Comparison of the performance of different HPV genotyping methods for detecting genital HPV types

Classification of high-risk HPV types for cervical cancer screening depends on epidemiological studies defining HPV type-specific risk. The genotyping tests that are used, are however, not uniform with regard to type-specific detection rates making comparisons between different studies difficult. To overcome the lack of a "gold standard" four tests were evaluated crosswise using 824 cervical smears pretested by HC2. The tests evaluated were the L1-PCR-based assays PGMY09/11 LBA, HPV DNA Chip and SPF LiPA and an E1 consensus PCR followed by cycle sequencing (E1-PCR). A subset of 265 samples was tested in addition with the GP5+/6+ reverse line blot assay. Differences were noted in the sensitivity and range for specific HPV types, e.g. with detection rates for HPV53 ranging from 2.3% to 11.6%. HPV16 was the most prevalent type detected by all tests except for the SPF-10 LiPa, which detected HPV31 more often. Kappa values calculated ranged from poor (k = 0.20) to intermediate (k = 0.54) for HPV positivity, but were higher for high-risk type positivity (k = 0.31 -0.61) and best for recognition of HPV16 (k = 0.53-0.72). The analytical sensitivity of the tests ranged between 15% and 97% for individual types and specificity was highly dependent on which test system was used as "gold standard" for the analysis. The results of histology were used for calculation of clinical sensitivity and specificity. E1-PCR, PGMY09/11 LBA and SPF-10 LiPA had a high clinical sensitivity (>95%) for the detection of cervical intraepithelial neoplasia 2 or higher, whereas the HPV DNA Chip reached only 84.1%.