Optimising risk stratification for atypical glandular cells: insights from a large Chinese gynaecologic cohort

Aims To optimise risk-stratified management of atypical glandular cells (AGC) by evaluating risk factors for cervical and endometrial pathology. Methods We retrospectively analysed 2460 AGC patients without prior malignancy (2007–2024), examining AGC subtypes, age, high-risk human papillomavirus (hrHPV) genotypes, ultrasound findings and histological outcomes. Results Atypical endometrial cells (AEM) predicted endometrial lesions (30.3%, 95% CI 25.9% to 35.1%) versus other subtypes (5.4%, 95% CI 4.2% to 6.9%, p<0.001). AGC endocervical cells, favour neoplastic (AEC-FN), and AGC, favour neoplastic (AGC-FN) had higher significant lesion risks than non-neoplastic AGC (69.8%, 95% CI 62.9% to 75.9% vs 21.4%, 95% CI 19.2% to 23.8%, p<0.001). Age ≥45 plus positive ultrasound improved endometrial prediction (AUC 0.825 vs 0.732, p<0.001). HPV18 and HPV16 conferred highest cervical risks (68.8%, 95% CI 58.3% to 77.8% and 50.3%, 95% CI 42.3% to 58.3%), with HPV18 linked to cervical glandular lesions (58.1%, 95% CI 47.4% to 68.2%). hrHPV-negative patients with significant lesions predominantly had endometrial pathology (69.2%, 95% CI 61.6% to 76.0%). Conclusions AGC subtyping provides clinically meaningful risk stratification: AEM indicates endometrial evaluation; AEC-FN and AGC-FN indicate cervical evaluation. Age and ultrasound may refine endometrial risk assessment. HPV genotyping guides cervical evaluation but endometrial assessment remains important in hrHPV-negative patients. These findings suggest a potential approach requiring prospective validation before clinical application.