Skip to main content

Research Briefs

Test and treat may not be best approach in South Africa, according to new study

With the availability of effective antiretroviral (ARV) drugs that have fewer side effects and are easier to take than ever before, calls for the implementation of “test and treat” strategies against the AIDS pandemic are growing in volume and frequency. The idea is to test people regularly and immediately offer antiretroviral therapy (ART) to those who test positive. When World Health Organization (WHO) researchers recently modeled the impact of such an approach on the South African epidemic, they concluded that annual test and treat could eliminate new infections in just 10 years. On the other hand, “universal access”—offering ART to those who are already known to be HIV infected and require treatment—would not only fail to end the epidemic but would cost US$10 billion more over the next 40 years, according to their calculations (1; 2).

Those conclusions are now being disputed by Sally Blower and Bradley Wagner of the UCLA Center for Biomedical Modeling, who argue on the basis of their own mathematical modeling that universal access is the better strategy for South Africa: Not only would it come close to ending South Africa’s epidemic, they say, but would do so at a lower cost than an annual test and treat program (2).

One million South Africans currently receive ART, and an additional 1.6 million need it because their CD4+ cell counts are below 350. Blower and Wagner suggest giving ART to the 1.6 million who are known to lack access today, rather than testing all 30 million South Africans and offering ART to all five million people who would be expected to test positive. They agree that test and treat could indeed end the South African epidemic. But they find that universal access would also almost eliminate the epidemic in 40 years, at a cost of $12 billion less than annual test and treat.

Blower says the UCLA model assumes that each year the HIV in 3% of the patients who take ARVs accumulates resistance mutations and that these patients will therefore need to take second-line drugs. As a result, the total number of patients who require such therapies will keep increasing. “We put that into the model because [resistance] is a very serious problem in the US and in Europe,” Blower says, adding that in contrast, the WHO model assumed that the percentage of people who require second line drugs remains constant over the years.

The WHO model also assumed that drug treatment makes people 99% less infective, a level Blower and Wagner say is too high because many people don’t take their drugs as prescribed or develop resistance mutations. When the UCLA researchers lowered that estimate, their calculations indicated that infected people would have to remain on the drugs for several decades to ensure the epidemic is in fact eliminated.

The UCLA model also assumed that putting people on treatment if they drop below a CD4+ T-cell count of 350 would allow them to live decades longer, much longer than the additional six years the WHO model assumes, Blower says. This results in higher treatment costs and resistance levels for both the test and treat and the universal access strategy. But because test and treat requires putting more people on treatment in general than universal access, these effects are bigger in the test and treat model. This is why that approach ends up being more expensive than universal access in the UCLA model.

Test and treat is “completely unrealistic,” Blower says. “If we haven’t got the money, talking about these kinds of strategies is nuts. It’s pie in the sky both in terms of money and also in terms of actually doing it.” Even if the money were available, she argues, test and treat would be logistically difficult in places like South Africa and, given the side effects, not everyone who tests positive would even agree to take the drugs. Providing universal access now, Blower argues, is much better than trying test and treat later. “These are the people who are dying right now,” she says.

But Brian Williams, who participated in the WHO modeling, says one flaw of the UCLA model is that the universal access strategy it uses assumes that people should be started on therapy below a certain threshold of CD4+ T-cell counts of 350. CD4+ T-cell counts “are extremely misleading,” he says, because in countries like South Africa, CD4+ T-cell counts vary so much that they don’t have much to do with how urgently someone needs treatment, or how long someone has been infected.

Williams adds that, contrary to what Blower says, the WHO model assumes that every year, an additional 3% of patients will need to go on second-line drugs. He also does not agree with Blower that the longer life expectancy of people on ART in the UCLA model will necessarily result in an increase in resistance and cost. “If treatment as prevention becomes a reality and there is a market to keep 30 million people on ART, prices will come down," he says. In addition, he says, while the proportion of people on second-line drugs will undoubtedly increase over time, the development of new improved drugs will reduce the problem of resistance.

In the real world, Williams says, drug resistance is not as large a problem as Blower and Wagner assume, referring to studies by Julio Montaner that show that rolling out test and treat in high risk groups such as intravenous drug users (IDUs) drastically reduced the occurrence of resistance. If monitored carefully, Williams says, resistance is manageable, especially since the combination of drugs used in ART today is much less likely to induce resistance than did previous therapies.

Montaner says he has seen an overall 90% decline in the rate of new cases of resistance in the 13 years he has made ART immediately available for free to everyone infected in British Columbia (3), in a program that even involved giving IDUs HIV tests and ART at needle exchange centers. “We actually did it. We didn’t model it mathematically,” Montaner says. “Test and treat is the way to go. It works, but you have to do it properly: It has to be free, it has to be supported, and it has to be well done.” —Andreas von Bubnoff

1. Lancet 373, 48, 2009
2. PLoS One 7, e30216, 2012
3. Clin. Infect. Dis. 50, 98, 2010

Eliciting elite control—in monkeys

An animal model designed specifically to study elite control of HIV replication has shown that high frequencies of vaccine-induced CD8+ T-cell responses against epitopes on the Vif and Nef proteins of simian immunodeficiency virus (SIV) control viral replication. These responses might be inducible by an appropriate immunization regimen in humans (1). The study, led by David Watkins, professor of pathology at the University of Miami Miller School of Medicine, may also provide a way to identify what exactly constitutes an effective T-cell response—not just against SIV, but HIV as well.

The study was conducted in Indian rhesus macaques that express the major histocompatibility complex (MHC) Mamu-B*08 allele, whose peptide binding motif is similar to that of human leukocyte antigen (HLA)-B*27, which is enriched among human elite controllers. Eight rhesus macaques were vaccinated with SIVmac239 constructs from three immunodominant T-cell epitopes—Vif RL8, Vif RL9 and Nef RL10—that comprise more than 50% of the CD8+ T-cell responses in SIVmac239-infected Mamu-B*08+ elite controllers. The constructs were delivered by a vaccine regimen consisting of a recombinant yellow fever 17D (rYF17D) viral vector prime followed by a recombinant adenovirus serotype 5 (rAd5) viral-vector boost. A control group of eight macaques received an rYF17D/ rAd5 viral-vector prime-boost regimen lacking the genes for the three epitopes of interest.

The vaccine regimen induced robust CD8+ T-cell responses against the epitopes, while the control did not. Both groups of animals developed similar levels of total SIV-specific and CD4+ T-cell responses, as measured by interferon (IFN-y) enzyme-linked immunospot (ELISPOT).

Fifteen weeks after the final boost, all 16 animals were challenged intrarectally with a high dose of the highly pathogenic SIVmac239 virus. Four out of the eight animals in the experimental group and six out of eight controls became SIV-infected after the first homologous challenge. The remaining macaques were challenged again three weeks later and all but one became infected. (It took five challenges to infect the lone outlier.)

All eight macaques in the experimental group controlled viral replication during acute infection, while only two in the control group were able to do so. Six of the seven macaques in the experimental group that were infected after one or two challenges became elite controllers, defined in this study as having a set point viral load of less than 1,000 viral RNA copies per ml of blood.

Digging into the weeds a bit, researchers employed peptides and Mamu-B*08 tetramers to evaluate potential correlates of viral protection in the monkeys vaccinated with the regimen containing the Vif RL8, Vif RL9 and Nef RL10 epitopes. Peripheral blood mononuclear cell responses against the entire SIV proteome were equivalent in the two groups, but the experimental group with lower viral load had higher insert-specific and proteome-specific CD4+ T-cell responses.

The experimental group also exhibited earlier and higher magnitude Vif RL9- and Nef RL10-specific responses in the peripheral blood, lymph nodes and gut compared to the control group, though responses directed against the Vif RL8 were the same for both groups. The results suggest that the higher frequency of Vif RL9- and Nef RL10-specific responses in blood and tissue were directly responsible for viral control in the experimental group.

The virus did rebound in two of the eight vaccinated animals that received the three targeted epitopes. But sequencing data showed that it occurred following immune escape by all three epitopes, suggesting that the virus cannot replicate in the presence of these CD8+ T-cell responses.

But there are other possible mechanisms by which the macaques might be controlling SIV. Mauricio Martins, a post-doc in Watkins’ lab and a co-author of the study, cautioned that there are some notable differences between macaque and human models of elite control. Studies suggest that HIV controllers often have Gag-specific T-cell responses, while SIV-infected macaques do not appear to target Gag epitopes very often. “In fact, the two alleles that have the highest association with elite control [in macaques] do not restrict any immunodominant epitopes in Gag,” said Martins. The SIVmac239 strain is also more pathogenic than HIV—the median viral load is 1 million vRNA copies/ml compared to 30,000 vRNA/ml.

Bruce Walker, an expert on elite control and the director of The Ragon Institute of Massachusetts General Hospital, MIT and Harvard considers Watkins’ study important for at least two reasons. “It shows that a narrowly-directed cytotoxic T-cell response can be sufficient to control HIV infection, and it shows that manipulation of immunodominance with a vaccine prior to infection can dramatically impact outcome. In other words, it shows proof of principle that very targeted vaccinations can skew the immune response to a better outcome.” —Regina McEnery

1. Nature 2012, doi:10.1038/nature11443