Heterosexual HIV transmission more likely with sexually transmitted infections, early or late stage disease, no circumcision, or anal sex

Researchers conducting a meta-analysis of studies of the risk of HIV transmission during heterosexual sex have found that, in high-income countries prior to the introduction of combination therapy, the risk per sexual act was 0.04% if the female partner was HIV-positive, and 0.08% when the male partner was HIV-positive. However, these rates were considerably higher in lower-income countries, if the source partner was in either the very early or the late stage of HIV infection, or if one partner had genital ulcer disease, write the researchers in the February issue of The Lancet Infectious Diseases.

Marie-Claude Boily and colleagues attempted to identify all relevant observational studies of a sufficient methodological quality that provided empirical estimates of the transmission risk per sexual act (rather than the cumulative risk during an ongoing relationship with an HIV-positive person).

Looking for material on HIV-1 only, they found 43 relevant publications covering 25 different study populations. More than half were conducted in the USA or western Europe, with most of the others carried out in Africa, and a handful conducted in Thailand and Haiti. Although the researchers looked for studies published up to September 2008, almost all the reports used data that was collected in the 1980s or early 1990s, which means that the findings do not reflect combination therapy’s impact on transmission.

High-income and low-income countries

Pooling the data from studies in high-income countries, the researchers calculated that the risk of transmission from an HIV-positive man to his female partner was 0.08% per sexual act: in other words, it was likely to occur once every 1250 sexual acts. When it was the female partner who was HIV-positive, the male partner’s risk of acquiring HIV was 0.04% per sexual act – in other words, once every 2500 sexual acts.

The findings from each of these studies were broadly consistent, with the result that the 95% confidence intervals for the above figures were not too wide (for example, for transmission from men to women, 0.04% to 0.16%). This was not the case for the pooled data from studies in Africa, Thailand and Haiti.

In these countries, the researchers calculated a transmission risk of 0.19% when the male partner was HIV-positive, and 0.87% when the female partner had HIV. However the wider confidence intervals (for example, from men to women, 0.28% to 2.6%), could, the authors suggest, reflect poorer study quality, a wide variation in risk factors between study populations or some under-reporting of high-risk behaviour. They speculate that the overall higher apparent risk could be driven by higher rates of sexually transmitted infections or higher viral load levels.

Nonetheless, they point out that the figures from low-income settings suggest that there is a greater risk of transmission from women to men than the other way round, which is the inverse of the high-income country findings and is generally considered less biologically plausible.

One possibility could be that men in these settings might be more likely to have sex outside their primary relationship than women, and so what appear to be transmissions from the primary female partner are in fact infections acquired elsewhere.

Moreover, when the researchers excluded studies which involved sexual acts as part of commercial sex work (either as a client or a sex worker), the risk of female-to-male transmission decreased.

Cofactors

In fact, comparing populations involved in commercial sex work with those who were not (in any part of the world), the transmission risk was eleven times higher. The authors judge that this increased risk may be primarily driven by high rates of sexually transmitted infections in these populations.

Moreover, the per-act transmission risk when one of the partners had genital ulcer disease was 2.8%, or once every 36 sexual acts. Compared to situations when it was known that there were no sexually transmitted infections, the transmission risk was five times higher.

Two studies provided data on the circumcision status of male partners who were at risk of HIV infection. In both cases, the transmission risk was higher for men who were not circumcised, especially if they also had genital ulcer disease.

A few studies incorporated information on the source partner’s disease stage, focusing on early infection and late-stage disease, both periods when viral load is likely to be high. With an asymptomatic partner, the per-act risk was 0.07%, compared to 0.66% at early stage (nine times higher), and 0.55% at late stage (seven times higher).

However, because few studies have been conducted in the past decade, no estimates of the transmission risk were specifically provided for a partner taking antiretroviral treatment.

Just two studies provided data on the risk of transmission from anal sex, and the pooled estimate was 1.7% per act (i.e. about once every 60 sexual acts). However the 95% confidence interval was wide: 0.32% to 8.9%.

Having conducted a similar meta-analysis a few months ago in the same journal, Kimberley Powers and colleagues argued that focusing on a single figure as the per-act risk of transmission is misleading and frequently leads to the risk of transmission being underestimated. In the context of sexually transmitted infections, lack of circumcision, anal sex, acute infection or late-stage infection, she argued that: "The heterosexual infectivity of HIV-1 might exceed the commonly cited value of 0.001 [0.1%, one in 1000] by more than an order of magnitude. The vast extent of the current epidemic is more easily understood in the context of these biological cofactors, which create a more favourable environment for HIV transmission."

For her part, Marie-Claude Boily suggests that a figure of 0.07% or 0.08% may "represent the average per-vaginal-sex-act transmission in the absence of cofactors, during the asymptomatic stage."

She also notes the methodological challenges of these studies, and suggests that better quantification of per-act infectivity could help epidemiologists predict future patterns of the HIV epidemic as well as contributing to the development of more effective prevention strategies.