, public-private partnerships, and health system strengthening, HADm may evolve from a conceptual framework into a practical and equitable tool for accelerating global CC elimination.

Abbreviations

Ai-LBC, Artificial Intelligence-assisted Liquid-Based Cytology; ASC-US, Atypical Squamous Cells of Undetermined Significance; BV, Bacterial Vaginosis; CC, Cervical Cancer; CIN, Cervical Intraepithelial Neoplasia; CIN1, Cervical Intraepithelial Neoplasia Grade 1; CIN2, Cervical Intraepithelial Neoplasia Grade 2; CIN3, Cervical Intraepithelial Neoplasia Grade 3; CIN2+, Cervical Intraepithelial Neoplasia Grade 2 or Worse; CIN3+, Cervical Intraepithelial Neoplasia Grade 3 or Worse; CNN, Convolutional Neural Network; CT, Chlamydia trachomatis; DNAm, DNA Methylation; EGF, Epidermal Growth Factor; EPB41L3m, EPB41L3 methylation; FAM19A4m, FAM19A4 methylation; HADm, Holistic framework integrating Etiology, Morphology, Epigenetics, and Microecology; HC2, Hybrid Capture 2; HIV, Human Immunodeficiency Virus; HPV, Human Papillomavirus; HR-HPV, High-Risk Human Papillomavirus; HSV-2, Herpes Simplex Virus Type 2; HSIL, High-Grade Squamous Intraepithelial Lesion; IL-6, interleukin-6; IL-8, interleukin-8; JAM3m, JAM3 methylation; LBC, Liquid-Based Cytology; LMICs, Low- and Middle-Income Countries; LSIL, Low-Grade Squamous Intraepithelial Lesion; mNGS, Metagenomic Next-Generation Sequencing; N/C, Ratio Nuclear-to-Cytoplasmic; NG, Neisseria gonorrhoeae; NGS, Next-Generation Sequencing; PAX1m, PAX1 Methylation; PCR, Polymerase Chain Reaction; qPCR, Quantitative Polymerase Chain Reaction; Rb, Retinoblastoma Protein; TNF-α, tumor necrosis factor-alpha; UU, Ureaplasma urealyticum; WHO, World Health Organization; ZNF582m, ZNF582 Methylation.

Data Sharing Statement

Data sharing is not applicable to this article as no data were created or analysed in this research.

Ethics Statement

This study did not directly involve the testing of human and animal samples; ethical approval is not applicable.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare no conflicts of interest in this work.