Comparative Performance of Anyplex II HPV28 and Cobas 4800 Human Papillomavirus (HPV) Assays for High-Risk HPV Detection in Self-collected Anal Samples

Abstract We compared 2 human papillomavirus (HPV) assays to detect the 14 high-risk HPV (hrHPV) genotypes in self-collected anal samples. We found a good agreement and similar performance to detect HPV-16, HPV-18, and the 12 other hrHPV genotypes. The global performance to detect the 14 hrHPV genotypes was not significantly different between the 2 assays.

Around 30 000 new cases of anal squamous cell carcinoma (ASCC) are diagnosed each year worldwide [1].Key populations, such as men who have sex with men (MSM) and people with human immunodeficiency virus (HIV), have an increased risk of developing ASCC.For instance, HIV-infected MSM have the highest incidence rate of ASCC with 85 new cases per 100 000 person-years [2].Human papillomavirus (HPV) is involved in roughly 90% of ASCCs, of which 90% are HPV-16 positive.ASCC is preceded by precancerous lesions-that is, anal intraepithelial neoplasia 1, 2, and 3 (AIN1, AIN2, and AIN3, respectively), with AIN1 being identified as low-grade anal lesions and AIN2 and AIN3 as high-grade anal lesions (HGAIN).However for now, highrisk HPV testing (hrHPV) is not recommended for the screening of anal precancerous and cancer lesions, whereas it has been clinically validated with multiple hrHPV tests in cervical cancer screening [3,4].Although anal HPV-16 genotyping is not yet a widely accepted molecular marker in anal cancer screening, it is by far the most carcinogenetic type in the anus [5].Indeed, HPV-16 prevalence increases across the anal lesion from normal to precancerous and anal cancer [6], and in a recent prospective study, HGAIN HPV-16-positive cases are less likely to clear compared to those with other hrHPV genotypes [7].
In this study, we aim to compare the agreement and performance of the Cobas 4800 HPV test (Roche) and the Anyplex II HPV28 Detection assay (Seegene Inc) to detect hrHPV in selfsampling anal swabs from asymptomatic MSM and transgender women (TGW) in Phnom Penh, Cambodia, especially for HPV-16 detection, which was the most prevalent HPV genotype found in this population [8].

Study Population
A cross-sectional retrospective study was conducted between 1 October 2021 and 30 November 2021 in different clinics in Phnom Penh, Cambodia, and under the scope of activities of Men's Health Cambodia, a nongovernmental and nonprofit organization working with men, MSM, and TGW in Cambodia.During consultation, a questionnaire was used to collect patient data, and anal collection (FLOQSwabs, Copan) was selfperformed using anal swabs discharged in viral transport media (VTM).They were then stored at −80°C at the Institut Pasteur du Cambodge (IPC) until further analysis.A comprehensive explanation of eligibility criteria and procedures has been published previously [8].

HPV Testing and Methods Comparison
First analysis was performed at IPC on 400 μL of sample with the Cobas 4800 HPV test (Roche), which allows the detection of 14 hrHPV types by polymerase chain reaction (PCR) and nucleic acid hybridization.This test specifically identifies HPV-16, HPV-18, and 12 other hrHPV types in the same channel including HPV types 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68.Then, 1 mL of frozen VTM was sent to the Virology Department of the Pitié-Salpêtrière Hospital, Paris, France.DNA extraction (from 200 μL of sample to 100 μL) was performed with the E-mag device (bioMérieux), and both samples and extracted DNA were stored at −80°C.Then, HPV testing was performed with the Anyplex II HPV28 Detection assay (Seegene Inc), which detects and differentiates 28 distinct HPV types including the 14 hrHPV types detected by the Cobas 4800 HPV test and 14 other HPV types (not analyzed in this work).Both HPV tests also detect a human gene as internal control to assess the quality of the sampling and the efficiency of the PCR amplification.If internal control was not amplified, samples were referred as invalid.
Statistical analysis was performed with the package rstatix from R software.McNemar and kappa indices were calculated to compare the proportion of positive samples and the degree of agreement between the 2 methods to detect the 14 hrHPV types, respectively.A 2-tailed P value of <.05 was considered significant.

RESULTS
Overall, 162 participants were included with a median (IQR) age of 28.1 (24.4-31.8)years, of whom 56% were MSM and 8% were HIV infected.None of them received any HPV vaccine.
Further statistical analyses for method comparison were performed on 97 samples, 65 samples being referred as invalid in 1 or both of the 2 methods.The concordance of the results for detection of HPV-16, HPV-18, and other hrHPV type was 97%, 100%, and 88%, respectively (Table 2).Fifteen samples were discordant: (i) 3 for HPV-16 positivity, of which 2 HPV-16 were missed by Roche and 1 missed by Seegene; and (ii) 12 for other hrHPV positivity, of which 8 hrHPV types were missed by Roche and 4 by Seegene (Supplementary Table 1).HPV-52 was the most common hrHPV type missed by Roche (6/10 cases), followed by HPV-16 and HPV-56 (2/10 cases each).For most of the hrHPV types missed by Roche, they were detected tardily by Seegene, after 40 cycles of amplification.

DISCUSSION
To our knowledge, this is the first study conducted in selfcollected anal samples that compared the Cobas 4800 HPV test (Roche) and the Anyplex II HPV28 Detection (Seegene) tests for hrHPV detection.
First, comparing the 2 methods on 97 samples revealed a good agreement between them and comparable worldwide ability to identify any of the 14 hrHPV genotypes currently recommended for cervical cancer screening.This result was also consistent for HPV-16 alone, HPV-18 alone, and the other hrHPV genotypes detected with the 2 methods.However, compared to the Seegene approach, the Cobas test missed more anal hrHPV infections.In term of hrHPV prevalence in normal cervical tissue, a difference was also identified between the Cobas method and GP5 + /GP6 + PCR testing followed by genotyping with Luminex [4].However, equal frequency was showed in samples with abnormal cytology.In our work, we dealt with anal samples and we did not assess the association with cytology  results; thus, these 2 factors may have an impact on the difference between the 2 procedures.The prevalence of HPV-16 was the highest in our high-risk population, as expected, and the 2 methods showed high agreement and performance to detect the most oncogenic genotype in the anal site.This is of importance as HPV-16 testing could be one of the biomarkers used as anal cancer screening in key populations in the near future.For instance in France, new guidelines propose to use HPV-16 testing in anal cytology specimen collected by physicians as an anal cancer screening marker in MSM living with HIV [9].If positive, a reflex cytology is performed on the same sample and patient is referred to a proctologist.
Similarly to self-collected vaginal swabs for cervical cancer screening [10], self-collected anal samples could be of interest to reach patients outside the healthcare system [11].In our work, 41 (25%) and 59 (36%) samples failed to be amplified by the Anyplex II HPV28 Detection assay and Cobas 4800 HPV test, respectively.Low cellularity due to inappropriate sampling and misunderstanding of the participants may explain this high rate of invalid results.If implemented in this population, self-collection instructions should be improved.
In conclusion, both the Anyplex II HPV28 Detection assay and the Cobas 4800 HPV test are suitable to detect the 14 hrHPV genotypes, in particular HPV-16, in self-collected anal samples.

Table 2 . Agreement and Performance of the 2 Methods to Detect Human Papillomavirus (HPV) Types 16 and 18 and Other High-Risk Types
a Sixty-five samples being referred as invalid in 1 or both HPV tests.b