The ability of antiretroviral therapy to reduce sexual infectiousness has not been enough to compensate for increases in risky sex among gay men in the Netherlands, according to a recently published study in the journal AIDS. The investigators, who used mathemetical modelling to explain recent increases in HIV diagnoses, conclude that sexual risk behaviour must be reduced to pre-1996 levels for treatment to have an impact on the Dutch gay HIV epidemic.
The impact of antiretroviral therapy (ART) on sexual transmission has been the subject of intense debate this year, following publication of guidance from the Swiss Federal AIDS Commission (EKAF) stating that an individual with a blood plasma undetectable viral load is not infectious.
Although some of the concerns have been about the scientific accuracy of that statement, much of the debate has been due to concerns about increased sexual risk-taking on a population level cancelling out any gains in ART’s effect on infectiousness.
In 2000, a mathematical modelling study from San Francisco suggested that the availability of ART had resulted in rates of risky sexual behaviour amongst gay men equal to, or greater than, the beneficial impact of ART on per contact HIV transmission. Another mathematical model, focusing on gay men in Australia, came to similar conclusions in 2004.
However, a more recent mathematical model examining the impact of ART in resource-limited settings suggested that it may have a positive impact on the HIV epidemic, particularly in population where these is less frequent partner change, and a similar model also appeared to show benefits when examining the epidemic in in the Canadian province of British Columbia (B.C.). So much so, in fact, that last week, B.C. announced that it was planning to expand access to ART as a prevention tool.
To determine the impact of a decade of potent ART on the HIV epidemic investigators from Amsterdam and London evaluated the impact of sexual risk behaviour, HIV testing behaviour and ART uptake on the HIV epidemic between 1980 and 2004 using a mathematical model fitted to data from several national databases that provided them with extensive information on epidemic trends amongst gay men in the Netherlands.
The model’s most important factor was its estimation of the prevalence of infectious individuals, weighted by their relative infectiousness – which depended on their stage of infection, whether they were diagnosed, and whether or not they were on fully suppressive ART – and the incidence of new infections.
They did this by using data from 130 gay men from the Amsterdam Cohort who were identified during seroconversion prior to 1996. They estimated that there were six periods of HIV infectiousness, starting with primary infection (which lasted an average of 0.24 years, or just under three months) followed by five more periods (undiagnosed untreated; diagnosed untreated; time after first treatment failure; time after second treatment failure; and time after third treatment failure, or AIDS) each lasting an average of 1.89 years.
They also assumed that time on successful treatment did not result in being sexually infectious, and also accounted for the relative infectiousness of different stages of HIV disease (where primary infection and AIDS were the periods of highest viral load and transmission risk) based on data from heterosexual couples in Uganda practicing vaginal sex.
They then fitted the model to data from the ATHENA cohort that included annual new HIV diagnoses and annual new AIDS cases, as well as country of infection (they calculated that 14% of diagnosed infections in gay men did not take place in the Netherlands).
Finally, they analysed four time periods: 1980-1983; 1984-1995; 1996-1999; and 2000-2004. Their calculations suggest that new HIV infections peaked in 1983, fell again due to the widespread uptake of 'safer sex', and started rising again in 2000 following increased sexual risk behaviour among gay men.
Their model produced a ‘reproduction number’ for these time periods, which was the average number of people an HIV-positive gay man could infect over his whole infectious lifespan, and which incorporated sexual risk behaviour, the impact of diagnosis, and the impact of ART. A reproduction number above 1 would increase the HIV epidemic and one below 1 would decrease the epidemic.
They found that after 1996, once potent ART was introduced, the reproduction number declined to 0.76 although this was not as great as it could have been due to an estimated 18% increase in the risk behaviour rate.
Even though a number of factors should have reduced the reproduction number further between 2000 and 2004 – including a reduction in the estimated time between infection and diagnosis, and widespread uptake of treatment – a further increase in risk behaviour during this period (an estimated 66% increase, just 29% lower than before the concept of ‘safer sex’ was created in 1984) meant that the estimated reproduction number for 2000-2004 was 1.04.
This, they write, is “near or above the critical epidemic threshold, and thus indicat[es] that HIV may once again be spreading epidemically among MSM in the Netherlands."
Their model estimated that in 2005, just under one in four (24%) gay men with HIV in the Netherlands were unaware of their infection, and that these 24% accounted for 90% of sexual HIV transmission. However, they had assumed that gay men reduced sexual risk-taking by 50% following diagnosis, based on data from a 2005 meta-analysis of previous studies.
A more recent UK study found that diagnosed gay men were between 1.6-times and 3.29-times more likely to have sex that risked onward HIV transmission than undiagnosed gay men. It is possible, therefore, that the investigators may have overestimated the impact of undiagnosed individuals on transmission.
"On the basis of these model estimates," they write, "we conclude that HAART has played an important role in limiting transmission but that any gains made have been more than offset by increases in the risk behaviour rate. Had these increases not occurred in the HAART era, the reproduction number would have declined to 0.6, and the epidemic would have been in convinced decline.”
Although mathematical models are notoriously difficult to reliably estimate the dynamics of HIV epidemics, due to the use of so many assumptions, the investigators argue that their model is more robust than others because it used “several national databases recording diagnoses of HIV infection and AIDS, and deaths, allowing the diagnosis rate to be estimated reliably.”
They add that although gay men in the Netherlands are testing more frequently than before, their model suggests that testing frequency alone does not explain recent increases in diagnoses, but rather “a substantial increase in transmission. Our estimates were corroborated by changing trends in CD4 cell count at diagnosis, where a recent increase in the proportion of newly diagnosed individuals with high CD4 cell counts is apparent.”
They conclude that their model “suggests that the only way to reverse epidemic spread...is to reduce the risk behaviour rate from current levels” and that “the most effective intervention is to bring risk behaviour back to pre-HAART levels.”