Targeted Next Generation Sequencing for Human Papillomavirus Genotyping in Cervical Liquid-Based Cytology Samples

Simple Summary Testing for Human Papillomavirus (HPV) is currently being implemented as part of cervical cancer (CC) screening in several countries. However, infections with all but one of the HPV types classified as possibly carcinogenic cannot be detected by the assays used for CC screening today. The aim of our study was to demonstrate the use of a targeted next generation sequencing (NGS) HPV panel for CC screening—both general practitioner-collected samples and self-samples. We here show that the targeted HPV panel can detect HPV with a sensitivity and specificity similar to commercial HPV assays, one of which is used for CC screening today. However, the targeted HPV panel possess several advantages compared to the screening assays, as it enables specific detection of all relevant HPV types and can identify viral integration, variants in the HPV genome, and dominant HPV types in multi-infected cases. Abstract At present, human papillomavirus (HPV) testing is replacing morphology-based cytology as the primary tool for cervical cancer screening in several countries. However, the HPV assays approved for screening lack detection for all but one of the possibly carcinogenic HPV types and do not genotype all included HPV types. This study demonstrates the use of a targeted HPV next generation sequencing (NGS) panel to detect and genotype all 25 carcinogenic, probably carcinogenic, and possibly carcinogenic HPV types as well as the low-risk types HPV6 and HPV11. The panel was validated using a cohort of 93 paired liquid-based cytology samples (general practitioner (GP)-collected cervical samples and cervico-vaginal self-samples (SS)). Overall, the targeted panel had a sensitivity (GP = 97.7%, SS = 92.1%) and specificity (GP = 98.0%, SS = 96.4%) similar to the commercial HPV assays, Cobas® 4800 HPV DNA test (Roche) and CLART® HPV4S assay (GENOMICA). Interestingly, of the samples that tested positive with the NGS panel, three (6.4%) of the GP-collected samples and four (9.1%) of the self-samples tested positive exclusively for HPV types only included in the NGS panel. Thus, targeted HPV sequencing has great potential to improve the HPV screening programs since, as shown here, it can identify additional HPV positive cases, cases with HPV integration, variants in the HPV genome, and which HPV type is dominant in multi-infected cases.


Introduction
Persistent infection with high-risk human papillomavirus (HPV) is the main cause of cervical cancer development [1]. Several HPV types are known to cause cervical cancer self-samples collected at home were selected (median no. of days between samples: 43, range: 13-95 days). Both GP-collected samples and self-samples from each woman in the study cohort had been analyzed with Cobas ® 4800 (Roche, Hvidovre, Denmark) and CLART ® HPV4S (GENOMICA, Madrid, Spain) for use in previous publications, and the recruitment procedure as well as the sample collection, processing, storage, and analysis have been described elsewhere [33,34].
For each of the 27 HPV types included in the NGS panel, eight amplicons were designed. Two amplicons were designed for each of the E6 and E7 oncogenes and, for comparison with the L1-detecting analyses such as Cobas ® 4800 and CLART HPV4S, one amplicon was designed for L1. Additionally, three amplicons were designed for E2, whereof two were in the E2 region called the hinge-region [35]. E2, and particularly the hinge-region, was covered by amplicons, since this is a hotspot for integration of HPV into the host genome resulting in the deletion of this part of the viral genome [6,36,37].
To confirm the presence of DNA in each sample, an amplicon for each of five human reference genes (BTF3, PABPN1, PPIE, RAB1B, SRSF3) were included in the panel as well. How to select appropriate human reference genes is described in a previous publication [32]. The panel was designed with short amplicons of approximately 90 base pairs using the circulating free DNA (cfDNA) pipeline of the AmpliSeq Designer.
The NGS panel was validated using DNA from paired GP-collected cervical samples and cervico-vaginal self-samples. The DNA had been purified for a previous study using the Cobas x480 Instrument (Roche, Hvidovre, Denmark) and analyzed using the Cobas ® 4800 and CLART ® HPV4S assays [33,34].

Preparation of Samples and Sequencing
Excess DNA, purified for Cobas ® 4800 as per routine in the cervical cancer screening program, was used for the NGS assay. The samples were prepared for sequencing using the Ion AmpliSeq Library Kit 2.0 (Thermo Fisher Scientific, Roskilde, Denmark) according to the manufacturer's protocol, except that the reactions were performed in half volume. Input DNA was in the range of 10-60 ng for the self-samples. Due to limited material, the input DNA concentration was not measured for the GP-collected samples. However, the DNA concentrations were measured for six comparable GP-collected samples, not used for NGS, from the original cohort of 213 women. The DNA concentrations of these samples were very low (mean: 2.42 ng/µL, range: 0.38-6.95 ng/µL), indicating very low DNA concentrations of the GP-collected samples included in this study as well. Therefore, 6 µL of undiluted DNA was used for library preparation of the GP-collected samples. The GP-collected samples were labelled with Ion Xpress Barcode Adapters (Thermo Fisher Scientific, Roskilde, Denmark), but in an attempt to reduce the amount of background, the self-samples were labelled with Ion Torrent Dual Barcodes (Thermo Fisher Scientific, Roskilde, Denmark). Before pooling the libraries for template preparation and sequencing, the libraries were quantified using the Ion Library TaqMan Quantitation Kit (Thermo Fisher Scientific, Roskilde, Denmark) and diluted to 50 pM. Template preparation was performed using the Ion 510 & Ion 520 & Ion 530 Kit-Chef (Thermo Fisher Scientific, Roskilde, Denmark) on the Ion Chef System (Thermo Fisher Scientific, Roskilde, Denmark).
Sequencing was performed using the Ion GeneStudio S5 System (Thermo Fisher Scientific, Roskilde, Denmark) with the coverage analysis plugin set to a minimum amplicon length of 50 base pairs.

NGS Data Analysis
The sequencing data from each sample were aligned to a custom reference sequence composed of the human reference genome (hg19) and the genomes of 286 HPV types. The applied HPV reference genomes have been used in a former study [32], where a complete list of the accession numbers can be found. The data, generated by the S5 Torrent Server (Thermo Fisher Scientific), were delivered as a coverage analysis Excel file containing the total number of reads for each amplicon, including both HPV amplicons and amplicons of the five human reference genes for all samples. The inclusion of the five human reference genes were to ensure that an HPV negative sample was indeed negative and not the result of a flawed analysis; thus, all five genes should be amplified and sequenced for an HPV negative sample to be valid. Based on the reads of the E6 and E7 HPV16 amplicons, a cut-off was calculated. Since HPV16 was the most prevalent HPV type and the HPV type with the highest number of reads, the cut-off was based on this type, creating the most background reads. Since both E2 and L1 have previously been reported to be deleted following integration [6,32,38,39], the cut-off was based solely on the E6 and E7 amplicons. The amount of background of the E6 and E7 HPV16 amplicons was assessed using the samples that had tested HPV16 negative in both the Cobas ® 4800 and CLART ® HPV4S assays, since the HPV16 reads present in these samples would correspond to background reads. The cut-off was defined as the mean of the HPV16 E6 and E7 amplicons' limit of detection (LoD). The LoD for each HPV16 E6 and E7 amplicon was calculated as follows: LoD = mean blank + (2 * SD blank ). To render a woman HPV positive for a given HPV type, the number of reads for all four E6 and E7 amplicons had to be above the cut-off. However, if the number of reads for three out of four of the E6 and E7 amplicons were high (>500 reads), the woman was rendered HPV positive as well. Poor quality amplicons, i.e., amplicons with read depth of <1% of the total amount of reads on the specific HPV type in at least 25% of the HPV cases positive for that given HPV type, were not included when applying the cut-off to different HPV types.

Statistics
The NGS, Cobas ® 4800, and CLART ® HPV4S assays were compared regarding sensitivity and specificity for HPV detection in general as well as for HPV16 detection. Each assay was compared to the two other assays to assess a "true/consensus" HPV result, which means that when at least two of the assays showed the same HPV result, this result was interpreted to be correct. For example, if the NGS assay did not detect any HPV types/HPV16, but the two other assays detected HPV/HPV16, the result of the NGS assay was interpreted as being false negative. The three assays were compared only for the HPV types detectable with all three HPV assays (HPV16, 18,31,33,35,39,45,51,52,56,58,59,66, and 68). Sensitivity, specificity, and corresponding 95% confidence intervals (CI) were calculated for each of the three HPV assays for both GP-collected samples and self-samples.
The specific HPV types identified by the NGS assay in the study cohort were compared to the HPV types identified by CLART ® HPV4S. The Cobas ® 4800 assay was excluded from this comparison, as this assay only genotypes HPV16 and HPV18. The NGS and CLART ® HPV4S assays were compared only for the HPV types detectable with both HPV assays (HPV6, 11,16,18,31,33,35,39,45,51,52,56,58,59,66, and 68). The degree of agreement between the NGS assay and the CLART ® HPV4S assay was categorized as follows: Agreement was defined as detection of the exact same HPV types in the NGS assay and in CLART ® HPV4S. Partial agreement was applied in samples infected with multiple HPV types and defined as detection of at least one of the same HPV types in the NGS assay and CLART ® HPV4S. Disagreement was defined as detection of none of the same HPV types between the NGS assay and CLART ® HPV4S. The percentage of agreement between NGS and CLART ® HPV4S was calculated as the number of samples with agreement divided by the total number of samples.

Ethical Approval
According to the EUs General Data Protection Regulation, the project was listed at the record of processing activities for research projects in the CDR (journal No.: 1-10-72-69-15). The study was approved by the local Ethical Committee of the Central Denmark Region (journal No.: 1-16-02-209-15).

HPV Types GP-Collected Cervical Samples, n (%)
Cervico-Vaginal Self-Samples, n (%) 6 1 (1. Several possibly carcinogenic HPV types only included in the NGS assay were identified (HPV30, 53, 67, 69, 70, 73, and 82) (Table S1). One or more of these types were detected in 16 (34.0%) of the HPV positive GP-collected samples and in 18 (40.9%) of the HPV positive self-samples. Interestingly, out of the samples that tested positive with the NGS assay, three (6.4%) of the GP-collected samples and four (9.1%) of the self-samples tested positive exclusively for HPV types only included in the NGS assay (HPV30, 53, 67, and 70). Thus, these were not detectable with the Cobas ® 4800 or CLART ® HPV4S assays (patient No. 64, 92, and 211 for the GP-collected samples and patient No. 21, 107, 130, and 211 for the self-samples). However, two of the three mentioned GP-collected samples had a Cobas ® 4800 HPV positive result anyway. For these two samples, one of them tested HPV16 positive using Cobas ® 4800, but since HPV16 was not detected neither by any other assay nor by Cobas ® 4800 itself in the self-sample; this is most likely a false positive result. The other sample tested positive for an unspecified HPV type using Cobas ® 4800. Since the self-sample from this patient tested HPV positive using the three different HPV assays, it is difficult to conclude whether this is a false positive sample or not. However, Cobas ® 4800 has been observed to cross-react with HPV types not included in the assay [40]. This includes cross-reactivity for HPV70, which is present in both of the two mentioned samples when tested with NGS.

Concordance in HPV Genotyping
The specific HPV types genotyped using the NGS assay were compared to the CLART ® HPV4S genotyping results. Among the GP-collected samples, 75 (80.7%) were categorized as agreement, ten (10.8%) as partial agreement, and eight (8.6%) as disagreement. For the self-samples, 78 (83.9%) showed agreement, eight (8.6%) showed partial agreement, and seven (7.5%) showed disagreement ( Table 3). Two of the disagreement cases in the GP-collected samples were caused by detection of HPV only when using CLART ® HPV4S while the other six were caused by HPV detected only when using the NGS assay. The Cobas ® 4800 assay agreed with the NGS assay in detection/no detection of HPV in six of the disagreement cases and with the CLART ® HPV4S assay in the two other disagreement cases. For the self-samples showing disagreement, HPV was detected solely by CLART ® HPV4S in three samples and solely by NGS in the other four samples showing disagreement.
Here, the Cobas ® 4800 assay agreed with the NGS assay in detection/no detection of HPV in one case and with the CLART ® HPV4S assay in five of the disagreement cases. The last disagreement case in the self-samples showed detection of HPV6 using NGS. Since this HPV type is not included in the Cobas ® 4800 assay, Cobas ® 4800 could not be used for comparison in this case. The detected HPV types for each assay are presented in Table  S1 for both GP-collected samples and self-samples. Table 3. Degree of agreement between the NGS assay and CLART ® HPV4S in GP-collected cervical samples and cervico-vaginal self-samples.

GP-Collected Cervical Samples
Cervico-Vaginal Self-Samples 1 Agreement was defined as detection of the exact same HPV types in the NGS assay and CLART ® HPV4S. 2 Partial agreement was applied in samples infected with multiple HPV types and defined as detection of at least one of the same HPV types in the NGS assay and CLART ® HPV4S. 3 Disagreement was defined as detection of none of the same HPV types between the NGS assay and CLART ® HPV4S. GP: General practitioner.

Sensitivity and Specificity for HPV16 Detection
Since HPV16 is genotyped by all three assays, the sensitivity and specificity for detection of HPV16 were assessed as well. Moreover, HPV16 is the most frequent HPV type and accounts for approximately 60% of cervical cancers [41]. Due to few HPV16 positive samples in the relatively small study cohort, the GP-collected samples and self-samples were pooled. The samples were independent of each other, taken on different days and from different areas of the cervix and vagina. The sensitivity of HPV16 detection in the NGS, Cobas ® 4800, and CLART ® HPV4S assays was 100.0% (95% CI: 81.5-100.0%), 100.0% (95% CI: 81.5-100.0%), and 83.3% (95% CI: 58.6-96.4%), respectively. The specificity of HPV16 detection using the same assays was 98.8% (95% CI: 95.8-99.9%), 95.8% (95% CI: 91.6-98.3%), 100.0% (95% CI: 97.8-100.0%), respectively (Table 4). Table 4. Sensitivity and specificity for detection of HPV16 in the NGS assay, Cobas ® 4800, and CLART ® HPV4S assay on GP-collected cervical samples and cervico-vaginal self-samples 1 using a consensus * as comparison. The general practitioner-collected cervical samples and the cervico-vaginal self-samples were pooled to obtain a larger cohort. The two samples were taken independent of each other on different days and from different areas of the cervix/vagina. * Each assay was compared to the two other assays to assess a consensus result, meaning when at least two of the three assays showed the same HPV result (HPV16 positive or HPV16 negative) this result was interpreted as the consensus result. GP: General practitioner.

Assessment of the Viral Integration State
A former study published by our group showed the ability to detect an episomal as well as an integrated HPV16 viral state [32]. As described earlier, integration of HPV into the host genome often occurs in the E2 hinge-region, and therefore, in the NGS panel designed for this study, E2 amplicons were designed for each of the 27 included HPV types. Since the absence of E2 amplicons indicates HPV integration, this enables an assessment of whether the sample has episomal or integrated HPV. The HPV18 positive patient no. 96 showed a clear integration pattern with no reads in any of the three included E2 amplicons of HPV18 as depicted in Table 5. The patient no. 96 depicted in Table 5 was further positive for a non-integrated HPV31. The lower number of reads in the self-sample (Table 5) was a trend we saw for the vast majority of paired samples and might be due to the general practitioner sampling more precisely from the cervix compared to self-sampling. This illustrates the NGS assay's ability to detect viral integration also in HPV types different from HPV16. However, this panel is designed with the main purpose of HPV genotyping, thus only containing few regions of each HPV type to ensure high coverage at low cost. Since studies have shown that integration can cause breakpoints at various sites of the HPV genome not covered in this NGS panel [6,38,42], a panel with the main purpose of assessment of the viral integration state should contain additional regions of each HPV type. The upper five amplicons are included human reference genes and the eight lower amplicons cover the HPV18 genome. GP: General practitioner.

Detection of HPV Variants
When a variant caller was applied to the data, the NGS assay detected multiple genomic variants of the different HPV types included (Table S2). The number of variants detected for each HPV type ranged from one (HPV35) to 13 (HPV66) variants. By detection of different HPV variants, determination of HPV sub-lineages may be possible. However, this NGS panel was not designed for determination of HPV sub-lineages, and thus, the included regions of each HPV type do not necessarily contain the genomic positions for determination of sub-lineages of the identified HPV types.

Discussion
Our NGS assay is a semi-quantitative method that can detect and genotype HPV and includes all 25 carcinogenic, probably carcinogenic, and possibly carcinogenic HPV types as well as the low-risk HPV types HPV6 and HPV11. When comparing the sensitivity and specificity of HPV detection using the NGS, Cobas ® 4800, and CLART ® HPV4S assays, the three assays showed a similar performance both when examining GP-collected samples and self-samples. The same was observed for specific detection of HPV16 when examining both sample types. The specific HPV types genotyped with NGS showed good concordance with the CLART ® HPV4Sassay, where agreement was obtained in 80.7% of GP-collected samples and 83.9% of self-samples. Furthermore, our NGS assay detected viral integration in an HPV18 positive sample and multiple genomic variants of the different detected HPV types. Interestingly, of the samples that tested positive with the NGS assay, three (6.4%) of the GP-collected samples and four (9.1%) of the self-samples tested positive exclusively for HPV types only included in the NGS assay (HPV30, 53, 67, and 70).
Since multiple studies have shown carcinogenicity of the above-mentioned HPV types HPV53, 67, and 70 as well as the HPV types HPV26, 73, and 82 [43][44][45] whereof all are classified as possibly carcinogenic, inclusion of these HPV types in HPV assays approved for screening should be considered. Especially since several countries are currently changing their cervical cancer screening programs to primary screening for HPV rather than cytological detection of cervical dysplasia [13,14,17,18]. Additionally, with the introduction of HPV testing as the primary screening tool, a change in the screening interval from every 3rd year to every 5th or 10th year has been proposed [14,16]. This is because HPV testing is believed to increase the sensitivity of high-grade cervical intraepithelial neoplasia detection in screening [12,16]. Thus, it becomes important to minimize the number of false negatives. However, the HPV assays approved by the U.S. Food and Drug Administration (FDA) only cover 13-14 HPV types and thus all lack detection of 11-12 HPV types classified as possibly carcinogenic. This includes the HPV types HPV30, 53, 67, and 70, which in our study cohort were detected in some patients without detection of any of the HPV types included in screening assays today. Furthermore, the FDA-approved assays are not able to distinguish all detectable HPV types from each other, i.e., the assays cannot genotype all included HPV types [46]. The lack of possibly carcinogenic HPV types and HPV genotyping in the FDA-approved assays makes it difficult to differentiate between women with same genotype persistence and genotype switch. Differentiation of these two groups of women is important, since only persistent HPV infections can develop to cervical cancer [1,47], and this is enabled by our NGS assay with the inclusion and genotyping of all carcinogenic, probably carcinogenic, and possibly carcinogenic HPV types.
Since an increase of colposcopy referrals have been observed with the shift to HPV detection in cervical cancer screening [13,48], an introduction of all possibly carcinogenic HPV types into cervical cancer screening could potentially result in an additional increase of colposcopy referrals. The reason for this is that most HPV infections clear spontaneously without the need of any treatment. The dilemma of detecting and preventing more cervical cancer cases at the expense of increased colposcopy referrals could potentially be solved by better risk stratification of HPV positive women. Our NGS assay can potentially help solve this problem with its ability to genotype the 25 carcinogenic, probably carcinogenic, and possibly carcinogenic HPV types and, in addition, assess the viral integration state and detect viral variants while being semi-quantitative. Since the NGS assay is semiquantitative, it gives an indication of which HPV type is dominant in women with multiple HPV infections and can, moreover, be used as a measure of the viral load in cervical biopsies. Viral load is influenced by factors such as sampling, age, infection with multiple HPV types, and area of cervical lesion [49] but a high viral load has been related to highgrade cervical dysplasia [7,8]. The same has been observed for the viral integration status where viral integration has been linked to increased risk of progression of cervical dysplasia to cervical cancer [7][8][9]. Moreover, HPV types can be classified into lineages (that diverge by 1-10% in sequence) and sub-lineages (that diverge by 0.5-1% in sequence) [50] and specific HPV sub-lineages have been linked to persistent infection and increased risk of cervical cancer development as well [51][52][53][54]. Further, the risk of precancer/cancer for specific sub-lineages have been reported to depend on the women's race/ethnicity where a race/ethnicity matching that of the infecting HPV16 sub-lineage increased the risk of CIN3+ [51]. With the NGS assay's ability to detect viral variants, amplicons could be designed to cover specific areas that would make sub-lineage classification possible. Lastly, some specific HPV variants have been related to persistent infection and increased risk of progression to high-grade cervical dysplasia or cervical cancer as well [53,54] and these specific variants could, therefore, also be covered by our NGS assay. Thus, the NGS assay performs as well as Cobas ® 4800 and CLART ® HPV4S in sensitivity and specificity for HPV detection while being semi-quantitative and being able to assess the viral integration status as well as detect viral variants.
Besides the above-mentioned features, our NGS assay detects several genomic regions of each included HPV type in the oncogenes E6 and E7 as well as E2 and L1. Both Cobas ® 4800 and CLART ® HPV4S, as well as several other HPV assays, detect only a region of the L1 gene [30]. However, studies have shown the possibility of L1 deletion following viral integration [38] which creates the risk of false negative HPV results in HPV assays only detecting L1 [32,55,56]. The fact that both Cobas ® 4800 and CLART ® HPV4S detect L1 poses the possibility that these two assays might be more likely to agree with each other compared to the NGS assay detecting several regions of the HPV genomes. This is a disadvantage of the comparison method used in this study where the consensus HPV results were based on HPV detection agreement between at least two of the assays. This could potentially result in a false negative consensus HPV result if both Cobas ® 4800 and CLART ® HPV4S do not detect any HPV due to L1 deletion. However, this does not seem to be the case in our cohort, which could be due to a low number of integrated cases. Since viral integration often occurs during progression of cervical dysplasia to cancer [6][7][8][9]57] and all samples analyzed in this study were ASC-US samples, the detection of a clear integration pattern in only one sample included in this study cohort was not a surprise. However, integration has been observed in several cervical cancer samples analyzed with the same NGS assay for another, yet unpublished, study in our group. Even though viral integration often results in deletion of part of the E2 gene, deletion can occur at various sites of HPV [6,38,42]. Therefore, to ensure detection of viral integration, a panel with the main purpose of assessment of the viral integration state should contain additional regions of each HPV type. The same would be the case if classification of sub-lineages should be achieved. However, this would increase both the cost and the analysis time, which may not be optimal in a screening setting. As long as the panel size is kept small, we have previously estimated a price of around 75 EUR per sample, which is within the range of PCR-based HPV analyses [32]. Despite this, a larger NGS panel could be relevant in specific HPV positive cases for further risk stratification.
As previously mentioned, because the NGS assay can detect and genotype all relevant oncogenic HPV types, it enables establishment of whether a patient has same genotype persistence or genotype switch in successive cervical samples [47]. This contrasts with the HPV assays approved for screening today where only few HPV types are genotyped. Hereby, it is not possible to confirm whether two consecutive HPV positive screening results are caused by a persistent infection or two independent infections. However, with the development of a larger NGS panel, additional variants for each HPV type could help additionally in differentiating between persistent and transient infections and thus help to avoid overtreatment of transient infections thought to be persistent.
A limitation of our study is the relatively small study cohort, which resulted in few HPV16 positive samples even after pooling the GP-collected samples and the self-samples, whereby each HPV16 positive sample had a huge impact on the calculated sensitivities for HPV16 detection. The relatively low sensitivity of 83.3% (95% CI: 58.6-96.4%) observed for CLART ® HPV4S was thus caused by only three HPV16 false negative samples, and the study cohort should therefore be expanded for more precise sensitivity and specificity estimations, especially regarding HPV16 detection.
Besides the aforementioned features, a strength of our NGS assay is its ability to work on both self-samples and GP-collected samples. Previous studies have shown an increase in cervical cancer screening attendance using self-sampling, and several countries have already implemented or are in the process of implementing self-sampling in their screening programs [14,25,29,58,59] which makes the ability of HPV detection in selfsamples important in future HPV screening assays.

Conclusions
In conclusion, the NGS assay enables detection and genotyping of all carcinogenic, probably carcinogenic, and possibly carcinogenic HPV types while performing as well as the Cobas ® 4800 and CLART ® HPV4S assays in sensitivity and specificity for HPV detection and HPV16 detection in both GP-collected cervical samples and cervico-vaginal self-samples. The NGS assay could, thus, potentially decrease the number of false negative screening results caused by infections with HPV types not currently screened for whereby more cervical dysplasia and cervical cancer cases could be avoided. Moreover, the NGS assay gives valuable additional information about the dominating HPV type in cases with multiple HPV infections, integration status, and variants in the HPV genome which are features that could be valuable in future stratification of HPV positive women. Thus, we believe the additional information provided by NGS-based HPV analysis outweighs the longer analysis time (2-3 days) compared to PCR-based HPV analysis (hours).

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cancers14030652/s1, Table S1: HPV types detected in the NGS assay, Cobas ® 4800, and CLART ® HPV4S in general practitioner-collected cervical samples and cervico-vaginal self-samples of women with atypical squamous cells of undetermined significance (ASCUS); Table S2: Genomic variants of HPVs detected by the NGS assay in general practitionercollected cervical samples with atypical squamous cells of undetermined significance (ASCUS).