Circumcision of men with HIV

All the RCTs studied the effects of circumcision in HIV-negative men, and having HIV was an exclusion criterion. But Maria Wawer, co-Principal Investigator of the Rakai RCT, pointed out at CROI 2008: “It is inevitable that some HIV-positive men will seek circumcision. It is the only HIV prevention modality that leaves a mark, and no one wants to be the only guy in the village who is uncircumcised if it becomes regarded as a mark of HIV.”¹

Is the operation safe for men with HIV and does it reduce their risk of transmitting HIV? The answers appear to be ‘yes’ and ‘no’: the rate of adverse events observed in men with HIV seems no higher, though healing may be slightly slower. On the other hand, circumcising men with HIV has resulted in a somewhat higher rate of HIV infection in their partners. This appears linked to resuming sex before proper wound healing has taken place.

In a presentation of results from the Rakai study in HIV-positive men at CROI 2008,¹ lead investigator Maria Wawer said that there was a trend towards higher HIV incidence in the wives of HIV-positive men who were circumcised, compared with wives of men left uncircumcised.

In this Gates Foundation-funded study, 1015 HIV-positive men were randomised either to immediate circumcision or circumcision delayed by two years. Of these, 770 were married and were asked to invite their wives into the study; 566 wives were enrolled of whom 245 (43%) were HIV-negative and therefore in a serodiscordant relationship.

The annual HIV-incidence rate in the wives of the men who were circumcised was 14.4% over two years of follow up compared with 9.1% in women whose partners remained uncircumcised (relative risk = 1.58). This result may be due to chance as it was not statistically significant, but Maria Wawer described it as “unexpected and somewhat disappointing”. It was not due to behavioural disinhibition as condom use was the same in both arms.

If the increased incidence in the partners of circumcised HIV-positive men is real and not due to chance, it may largely have been due to men resuming sex before their circumcision wound was certified as having healed.

In wives of men who resumed sex more than five days prior to certified wound healing, 5 out of 18 (28.8%) became HIV-positive themselves. In contrast, 6 out of 63 wives of men who resumed sex no earlier than five days prior to certified wound healing were infected (9.5%), and this was statistically equivalent to 6 out of 68 wives of men who remained uncircumcised (8.8%).

After six months, HIV incidence declined to 5.7% a year in partners of circumcised men and 4.1% in wives of uncircumcised men, a difference which was also not statistically significant. The results may be partly due to HIV-positive men tending to heal more slowly from circumcision than HIV-negative men.

An early study from Rakai² found statistically significant evidence that circumcision might have a role in protecting the female partner – but that the effect of viral load may be crucial too. This suggests that the effects of circumcision and antiretroviral therapy might be synergistic in reducing HIV transmission.

This study showed an overall protective effect to the female partner of HIV-positive circumcised, compared with uncircumcised, men of 26%. This was statistically significant (p=<0.05).

But this depended on the man’s viral load. If men had viral loads of over 50,000, circumcision had no protective effect for women. But among men with viral loads under 50,000, there were zero transmissions from circumcised men, compared with 26 transmissions from uncircumcised men, and this was highly statistically significant. Circumcision also appeared to protect women from acquiring bacterial vaginosis and trichomoniasis, but not other STIs.

Another Rakai study had suggested that circumcision of HIV-positive men may have an indirect protective benefit to women in the long term - more than six months after surgery.

In a study presented to CROI in 2006,³ Ronald Gray said that there was indirect evidence that circumcision might reduce HIV in female sexual partners by approximately one third.

In an observational study from Rakai, the incidence of HIV infection was measured in 44 women with circumcised HIV-positive partners and 299 women with uncircumcised HIV-positive partners.

The incidence of HIV infection was just under 7 infections per 100-person years for women with circumcised male partners compared to 10 infections per 100-person years for women with uncircumcised partners. However, the difference was not statistically significant (p = 0.22).

In 2010, a substudy⁴ of the Partners in Prevention study, a large RCT of herpes prophylaxis as a possible HIV-prevention measure, found that male circumcision reduced the risk of an HIV-positive man transmitting HIV to a female sex partner by 40%. However, this reduction was not statistically significant (p=0.1). The men in this study had undergone circumcision in childhood, so it was able to determine the effects of circumcision on HIV-transmission risk after full wound healing.

One study⁵ of the effect of circumcision on male-to-female HIV transmission relied on asking women about the circumcision status of their partners. This study, published in August 2007, was part of a non-randomised cohort study of the effect of hormonal contraception on women’s risk of acquiring HIV.

This two-year study found that male circumcision had “little influence” on the incidence of HIV in women. But there was a suggestion that women with high levels of sexual risk were slightly less likely to contract HIV if their partners were circumcised, and the investigators suggest that this finding should be explored in further studies.

The study population comprised over 4000 Ugandan and Zimbabwean women aged 18 to 35, including 393 Ugandan women with high HIV risk.

At baseline, almost three  quarters of the women (74%) reported that their partner was uncircumcised and 22% said their partner was circumcised.

A total of 210 women became infected during follow up. The unadjusted HIV incidence was 2.03 per 100 person years for women with circumcised partners, 2.96 per 100 person years for women with uncircumcised partners, and 3.51 per 100 person years from women who did not know if their partner was circumcised.

Was this because women with uncircumcised partners also took more HIV risks? No,  rather the opposite. It was the women with circumcised partners who took more risks; there were statistically significant differences in terms of number of unprotected sex acts, sexually transmitted infections and having sex with a ‘risky’ sexual partner (a man with HIV or with STI symptoms).

The investigators then performed a number of statistical analyses. In their first unadjusted model, they found that the risk of HIV infection was reduced 31% for women with circumcised partners compared to women whose partners were uncircumcised. This difference was of borderline statistical significance (p = 0.06). When the investigators adjusted their model to take into account confounding factors such as age and number of sexual partners, there was no statistically significant difference.

In further analysis, whether women were assessed as being ‘low risk’ or ‘high risk’ for HIV acquisition was taken into consideration. ’High-risk’ women included patients from sexually transmitted infection clinics, sex workers and military wives. The investigators found that low-risk Ugandan and Zimbabwean women had a similar risk of HIV infection, regardless of their partner’s circumcision status.

The high-risk Ugandan women were considerably less likely to get HIV if their partner was circumcised. But because numbers were small (only 19 infections, 2 amongst partners of circumcised men), this was not statistically significant (HR 0.16; 95% CI, 0.02 to 1.25).

The investigators comment: “After adjustment, we did not observe a significant protective effect of male circumcision overall. For a small group referred through high-risk settings, we found a suggestion of a lower HIV risk for women with circumcised partners, but the suggestion that male circumcision may be protective for these high-risk women must be interpreted very tentatively.” Nevertheless, they conclude that this finding “warrants further investigation.”