Abstract
BACKGROUND: Identification of human papillomavirus (HPV) DNA in cervical tissue is important for understanding cervical carcinogenesis and for evaluating cervical cancer prevention approaches. However, HPV genotyping using formalin-fixed, paraffin-embedded (FFPE) tissues is technically challenging. We evaluated the performance of four commonly used genotyping methods on FFPE cervical specimens conducted in different laboratories and compared to genotyping results from cytological samples. METHODS: We included 60 pairs of exfoliated-cell and FFPE specimens from women with histologically confirmed cervical intraepithelial lesions grade 2 or 3. Cytology specimens were genotyped using the Linear Array assay. Four expert laboratories processed tissue specimens using different preparation methods and then genotyped the resultant sample preparations using four different HPV genotyping methods: SPF10-PCR DEIA LiPA25 (version 1), Inno-LiPA, Linear Array and the Onclarity assay. Percentage agreement, kappa statistics and McNemar's chi-square were calculated for each comparison of different methods and specimen types. RESULTS: Overall agreement with respect to carcinogenic HPV status for FFPE samples between different methods was: 81.7, 86.7 and 91.7% for Onclarity versus Inno-LiPA, Linear Array and SPF-LiPA25, respectively; 81.7 and 85.0% for Linear Array versus Inno-LiPA and SPF-LiPA25, respectively; and 86.7% for SPF-LiPA25 versus Inno-LiPA. Type-specific agreement was >88.3% for all pair-wise comparisons. Comparisons with cytology specimens resulted in overall agreements from 80 to 95% depending on the method and type-specific agreement was >90% for most comparisons. CONCLUSIONS: Our data demonstrate that the four genotyping methods run by expert laboratories reliably detect HPV DNA in FFPE specimens with some variation in genotype-specific detection.