Objectives: To investigate whether combining a semi-quantitative scoring method with the immunohistochemical expression of CEA, MIB-1 and p16, would improve the diagnostic accuracy of endocervical glandular lesions.
Methods: The hematoxylin and eosin-stained sections of 95 cervical biopsies were examined by 4 different observers and were grouped into three categories, benign, dysplasia and adenocarcinoma in situ, depending on the degree of nuclear stratification, nuclear atypia and the number of mitosis and apoptotic figures. Each case was also stained immunohistochemically with antibodies against CEA, Ki-67 (MIB-1) and p16. Staining was graded as negative, weak and positive. The accuracy of the scoring method alone was compared to the accuracy of combining the score with the immunostaining results.
Results: Using the semi-quantitative scoring system, most of the cases that were initially diagnosed as atypical hyperplasia or tuboendometrial metaplasia fell into the benign category. This scoring system discriminates effectively (Kruskal-Wallis, p<0.001) between the three categories (benign, endocervical glandular dysplasia and adenocarcinoma in situ). When analyzing the immunohistochemical score, only Ki-67 staining seems to be effective mostly in discriminating between normal glands or glands with atypical hyperplasia and epithelial glandular dysplasia. Ki-67, CEA and p16 failed to discriminate between tuboendometrial metaplasia and epithelial glandular dysplasia. Combining the semi-quantitative scoring system with the immunohistochemical results discriminates between the three categories equally well as the semi-quantitative scoring system alone (Kruskal-Wallis, p<0.001). Nevertheless, the proportion of cases that were classified similarly to the prestudy diagnosis was higher when the combined score was used.
Conclusions: Combining a semi-quantitative scoring scheme with the immunohistochemical expression of CEA, MIB-1 and p16 seems to be of value in classifying some endocervical glandular lesions.