Integrating scRNA-Seq and Spatial Transcriptomics Reveal Novel Risk Stratification Signatures in Cervical Carcinoma

Cervical cancer is the fourth most common malignancy in women worldwide. Biomarker-assisted screening may improve diagnostic accuracy, but biomarkers predicting progression of low-grade squamous intraepithelial lesions (LSIL) remain limited. We integrated single-cell RNA sequencing and spatial transcriptomics to identify prognostic biomarkers associated with LSIL outcomes, followed by validation using improved in situ sequencing (IISS). Analysis of 341,978 single cells identified eight major cell populations and twelve epithelial subclusters, including a previously unrecognized subcluster, epi8. Although mainly derived from normal cervix, epi8 exhibited cancer-associated features and shared molecular characteristics with the prognostic biomarkers FAM19A4 and miR124-2. In situ sequencing showed that epi8 localized adjacent to human papillomavirus-infected basal cells, with stage-dependent spatial changes during disease progression. Retrospective analysis further demonstrated that epi8-associated markers stratified LSIL into progression, persistence, and regression groups, with CRYAB^high/GAS5^high associated with progression and RPLP1^high/SNHG5^high associated with regression. We identified a previously unrecognized epithelial subcluster associated with early cervical carcinogenesis and demonstrated that its molecular markers enable risk stratification of LSIL, providing candidate biomarkers for precision triage in cervical cancer screening.