Circumcision status and risk of HIV seroconversion in the HIM cohort of homosexual men in Sydney David Templeton1,2, Fengyi Jin1, Garrett Prestage1, Basil Donovan1,3, John Imrie4, Susan Kippax4, John Kaldor1, Andrew Grulich1 1National Centre in HIV Epidemiology and Clinical Research, University of New South Wales 2Sexual Health Service, Sydney South West Area Health Service 3Sydney Sexual Health Centre 4National Centre in HIV Social Research, University of New South Wales

Overview Background Methods: the HIM study Related analyses in HIM Demographic predictors of circumcision Validation of circumcision status Circumcision & HIV seroconversion Univariate Multivariate In those reporting no receptive UAI First, I’ll briefly outline the methods of the Health in Men cohort study, and then discuss related background analyses including demographic predictors of circumcision status, the sub study validating participants' self-report of circumcision status by clinical examination, and multivariate analyses of risk factors for HIV seroconversion in HIM. Finally, I’ll present results of our analyses of the association of circumcision status with incident HIV infection in HIM

Background Circumcision reduces HIV acquisition among heterosexual African men Few data in homosexual men X-sectional studies – conflicting1,2,3 Prospective cohort 4 - HIV risk in uncircumcised - AOR 2.0 (95% CI 1.1-3.7) There are now three randomised trials from Africa which have found HIV incidence is reduced by approximately 60% among circumcised heterosexual African men. However, data on the association of circumcision with HIV in homosexual men is limited and conflicting. While one US cross-sectional study in the early 90’s found uncircumcised homosexual clinic-attendees to be more than twice as likely to be HIV-infected than circumcised men, there was no association between circumcision status and prevalent HIV infection among San Francisco STD clinic attendees, nor among a smaller group of Australian gay men who had recently seroconverted to HIV and reported insertive unprotected anal sex as their highest HIV exposure risk. One prospective cohort study published in JAIDS a couple of years ago, reported a doubling of risk associated with being uncircumcised after adjusting for demographic factors, drug use and sexual behaviour 1. Kreiss et al. J Infect Dis 1993:168:1404-8 2. Grulich et al. AIDS 2001;15:1188-9 3. Klausner et al. National STD prevention conference, Florida, May 2006 4. Buchbinder et al. J Acquir Immune Defic Syndr 2005;39:82-9

Methods HIM cohort HIV-neg homosexual men recruited 2001-2004 Annual face-to-face & annual telephone interviews Risk factors for HIV Detailed behavioural data Annual STI testing & self-reported STIs last 12 months Circumcision status Self-reported; subgroup validated by examination HIV identification Annual HIV testing & match against National HIV Register Statistical analysis Association between circumcision & HIV seroconversion assessed using Cox regression The Health in Men study recruited Sydney gay community attached men between 2001 and 2004. At recruitment participants underwent an HIV test and HIV negative men were enrolled into the study. Detailed sexual behavioral data was collected every six months, at alternating face-to-face and telephone interviews. All participants were offered STI testing at face-to-face interviews, and approximately 90% agreed to testing each year. In addition, participants self-reported STI diagnosed outside the study in the previous 12 months. Circumcision status was self-reported at baseline and validated by clinical examination in a sub group of participants. All participants underwent annual HIV testing, and an annual match with the National HIV register was performed to identify participants who were lost to follow-up from the study, whose HIV infections were diagnosed elsewhere in Australia

Source of recruitment % Gay community events 55 Word of mouth 13 Gay venues 7 Previous study 5 ACON or other gay organisation Internet 4 Clinics Gay press 3 Sydney periodic surveys 2 Other A diverse recruitment strategy was used for the HIM study. As you can see, most participants were recruited at gay community events, with only a small number recruited at clinics. Therefore our study population was considered to be truly community-based

Retention at two years: 80% Median age: 35 yrs (range 18-75) Demographics 1427 men enrolled Retention at two years: 80% Median age: 35 yrs (range 18-75) Median follow-up time: 3.0 years Circumcision rates similar at baseline and 3rd year of follow-up (66% vs. 69%, p=0.15) 95% self-identified as gay or homosexual Over 1400 men were enrolled with an 80% retention rate at 2 years. Median age of participants was 35 years and median follow-up time was 3 years. Participants lost to follow-up did not differ significantly in terms of circumcision status compared with the entire cohort at baseline. 95% of participants self-identified as gay or homosexual.

Demographic predictors of circumcision 66% circumcised 92% <1 yr old 6% 1-18 yrs old 3% >18 yrs old Independent predictors of circumcision Age ( p-trend<0.001) Ethnicity (p<0.001) Country of birth (p<0.001) Critics of some past research on this topic have cited inadequate control for demographic factors which could confound any association of circumcision status and HIV seroconversion. A previous analysis sought to elucidate these factors which I will briefly present. Two-thirds of participants reported being circumcised at baseline with the vast majority circumcised as an infant. On multivariate analysis, older age was strongly associated with being circumcised, as were ethnicity and country of birth. Excluding the seventeen Jewish and Moslem participants, all of whom were circumcised, there was no association between religious affiliation and circumcision status. There was also no association with level of education or income. Templeton et al. Sex Health 2006;3:191-3

Validation of circumcision status Self-reported circumcision status at baseline in HIM Examination by trained nurse - blinded to participants’ reported circumcision status n = 247 87% participation rate 100% concordance between baseline self-report and examination Correlation of clinical examination findings with self-reported circumcision status has been reported by some authors to be poor We were unable to find any published research among gay men and therefore sought to validate participants self-reported circumcision status. From February to June this year, 87% of HIM participants approached at their annual face-to-face interview agreed to be examined by the study nurse who was unaware of their reported circumcision status at the baseline interview. There were no cases of misclassification of self-reported circumcision status among the 247 men examined.

Circumcision & HIV: univariate 49 HIV seroconversions identified in HIM Incidence 0.80 per 100 PY n Incidence (per 100PY) HR 95% CI P value Circumcision No (n = 488) 13 0.87 1 --- 0.835 Yes (n = 938) 29 0.93 1.07 0.56-2.06 Overall to the end of 2006 there were, 49 HIV seroconversions in HIM, an incidence of 0.8 per 100 person-years. PY, person-years; HR, hazard ratio; CI: confidence interval

HIV Seroconversion: multivariate Adjusted HR 95% CI P value Circumcision 0.705 No 1 --- Yes 0.88 0.44-1.73 Age 0.653 Per year increase 1.01 0.97-1.04 UAI according to partners’ HIV status <0.001 No UAI With HIV negative only 2.98 1.07-8.31 With HIV status unknown 5.07 1.75-14.66 With HIV positive 20.14 6.98-58.12 Anal Gonorrhoea (NAAT) 0.007 7.52 1.74-32.45 Anal warts (self-report) 0.016 2.97 1.22-7.21 Because of its strong association with being circumcised, we forced age into the multivariate model despite a lack of univariate association with HIV incidence. Of the other demographic factors associated with circumcision in HIM, neither ethnicity nor country of birth were associated with HIV seroconversion, so were not included in the multivariate model. We also controlled for unprotected receptive anal intercourse according to the reported status of participants’ sexual partners, as this was the single sexual behavioral variable most strongly associated with HIV seroconversion on univariate analysis. Finally, we controlled for self-reported anal warts and study-confirmed anal gonorrhoea as these were the only STIs significantly associated with HIV seroconversion. After controlling for age, unprotected anal sex by partners serostatus, and these anal STIs, there remained no association between HIV incidence and being circumcised.

Circumcision & HIV: stratified analysis Participants not reporting receptive UAI Incidence (per 100PY) Univariate Age-adjusted n HR 95% CI p Circumcision 0.989 1.00 No 3 0.36 1 --- Yes 6 0.35 0.99 0.25 - 3.96 0.24 - 4.09 As most gay men are infected with HIV via receptive unprotected anal intercourse, and thus there is no potential for their own circumcision status to protect them against HIV, we performed a stratified analysis on participants more likely to have contracted HIV by insertive anal sex. However there were very small numbers of seroconverters who reported no unprotected receptive anal sex, and analysing these nine there remained no association, even when controlling for age PY, person-years; HR, hazard ratio; CI: confidence interval

Adequate control for confounders Weakness Lack of power Conclusions No association of circumcision with HIV seroconversion among homosexual men in the HIM study Strengths Prospective design Validation Adequate control for confounders Weakness Lack of power Premature to promote circumcision as HIV prevention intervention in homosexual men We found no association of circumcision status with incident HIV infection in a cohort of community-attached homosexual men in Sydney. This lack of association is not surprising given that multivariate analysis of anal sexual behaviours as risk factors for HIV in HIM found an association with receptive, and not insertive, anal sexual practices. The strengths of our study design include validation of circumcision status, investigation of & subsequent control for potential confounding factors. However, the low HIV incidence, especially in those reporting no unprotected receptive anal sex, limits firm conclusions that can be drawn. Despite convincing evidence of the effects of circumcision on HIV incidence in heterosexual African men, the few and conflicting data that exist in homosexual men suggests it would be premature to consider promoting circumcision as an HIV prevention intervention in this group.

Acknowledgements HIM participants and the HIM study team Collaborators and colleagues at NCHECR & NCHSR Dr Catherine O’Connor, SSWAHS Funding: NCHECR & NCHSR are funded by the Australian government Department of Health and Ageing. HIM project funded by the Australian Government Department of Health and Ageing (Canberra), the New South Wales Health Department (Sydney), the National Institutes of Health, a component of the U.S. Department of Health and Human Services (NIH/NIAID/DAIDS: HVDDT Award N01-AI-05395), the National Health and Medical Research Council (NHMRC) (#400944). Dave Templeton is supported by an NHMRC Public Health Scholarship (#351044). Becton Dickinson provided GC/CT testing materials *Jin et al. Anal STIs as risk factors for HIV seroconversion: data from the HIM cohort. IAS 2007: Poster TUPEC010