Summary: is treatment prevention?

Evidence has been accumulating ever since the advent of combination antiretroviral therapy that people with suppressed or undetectable HIV viral loads, especially those on therapy, are a great deal less likely to transmit HIV than untreated persons.

In the last two years, two pivotal studies in heterosexual couples where one of the partners has HIV have demonstrated that HIV-positive people who are on antiretroviral (ARV) treatment are about 20 times less likely to transmit the virus to their partners than people who are not taking treatment.

First, in February 2010, the Partners in Prevention study, which was designed to find out if treating herpes might reduce HIV transmission, found that the minority of its participants who started taking ARVs were 92% less likely to transmit HIV to their partners once they started therapy.¹ However Partners in Prevention was not a randomised controlled trial of the prevention potential of ARV treatment and so it could not prove beyond a shadow of doubt that this was the true reduction in risk.

The definitive proof came in May 2011 when the HPTN 052 study was stopped three years earlier than planned. This study randomised the positive partner in heterosexual couples either to start taking ARVs immediately, at an average CD4 count of 436 cells/mm³, or to delay taking them till their CD4 count fell below 250 cells/mm³.

The study found that people on treatment were 96% less likely to transmit HIV to their partners than untreated people. Michel Sidibé, the Executive Director of UNAIDS, commented that, “This breakthrough is a serious game changer and will drive the prevention revolution forward. It makes HIV treatment a new priority prevention option.”²

HPTN 052 shows that, if a sufficiently large proportion of the HIV-positive population could be treated and their viral loads brought down to an undetectable level, transmission might become rare enough for the epidemic to stop in its tracks.

On a population level this can be expressed as a reduction in the ‘community viral load’, the viral load averaged over the whole population, a term first coined in a paper from San Francisco in 2004.³

This paper suggested a new approach to HIV prevention. At the start of the epidemic the only proven ways to reduce the risk of infection were needle-exchange programmes, the provision of condoms (and female condoms from 1993 on), and behaviour-change programmes to reduce sexual risk and encourage the use of condoms. With the exception of needle exchange these have all had limited, and unpredictable, success.

Since then two approaches were scientifically proven to reduce HIV infection, but they can only benefit certain sections of the population. One is the prevention of mother-to-child transmission, a largely ARV-based strategy which started to be adopted in the late 1990s. The other is male circumcision which, though proven to be effective in 2005, is only of immediate benefit to heterosexual men (though female partners will benefit in the longer term) and is only likely to help reduce HIV incidence on a population level in countries with generalised epidemics.

Evidence for the use of ARVs as prevention methods in HIV-negative people, either as orally-dosed pre-exposure prophylaxis (PrEP) or as a topical microbicide, has been strengthened by positive results from the CAPRISA 004 study of a tenofovir gel microbicide for women⁴ and the iPrEx study of oral tenofovir/FTC as PrEP in gay men.⁵ However the overall efficacies demonstrated (39% in CAPRISA and 43% in iPrEx) were not enough for the use of these methods to be licensed without question. The studies made it clear that a number of advances in formulation, adherence support and user targeting would need to be made if they are going to be widely adopted.

Disappointment in the field of vaccine research showed that developing a vaccine against a retrovirus like HIV will, at the very least, require fundamentally new technologies.

Until now, the epidemic has therefore lacked a generally applicable biomedical prevention method which is less dependent on the vagaries of human sexual behaviour than barrier methods.

In the last four to five years, however, mass treatment of people with HIV as a prevention method has gained credibility as a way of stopping or slowing the epidemic.

Previous to this, there were a number of reasons why it was not seen as viable:

• The available ARVs, especially the cheaper ones used mainly in the lower-income world such as d4T, were sufficiently toxic that providing them to people with high CD4 counts who were not in immediate clinical need was likely to do more harm than good.
• Viral-suppression rates were less than 80% in 2000 even in patients with near-perfect adherence.⁶ These improved with the advent of boosted protease inhibitors, but were still only in the order of 80-85%. For this reason, and because of fear that people with high CD4 counts might have poorer adherence, there was concern (and some mathematical models appeared to show) that population-based ARV provision risked generating huge amounts of drug-resistant HIV.
• Even providing ARVs to stop people dying of AIDS has required an unprecedented expansion of health programmes and funding for them in lower-income countries, and is expensive even in richer ones. In the absence of strong cost-benefit data, population-wide ARV provision was seen as financially unviable.

Several developments, however, even before the Partners in Prevention and HPTN 052 results, had contributed to making population-level ARV provision seem more effective and feasible as a strategy:

• In 2006 the SMART study⁷ found that patients who kept taking ARVs at CD4 counts of over 350 cells/mm³ were about half as likely to die as those who stopped. This result, unexpected at the time, reset the perceived risk/benefit ratio for taking HIV treatment in favour of taking it. Other studies have since confirmed this, with some finding that the benefit of taking HIV therapy applies even to people with CD4 counts over 500 cells/mm³.⁸ Newer-generation HIV drugs like tenofovir and some of the newer protease inhibitors appear less toxic than the previous-generation drugs such as d4T, AZT and indinavir, and are less associated with stigmatising fat-redistribution effects.
• Rates of viral suppression seen in people taking ARVs are now in the region of 90% or more in many settings,⁹ and HIV drug resistance has become considerably rarer as the number of people taking ARVs with an incompletely suppressed viral load has fallen.¹⁰
• Although universal provision of antiretrovirals is still a very ambitious goal, especially in the context of global recession when clinical ARV drug programmes in lower- and middle-income countries are facing rationing,¹¹ the achievement of high treatment coverage for those in clinical need in countries such as Botswana has made universal coverage at least seem within the realm of possibility.
• Studies have provided evidence that people taking antiretrovirals are more than ten times less likely to transmit HIV to their partners than those not on treatment.¹ This means that HAART has the highest efficacy so far seen for any ‘real-world’ HIV-prevention intervention. (Condoms have 95-99% efficacy in ideal use, but in real-world settings 100% attempted use has an efficacy of no more than 85% - see Condoms and lubricants.)
• There is some tentative evidence that the reduction in infectiousness at a population level produced by very high ARV coverage rates in some communities has started to produce declines in HIV incidence.^12,13

A peripheral development, though one that signalled an important shift in thinking in HIV prevention, was the issuing of a statement by the Swiss Federal AIDS Commission (SFAC) in January 2008.¹⁴ What is now known as the “Swiss Statement” was the first time a group of clinicians had given backing to something that was already suspected to be the case amongst certain sections of the HIV-affected community. This was that, under certain carefully defined conditions, some people with undetectable viral loads were essentially uninfectious.

This statement has been revised and qualified since then, but although it was issued primarily in reaction to the criminal prosecution for ‘HIV exposure’ (i.e. a person with HIV having unprotected sex, even with disclosure of HIV status) of people who were on antiretrovirals, it served to raise awareness within the HIV community and elsewhere that ARVs could profoundly reduce the likelihood of someone transmitting HIV. However it also caused alarm amongst some people who felt it might serve as justification for abandoning safer sex.

Moreover, in 2006, Julio Montaner and his team from the British Columbia Centre for Excellence in HIV/AIDS had published an article in The Lancet¹⁵ that used a mathematical model to forecast that providing antiretroviral therapy to every HIV-positive person in the world would stop the HIV epidemic within 50 years at an average cost of $7 billion a year. At the time, this model attracted relatively little attention and was primarily seen as an advocacy tool.

However, when, in 2008, after the release of the Swiss Statement, the same team published an article in the Journal of Infectious Diseases applying the same mathematical model to British Columbia alone, and predicting that immediate and considerable declines in HIV diagnosis would follow from a 50-100% increase in ARV coverage,¹⁶ the Health Minister of the province announced a new, aggressive policy of diagnosing and treating the maximum number of individuals possible in order to reduce the HIV-infection rate.

This was the first public endorsement of a strategy that is sometimes called ‘Universal Test and Treat’ (UTT). The part of the strategy that requires an expansion in testing and diagnosis is described in ‘The Role of HIV Testing in HIV Prevention’.

In January 2009, the World Health Organization published its own mathematical model, based on the British Columbia one, that calculated that HIV transmission could be virtually eliminated by 2020 in countries with high levels of HIV prevalence, such as South Africa, if it were possible to persuade everyone in the community to test for HIV infection once a year and then provide antiretroviral therapy to all who tested HIV-positive.¹⁷

This model caused considerable controversy amongst experts, both in terms of the practicality of putting anything like universal testing and treatment into practice, and also from human rights activists who were concerned that such programmes could become coercive. The controversy was such that the WHO hosted a consultation in November 2009 (see www.who.int/hiv/events/artprevention/en/index.html) to look at the issues of implementing expanded testing and treatment.

Subsequent work at Imperial College, London has refined the WHO’s rather simple model, finding that in some circumstances, where the rate of sexual mixing and partner change is not too high, universal treatment would not necessarily be required but that treating in accordance with BHIVA and EACS guidelines and testing every four or five years would be sufficient to produce an 85% reduction in new infections.¹⁸

However, this model was extremely sensitive to sexual behaviour - much higher and probably unrealistic rates of testing and treatment would be needed for populations with high rates of sexual-partner change.

These models have not been received with universal approval by some HIV and prevention activists. Some pointed out that the evidence backing up the treatment-as-prevention model is weak and that several crucial areas are under-researched.

These include:

• whether we need to gather more evidence on the long-term toxicities of ARVs before making final decisions on long-term universal treatment programmes
• whether someone with a sufficiently low blood viral load can ever be considered completely uninfectious. People who are ‘undetectable’ by standard tests actually maintain a persistent blood viral load averaging about three copies/ml¹⁹
• whether blood plasma viral load is correlated enough with viral load in sexual fluids to guarantee that a viral load test is a true indicator of infectiousness
• whether expanded treatment would reduce transmission via anal sex as much as it would via vaginal sex
• the exact degree to which sexually transmitted infections exacerbate HIV infectiousness and whether treating them reduces infectiousness
• whether expanding treatment might take too much attention away from research into, and advocacy for, other HIV-prevention methods and programmes
• whether increasing risk behaviour by people believing they, or their partners, are non-infectious might cancel out any benefits
• whether people in relatively good health will be sufficiently motivated to adhere to long-term treatment
• the role played in the epidemic by transmission from people in primary HIV infection, who are less likely to be diagnosed
• the overall practicality and cost effectiveness of upscaling testing and treatment in this way.

Concerns were also expressed that over-promotion of the concept might lead to compulsory or coercive testing and treatment programmes in countries with authoritarian governments, and that the WHO and other modellers had underestimated the importance of HIV stigma as a disincentive to test.²⁰ The WHO conducted its own consultation in 2009 to discuss some of these issues.

Evidence that widespread antiretroviral coverage is starting to bring down HIV prevalence and incidence is hard to find. This is not because the concept is inherently flawed. It is because the proportion of the HIV population that needs to be diagnosed and treated in order to bring down the community viral load sufficiently far to make a difference is so high that it has rarely been achieved anywhere.

Situations of falling HIV incidence have generally been ones where a population spontaneously, or in response to campaigns, curbs its own behaviour risk. Falls in incidence may take a long time to translate into falls in prevalence.

(Prevalence is the proportion of the population who, at a specific time, have the disease: so, for instance, if HIV prevalence in a population is 12.5%, then one in every eight people has it. Incidence is the rate of infection: the proportion of people who develop the disease within a specific time frame. In a chronic disease like HIV this is usually expressed as the number of infections per 100 people per year (or per, say, 100,000 for a rarer condition). So in a community with an annual HIV incidence of 2%, then one in every 50 people would be infected with HIV in a year.)

Initially, testing more people for HIV and putting more of them on successful ARV treatment will tend to produce a rise in HIV prevalence. In places where there has been high AIDS-related mortality, this is because the death rate starts to fall faster than the infection rate, and more people survive to infect others. In developed-world settings, where the new diagnosis rate currently far exceeds the death rate in people with HIV, prevalence will inevitably tend to rise, at least until large numbers of people with HIV start to reach the age at which deaths start to match diagnoses, unless a very high proportion of the population is virally suppressed. It will be even harder to reduce prevalence if a rise in the proportion of people who are virally suppressed is accompanied by a rise in the proportion having risky sex.

The UK is a good example of a country with a relatively high rate of diagnosis and very high rates of successful viral suppression, but where HIV prevalence has nonetheless continued to climb inexorably. Another is the city of Washington, DC where a nearly fivefold expansion in testing in the last five years has led to a 20% rise in new diagnoses.²¹

However, in populations with very high testing rates, a low proportion of people with HIV who are undiagnosed, and very high levels of ARV coverage and viral suppression, the effect of reducing the community viral load can start to be seen. The best evidence we have so far appears to come from San Francisco. Here amongst gay men, which in this city form 85 to 90% of the HIV-positive population, frequent testing, a low rate of undiagnosed cases, and high rates of ARV coverage have produced a 40% fall in the community viral load between 2004-2008, a fall in new diagnoses of 45%, and what appears to be a 33% decline in HIV incidence between 2006-2008, though as yet this has not reached statistical significance.²² Similar falls have been seen as a result of an intensive testing and prevention campaign amongst injecting drug users in Vancouver, British Columbia, though this has not as yet been seen to be happening in the rest of the province.¹³

In March 2010, San Francisco became the second area after British Columbia to officially adopt a policy of aggressive expansion of testing, especially for underserved populations, and universal treatment regardless of CD4 count for those diagnosed.^23,24 Epidemiologists in Switzerland have urged the provision of early, continuous antiretroviral therapy for sexually active HIV-positive gay men in the country.²⁵