Biomedical interventions
Research on biomedical interventions to interrupt sexual transmission have been disappointing, although there have been some signs of hope. Of 26 randomized controlled trials of different interventions, including four vaccine, 10 microbicide and three herpes suppression trials, 22 failed to show efficacy [
12‐
14]. The four positive trials included three on male circumcision, and the sexually transmitted infection intervention trial in Mwanza, Tanzania, more than a decade ago, of limited generalizability [
14].
At the Conference on Retroviruses and Opportunistic Infections (CROI 2009) in Montreal, Canada, in February 2009, Karim and colleagues presented results from HPTN 035 assessing the microbicide gel, Pro 2000 [
15]. Though not reaching statistical significance, a 30% reduction in HIV incidence was observed in gel users. Pre-exposure prophylaxis (PrEP) can be provided through topical, as well as oral antiretroviral agents [
15]. Encouraging animal data suggest that a HAART-containing gel may provide pre-exposure protection [
16]. The efficacy of oral PrEP is being assessed in 10 ongoing or planned randomized controlled trials involving some 20,000 participants internationally; the first results are expected in late 2009 or 2010. Although PrEP is promising, it may take unusual persuasiveness to convince a decision maker to give drugs to HIV-uninfected persons when many with declared HIV disease are dying from lack of access.
Recent results from the vaccine trial in Thailand [
17] remind us that while implementing evidence-based, near-term prevention solutions, work must continue to develop a vaccine that could be used in the future to mitigate the HIV epidemic. The overall situation has prompted many people to consider the potential prevention role of HAART.
Scientific evidence for HAART as prevention
Before considering the potential impact of HAART for prevention, it is important to consider the scientific evidence for the basic assumptions behind HAART for prevention. These assumptions include the obvious but often overlooked fact that HIV transmission occurs only from people with HIV [
9].
There are studies supporting the assumption that viral load is the single greatest risk factor for all modes of transmission. Quinn's landmark study in Rakai, Uganda, showed that patients with less than 400 copies of HIV RNA per millilitre. have the lowest rate of HIV transmission, and demonstrated a stepwise increase in transmission rates for higher RNA levels [
18]. Lowering viral load is essential to interrupting transmission, and HAART can lower viral load to nearly undetectable levels. A 2009 meta-analysis that included 11 cohorts and 5021 heterosexual couples concluded that there was zero risk of sexual transmission while on HAART for those with HIV-1 RNA below 400 copies with an upper confidence limit of 1.27 per 100 years [
19].
Prevention of mother to child transmission (PMTCT) offers proof of the concept of HAART interrupting HIV transmission. Although some would argue that lessons from PMTCT may not be applicable to preventing sexual transmission, perinatal AIDS cases have been virtually eliminated in the United States [
20]. This is likely due to the implementation of Public Health Service guidelines for the universal counselling, voluntary HIV testing and HAART for pregnant women and newborn infants [
20]. Recent trial data from sub-Saharan Africa support HAART to block PMTCT, with one study showing a decrease of transmission to 1% [
21].
Observational studies illustrate the potential for HAART for prevention of HIV transmission [
22]; Bunnell and colleagues in Uganda showed that when HAART is added to couples counselling, transmission during sex can be reduced by 98% [
23]. Numerous studies suggest a potential for the community-level impact of HAART on HIV transmission. In British Columbia, public health scientists showed a decrease in community plasma HIV RNA concentrations and HIV incidence among injecting drug users associated with HAART use [
24]. Work from Taiwan found a 53% reduction in new HIV cases associated with the provision of free access to HAART [
25]. Additional data from areas with high antiretroviral treatment (ART) coverage do exist, and additional community-based studies examining the impact of HAART on HIV transmission are in preparation.
When to start?
Although knowing one's HIV status is key for prevention, it is not known with certainty how early to start HAART. People living with HIV in sub-Saharan Africa start HAART at a median CD4 count of around 100 cell/mm
3, which is much later than in the north [
26]. This lack of access to HAART until very late in the course of the disease is decreasing in areas as people learn their HIV status earlier, guidelines change and services expand. However, mortality is still markedly higher in sub-Saharan Africa compared with other contexts [
26,
27].
At CROI 2009, Lawn and colleagues from South Africa showed a steeply increasing risk of death in patients on HAART associated with the time spent below 200 CD4+ cells/mm
3 [
28,
29]. Mortality reached almost 40/100 person years when the CD4+ count was less than 50 [
28,
29]. Although mortality rates at the higher CD4 levels may be relatively low, when applied to large numbers of people living with HIV, this lower risk converts into a significant impact on mortality. Observational data from North American cohorts showed that among 8362 patients with a CD4+ count of 351 to 500 cells/mm
3, deferral of therapy until the CD4+ count had fallen to 350 cells/mm
3 or less was associated with an increase of 69% in the risk of death, as compared with patients who initiated therapy when their CD4+ count was within the designated range (relative risk in the deferred-therapy group, 1.69; 95% confidence interval [CI], 1.26 to 2.26; P < 0.001) [
30].
Similarly, among 9155 patients with a CD4+ count of more than 500 cells/mm
3, deferral of therapy until the CD4+ count fell below 500 cells was associated with a significantly increased risk of death of 94% (relative risk, 1.94; 95% CI, 1.37 to 2.79; P < 0.001) [
30,
31]. Although not a randomized clinical trial and relying on relatively few events, it nevertheless presents further evidence for the potential benefits of starting earlier and is in line with other observational data [
26]. Sax and Baden, in their accompanying editorial, suggest that we may be moving to an era when most people will choose to start HAART when they are ready [
31]. The When to Start Consortium analyzed data from 18 cohorts in Europe and North America and included more than 40,000 patients [
26]. Starting treatment at higher CD4 counts reduced the probability of AIDS or death, with those starting before reaching a level of 450 cells/mm
3 having the most benefit [
26]. There are numerous other cohort studies that also suggest that starting earlier is better [
32‐
34].
Trial data, although limited, also point in the direction of an earlier start. The National Institutes of Health-supported randomized clinical trial, CIPRA HT 001, in Haiti was stopped by the Data Safety Monitoring Board. Patients who started treatment earlier, with CD4 counts of between 200 and 350 cells/mm3, had significantly fewer deaths and fewer cases of TB compared with those who had deferred treatment to less than 200 cells/mm3 or when an AIDS-defining illness occurred.
ACTG A 1564, conducted predominantly in the United States, compared early with deferred HAART in patients treated for various opportunistic events (but not TB) [
35]. Survival curves show a 47% reduced progression or death in patients receiving immediate as opposed to deferred HAART [
35]. There is increasing evidence that suggests the damaging effects of HIV, even at higher CD4 count levels [
36,
37]. The SMART trial showed not only that starting earlier provided superior outcomes; it also suggested that HIV may be associated with serious non-AIDS-defining events, including cardiovascular, renal and liver disease and non-AIDS malignancies [
36,
37]. For some, this evidence of HIV infection as a chronic inflammatory disease process provides additional rationale for an earlier start of HAART.
TB is the major killer for most people living with HIV in sub-Saharan Africa. Studies from Cape Town demonstrate increasing TB incidence, and the group has coined the phrase, "TB death zone", to describe living below 500 cells/mm
3 [
28,
29]. There is increasing recognition among people working on HIV-related TB that HAART has a significant role to play in preventing TB morbidity, transmission and mortality. Preliminary results of the South African SAPIT trial, presented at CROI 2009, compared outcome of HAART integrated with TB treatment versus deferred treatment until TB therapy was completed in patients with CD4+ counts below 500 cells/mm
3 [
15]. There was a 56% reduction in mortality in the integrated group, and this applied across the whole CD4+ spectrum in a stratified analysis [
15].
The emerging evidence suggests that HAART should be initiated as soon as possible in acute illness. The evidence increasingly suggests that if the future is to be different, we have to intervene with treatment far earlier, before people living with HIV fall into or spend too long in the CD4+ "death zone" for tuberculosis.
Under universal access, most people living with HIV will eventually become eligible and require HAART. How early to start HAART is a matter of perspective, and the crux of the issue is time from HIV infection. A recent study gathered CD4 and RNA data from 30 international studies and 16 cohorts of untreated adults with HIV [
38]. The analysis of median CD4 counts over time found relatively low starting levels and a relatively rapid progression to commonly discussed CD4 thresholds, such as 500, 350 and 200 [
38].
The time to reach commonly used CD4 eligibility thresholds for HAART was variable and in some settings, only a few years after HIV infection [
38]. Data suggests that on average it takes about two to four years to reach a CD4 count of 500 cells/mm
3, depending on one's starting point; from this perspective this represents a significant amount of time. However, if we consider that people will live for tens of years on HAART, then the two to four years it would take to reach a CD4 level of 500 cells/mm
3 makes up only a small fraction of the remaining years of life. In other words, optimally most people will have access to HAART, and when discussing whether to start at 200, 350 or 500, we are really discussing the difference of a few years earlier in the course of a much longer life span on HAART.
Guidelines are written for the context within which they are meant to be applied, and European, North America and World Health Organization (WHO) recommendations are divergent on a number of important issues. In addition to clear differences in formulary and laboratory support, in countries with more resources, people are started earlier, and factors, such as viral replication status, CD4 decline and being in a discordant couple, are now appearing as potential start criteria, even at higher CD4 counts.
Consideration of the durability of currently used regimens in resource-constrained settings will be central in evaluating using HAART for prevention. Our ongoing scientific and moral challenge will be to continue to narrow the treatment gulf between north and south and to ensure that we do not accept the establishment of two standards of care: one for the richer countries and the other for the poorer, for lack of better terms.
Essentially all people living with HIV will eventually require HAART for clinically progressive, ultimately fatal immunodeficiency. Regardless of when we decide that people should start HAART, we will not reach universal access to prevention, care and treatment unless millions of people with HIV learn their HIV status. Despite considerable efforts to expand access to HIV testing, an estimated 80% of people living with HIV in sub-Saharan Africa do not know their status and 90% do not know their partners' status [
1].
Kenya is a leader in improving access for HIV counselling and testing. However, data from the 2007 Kenya AIDS Indicator Survey shows that of people eligible for HAART (with a CD4 count of less than 250), 57% have no idea that they have HIV [
39]. Knowing one's HIV status was a key determinant in accessing treatment in Kenya: 92% of those who knew their status and were eligible were on HAART [
39]. Susan Allen, working in the late 1980s and '90s in Rwanda and Zambia, and many others have built the scientific evidence base to show that HIV counselling and testing itself [
23,
40,
41], particularly when it involves couples counselling, can be a remarkably effective prevention intervention [
23,
41,
42].
Community-based efforts, including home-based couples counselling and testing, have considerable promise. Through a seven-day campaign designed to prevent HIV, diarrheal disease and malaria in a district in western Kenya, a private sector company working with local non-governmental organizations, Centers for Disease Control Kenya, and the Ministry of Health was able to test 41,040 or 80% of the men and women between 15 and 49. The total population reached was 47,311, out of which 97% were voluntarily tested and counselled for HIV. Of those reached, 18,300 or 38% were men [
43].
Of course, knowing one's HIV status is not enough: the cornerstones of HIV counselling and testing scale up must include improved protection from stigma and discrimination, as well as improved access to integrated prevention, treatment and care services. A human rights approach, based on the "3 Cs" of HIV testing (confidentiality, counselling and informed consent) is a prerequisite for success [
44].
Modelling results
This quote from George Box, one of the most influential statisticians of the last century, puts it nicely: "Essentially, all models are wrong, but some are useful." Models help us to better understand what we think we know and perhaps most importantly, what we need to find out. Our model [
45] builds on and extends earlier analyses suggesting that rapid scale up of conventional HAART approaches could significantly reduce mortality [
46] and have a substantial impact on HIV incidence [
47,
48].
R
0 or reproductive ratio is the average number of secondary cases of infection to which one primary case gives rise throughout its infectious period. We focused on a generalized HIV epidemic setting largely driven by heterosexual sex, and used data from South Africa, Uganda, Malawi and elsewhere [
45]. The South Africa HIV surveillance data shows an initial doubling time of 1.25 years, which means that on average each person with HIV infects another person once every 1.25 years. Life expectancy for people with HIV was 10 years, which allows us to estimate an R
0 of around 7. Therefore, we assumed that cutting transmission by a factor of more than 8 would reduce R
0 < 1 and eliminate HIV infection.
We used a stochastic model for estimating R
0 and we examined phases and relative infectivity with time. The parameter values for infectivity are not known precisely, but we used the calculated R
0 and the literature for our assumptions. The relative importance of the acute phase and concurrency has been the subject of some debate among modellers, epidemiologist and others. We conducted sensitivity analyses by varying the degree of infectiousness and the duration of the acute, chronic and final phases [
45].
We used a stochastic model to examine the frequency of HIV testing and CD4 level needed to get R
0 < 1. We would need to test people on average about once per year and need to start them on immediate HAART when their CD4 count was around 1000. In the South African setting, the average CD4 count after conversion is 884. Therefore to reach R
0 < 1 for most people, we need to start HAART immediately, irrespective of CD4 count. We also see that annual testing and a CD4 threshold of 200 gives you a R
0 of 4 - a significant reduction in transmission but not elimination, which we defined as 1000 cases per million per year [
45].
For the model, we used the 2007 estimate that 5% of people living with HIV were already on HAART and that programme coverage increases logistically. In other words, programme coverage reaches 50% in 2012 and 90% in 2016 [
45]. The model's programme start date can be altered accordingly, with more rapid implementation translating into shortened time to maximal impact. We added in a 40% prevention effect whereby transmission would be additionally decreased by 40% over the time period. For some, this is optimistic for a combined prevention approach, and for others, not high enough, but we felt that if current prevention efforts beyond HAART could decrease transmission by 40%, that would be reasonable.
Although this combined approach dramatically reduces transmission, it does mean that a large cohort of people will be on HAART for a long time; of course, we thought that this was not a bad thing since they would at least be alive, and it is similar to accelerated universal access, but nevertheless represents a significant challenge. Modelling showed that the universal voluntary HIV testing and immediate HAART strategy with combined prevention interventions resulted in a 95% reduction in incidence or new HIV cases in 10 years. The theoretical strategy reduced incidence from 15-20,000 per million population to 1000 per million and the prevalence to less than 1% by 2050 [
45].
The model suggests that mortality would decline rapidly and the epidemic would become concentrated with particular populations remaining at risk. Health services could then switch focus from making HAART available to those in greatest need to providing support and services for those who are on HAART. While other prevention interventions, alone or in combination, could significantly reduce HIV incidence, the model suggests that only universal HIV testing and immediate initiation of HAART could reduce transmission to the point where elimination might be feasible by 2020 for a generalized epidemic, such as that in South Africa. The less than 350 strategy has a major impact, but we are left with a persistent epidemic. In summary, the universal voluntary HIV testing and immediate treatment strategy drives down incidence, prevalence and mortality towards an elimination phase [
45].
The different strategies do come with different impacts: the less than 350 on HAART strategy could save nearly 2.41 million lives, while the universal voluntary HIV testing combined prevention approach nearly triples that number to 7.35 million [
45]. These dramatic results beg the question: how much will it cost? We did a rough costing for the strategies based on the estimated costs of delivering ART [
45]. The funding needed to implement the theoretical strategy for an epidemic of South African-type severity peaks in 2015 at $3.4 billion per year (range: $2.2 billion-$5.3 billion).
Although the initial yearly cost of the theoretical strategy is higher than the present strategy, it is within UNAIDS projections of the $8.84 billion needed every year for universal access to prevention, care, and treatment in a South African-type situation in 2015 [
45]. The "front-loaded" universal voluntary strategy is initially more expensive, but becomes cost saving around 2030 as the HIV epidemic moves toward elimination [
45]. A programme of this size would require considerable initial effort. However, over time, the decreasing HIV incidence would free scarce health care resources that are currently overwhelmed by the immediate demands of the HIV epidemic.