Performance of the linear array HPV genotyping test on paired cytological and formalin-fixed, paraffin-embedded cervical samples

Detection and genotyping of human papillomavirus (HPV) from formalin-fixed, paraffin-embedded (FFPE) samples may be difficult when using assays based on amplification of large fragments. The objective of the present study was to investigate the performance of the Linear Array HPV Genotyping Test (Linear Array) on FFPE cervical cone biopsy specimens using paired cytologic samples obtained immediately before the conization as a criterion standard. Thirty-nine samples of grade 2 or higher cervical intraepithelial neoplasia were selected; all of the corresponding cytological samples were positive by the Linear Array and had a report of atypical squamous cells of undetermined significance or worse. A valid Linear Array test result was obtained for 38 FFPE specimens (97.4%, 95% CI 88.0 to 99.9). Specifically, 34 were HPV-positive (89.5%, 95% CI 76.5 to 96.9) and 4 were HPV-negative (10.5%, 95% CI 3.4 to 23.5). The overall agreement of the results obtained for the cytologic and histologic paired samples was good (Cohen's κ = 0.85, SE = 0.082, P = 0.000). Further analysis of samples with negative or invalid Linear Array test results, both modifying the nucleic acids extraction protocol and using the INNO-LiPA assay, suggested that failure of the Linear Array test in HPV detection from tissues was probably due to DNA fragmentation. Parallel analysis of paired FFPE and cytologic samples is extremely useful for evaluation of the efficiency of PCR-based assays in HPV detection and genotyping from tissue samples. In the present study, false-negative results were obtained in a limited percentage of cases, our data depicting the successful performance of the Linear Array test on FFPE samples.