Prospective Evaluation of Cytology, CINtec® and PD-L1 for the Detection of Cervical Intraepithelial Neoplasia: A Single-Center Study.

Background/

Objectives

This study evaluated the diagnostic accuracy of cervical cytology, CINtec® (p16/Ki-67 dual staining), and PD-L1 immunohistochemistry, individually and in combination with high-risk HPV (HR-HPV) testing, for identifying histologically confirmed cervical lesions ranging from CIN1 to invasive carcinoma.

Methods

We conducted a prospective cross-sectional study including 114 patients who underwent cervical cytology, CINtec®, PD-L1 staining, HPV genotyping, and histopathologic confirmation at a tertiary clinical center between September 2024 and September 2025. Sensitivity, specificity, PPV, NPV, and ROC performance were calculated for each test across lesion categories. Multivariable logistic regression models incorporating HR-HPV status were used to assess added predictive value.

Results

All tests showed poor performance for CIN1 (cytology AUC 0.488; CINtec® 0.374; PD-L1 0.366). Diagnostic accuracy improved markedly with lesion severity. For CIN3, CINtec® demonstrated the highest discriminative ability (AUC 0.826), with cytology and PD-L1 also performing well (AUC 0.820 and 0.753). Cytology achieved the strongest ROC performance for CIN2+ (AUC 0.937), CIN3+ (0.913), and invasive carcinoma (0.887). PD-L1 consistently showed lower accuracy across categories. Cytology + HR-HPV demonstrated the highest AUC across all lesion categories.

Conclusions

Cytology and CINtec® exhibited strong diagnostic accuracy for high-grade lesions, while PD-L1 showed limited utility as an independent screening marker. Combining cytology with HR-HPV testing enhanced predictive performance across all lesion categories. These findings support the continued use of cytology-based triage and highlight CINtec® as a valuable adjunct for high-grade disease detection. Because this study used a high-prevalence referral cohort, specificity may be overestimated and not representative of population-based screening.