Comparison of the analytical and clinical performance of five tests for the detection of human papillomavirus genital infection

HPV-based screening provides greater protection against cervical cancer (CC) than cytology-based strategies. Currently, several molecular diagnostic assays for the detection of human papillomavirus (HPV) are available. In this study, we analyzed 5 different HPV testing and genotyping techniques (Hybrid Capture 2 [HC2; Qiagen, Hilden, Germany], AnyplexTMII HPV28 [Anyplex; Seegene, Seoul, Korea], Linear Array [Roche, Branchburg, NJ, USA], GP5+/6+ PCR-EIA-RH [Labo Bio-medical Products, Rijswijk, The Netherlands] and CLART2 [Genomica, Madrid, Spain]) in 295 women referred to the hospital Colposcopy Clinic from 2007 to 2008 due to positive HPV test results or an abnormal Pap test. DNA extraction for HPV genotyping was performed in cervical sample specimens after Pap test and HPV detection by HC2. The inclusion criteria were: (1) adequate cervical sampling with sufficient material for the Pap test and HPV detection and genotyping, and (2) colposcopically-directed biopsy and/or endocervical curettage. HC2 showed the highest sensitivity for high-grade squamous intraepithelial lesion and CC (HSIL+) detection (96.1%), but all the HPV genotyping tests showed a higher specificity. (Anyplex 86.8%; Linear Array 86.0%; GP5+/6+ 78.8%; CLART2 76.5%). The agreement between HC2 results and the other techniques was similar: 82.4%, kappa = 0.650 for Anyplex; 83.4%, kappa = 0.670 for Linear Array, 79.93%, kappa = 0.609 for GP5+/6+ and 82.4%, kappa = 0.654 for CLART2. HPV 16 and/or 18 infection was a risk factor for underlying HSIL+ in the univariate analysis. Anyplex showed the highest risk of underlying HSIL+ after positive HPV 16 and/or 18 tests (OR 31.1; 95% CI 12.1-80.0).