Homosexual men are at greatly increased risk of human papillomavirus (HPV)-related anogenital disease, including anal cancer. Before the HIV epidemic the incidence of anal cancer in homosexual men in the United States was estimated at 35 cases per 100 000 person-years,1 a rate similar to that for cervical cancer before screening. The incidence of anal cancer in people with HIV infection in Australia, the majority of whom are homosexual, was 37 times that in the general population during the 1980s and 1990s.2 Anal cancer is increasingly being recognised as an immune deficiency-related cancer,3 4 and there is strong evidence that incidence is stable or even increasing in homosexual men and people with HIV infection, even in the highly active antiretroviral therapy (HAART) era.5 6

HPV infection is causally associated with anal cancer.7 HPV-16, a high-risk genotype for cancer of the cervix, is isolated from most (70-80%) population-based anal cancer specimens.3 6 8^–11 The prevalence of the next most common genotype, HPV-18, is an order of magnitude lower (7%). Of note, HPV was detected in virtually all anal cancers diagnosed in homosexual men.6 11 Despite these striking findings, data on the prevalence of anal HPV infection and the range of HPV genotypes in the homosexual population are limited, and most data come from clinic-based rather than community-based populations.

Here we report results from a study of anal HPV infection in community-based cohorts of HIV-positive and HIV-negative homosexual men in inner Sydney, Australia.

METHODS

Study design

We performed a cross-sectional study of HPV prevalence in anal swab specimens.

Study population

Consecutively presenting homosexual men (n = 332) from the positive Health (pH; HIV-positive)12 and Health in Men (HIM; HIV-negative)13 cohorts in inner Sydney, Australia, between February and August 2005. Cohort participants were recruited from a wide range of community-based settings, including various outdoor gay events and gay community events. Men were eligible if they reported sex with at least one man during the previous 5 years and lived in Sydney or participated regularly in its gay community.

Sample collection and processing

Anal samples were collected by a trained nurse using a sterile swab, moistened with sterile saline, without direct visualisation.14 The swab was inserted 3–5 cm into the anus, rotated for 1 minute and slowly removed. The swab was placed in 20 ml of a methanol-based fixative (PreservCyt; Cytyc Corporation, Marlborough, MA, USA) and transported to a single laboratory for processing. After ThinPrep processing, HPV testing was performed on all samples using the Hybrid Capture-2 (HC-2; Digene Corporation, Gaithersburg, MD, USA) assay for the detection of both LR (type 6, 11, 42, 43 or 44) and HR (type 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, or 68) genotypes.15

Testing by Roche Linear Array (LA; Roche Molecular Systems, Alameda, CA, USA) to detect 37 individual HPV genotypes was also performed using a modified method16^–18 on 133 of the men in whom samples before cytology processing were available. This assay was utilised to detect LR (type 6, 11, 26, 40, 42, 53, 54, 55, 61, 62, 64, 66, 67, 69, 70, 71, 72, 73, 81, 82, 83, 84, IS39, or CP6108) and HR (type 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, or 68) genotypes.19 To verify the presence or absence of HPV-52 among samples testing positive for HPV-33, HPV-35 and/or HPV-58, two samples were tested by a real-time supplementary polymerase chain reaction (PCR) test.18 CD4 counts were self-reported by HIV-positive men; Australian guidelines state that CD4 count should be monitored 3-monthly in patients with HIV.

Statistical analysis

The demographic and lifestyle characteristics of the HIV-positive and HIV-negative homosexual men were compared by use of Pearson’s χ² test for categorical variables or t test for continuous variables.

Samples of sufficient quality for analysis were classified as positive or negative for LR-HPV or HR-HPV DNA, and the HPV genotype detected using LA was also recorded. Results were stratified by HPV genotype and HIV serostatus; individuals co-infected by more than one type were included in counts of each category or specific type. HPV prevalence by HIV status was compared by calculating an unadjusted odds ratio (OR) and 95% confidence interval (CI), and by using Pearson’s χ² test. The mean number of HPV genotypes by HIV status was compared using the independent samples t test. p Values for trend were calculated using Pearson’s correlation coefficient. All statistical tests were two-sided, and the level of statistical significance was set to α = 0.05.

Potential risk factors for anal HR-HPV detection using the HC-2 assay were examined separately for HIV-positive and HIV-negative men. Logistic regression modelling was used to determine the OR and 95% CI for each variable of interest. Risk factors examined included demographic, lifetime and recent sexual behaviours, cigarette smoking and clinical findings, and for HIV-positive men only, CD4 count and history of HAART. All variables with two-sided statistical significance at the p<0.10 level in univariate analysis were considered multivariately. The final multivariate model included only those covariates with two-sided statistical significance at p<0.05 after adjustment for each other.

Analyses were performed using Stata software version 10.0 (Stata Corporation, College Station, Texas, USA).

RESULTS

LR and HR HPV DNA detection using Hybrid Capture 2

After the exclusion of 16 samples with inadequate volume after ThinPrep processing, HC-2 results were available for 316 participants (193 HIV-negative, 123 HIV-positive). Compared to HIV-negative men, the HIV-positive men were more likely to smoke (p = 0.001) and to have had a sexually transmitted infection in the past 12 months (p = 0.006; table 1).

Any HPV infection was detected in 79% of men, and the prevalence was higher in HIV-positive compared to HIV-negative men (94% vs 70%, p<0.0001). The detection rate for all men was 55% for LR-HPV and 69% for HR-HPV. The rates of both LR-HPV and HR-HPV were significantly higher in HIV-positive compared to HIV-negative men (73% vs 44%, p<0.0001 and 88% vs 56%, p<0.0001 respectively). In HIV-positive compared to HIV-negative men, LR genotypes were 3.5 (95% CI 2.1 to 5.7) times more likely to be detected, and HR genotypes were 5.5 (95% CI 3.0 to 10.2) times more likely.

Among HIV-negative men, 50 (26%) reported no casual sexual partners in the preceding 6 months. The prevalence of HPV infection was no different in these men from those reporting one or more casual sexual partners (LR-HPV 40% vs 46%, p = 0.45; HR-HPV 50% vs 59%, p = 0.28). Among HIV-positive men, 28 (23%) reported no casual sexual partners in the preceding 6 months. The prevalence of HPV infection was no different in these men from those reporting one or more casual sexual partners (LR-HPV 71% vs 73%, p = 0.90; HR-HPV 96% vs 81%, p = 0.11). Few men (n = 8 HIV-positive, n = 4 HIV-negative) reported no casual or regular sexual partners in the past 6 months.

In both HIV-negative and HIV-positive men, there was no consistent trend in HPV prevalence beyond 34 years of age (fig 1A and 1B). For HIV-positive men, there was no difference in LR- or HR-HPV prevalence by category of CD4 count (⩽350 vs >350; p = 0.35 and p = 0.99 respectively).

Risk factors for detection of HR-HPV by HC-2

The only factor significantly associated with risk of anal HR-HPV detection in HIV-positive men was anal bleeding in the past 12 months (OR 8.36, 95% CI 1.05 to 66.0). Univariately, risk of anal HR-HPV infection in HIV-negative men was increased by a lifetime history of anal warts (OR 3.26, 95% CI 1.45 to 7.33), infection by anal gonorrhoea (all five cases; p = 0.068) or anal chlamydia (OR 3.06, 95% CI 0.83 to 11.35) in the past 12 months, testing seropositive to herpes simplex virus type 2 (OR 1.86, 95% CI 1.01 to 3.45), more than 500 lifetime male partners (vs <50 partners; OR 2.22, 95% CI 1.01 to 4.87), and having no preference for either insertive or receptive anal intercourse (vs insertive; OR 2.09, 95% CI 1.09 to 4.01). In multivariate modelling, only history of anal warts independently predicted risk (OR 3.00, 95% CI 1.30 to 6.81).

HPV DNA detection and genotyping using LA

Of the 133 samples tested using LA, four tested β-globin negative, leaving 93 HIV-negative and 36 HIV-positive assessable samples. Any HPV genotypes were detected in 95% of men, and the prevalence was similar for HIV-positive (97%) and HIV-negative (94%) men. The detection rate for all men was 88% for LR-HPV and 79% for HR-HPV. The prevalence of LR HPV did not differ between HIV-positive (89%) and HIV-negative (87%) men (p = 0.78). The prevalence of HR-HPV infection was significantly higher in HIV-positive (94%) compared to HIV-negative (73%) men (p = 0.008).

Of the LR types, type 6 or 11 were present in 37% of men, and there were 13 individual LR types for which the prevalence was at least 10% (table 2). A significantly higher proportion of HIV-positive men had HPV types 6 or 11 (58% vs 29%, p = 0.002). Of the HR types, types 16 or 18 were present in 39% of men, and there were 11 individual HR types for which the prevalence was at least 10% (table 2). A similar proportion of HIV-positive and HIV-negative men had HPV-16 (36% vs 27%, p = 0.30) and HPV-16 or 18 (39% vs 39%).

A significantly higher number of genotypes were detected in HIV-positive compared to HIV-negative men. The mean (SD) number of HPV types was 7.1 (3.7) versus 4.2 (3.1) (p<0.0001) for any type, 3.5 (2.3) versus 2.2 (1.8) (p = 0.001) for LR and 3.6 (2.3) versus 2.0 (1.9) (p<0.0001) for HR. More than one HPV type was detected in 86% of all men; this proportion was 62% when only HR-HPV types were considered.

There was no significant linear trend in HPV prevalence beyond 34 years of age for either HIV-negative (fig 1C) or HIV-positive men (fig 1C and 1D). There was no difference in LR-HPV or HR-HPV prevalence by category of CD4 count (⩽350 vs >350) for the HIV-positive (p = 0.87 and p = 0.26, respectively).

DISCUSSION

In this community-based population of sexually active homosexual men, anal HPV infection was nearly universal. A wide variety of HPV genotypes were detected, and co-infection with multiple genotypes was common. An alarming one-third of participants tested positive for HPV-16, the primary causal agent of anal cancer. The data indicate a greater level of HPV infection than previously described in community-based samples of homosexual men. HIV infection was associated with a greater prevalence of HR-HPV, and double the number of genotypes. The lack of a consistent decline in HPV prevalence with increasing age, and the similarity in rates for those with and without casual sexual partners in the preceding 6 months, suggest that it may be persistent infection and impaired clearance rather than new infection. Risk of HR-HPV was predicted by anal bleeding in the preceding 12 months for HIV-positive men and a history of anal warts for HIV-negative men.

Our data confirm previous findings in community-based homosexual populations that anal HPV infection is common, that a wide variety of genotypes are present, and that multiple infections are also common.20^–22 However, compared to these US studies, the rate of anal HPV infection in our population of HIV-negative men was higher. The larger of these studies, the US EXPLORE Study of HIV-negative homosexual men, reported PCR-detected HPV infection in 57%, HR-HPV in 22% and LR-HPV in 22%.20 HPV-16 was detected in 12% of men and the mean number of HPV types was 1.8. These differences may be a consequence of behavioural differences; while the proportion of men reporting receptive anal intercourse in the previous 6 months was comparable (81% in our population and 77% in the US population), our population was less likely to always use a condom (22% vs 38%).

Cervical infection by more than one HPV type is detected by PCR in around 30% of young women.23 Our data support the earlier evidence that a greater proportion of homosexual men exhibit anal infection by multiple genotypes.20 21 It has not yet been established whether co-infection with multiple HPV genotypes impacts on the persistence of a given HPV genotype or on progression of intraepithelial abnormalities.24 However, in one study of homosexual men, anal infection with multiple HPV genotypes was associated with prevalent anal intraepithelial neoplasia (AIN) and progression to higher grade AIN over time.25

Cervical HPV infections in high-income countries peak in women around 20 years of age and decline markedly thereafter.24 26 This characteristic age distribution is a consequence of the processes of HPV acquisition and clearance. The lack of decline in the anal HPV age-specific prevalence rates in homosexual men, as shown in this study and the EXPLORE Study,20 may be the result of continuing newly acquired infections, the impaired clearance of existing infections, or the re-activation of latent infection.27 The latter is particularly relevant in the context of HIV infection. The sexual behaviour profile of these homosexual men, most of whom reported at least one new sexual partner in the preceding 6 months, supports a role for newly acquired infection. However, recurrent infection of the same type after clearance is believed to be uncommon24 and the similarity in HPV prevalence for those with and without new sexual partners in the past 6 months suggests a contribution by persistent infection.

Our data confirm the previous community-based finding that anal HPV infection is more common and involves more genotypes in HIV-positive compared to HIV-negative homosexual men.21 While the exact biological mechanism is not known, immunosuppression may result in a loss of HPV-specific immunity, and reactivation or upregulation of pre-existing HPV infection.21 28 The greater number of HPV genotypes in HIV-positive compared to HIV-negative men also likely indicates an inability to clear infection. Higher sexual risk behaviour in HIV-positive men may also be a factor.29

Previously reported risk factors for anal HPV infection in community-based homosexual populations include receptive anal intercourse and the number of male partners during the previous 6 months in HIV-negative men,20 and lower CD4 levels in HIV-positive men.21 Our risk factor data for HR-HPV infection in HIV-negative men were supportive of an association with lifetime sexual behaviour, but not after adjustment for history of anal warts.

The study had the strength of being community-based, and thus the participants can be considered to be broadly representative of HIV-negative and HIV-positive homosexual men involved in the gay community in inner Sydney. We also ascertained both lifetime and recent sexual behaviour history, allowing an assessment of the role of recent HPV acquisition. In addition, we tested for HPV infection using two assays. As expected, however, we did not observe exact concordance between the assays because the HC-2 probes detected fewer LR types and the PCR assay detected infections at a lower viral load threshold.30 The study was limited by the modest numbers of participants, particularly in the subgroup analyses.

In summary, our study confirms that a range of HPV genotypes, typically multiple genotypes, are detectable in the anal canal of homosexual men, and that the HPV burden is greater for those with HIV infection than those without. It indicates a higher prevalence of HPV infection in HIV-negative homosexual men, and a higher prevalence of HPV-16, the genotype responsible for most anal cancers, than previously recorded. The possibility of persistent or newly acquired HPV infection in this high-risk population supports the urgent need for large-scale cohort studies to assess the natural history of anal HPV infection and cofactors for anal HPV infection in homosexual men. The utility of the currently available HPV vaccines would also be informed by further population-based data on the HPV types isolated from anal cancers in homosexual men.