The few studies that have analysed concurrent or sequential oral and anogenital HPV infections have examined similar study populations: high-risk sex workers, women with a history of cervical HPV infection, and HIV-positive women [15–18]. In the present longitudinal study conducted among HIV-infected MSM, we have found a high prevalence of HPV infection in the anal area, both at initial screening (88.6%) and follow-up (86.3%). These findings are consistent with those of two other longitudinal studies conducted in HIV MSM, suggesting that HIV infection not only increases susceptibility to persistent HPV but also increases the risk of acquisition of new HPV infections and the risk of reactivation of latent infections [9, 28]. Our overall HPV anal prevalence data are consistent with those of earlier studies conducted in HIV-infected populations [7, 29–31]. However, compared to previous studies, we have found a lower rate of HR types and a higher prevalence of LR or possible HR types. This observation is not surprising because the prevalence of infection with HR HPV types may differ substantially by geographic location . When analysing specific HPV types, we found that the low-risk HPV-6 type was the most common HPV type (13% of specimens at screening and 16% at follow-up), followed by HPV-16 (13 and 7%), HPV-11 (12 and 3%), HPV-66 (11 and 7%) and HPV-61 (12 and 5%). A higher prevalence of HR types has been found in anal samples from HIV-infected patients in previous studies, with rates of up to 70% observed . Conversely, low-risk genotypes have been found in higher proportions in heterosexual men with no other sexually transmitted diseases and in a small dataset of HIV-infected patients [8, 33]. A number of subjects had anal infections with an HPV strain with an undetermined risk level (17.8% at baseline and 14.2% at follow up among genotyped strains).
In a more recent report on HIV-infected MSM , most patients (90.9%) were infected with multiple HPV types in the anal canal, with a median number of 5 HPV types per sample (range: 0-18). The median number of HR HPV types was 5 per sample (range: 0-12). The different methods used for HPV typing by the authors of these studies makes it difficult to compare their results with the results of the present study. These authors obtained genotypes using the reverse line-blot detection system  that is able to identify 36 genital HPV types; samples that were not positive for any of these types were considered HPV negative. This method enabled the authors to demonstrate the presence of co-infection. We chose a population sequencing method, which is less able to demonstrate viral co-infections, to be able to find rare genotypes. However, minority genotypes are easily overlooked. Nevertheless, in our study, multiple HPV infections, which can be identified sometimes during editing of the electropherograms as multiple sequences, appear to occur less frequently than in the HIPVIRG study.
Current data on the spread of HPV infection to various body parts implicated in sexual practices in both MSM and heterosexual individuals are limited. Oral sex behaviours have been associated with oral HPV infection and the transmission of other viral infections such as HSV [13, 35]. Moreover, sexual behaviours are associated with a risk of cancer at the head and neck subsites that have previously been associated with HPV infection . In the present study, the prevalence of HPV in oral samples was 20.1% at screening and 21.3% at follow-up. There are limited data in the literature concerning the relationship between oral and cervical HPV infection, an emerging area of research given the potentially increased risk of HPV-associated oropharyngeal cancers in HIV-positive individuals . In a study of 30 HIV-positive women from South Africa who were naïve to ART and had a CD4 count of less than 300, oral HPV DNA was detected in 20% of oral specimens from the women and in 97% of the cervical specimens from the same women . A limited correlation between oral HPV types and those identified in the cervical mucosa was found. In a larger study of ART-experienced women from the U.S.A., HIV-positive women were more likely to have detectable oral HPV infection than HIV-negative women (33 vs. 15%, p = 0.016) . However, the prevalence of oral HPV infection over six months was significantly less than for cervical infection (p < 0.0001). These data suggest that while oral HPV infection is more common among HIV-positive women compared with HIV-negative women, the incidence and natural history of HPV infection at these two mucosal sites are not highly correlated. A similar relationship was noted in the past between cervical HPV infection and anal HPV infection in HIV-positive women ; while anal infection was more common than cervical infection, the HPV types detected at the two sites were different in the majority of the women.
There are fewer studies on HPV prevalence in oral specimens from men and these studies are difficult to compare. HPV infection was detected in 4.8% of 332 control patients from an outpatient clinic and in 2.9% of 210 college-aged men (age range: 18-23 years) . Oral sex and open-mouthed kissing were associated with the development of oral HPV infection. Oral HPV infection is strongly associated with oropharyngeal cancer among subjects with or without the established risk factors of tobacco and alcohol use . Data from another study show that HPV is detectable in clinically normal oral mucosa in approximately one third of the general population, while the detection rate observed in HIV-positive individuals is up to 60% . We found the prevalence of HPV in oral specimens of HIV-infected MSM to be around 20% and we did not find genotype concordance between oral and anal HPV genotypes. Taking into account potential bias due to the lack of information from the 14.5% inadequate samples, this observation is higher than that seen in other studies (5.5% reported by Fakhry in 2006), and our data are similar to those reported by Palesky and colleagues in HIV-positive women .
Our findings show that the prevalence of HR HPV genotypes was highest in the 43-50 year old age group, where it was over 25%. The persistence of HR genotypes (71.4%) did not exceed the persistence of low-risk genotypes (76.7%), although further surveillance of this cohort will be necessary to understand of the determinants of HPV persistence. The presence of an HPV-positive anal swab was negatively related to age, CD4 count and antiretroviral treatment, and positively related to HIV viral load. The negative correlation between anal HPV infection and CD4 count, albeit expected, warrants further study and corresponds with the positive correlation with HIV plasma viral load. These observations support a role for HIV viral replication and immunodeficiency in affecting susceptibility to infection with all genotypes of HPV infection in the anal area or the ability to clear HPV.
The relationship between anal HPV infection and age was negative (decreasing) both by linear regression (using age as a dependent variable) and by logistic regression. However, we interpret this finding to be somewhat artifactual, as the age-related prevalence of all HPV genotypes shows a plateau followed by a decline in the group older than 50 years, whereas the high-risk HPV genotype infections show a peak in the 43-50 years age group, in partial agreement with Nyitray . Moreover, two subpopulations were detected: a larger and younger population with a lower antiretroviral treatment rate, less control of HIV replication, lower CD4 levels, and a higher prevalence of anal HPV infection, including high-risk genotypes, and a smaller, older subpopulation with opposite features.
The correlation between HPV infection and T. pallidum infection was close to the significance level and became significant when the possible high-risk genotypes 53 and 66 were included in the analysis. Syphilis (active, latent or treated) was positively related to age, HCV infection, and, unexpectedly, lower HIV plasma viral load and higher CD4 count. The latter two correlations are probably explained by the characteristics of the older HIV patients, as the older patients were more often receiving ART and as a consequence, they had lower viral loads and higher CD4 cell counts. The association between syphilis and HCV infection is somewhat unexpected, since recent data suggests that permucosal HCV transmission might result from high-risk sexual and non-injecting drug use behaviours among MSM, raising the question of the importance of sexual transmission of HCV .
HHV-8 oral shedding was positively related to HPV infection in anal area: HPV infection rates were 63.6% in the HHV-8-positive and 30.0% in HHV-8-negative group. It has been demonstrated that the risk of HHV-8 infection among men in the United States and Western Europe is closely related to the number of male sexual partners, and in these populations sexual intercourse is the major mode of HHV-8 transmission [46, 47]. A significant correlation was also found between the persistence of HHV-8 shedding and HIV viral load. We have previously demonstrated that HHV-8 viral load persists for up to 6 months in the majority of patients with concurrent HIV viral load, whereas all of the patients on ART with undetectable HIV viral load cleared HHV-8 from the plasma after a mean period of 12 months after the first positive blood specimen .
Some limitations of our study should be addressed. Our data lack information regarding a number of subjects and the robustness of our correlations should be confirmed with further observations. HPV genotyping was performed using a population sequencing method that does not allow us to fully characterise co-infections and to correctly trace all genotypes found during the double sampling. Nevertheless, this method allowed us to demonstrate a high prevalence of infection by undetermined-risk HPV types.
In conclusion, the negative relation between anal HPV infection and CD4 count detected by this multivariate logistic procedure, albeit expected, deserves interest and is in agreement with the positive correlation with HIV plasma viral load found using a bivariate analysis.