Endocervical glandular cell abnormalities in conventional cervical smears: evaluation of the performance of cytomorphological criteria and HPV testing in predicting neoplasia
Version of Record online: 15 JAN 2008
Volume 19, Issue 1, pages 34–43, February 2008
How to Cite
Rabelo-Santos, S. H., Derchain, S. F. M., Do Amaral Westin, M. C., Angelo-Andrade, L. A. L., Sarian, L. O. Z., Oliveira, E. R. Z. M., Morais, S. S. and Zeferino, L. C. (2008), Endocervical glandular cell abnormalities in conventional cervical smears: evaluation of the performance of cytomorphological criteria and HPV testing in predicting neoplasia. Cytopathology, 19: 34–43. doi: 10.1111/j.1365-2303.2007.00466.x
- Issue online: 15 JAN 2008
- Version of Record online: 15 JAN 2008
- Accepted for publication 19 December 2006
- atypical glandular cells;
- adenocarcinoma in situ;
- squamous neoplasia;
- glandular neoplasia;
- conventional cytology;
- cervical intraepithelial neoplasia
Objective: To analyse the correlation between cytomorphological criteria in smears with atypical glandular cells (AGC) or adenocarcinoma in situ (AIS) and human papillomavirus (HPV) reflex test results with different neoplastic histological diagnoses, particularly to distinguish between glandular and squamous neoplasia.
Methods: A series of 155 women with glandular abnormalities in their conventional cervical smears was included: 106 with AGC, 35 with AGC associated with high-grade squamous intraepithelial lesion (HSIL) and 14 with AIS. Two reviewers evaluated 35 cytomorphological criteria and hybrid capture II (HCII) was performed in all cases. Colposcopy was carried out in all cases and biopsy in 126/155. For statistical purposes, predictive values and odds ratio (OR) were calculated, followed by chi-square automatic interaction detection.
Results: Histology detected 56 cases of squamous and 17 of glandular intraepithelial or invasive neoplasia. Predictive values of the papillary groups and feathering criteria for glandular neoplasia were, respectively, 80.0% and 73.3%. Feathering was the criterion with the highest OR for distinguishing glandular from squamous neoplasia and also for distinguishing between glandular and non-neoplastic diagnosis. Rosettes and pseudostratified strips did not perform as well. Multivariant Classification and Regression Trees analysis identified feathering as the best criterion for distinguishing between glandular, squamous and non-neoplastic diagnoses regardless of HPV status.
Conclusions: Feathering was the best criterion for predicting glandular neoplasia.