The incidence of anal cancer is low among men in the general population, but is considerably increased among MSM. Prior to the HIV epidemic, the incidence of anal cancer among MSM was estimated to be as high as 37 per 100 000 , which is similar to the incidence rate of cervical cancer among women in western countries before routine cervical cytology screening was introduced . Among MSM who are also HIV-positive, the incidence rates have been reported to be even higher at 131 per 100 000 . Unlike other virus-associated cancers, the risk of anal cancer has not declined among HIV-positive MSM since the introduction of HAART [4–7]. Instead, the incidence of anal cancer has continued to increase , and it is possible that this trend may continue as the HIV-positive population continues to age.
Similar to cervical cancer, anal cancer is associated with human papillomavirus (HPV) infection. Approximately 90% of all anal cancers are attributed to HPV infection . HPV 16 and HPV 18 are responsible for 76 and 9% of anal cancers, respectively .
In prevalence studies, anal HPV infection is almost universal among HIV-positive MSM, with reported prevalence estimates between 87 and 98% [9–12]. HPV 16 is among the most common types detected, with a greater than 30% prevalence [9–12]. Risk factors for prevalent anal HPV infection include receptive anal intercourse , a higher number of male partners [9,10], inflammation [9,12], and a lower CD4+ cell count [9,10,13]. However, cross-sectional studies are limited in their ability to examine risk factors, given that the prevalent infections are a mix of new and persistent infections. Risk factors for prevalent infection may be factors associated with persistence and not acquisition of new infection.
We conducted a prospective cohort study to assess the natural history of anal HPV infection in MSM in the antiretroviral therapy (ART) era. The goal of this analysis is to report the incidence of detection of type-specific anal HPV infection and to evaluate risk factors for infection.
HIV-positive MSM were recruited into a prospective cohort study conducted by the University of California, San Francisco (UCSF). Enrollment was completed between February 1998 and January 2000, and participants were followed every 6 months for at least 2 years. At each visit, participants completed an interviewer-administered questionnaire and a clinical examination, including collection of an anal swab for HPV testing. Blood was collected for CD4+ lymphocyte cell counts that were measured using standardized two or three-color fluorescence methods. HIV plasma viral load (HIV-VL) was measured using the branched-chain Chiron assay (Chiron, Emeryville, California, USA). Written informed consent was collected from participants and the study was approved by the Committee on Human Research at UCSF.
Human papillomavirus testing
Testing for anal HPV infection was performed as described previously using the PCR with L1 consensus primers and probes specific for 29 individual HPV types and a mixture of 10 other types . Beta-globin-negative (indicating insufficient DNA) samples were excluded from analysis. Detection of incident HPV infection was computed for each HPV type individually. Incident HPV infection was defined as a negative test result at baseline for an individual HPV type followed by at least two consecutive positive tests for that type during the follow-up period. Two consecutive positive tests were required because detection on a single occasion is less likely to represent true infection than is two consecutive positive tests 6 months apart, and is consistent with the definition of infection commonly used in vaccine studies [14,15]. Requiring two positive tests reduces potential outcome misclassification and provides a conservative estimate of incidence. Because of this requirement, participants were included in this analysis only if they had at least two follow-up visits or 1 year of follow-up.
Assessment of potential risk factors
Demographic factors, lifestyle characteristics, medical history including ART use, history of treatment for high-grade anal intraepithelial neoplasia (HGAIN), and history of sexual behavior were collected as part of the baseline questionnaire. At each follow-up visit, the interviewer collected information on factors that may have changed since the last interview. Two time periods were included: approximately 6 months prior and in the past 30 days.
We asked if the participant had engaged in each sexual behavior (yes/no); the number of partners with whom he had engaged in the behavior; the number of ‘new’ partners with whom he engaged in the behavior; and the frequency of the behavior. Men were queried about sex with men and women, ‘insertive’ anal intercourse (participant inserts his penis into partner's anus) and ‘receptive’ anal intercourse (participant receives his partner's penis into his anus); oral–anal contact (‘rimming’, participant's anus receives contact from partners mouth).
The rate of detection of incident anal HPV infection was computed as number of participants with incident infection detected divided by person-years of follow-up. An incident infection was assumed to have taken place halfway between the previous visit and the current visit, and person-time was assigned as (current visit date minus previous visit date)/2. Exact Poisson confidence intervals (CIs) were computed for each incidence rate .
Participants who had an incident type-specific HPV infection were censored from that type-specific analysis at future time points, but remained in the analyses for other HPV types. Participants could have ‘any incident HPV’ at multiple visits if they had different type-specific infections at different visits.
We evaluated the bi-variable association of baseline characteristics with any incident infection in participants (n = 376 participants) using the chi-square test for independence for categorical variables, and analysis of variance (ANOVA) or ranked ANOVA for continuous variables, as appropriate. We evaluated bi-variable associations of incident HPV infection with time-dependent characteristics by participant visit (n = 888 participant visits). P-values were generated from generalized linear models estimated through repeated measures with generalized estimating equations (GEE) , adjusting for potential correlation from taking repeated measures on participants, as well as reporting robust standard errors. We note that the measure of association obtained from this procedure is the odds ratio (OR) related to the hazard of new infection.
To evaluate the effect of risk factors for incident anal HPV infection, we constructed logistic regression multivariable models estimated through repeated-measures GEE . Four different model sets were created for each risk factor. Model 1 is the unadjusted model and includes only the risk factor. Model 2 additionally includes the baseline value of the risk factor. Model 3 adds six baseline factors identified a priori as potential confounders (baseline age, race, education, smoking 100+ lifetime cigarettes, baseline CD4+ cell count, and HPV status at baseline). Model 4 adds the potential time-dependent confounder of CD4+ cell count measured at the previous visit. In all models, the main risk factor is treated as a time-dependent variable.
We also conducted a sub-analysis to determine if treatment for low-grade anal intraepithelial neoplasia (LGAIN) or HGAIN had an effect on our models. We added a variable for treatment at any time during the study to our “Model 4's” (described above). All analyses were conducted using SAS 9.2 (SAS, Cary, North Carolina, USA).
Participants excluded from analysis
Four hundred and ninety-three HIV-positive MSM enrolled in the study at baseline. Of those, 41 (8%) were found to be beta-globin-negative and were excluded from further analyses. Of the remaining 452 considered eligible for our follow-up analysis, 83 (18%) did not return for at least two follow-up visits. The participants who did not return for at least two visits did not differ significantly from those who did by race/ethnicity, education, lifestyle factors, sexual risk behaviors, or HPV prevalence at baseline. Those who did not return were younger in age (41 vs. 45 years; P = 0.001). There was also an important difference between the indicators of HIV disease status between the two groups. Those who did not return had a lower mean CD4+ cell count (365 vs. 459 cells/μl; P = 0.001), a higher mean HIV-VL (40 242 vs. 10 772 copies/μl; P < 0.001), and were less likely to be taking any ART (67 vs. 86%; P = 0.001). Table 1 contains a description of the study population.
Incidence of human papillomavirus infection
One hundred and twenty-two men had a detectable type-specific incident anal HPV infection over the 2-year follow-up period and the overall incident detection rate was 21.3 per 100 person-years (95% CI 17.7–25.4) (Table 2). The most common incident HPV type was HPV 18, with an incidence rate of 3.7 per 100 person-years. The incidence of any oncogenic HPV infection was 13.1/100 person-years (95% CI 10.5–16.6).
Of the 122 men with one or more incident type-specific infections, 97 (80%) had an incident infection with only one HPV type, 19 (16%) had two different HPV type infections, four (3%) had three HPV type infections, one (0.8%) participant had four HPV type infections, and one (0.8%) participant had six different HPV type infections.
Men who had an incident HPV infection were less likely to have had a prevalent HPV infection of any type at baseline (88 vs. 94%; P = 0.04) and also were less likely to have had a prevalent HPV 6 infection (32 vs. 43%; P = 0.04), the most prevalent HPV type at baseline.
Bi-variable associations between risk factors and incident human papillomavirus infection
Of the demographic, lifestyle, sexual risk, and medical history factors measured over a participant's lifetime, only two were significantly associated with having an incident HPV infection during the follow-up period. Having smoked at least 100 lifetime cigarettes was more common among those without an incident HPV detection (59 vs. 47%; P = 0.03). The only other factor associated with incident HPV infection in bi-variable analyses was having been diagnosed with HIV in the past 12 months. Of the men with an incident HPV infection, 4% had a recent diagnosis of HIV infection compared with only 0.8% in those who had been diagnosed with HIV more than 12 months ago (P = 0.03). Many sexual behaviors in recent time periods were associated with incident anal HPV infection in bi-variable analyses (Table 3).
Multivariable analyses: effect of recent behaviors on incidence of anal human papillomavirus infection
The results of the four multivariable models are included as Supplemental Digital Content Table 1 (http://links.lww.com/QAD/A502). The results of Model 1 (unadjusted estimates) and Model 4 (fully adjusted estimates) are included in Table 4.
Of the sexual behaviors reported in the past 6 months, many remained significantly associated with incident anal HPV infection after adjusting for the baseline value of the factor, potential baseline confounders and time-dependent confounders. Number of receptive partners (partners with whom the participant was the receptive partner – 8+ vs. 0–1) increased the risk of anal HPV infection by 2.9 [(1.9–5.5), P = 0.002] in the fully adjusted model. The frequency of receptive intercourse also was associated with an OR of 2.6 (1.6–4.6, P = 0.004) when 1+ times per week was compared with no acts. Having a higher number of new oral–anal contact partners also increased the odds of HPV infection by a small but significant amount per new partner [OR 1.1 per partner (1.03–1.1)]. The number of insertive partners (partners with whom the participant was the insertive partner) was not associated with incident HPV infection in the fully adjusted models.
The only behavior reported in the past 30-day recall period that remained significantly associated with incident HPV infection in the fully adjusted model was frequency of receptive intercourse in the past 30 days, which increased the odds of incident HPV infection [OR 1.1 (1.03–1.1) per partner].
A small number (∼1%) of patients in our study were treated for their LGAIN or HGAIN, However, the inclusion of this variable in our multivariable models did not substantively change our findings.
Our study represents one of the few natural history studies of a cohort of HIV-positive MSM, followed prospectively for anal HPV infection, together with a detailed evaluation of risk factors for infection during the post-ART era. We had a high rate of follow-up with 82% of our participants returning for at least two visits and at least 1 year of follow-up.
Overall, the incidence of any anal HPV infection (21.3/100 person-years) and of oncogenic anal HPV infection (13.3/100 person-years) was greater than the reported incidence of other sexually transmitted infections (STIs) among HIV-positive MSM [18–20]. Rates of infection with individual anal HPV types were similar to the increased rates of other STIs commonly seen among MSM. The most common anal HPV type detected was HPV 18, followed closely by HPV 16. These two HPV types are the types most commonly associated with anal cancer and are usually found with the highest prevalences in cross-sectional studies of MSM [9,21]. Three other oncogenic HPV types (31, 33, and 58) also had high incidence rates (>2.0/100 person-years). Although these types have rarely been detected in anal cancer specimens, they are known to be associated with cervical cancer and may be important in anal disease in HIV-positive populations. The high incidence rates of oncogenic HPV types imply that the prevalences of these types found in cross-sectional studies are not only the result of persistence of infection but also represent an increased rate of acquisition of new infections. Also important to note was that 20% of the men with an incident HPV infection had more than one new HPV type detected during follow-up and many had multiple oncogenic types. An increased number of HPV types have been associated with a greater likelihood of HPV-associated disease .
Few published studies are available for comparison with the rates of incident anal HPV infection. Critchlow et al.  followed HIV-seropositive MSM for 2 years and found that 38% of men positive at baseline for anal HPV had anal HPV DNA of a new type detected during the follow-up period. Our cumulative incidence of any anal HPV (33%) is comparable with this finding. A recent study of HIV-positive MSM in Canada had type-specific incidence rates that were much greater than the rates we found. However, they defined an incident infection as HPV DNA detected at any single visit after a negative test at baseline. In that study, the incidence rate of HPV 16 was 13.0/100 person-years, compared with our finding of 3.5/100 person-years. When we re-computed our incidence rates using their definition (data not shown), our rates of incident detection were comparably increased (e.g. the rate of HPV 16 infection was 14.8/100 person-years). Because HPV DNA detected from an anal swab sample could be either from an infection in the participant or the result of HPV virus deposited in the anus during anal intercourse, we believe that our stricter definition of an incident infection used in our analyses minimizes potential misclassification of the outcome, which is incident anal HPV. However, we also may be underestimating the true incidence of anal HPV infection.
In our multivariable repeated-measures analyses, recent sex was an important predictor of incident anal HPV infection. Almost all indicators of receptive anal intercourse were significantly associated with incident anal HPV infection. This is to be expected, as receptive anal intercourse is a plausible means of exposure to HPV. However, the association has proved difficult to demonstrate in cross-sectional studies of HIV-positive MSM  because most MSM have a history of receptive anal behaviors and there is little variation in the variable. In studies of prevalence, risk factors for prevalent HPV infection may be factors associated with persistence of HPV infection. The findings of these analyses offer convincing evidence that receptive anal intercourse increases the risk of acquiring anal HPV infection among HIV-positive MSM. The number of new oral–anal contact partners, in which the participant's anus receives oral contact, also was associated with incident anal HPV infection. This is an important finding suggesting another potential route of exposure to anal HPV infection. Oral–anal contact also could explain anal HPV infection among men (and women) who deny receptive anal intercourse.
Our study has a number of limitations. We required at least two follow-up visits to be included in our analysis. This produced an important difference between participants included in our analyses and those not included. Men who were excluded had worse measures of HIV disease status than men who were included. It could be that this inclusion criterion induced some selection bias in our study. This may help explain the diminished association between CD4+ cell count and anal HPV incidence seen in our study in comparison with other studies of HIV-positive populations [23–25] given that men with lower CD4+ levels were excluded from the study. Also, men included in our study who had a single positive anal HPV detection were classified as not having an incident anal HPV. If these men had true incident infections that resolved before the following study visit, they would have been misclassified as not having an incident anal HPV infection. It is unlikely that this misclassification is associated with any of the main risk factors and the bias to our estimates would be towards the null value. Another limitation of our study is that we did not collect information on condom use. However, the likely effect of condom use would be to reduce the effect size (by preventing infection in those who have receptive sex). Thus, our results represent a conservative measure of the true association. Also, this study was conducted in the early ART era and treatment of HIV infection has improved since that time. Therefore, our results may not be generalizable to a population of HIV-positive MSM taking improved ART, who may have better health than our participants, and therefore lower incidences of anal HPV infection.
Our current prospective analyses add to the existing evidence from prevalence studies that anal HPV infection is common among HIV-positive MSM, and that the type distribution of incident infection is similar to that seen in prevalence studies. Recent receptive anal intercourse is an important predictor of anal HPV infection and oral–anal contact is also a risk factor for incident infection. HIV-positive MSM should be counseled about anal cancer and risk factors for HPV infection. They also should be counseled about primary prevention measures such as condom use and the HPV vaccine that was recently approved for prevention of anal HPV infection and HPV-associated disease in men aged 9–26 years .
Author contributions: All authors contributed to the interpretation of data and the overall intellectual content of the paper, and drafting and editing of the manuscript. E.H. and J.M.P. contributed to the study design and data collection. A.L.H., J.E. and J.M.P. contributed to data analysis.
The authors thank A.L.H.'s dissertation committee for review and editing of the manuscript; the anal cohort study staff, the UCSF HPV laboratory, and the study participants for their time and dedication.
Sources of support: The clinical portion of the study was conducted in the General Clinical Research Center, University of California, San Francisco with funds provided by the Division of Research Resources 5 M01-RR-00079, US Public Health Service. The overall research project was supported by NCI grant number R01CA54053. A.L.H. was also supported through the University of California, Berkeley, Fogarty International AIDS/HIV International Training and Research grant (AITRP) 1-D43-TW000003.
Conflicts of interest
Financial disclaimer: A.L.H. – no potential conflict of interest to disclose. J.T.E. – no potential conflict of interest to disclose. E.A.H. – no potential conflict of interest to disclose. J.M.B. – no potential conflict of interest to disclose. N.J. – no potential conflict of interest to disclose. J.M.P. – no potential conflict of interest to disclose.
1. Daling JR, Weiss NS, Hislop TG, Maden C, Coates RJ, Sherman KJ, et al. Sexual practices, sexually transmitted diseases, and the incidence of anal cancer
. N Engl J Med
2. IARCHuman papillomavirus
. IARC monographs on the evaluation of carcinogenic risks to humans
3. Silverberg MJ, Lau B, Justice AC, Engels E, Gill MJ, Goedert JJ, et al. Risk of anal cancer in HIV-infected and HIV-uninfected individuals in North America
. Clin Infect Dis
4. Piketty C, Darragh TM, Heard I, Da Costa M, Bruneval P, Kazatchkine MD, et al. High prevalence of anal squamous intraepithelial lesions in HIV-positive men despite the use of highly active antiretroviral therapy
. Sex Transm Dis
5. Piketty C, Selinger-Leneman H, Grabar S, Duvivier C, Bonmarchand M, Abramowitz L, et al. Marked increase in the incidence of invasive anal cancer among HIV-infected patients despite treatment with combination antiretroviral therapy
6. Palefsky JM, Holly EA, Efirdc JT, Da Costa M, Jay N, Berry JM, et al. Anal intraepithelial neoplasia in the highly active antiretroviral therapy era among HIV-positive men who have sex with men
7. Palefsky JM. Anal squamous intraepithelial lesions in human immunodeficiency virus-positive men and women
. Semin Oncol
8. Crum-Cianflone NF, Hullsiek KH, Marconi VC, Ganesan A, Weintrob A, Barthel RV, et al. Anal cancers among HIV-infected persons: HAART is not slowing rising incidence
9. Palefsky JM, Holly EA, Ralston ML, Jay N. Prevalence and risk factors for human papillomavirus infection of the anal canal in human immunodeficiency virus (HIV)-positive and HIV-negative homosexual men
. J Infect Dis
10. Critchlow CW, Hawes SE, Kuypers JM, Goldbaum GM, Holmes KK, Surawicz CM, et al. Effect of HIV infection on the natural history of anal human papillomavirus infection
11. de Pokomandy A, Rouleau D, Ghattas G, Vezina S, Cote P, Macleod J, et al. Prevalence, clearance, and incidence of anal human papillomavirus infection in HIV-infected men: the HIPVIRG cohort study
. J Infect Dis
12. Vajdic CM, van Leeuwen MT, Jin F, Prestage G, Medley G, Hillman RJ, et al. Anal human papillomavirus genotype diversity and co-infection in a community-based sample of homosexual men
. Sex Transm Infect
13. Parisi SG, Cruciani M, Scaggiante R, Boldrin C, Andreis S, Dal Bello F, et al. Anal and oral human papillomavirus (HPV) infection in HIV infected subjects in Northern Italy: a longitudinal cohort study among men who have sex with men
. BMC Infect Dis
14. Castellsague X, Munoz N, Pitisuttithum P, Ferris D, Monsonego J, Ault K, et al. End-of-study safety, immunogenicity, and efficacy of quadrivalent HPV (types 6, 11, 16, 18) recombinant vaccine in adult women 24–45 years of age
. Br J Cancer
15. Giuliano AR, Palefsky JM, Goldstone S, Moreira ED Jr, Penny ME, Aranda C, et al. Efficacy of quadrivalent HPV vaccine against HPV Infection and disease in males
. N Engl J Med
16. Selvin S. Statistical analysis of epidemiologic data
. 2nd ed.Oxford:Oxford University Press; 1996.
17. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes
18. Rieg G, Lewis RJ, Miller LG, Witt MD, Guerrero M, Daar ES. Asymptomatic sexually transmitted infections in HIV-infected men who have sex with men: prevalence, incidence, predictors, and screening strategies
. AIDS Patient Care STDS
19. Bottieau E, Apers L, Van Esbroeck M, Vandenbruaene M, Florence E. Hepatitis C virus infection in HIV-infected men who have sex with men: sustained rising incidence in Antwerp, Belgium, 2001–2009
. Euro Surveill
20. Wilkinson A, El-Hayek C, Fairley CK, Leslie D, Roth N, Tee BK, et al. Incidence and risk factors associated with chlamydia in men who have sex with men: a cohort analysis of Victorian Primary Care Network for Sentinel Surveillance data
. Sex Transm Infect
21. Damay A, Fabre J, Costes V, Didelot JM, Didelot MN, Boulle N, et al. Human papillomavirus (HPV) prevalence and type distribution, and HPV-associated cytological abnormalities in anal specimens from men infected with HIV who have sex with men
. J Med Virol
22. Chin-Hong PV, Vittinghoff E, Cranston RD, Browne L, Buchbinder S, Colfax G, et al. Age-related prevalence of anal cancer precursors in homosexual men: the EXPLORE study
. J Natl Cancer Inst
23. Jamieson DJ, Duerr A, Burk R, Klein RS, Paramsothy P, Schuman P, et al. Characterization of genital human papillomavirus infection in women who have or who are at risk of having HIV infection
. Am J Obstet Gynecol
24. Palefsky JM, Minkoff H, Kalish LA, Levine A, Sacks HS, Garcia P, et al. Cervicovaginal human papillomavirus infection in human immunodeficiency virus-1 (HIV)-positive and high-risk HIV-negative women
. J Natl Cancer Inst
25. Strickler HD, Palefsky JM, Shah KV, Anastos K, Klein RS, Minkoff H, et al. Human papillomavirus type 16 and immune status in human immunodeficiency virus-seropositive women
. J Natl Cancer Inst
26. MMWRFDA licensure of quadrivalent human papillomavirus vaccine (HPV4, Gardasil) for use in males and guidance from the Advisory Committee on Immunization Practices (ACIP)
. MMWR Morb Mortal Wkly Rep