Affiliations: Department of Pathology, Rogaland Central Hospital (SiR), Stavanger, Norway | Free University, Amsterdam, The Netherlands
Note: [] Corresponding author: Prof. Jan P.A. Baak, Department of Pathology, Rogaland Central Hospital (SiR), PO Box 8100, 4068 Stavanger, Norway. Tel.: +47 51 519378; Fax: +47 51 519910; E‐mail: [email protected].
Abstract: Aim: To assess the correlation between syntactic structure analysis (SSA) features, revised dysplasia grade and the presence of high‐risk human papillomavirus DNA in cervical intraepithelial neoplasia (CIN). Materials and methods: HPV polymerase chain reaction (PCR) was assessed in 101 consecutive biopsies and consensus in CIN grade between the experts occurred in 88 cases (CIN1=16, CIN2=27, CIN3=45). SSA was performed in the diagnostic histological section of the CIN lesions in these patients and SSA features were compared with the blind review CIN grade, and presence/absence of high‐risk HPV DNA. Results: One of the SSA features (points from which the surrounding surfaces has 4 edges, PECO‐4) was significantly different between all three consensus CIN grades. Many more features revealed significant differences between CIN1 and CIN2 or between CIN2 and CIN3 cases. With stepwise discriminant analysis, the best multivariate combination of features to distinguish the different CIN grades were the Maximum MST Line Length (MML) and the Area Disorder. Crude overall classification of the consensus grades with these features was 69%. The MML and the Area Disorder is also the best combination to distinguish cases with and without high‐risk HPV DNA (77% correct classifications). Conclusions: SSA features are correlated with both CIN grade and presence of high‐risk HPV DNA, but the discrimination power is not good enough to be used as a routine method for quality control of subjective grade or as a surrogate marker for high‐risk HPV DNA presence.
