The objective of this study is to evaluate the analytical performance of three human papillomavirus (HPV) genotyping methods on cervical liquid based cytology (LBC) samples. HPV genotyping is primarily important for epidemiological research and vaccine surveillance, but is increasingly also used for clinical purposes.
Consecutive SurePath LBC samples with diagnosis ASC-US or LSIL from routine screening are included in the study (n=200). HPV genotyping methods to be compared are: A method based on MGP-PCR followed by hybridization (Luminex technology) detecting and genotyping 37 HPV types; Anyplex HPV28 (Seegene) detecting and genotyping 28 types; and HPV next generation MGP amplicon sequencing using the Illumina MiSeq desktop sequencer. Cobas 4800 HPV test results (clinical test detecting 14 high-risk types) are available for the samples, allowing a relative comparison of the non-clinical genotyping test results to clinical test results.
To date, all samples have been analysed with Luminex and Anyplex HPV28. In total, 168 samples (84%) were found positive for one or more HPV types; 32 samples (16%) were negative by both methods. A general observation is that the Anyplex assay has a tendency to detect more types than Luminex; up to eight types are reported in one sample. Altogether, 35 of the 38 genotypes covered by the two methods were identified. When comparing results for the 27 shared HPV types, a good overall agreement is observed. Considering the positive agreement, this is found low (<60%) for 9 of the 27 types. The negative agreement is good (well above 90%) for all types. Kappa values vary from 0.95 (0.90; 1.00) for HPV16 to 0.19 (0.003; 0.37) for HPV54. According to matrix correlation (RV) between Luminex and Anyplex, the measurements correlated moderately when considering all 27 types (RV=0.66), strongly among the 14 high-risk types (RV=0.79), and moderately among the remaining 13 types (RV=0.42). Of the 168 positive samples, 34 were negative with the Cobas test, of which 18 were identified with types other than the 14 included in the Cobas test.
For the two genotyping assays, a good correlation in terms of positive and negative samples is seen. Considering multiple infections, Anyplex HPV28 has a tendency to detect more types than the Luminex method. Both methods have a higher analytical sensitivity compared to the clinical Cobas 4800 HPV test. Correlations also including data from next generation amplicon sequencing will be presented.