Introduction

Chronic infection with oncogenic HPV is associated with precancerous lesions that may progress to invasive anal cancer (AC). Among MSM living with HIV, AC incidence is 131 per 100 000 person-year and the relative risk is 80 times higher than expected for the general population.1

Currently, the gold standard for the histopathological evaluation of suspect lesions is the biopsy directed by high-resolution anoscopy (HRA), a technically complex procedure that requires highly trained clinicians . Tests for HPV DNA detection are the recommended primary tools for cervical cancer screening,2 but their use in MSM living with HIV is under discussion mainly due to the high HPV infection prevalence.3 There are numerous molecular tests available for HPV detection. Hybrid Capture 2 (HC2; Qiagen, Germany) and Linear Array Genotyping Test (LA; Roche Diagnostics, USA) have been widely validated in cervical samples 2; however, few studies have compared their performance in anal samples. This study estimated anal HPV prevalence in MSM living with HIV by employing two HPV DNAdetection tests and compared their performance and prognostic value for the diagnosis of histologyproven highgrade intraepithelial anal lesions.

Methods

MSM aged 18 years or older living with HIV who attended their check-up visit were offered to participate in an AC prevention programme consisting of detection of HPV DNA detection, anal cytology, digital anorectal examination (DARE) and HRA between May 2014 and June 2017. Anal samples were collected with circulatory rotatory movements for at least 40 s and then introduced in ThinPrep® Pap test vial containing PreservCyt® liquid-based medium (Hologic, Marlborough, Massachusetts, USA). Samples were stored at room temperature. Anal cytology was performed using the automatised system ThinPrep method (Hologic) and classified as negative for intraepithelial lesion and malignancy, atypical squamous cells of undetermined significance, low-grade squamous intraepithelial lesion (LSIL), high-grade squamous intraepithelial lesion (HSIL), atypical squamous cells cannot exclude HSIL (ASC-H) and squamous cell carcinoma.4HRA was performed employing an anoscope coated with a lubricant inserted into the anal canal that was visualised using a colposcope. From 2014 until June 2016, HRA was performed a few days later after anal sample collection; from June 2016 onwards, HRA was performed at the same visit. Biopsies of any suspicious lesion were taken and were reported as negative, LSIL-AIN1, HSIL-AIN2 or HSIL-AIN3. HPV DNA detection was done using both HC2 and LA; discordant result were further tested with Anyplex II HPV28 (Seegene, Seoul, Korea). Agreement was determined using Cohen’s kappa (κ) index and McNemar’s χ2 test. Histology results were grouped as negative for a normal appearance on HRA or LSIL-AIN1 cases, and high-grade anal intraepithelial neoplasia (HGAIN) for HSIL-AIN2 and HSIL-AIN3 cases. The prognostic value of the assays was calculated employing HGAIN histology result as the end point.

Results

274 participants were enrolled in the study. All samples rendered a valid HPV detection result; 113 samples (48.5%) were positive by HC2 and 245 (89.4%) were positive by LA. Multiple HPV genotypes were detected in 207 (75.5%) participants; the median number of oncogenic and non-oncogenic HPV genotypes was 3 (IQR: 1–4) and 2 (IQR: 1–2), respectively. A valid anal cytology result was obtained in 269 samples (98.2%). Among 176 (64.2%) participants who underwent HRA, 95 (53.4%) had no anoscopically visible abnormalities and 81 (29.6%) were biopsied. Online supplemental table 1 shows the cytology and histology results. HPV prevalence by HC2 was 41.6% (64/154) in men with negative histology and 90% (18/20) in men with HGAIN (p<0.0001). HPV prevalence by LA was 85.7% (132/154) and 100% (20/20) (p=0.0806), respectively. HPV16 (70% (14/20)) and HPV6 (45% (13/20)) were the most prevalent genotypes in patients with HGAIN results. Differences regarding HPV type-specific prevalence according to HGAIN and negative results were identified for HPV16 (70% (14/20) vs 23.4% (36/154); p<0.0001), HPV18 (25% (5/20) vs 8.4% (13/154); p=0.0384), HPV39 (30% (6/20) vs 11% (17/154); p=0.0303), HPV42 (25% (5/20) vs 8.4% (13/154); p=0.0384) and HPV58 (30% (6/20) vs 11% (18/154); p=0.0374). Similarly, differences were observed for the prevalence of multiple oncogenic (85% (17/20) vs 54% (83/154); p=0.0081) and non-oncogenic (75% (15/20) vs 44.8% (69/154); p=0.0160) HPV genotypes. The agreement between assays was 75.2% (κ=0.51; 95% CI 0.42 to 0.60). Discordant results corresponded to 61 samples (22.3%) that tested negative by HC2 and positive by LA and 7 samples (2.6%) that tested a positive result by HC2 and negative by LA (p<0.001). Anyplex II HPV28 detection identified at least one oncogenic HPV genotype targeted by HC2 in 6/7 (85.7%) samples. The performance characteristics of the HPV DNA detection tests and the anal cytology in identifying HGAIN results are presented in table 1.

Table 1

Performance of HPV DNAdetection using HC2, LA and anal cytology for identification of HGAIN lesions results among HIV-infected men who have sex with men

Discussion

In our study of anal samples, the agreement between two HPV DNA detection assays was slightly lower than described for cervical samples, although the presence of lubricants, stool or even a lowerDNA levels in anal samples could interfere with their performance.5 Although most discordant results may be explained by the higher analytical sensitivity of the LA, there were also identified samples classified as analytically false positive or false negative. HC2 false positivity may be attributed to the cross-reactivity with genotypes not targeted by the probe, which may have been the case in our sample, in which HPV61 was identified. LA false negativity may be explained by a higher analytical sensitivity for the detection of specific HPV genotypes due to the overcoming of masking effects to the competition for the use of consensus primers versus specific primers (eg, Anyplex II HPV28).6 Moreover, the size of the amplicons would also be associated with increased detection because longer amplicons may result in more negative samples. Despite the fact that Anyplex II HPV28 test does not provide information about the amplicon size, LA amplifies a fragment that can be considered long (450 bp). Interestingly, four of these six LA false negative samples harboured HPV genotypes frequently reported as cross-reactive in HC2.

The role of HPV DNA testing in AC screening programmes in people living with HIV or MSM is still under debate.3 We identified that total HPV DNA detection demonstrated a high sensitivity but a low specificity for HGAIN, mainly as a consequence of the high prevalence of anal HPV .7 In contrast to earlier studies, we identified that the use of HC2 versus LA showed higher specificity, PPV and area under the curve (AUC), while having a slight impact on sensitivity.8 This better overall performance of HC2 may be explained by the analytical sensitivity and the association of a higher HPV viral load of persistent infection with a higher risk of HGAIN.7 Indeed, we observed that the median values of RLU/CO increased with the grade of abnormality in the histology results. Previous studies have reported how the inclusion of HPV genotyping versus total HPV DNA detection provided an increase in clinical specificity, and a decrease in the clinical sensitivity.7 Our findings support these results, and we found that the identification of selected combinations of HPV genotypes demonstrated higher sensitivity, specificity and NPV than individual genotyping. HPV16/18 genotyping showed the highest Youden’s index and AUC, which could be mainly explained because 90% of men histologically diagnosed with HGAIN enrolled in this study were positive for at least one of the two genotypes. In our study, the HPV detection-cytology cotesting decreased the sensitivity but increased the specificity and the PPV when we employed as the definition of an abnormal cytology a threshold of HSIL versus ≥ASC US, confirming that the anal cytology screening seems to underestimate the true level of histological dysplasia.9 Previous reports on cotesting demonstrate high sensitivity and a low specificity10 that improves with the HPV16 and/or HPV18 genotyping.7 In our population, global HPV detection by HC2 or HPV16/18 genotyping by LA showed similar clinical performance, and the use of a threshold ≥ASC US for abnormal cytology provided the highest diagnostic accuracy for detecting anal precancer, suggesting that either HPV DNA-based tests could be used for detection of HGAIN lesions.

This study is subject to several limitations. The prevalence of HGAIN results reported in our study is lower compared with other works. The expertise in the use of the anoscope to identify visually suspicious areas of the anal canal, characterised by a long learning curve, might have conditioned the number of biopsies. Moreover, not all men included in the study completed the protocol because, during the first period, the anal sample collection was performed in the first visit but the HRA was programmed some days later. Whether the low detection of lesions is attributable to a false low indication of biopsies remains to be seen in the future evaluation of the study subjects. In addition, all parameters evaluated are cross-sectional and it would be more relevant to have a prospective indication of the value of HPV DNA-based detection. Finally, the relatively small sample size might also have underpowered the study.

The advantage of this study was the use of fresh anal samples to compare the performance of two frequently employed HPV DNA-based assays along with cytology and HRA from a homogeneous population of MSM living with HIV. All results were tested blindly . To further our research, we are performing a long-term follow-up of the cohort to perform both longitudinal and cost-effectiveness analyses to discriminate potential HPV biomarkers with prognostic value to predict the persistent anal precancerous lesions.

Key messages

Anal HPV prevalence in a population of MSM living with HIV attending their check-up visit was very high employing either a total or type-specific HPV detection assay .

Oncogenic HPV16 and non-oncogenic HPV6 were the most prevalent genotypes among men histologically diagnosed with high-grade anal intraepithelial neoplasia.

The overall agreement between the two HPV DNA assays for the identification of anal HPV infection was moderate.

Abstract translationThis web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study was approved by Ethics Committee for Research of Bellvitge University Hospital (Barcelona, Spain) (ID: PR345/14). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

This paper is dedicated to Elena Ferrer, MD, who contributed to the implementation of the anal cancer screening programme of the HIV and STD Unit at Bellvitge University Hospital (Barcelona, Spain) and passed away on 5 September 2018.