Antiviral agents and HIV prevention: controversies, conflicts, and consensus

Abstract

Antiviral agents can be used to prevent HIV transmission before exposure as preexpo-sure prophylaxis (PrEP), after exposure as postexposure prophylaxis, and as treatment of infected people for secondary prevention. Considerable research has shed new light on antiviral agents for PrEP and for prevention of secondary HIV transmission. While promising results have emerged from several PrEP trials, the challenges of poor adherence among HIV-negative clients and possible increase in sexual risk behaviors remain a concern. In addition, a broader pipeline of antiviral agents for PrEP that focuses on genital tract pharmacology and safety and resistance issues must be developed. Antiretroviral drugs have also been used to prevent HIV transmission from HIV-infected patients to their HIV-discordant sexual partners. The HIV Prevention Trials Network 052 trial demonstrated nearly complete prevention of HIV transmission by early treatment of infection, but the generalizability of the results to other risk groups – including intravenous drug users and MSM – has not been determined. Most importantly, the best strategy for use of antiretroviral agents to reduce the spread of HIV at either the individual level or the population level has not been developed, and remains the ultimate goal of this area of investigation.

Introduction

Antiviral agents can be used to prevent HIV transmission in three ways: before exposure as preexposure prophylaxis (PrEP), after exposure as postexposure prophylaxis (PEP), and as treatment of infected people for secondary prevention [1–3]. PEP for HIV prevention has been well established but is not well suited to clinical research investigation. However, recent research developments in PrEP and secondary prevention provide a unique opportunity to highlight areas of advancement that have galvanized changes in HIV treatment and prevention and to highlight topic areas that remain undecided or controversial.

Assessing the prevention benefits of antiretrovirals: pharmaceutical and observational studies

With the development of antiviral agents in the early 1990s, a confluence of reasoning from two disciplines emerged. Clinical pharmacological studies demonstrated that HIV drugs penetrate the genital tract but with variable success [4–8]. Epidemiologic observational studies suggested that antiretroviral therapy (ART) might play a role in reducing the sexual transmission of HIV.

Intensive investigations of the pharmacology of antiretrovirals in genital secretions have demonstrated that several drugs in different therapeutic classes of antiretrovirals reliably concentrate in the male and female genital tract (Table 1). Drug penetration can be predicted for semen by the degree of antiretroviral protein binding in blood. However, predictors of drug penetration into the female genital tract remain unknown.

Table 1.

Exposures of antiretrovirals in genital tract fluid, cells, and tissues compared with blood plasma.

	Data under single-dose (multiple-dose) conditions
Drug	CVF	VCT	SP	Seminal mononuclear cellsa	Colorectal tissue
Nucleoside analogue reverse transcriptase inhibitors
Tenofovir	1.35 (0.75)b	5.8c,e/0.6d,e	4.4 (5.1)	9.4 (17.5)	33e
Abacavir	0.21 (0.08)b		(1.5)b
Lamivudine	2.41 (4.11)b		(6.70)	1.00
Zidovudine	3.71 (2.35)b		(2.30)	0.36
Didanosine	0.06 (0.21)b
Emtricitabine	1.11 (3.95)b	42c,e/7d,e			4.3e
Stavudine	0.04 (0.05)b		(3.50)
Protease inhibitors
Lopinavir (with RTV)	0.17 (0.08)b		(0.05)
Atazanavir (with RTV)	0.16 (0.18)b		(0.10)
Ritonavir (RTV)	0.18 (0.26)b		0.11 (0.07)b		5.8 (12.8)b
Indinavir (with RTV)	(4.60)		(1.9)b
Darunavir (with RTV)	(1.5)b		0.18 (0.20)b		1.26 (2.70)b
Amprenavir	(0.50) with RTV		(0.20) without RTV
Nelfinavir			(0.08)b
Saquinavir (with RTV)			(0.03)b
Nonnucleoside reverse transcriptase inhibitors
Efavirenz	0.005 (0.004)b		(0.03)
Nevirapine	(0.80)		(0.60)
Delavirdine	(0.40)
Etravirine	(1.3)b		0.17 (0.15)b		15.7 (7.5)b
Entry inhibitors
Maraviroc	1.9 (2.7)b	1.9b	0.6 (0.6)b		9 (28)b
Integrase inhibitors
Raltegravir	0.64 (0.93)b		(3.20)		39 (239)b

Comparisons are tissue/blood plasma (BP) paired samples unless otherwise noted. AUC, area under the concentration–time curve; BP, blood plasma; CVF, cervicovaginal fluid; SP, seminal plasma; VCT, vaginal and/or cervical tissue. Modified from [5].

^aSeminal mononuclear cells compared to peripheral blood mononuclear cells.

^bAUC_{CVF, SP, or tissue} : AUC_BP ratio.

^cCervical.

^dVaginal.

^eAUC_{tissue 1–14 days} : AUC_{BP 1–14 days} ratio.

The relationship between the penetration of drugs into the genital tract, suppression of genital tract replication with treatment, and the relevance of persistent HIV shedding to HIV transmission remains only partially understood. First, it is clear that persistent intermittent ‘shedding’ of HIV in both male [9–14] and female [15–17] genital secretions can be expected even when treatment has reduced the blood plasma viral load [18].

Several groups have reported detection of HIV in semen resistant to protease inhibitors [19,20], reflecting poor penetration of this class of agents. Conversely, Ghosn et al. [21] reported complete suppression of HIV in semen over 48 weeks with the lopinavir–ritonavir combination, even though these agents were only detectable in blood plasma and not in seminal plasma (<30 ng/ml). Inability to completely and consistently suppress replication of HIV in the genital tract in men and women demonstrates that these compartments do not get the full benefit of antiretrovirals. Indeed, quite recently, investigators have argued that the persistent low copy HIV replication might be ascribed to poor penetration or metabolism of ART in lymphoid tissues [22]. After a single dose of tenofovir, we have demonstrated discrepant penetration into cervical, vaginal, and rectal tissues, with cervical and vaginal tissue levels 10–100 times lower for tenofovir and tenofovir diphosphate than those achieved in rectal tissue [23]. Sexually transmitted diseases (STDs) can stimulate shedding of HIV in spite of antiviral therapy [24,25]. However, the results of observational reports and the HIV Prevention Trials Network (HPTN) clinical trial 052 (see below) suggest that HIV shedding (so readily observed in spite of treatment with antiviral agents) may not actually contribute much to HIV transmission.

HIV transmission among HIV-serodiscordant couples has been the subject of several observational studies [26,27]. A subset of these studies with information on index partner ART status (Table 2) allows us to consider the effects of these drugs on transmission risk [28–37]. Reduction of HIV transmission with treatment has been reported in all but one report from China. In the study from China, 1927 infected people in stable HIV-discordant couples were offered free ART [37]. Over 4918 person-years of follow-up time, similar proportions of participants who were taking antiretrovirals and participants who were not taking antiretrovirals infected their susceptible sexual partners (4.8 vs. 3.2%, P = 0.12). However, there is no way to judge whether, or how, the infected persons were taking their medications, and genetic tests were not performed to confirm linked transmissions within the pairs.

Table 2.

Studies assessing effects of antiretroviral treatment on HIV transmission.

Author	Study location	No. of couples	Study population	Conclusions
Retrospective cohorts
Musicco et al. [34]	Italy	436	HIV + clinic and HIV surveillance center clients and their seronegative partners	ART in HIV-infected men reduces, but does not eliminate, heterosexual transmission of infection
Castilla et al. [28]	Spain	393	HIV clinic patients and their seronegative partners	Combined ART applied according to current guidelines has a great potential for preventing HIV transmission to sexual partners
Prospective cohorts
Bunnell et al. [31]	Uganda	926	ART-naive HIV-positive adults enrolled in home-based ART program reporting on their partners (stable/nonstable, HIV-positive/negative)	ART, prevention counseling, and partner VCT associated with reduced estimated risk of HIV transmission during first 6 months of therapy
Del Romero et al. [29]	Spain	424	Couples recruited through HIV-positive patients at an HIV/STI clinic	Heterosexual infectivity of HIV in individuals taking effective antiretroviral treatment is low
Donnell et al. [30]	Botswana, Kenya, Rwanda, South Africa, Tanzania, Uganda, Zambia	3381	HIV-positive and HSV-positive individuals and their HIV-negative partners from the partners in prevention HSV/HIV Transmission Study	Provision of ART to HIV-infected patients could be an effective strategy to achieve population-level reductions in HIV transmission
Hernando et al. [32]	Spain	339	HIV-positive patients and their partners attending a HIV/STI clinic	Couples-based safe sex counseling and ART can reduce but not eliminate sexual HIV transmission
Melo et al. [33]	Brazil	93	HIV clinic patients and their seronegative partners	Transmitters showed significantly higher median viral loads, suggesting that heterosexual transmission of HIV is more a function of viral load than sex of index case. Antiretroviral use may play a role in the prevention of HIV heterosexual transmission
Reynolds et al. [35]	Uganda	250	Serodiscordant couples offered free ART if eligible	HIV transmission may be reduced among HIV-discordant couples after initiation of ART due to reductions in viral load and increased consistent condom use
Sullivan et al. [36]	Rwanda, Zambia	2993	Serodiscordant couples initiated on ART if eligible	ART was associated with a 94% reduction in transmission; ART initiation is a critical component of a package of biomedical and behavioral prevention services
Wang et al. [37]	China	1927	Former plasma donors and their seronegative spouses	Transmission events occurred with equal frequency in couples regardless of whether the partner was provided ART

ART, antiretroviral therapy; HSV, herpes simplex virus; STI, sexually transmitted infections; VCT, voluntary counseling and testing.

Attia et al. [26] undertook a meta-analysis that emphasized the limited data (five studies total, with a combined 1098 person-years of follow-up) available from observational studies for demonstrating transmission prevention related to ART. Nonetheless, these results, and modeling of transmission probabilities with ART (reviewed in [38]), led to the rather controversial 2007 ‘Swiss Statement’ [39]. In this declaration, the investigators concluded – based on available data – that under particular circumstances, treated, HIV-infected persons could engage in unprotected sex acts with minimal risk of transmission to an HIV-negative partner. The Swiss Statement requirements included informed consent from the HIV-negative partner, HIV-positive partner on ART suppressed to undetectable levels of blood plasma HIV for at least 6 months, and lack of any other STDs. The Swiss people took these recommendations seriously. In an analysis conducted among 7309 HIV-infected persons in the Swiss HIV Cohort Study, after the Swiss Statement was issued, participants were 1.24–2.04 (varied by risk group) times more likely to report unprotected sexual contacts [40].

Clinical trials in stable heterosexual couples and generalizability to other contexts

To determine the magnitude and durability of a prevention effect from ART, a randomized controlled trial was conducted by the HPTN [41]. In the HPTN 052 trial, 1763 sexually active HIV-discordant couples (in a stable relationship over the past 3 months reporting vaginal or anal intercourse at least three times during that period) were recruited from 13 sites in nine countries. At enrollment, the HIV-positive participant was required to be ART-naive and have a CD4 cell count between 350 and 550 cells/μl; individuals with active tuberculosis were excluded. Participants were randomized to start ART at study enrollment (early therapy) or after two consecutive CD4 cell count measures less than or equal to 250 cells/ml and more than 200 cells/μl (delayed therapy). The history of the HPTN 052 trial has been previously reported [42]. The HPTN 052 study is ongoing and scheduled for completion in 2015. However, because of the overwhelming benefits of ART observed in the first 1.7 years of average follow-up, the National Institutes of Health (NIH) Multinational Data Safety Monitoring Board (DSMB) recommended on 28 April 2011 that the sponsor (NIH) make the interim results available. The individuals in the early treatment arm had greater than 96% protection from HIV acquisition from their HIV-infected partner [41]. The single HIV transmission event in the treated group was ascribed to transmission before HIV suppression was possible (Jabara, Ping, Swanstrom, personal communication). Individuals in the early arm had reduced episodes of opportunistic infection, especially extrapulmonary tuberculosis. Further analysis has demonstrated that patients receiving early ART had delayed time to a primary HIV endpoint and significantly reduced secondary clinical endpoints (Grinsztejn, et al., abstract submitted, IAS 2012). Following the DSMB decision, all participants in the delayed treatment arm who had not already initiated treatment (due to CD4 cell count <250 cells/μl) were offered antiretroviral drugs. HPTN 052 is continuing so as to determine the durability of the prevention benefit and to monitor individuals in the delayed arm for adverse clinical events realized even after ART has been initiated; an observational study has suggested that delayed initiation of ART even at higher CD4 cell counts could lead to cardiovascular and other complications [43–45].

A key question from HPTN 052 is the generalizability to other contexts: heterosexual couples with CD4 cell counts lower and higher than those studied in HPTN 052, high-risk heterosexual individuals (e.g., sex workers and their clients), MSM, and intravenous drug users (IDUs). There are no data to address this issue directly. The PEPFAR Scientific Advisory Board concluded that for heterosexual transmission, there is no reason to believe that ART will not suppress HIV transmission regardless of pretreatment viral burden or stage of disease [46]. For MSM, a WHO Expert Committee concluded that there is no reason to presume that treated, HIV-infected MSM will not be rendered less contagious [47]. However, the biology of HIV transmission is sufficiently different in IDUs and MSM as compared with heterosexual transmission, warranting further consideration and study (K.E. Muessig, et al., under review).

For example, the number of HIV variants acquired and the efficiency of transmission are higher in MSM and IDU than in heterosexual transmission [48]. Additionally, whether ART reduces infectivity through anal sex by the same order of magnitude as for vaginal sex remains uncertain (K.E. Muessig, et al., under review). Although such an effect is widely assumed, this is a key missing piece of evidence. Although an individual-level clinical trial among MSM may be unethical in light of the results of HPTN 052, at least two observational studies are underway to assess the effects of ART on HIV transmission among MSM. The Partners of People on ART: A New Evaluation of the Risks (PARTNER) Study [49] and the ‘Opposites Attract’ study [50] will recruit serodiscordant male–male couples in Europe and Australia/Thailand, respectively. In these studies, serodiscordant couples who are engaging in anal intercourse, and in whom the HIV-infected partner is on ART, will be followed longitudinally to estimate the rate of HIV transmission.

Antiretroviral treatment and population-level benefits

The individual benefits of ART may translate to a population-level effect in some parts of the world; however, the data used to support this conclusion have been controversial in large part because of the limitations of the methods employed (Smith, et al., under review). Eight published ecological studies have examined trends in the HIV epidemic – in some cases using measures of HIV incidence – and concomitant availability of antiretrovirals (Table 3) [51–58]. These studies are interesting and provocative, but they are inevitably subject to considerable confounding and bias. The measurement of antiretroviral use is also problematic, as HIV transmission can only be prevented with reliable and durable suppression of HIV.

Table 3.

Ecological studies.

Author	Analysis method	Statistical analysis results	Conclusions
Castel et al. [51]	Negative binomial regression of new diagnoses on mean CVL	Lack of association reported (P = 0.11)	No association was found between trends in the mean CVL and newly diagnosed HIV/AIDS cases
Das et al. [52]	Poisson regression of new diagnoses on changes in total and mean CVL; meta-regression of estimated incidence on changes in total and mean CVL	Statistically significant trend with new diagnoses noted (P = 0.003); lack of association using estimated incidence (P > 0.30)	Reductions in CVL were associated with decrease in new HIV diagnoses but not with slight decrease in HIV incidence
Fang et al. [55]	Modified back-calculation to estimate reduction in transmission rate (new cases per prevalent case-year) between pre-ART and post-ART eras	Pre-ART transmission rate estimated as 0.391 new infections per prevalent case; post-ART transmission rate estimated as 0.184 new infections per prevalent case	Provision of free ART was associated with a 53% reduction in the estimated HIV transmission rate
Katz et al. [56]	Inferences drawn from observation of concurrent changes in HIV incidence rates, reported sexual behavior, STI diagnoses, and ART use among population in clinical care	–	ART impact on HIV transmission countered by increased reported risk behaviors
Law et al. [57]	Inferences drawn from predicted changes in prevalence of undetectable VL among population in clinical care and external reports of HIV incidence	–	Declines in predicted detectable viral load 1997–2009 coincide with reports of rising new diagnoses and estimated incidence in the same community
Montaner et al. [53]	Poisson regression of estimated new diagnoses on changes in median CVL and numbers receiving ART	Effect of 100 new patients receiving ART on estimated new diagnoses predicted as –0.97 (95% CI 0.96–0.98); effect of 1 log decrease median CVL on the estimated new diagnoses predicted as –0.86 (0.75–0.98)	Increased ART coverage and reduced CVL are associated with decreased number of new HIV diagnoses
Porco et al. [58]	Inferences drawn from trends in annual HIV incidence based on antibody testing and time period (pre-ART versus post-ART period) as indicator of ART use	–	Wider availability of ART appears to have slowed transmission in the study population
Wood et al. [54]	Unadjusted and adjusted Cox proportional hazards regression of time to seroconversion on median CVL in the preceding 6 months	Unadjusted hazard ratio for effect of median CVL on time to seroconversion estimated as 3.57 (2.03–6.27) per log10 CVL increase; adjusted hazard ratio for effect of median CVL on time to seroconversion estimated as 3.32 (1.82–6.08) per log10 CVL increase	Median CVL predicts HIV incidence independent of HIV risk behaviors

ART, antiretroviral therapy; CI, confidence interval; CVL, community viral load; IDUs, intravenous drug users; STI, sexually transmitted infection.

Ecological studies have had mixed results. In San Francisco, investigators argue that new diagnoses of HIV have fallen as a result of broader availability of ART and demonstrable reduction in viremia in some members of the ‘community’, broadly defined as people with recognized HIV who are receiving treatment [52]. In British Columbia, antiretrovirals have been related to reductions in new diagnoses of HIV among IDUs [53], a population that has received many other successful interventions [59,60] and in whom adherence to antiretrovirals may be poor [61]. HIV incidence among other groups such as MSM in the USA, Australia, France and Amsterdam may be rising, in spite of the wide availability of antiretroviral drugs [62–66]. In Canada, the portion of MSM making up the estimated number of incident infections increased between 2005 and 2008 [67]. Conversely, HIV incidence worldwide has been falling dramatically; between 2001 and 2009, HIV incidence decreased by more than 25% in 33 countries, 22 of which were in sub-Saharan Africa [68]. These declines are most readily ascribed to a complex set of interventions including antiretroviral drugs. So, although it is certainly possible that ART is reducing the incidence of HIV in communities with broad access to ART, the methods used in ecological studies, along with a large number of unaddressed issues and incongruent results, preclude a definitive answer to this hypothesis.

Most recently, investigators in South Africa reported results from a study in KwaZulu-Natal in which they compared the densityof ART coverage in different communities with HIV sero-conversion in a longitudinal cohort of 16 667 people. They adjusted their analysis to consider sexual behavior, socio-demographic variables associated with HIV acquisition, and HIV prevalence in the surrounding community (range <10 to >40%). In the adjusted analysis, each percentage point increase in ART coverage of all HIV-positive persons in the surrounding community resulted in a 1.7% (P < 0.001) reduction in risk of HIV acquisition for HIV-negative persons living in that community [69]. These results present the most compelling population-level evidence to date that treatment with ART can reduce incidence of HIV.

Community randomized trials

Given the need to understand the proper balance between antiretrovirals and other modes of HIV prevention, as well as the magnitude of benefit of combination prevention, population-level clinical trials are planned. At least 50 studies are ongoing or planned [70], some of which are quite ambitious. HPTN 065 is a study designed to determine whether people with HIV in New York City and Washington, District of Columbia, can be detected and efficiently linked to care [71]. Three studies are being supported by a variety of agencies including the Office of the Global AIDS Coordinator (PEPFAR), USAID, the Bill and Melinda Gates Foundation, the US Centers for Disease Control (CDC) and the US NIH, and one study by the French National Agency for Research on AIDS and viral hepatitis (ANRS). In a study in Botswana, the investigators intend to find and treat people with the highest viral loads, as they may be most contagious [72]. All of these studies are likely to use molecular phylogeny to better understand the spread of HIV, and new cases introduced from outside the target communities.

Treatment as prevention, though exciting, cannot be guaranteed success at the population level. Can enough people be detected, linked to care, and properly treated to make a difference [73–75]? Also, current HIV detection strategies cannot find people with acute and early HIV disease [74,75], who may or may not contribute greatly to the spread of HIV [76–79]. The biology and epidemiology of acute HIV infection have been extensively reviewed; the relative importance of acute HIV infection to the spread of HIV is the subject of current debate [80].

Preexposure prophylaxis

PrEP is used to prevent many infectious diseases (e.g., endocarditis, malaria). However, usage of PrEP is subject to universal questions: does the agent work (biological plausibility); can the agent be given at the right time to work (pharmacokinetics and dynamics); will at-risk individuals use the agent properly and reliably; will cost and toxicity outweigh the benefit(s) of the intervention; and is usage sufficiently limited? In essence, the right drugs must be used at the right time for the right duration.

Attempts to develop PrEP regimens for HIV have been challenging, and the results confusing. We have summarized the work from seven trials (Table 4) [41,81–86]. To date, a limited number of agents have been used, selected primarily because they are well tolerated and because they provided protection in a macaque model using either rectal or vaginal exposure to SIV [87,88]. However (as discussed below), the relative safety of daily oral tenofovir for HIV-negative people has been questioned [89].

Table 4.

Preexposure prophylaxis clinical trials.

Study	Study population	Study size	Location	Intervention	Outcome
CAPRISA 004 [81]	Heterosexual women (aged 18–40 years)	898	South Africa	Coitally dependent TFV 1% gel (two doses up to 12 h precoitus and postcoitus)	39% protection; 54% protection calculated in participants using >80% of doses
iPrEx [82]	MSM (mean age in TDF/FTC group 27.5 years)	2499	North and South America, Thailand, South Africa	Daily oral TDF/FTC	44% protection; 92% protection calculated for patients with detectable drug concentrations
TDF2 [83]	Sexually active adults (approximately 90% aged 21–29 years)	1200	Botswana	Daily oral TDF/FTC	63% protection
PIP [84]	Heterosexual HIV serodiscordant couples; (median age 33 years)	4747	Botswana, Kenya, Rwanda, South Africa, Tanzania, Uganda, Zambia	Daily oral TDF or TDF/FTC	62% protection with TDF alone; 73% protection with TDF/FTC
FEM-PrEP [85]	Heterosexual women (aged 18–35 years)	1951	Kenya, South Africa, Tanzania	Daily oral TDF/FTC	Trial discontinued for futility in April 2011
VOICE (MTN-003) [86]	Heterosexual women (aged 18–45 years)	5029	Uganda, South Africa, Zimbabwe	Daily oral TDF or daily oral TDF/FTC or daily topical TFV gel	Oral TDF group discontinued for futility in September, 2011; TFV 1% gel and placebo gel groups discontinued for futility in November 2011; oral TDF/FTC group continues
HPTN 052 [41]	Heterosexual and homosexual HIV serodiscordant couples (60% aged 26–40 years)	1726; 37	Botswana, Kenya, Malawi, South Africa, Zimbabwe, Brazil, India, Thailand	Immediate or delayed ART in HIV-infected partner	96% protection

ART, antiretroviral therapy; FTC, emtricitabine; TDF, tenofovir disoproxil fumarate; TFV, tenofovir; TFV-DP, tenofovir diphosphate.

Two trials have used 1% tenofovir gel intravaginally with different results. Although the populations were similar, the dosage schedules were different. In the CAPRISA 004 study [81] heterosexual women at high risk of infection, ages 18–40 years, used the gel in a coitally dependent manner: one dose of gel up to 12 h before sex and one dose of gel up to 12 h after sex, with no more than two doses in 24 h. In one arm of the VOICE trial [86], women ages 18–45 used a dose of gel daily, regardless of sexual activity. The CAPRISA 004 study demonstrated 39% protection against HIV acquisition by tenofovir gel, whereas the tenofovir gel arm of the VOICE trial was stopped for futility. The reasons for the failure of tenofovir gel in VOICE trial after the success of CAPRISA 004 have not yet been determined.

Five studies of oral PrEP have been undertaken (Table 4) [82–86]. iPrEx is the only study devoted to MSM. In this trial [82], 2499 MSM in South America and the USA were provided a daily fixed-dose combination pill of tenofovir with emtricitabine. The investigators reported a 44% reduction in HIV acquisition compared with placebo controls. Data were also analyzed to consider self-reported pill usage. Incidence was reduced by 73% if self-reported adherence was high (>90% of doses taken), 50% if adherence was intermediate (>50% of doses), and 32% if adherence was low (<50% of doses). Among those who reported good adherence (taking study drug 50% of days), 46% of men who remained HIV-negative and 92% of men who seroconverted had no drug detected in selected blood and cell samples (Grant et al., 2010, Supplemental Table 8). The iPrEx investigators also used a case–control design to measure the association between detectable antiretrovirals in blood plasma and peripheral blood mononuclear cells and incident HIV infection. The majority of incident infections in the study drug arm had pharmacokinetic data available (34/36), and only 9% of these individuals had detectable study drug levels in selected plasma and cell specimens as compared with 51% of those who did not become infected. On the basis of this result, the authors argue that PrEP resulted in a relative risk reduction of 92% [95% confidence interval (CI) 40–99%] comparing patients with detectable study drug levels to those without detectable drug levels. Despite these promising findings, both self-report and pharmacologic markers raise concerns for accurately and reliably measuring drug adherence, an issue we discuss below.

Three studies of oral PrEP have involved women. The TDF2 study [83] enrolled 540 women and 660 men randomized to receive a daily fixed dose combination pill of tenofovir–emtricitabine or placebo. Study participants were predominantly unmarried adults ages 21–29 years living in Botswana. In this study, tenofovir–emtricitabine offered 64% protection against HIV infection. However, the study numbers were too small to draw definitive conclusions about protection in men and women separately, and 30% of those enrolled did not complete the study. In the FEM-PrEP study [85,90], 2120 heterosexual women aged 18–45 years living in high-prevalence areas in Kenya, South Africa, and Tanzania were randomized to receive either daily tenofovir–emtricitabine (FTC/TDF) or placebo. This study was discontinued for futility in April 2011. The investigators recently reported an HIV incidence rate of 4.7/100 person-years among the FTC/TDF group and 5.0/100 person-years in the placebo group for a hazard ratio for infection of 0.94 (95% CI 0.59–1.52, P = 0.81) [90]. Adherence may have been a critical contributing factor as less than 50% of infected cases and uninfected matched controls had detectable study drug in their blood plasma. Finally, the VOICE (MTN-003) study [86] enrolled heterosexual women aged 18–45 years in high-prevalence areas of Uganda, South Africa, and Zimbabwe. Women were randomized to daily oral tenofovir or daily oral tenofovir–emtricitabine. In September 2011, the daily oral tenofovir arm was stopped for futility, whereas the daily oral tenofovir–emtricitabine arm continues.

The largest trial, Partners in Prevention, focused on 4758 discordant heterosexual couples in Kenya and Uganda (38% negative women and 62% negative men) who reported intensive condom use and received counseling, in addition to the seronegative partner receiving daily tenofovir or daily tenofovir–emtricitabine. At the 2011 International AIDS Society meeting, the investigators reported that daily tenofovir conferred 62% protection against HIV acquisition and daily tenofovir–emtricitabine conferred 73% protection against HIV acquisition [83]. At the 2012 Conference for Retroviruses and Opportunistic Infections, the authors updated their results reporting 67% protection from daily tenofovir and 75% protection from the combined regimen [91]. Only two individuals who were infected at randomization and one individual infected after randomization developed resistance (K65R or M184V mutation).

These trials differ in many ways. However, clearly, poor adherence would limit success as reported for FEM-PrEP and iPrEx. It is also possible that the drugs employed are not perfectly suited to PrEP, especially in women. Substantial differences in antiretroviral drug concentrations in mucosal tissues have been reported [23,92], which may help explain these discordant study findings [93]. As noted above, after the first dose of oral FTC/TDF, rectal tissue concentrations are far greater than cervical or vaginal tissue concentrations [23]. This could explain why tenofovir–emtricitabine conferred protective efficacy in iPrEx [82], despite only modest adherence. In the VOICE 003 trial [86], the lack of protection with oral TDF could reflect low tissue concentrations of the drug in the cervix and vagina. Furthermore, differences between study populations in risk behaviors and underlying rate of infection deserve consideration.

Two new studies (HPTN 067 and HPTN 069) sponsored by the HIV Prevention Trials Network (HPTN), one study sponsored by ANRS (IPERGAY), and an extension of the iPrEx study (iPrEx OLE) are addressing some of the limitations of the previous PrEP studies. Due to the adherence issues with daily PrEP dosing, the behavioral study HPTN 067 (the ADAPT study: Alternative Dosing to Augment PrEP Pill-Taking) is designed to test the hypothesis that recommending intermittent usage of oral tenofovir–emtricitabine, compared with recommending daily usage, will be associated with equivalent coverage of sex events before and after exposure dosing, lower number of pills needed for coverage, and decreased severity and frequency of self-reported side effects [94]. HPTN 067 is enrolling 180 MSM and 180 heterosexual women and aims to identify dosing regimens that foster healthy sexual practices and pill-taking behavior in people at high risk of infection. The study includes a 6-week lead-in period, which includes directly observed therapy at enrollment and weeks 1 through 4, followed by 1 week without dosing to determine individual steady-state pharmacokinetics. Participants are then randomly assigned to one of three dosage groups in a 1 : 1 : 1 ratio: daily dosing, time-driven dosing, and event-driven dosing to be completed over 24 weeks.

iPrEx OLE (open-label extension) is a continuation of the iPrEx study that is enrolling participants in 11 sites in Peru, Ecuador, Brazil, the USA, South Africa, and Thailand [95]. It is hoped that knowing with certainty that one is on the study drug (and not placebo) will lead to higher medication adherence. The 72-week study will also assess the long-term efficacy and safety of PrEP, changes in sexual behavior, drug resistance, changes in bone mineral density and fat distribution, and the impact on hepatitis infection.

The IPERGAY study sponsored by ANRS is a randomized controlled trial among HIV-negative MSM testing the prevention efficacy of Truvada (tenofovir–emtricitabine; Gilead Sciences Inc., Foster City, California, USA) in combination with regular HIV/STD testing, immunization against hepatitis A and B, postexposure treatment as needed, and condom distribution [96]. Enrollment is currently ongoing in France and will also be extended to Canada.

Due to the possible limitations related to efficacy and toxicity (see below) of tenofovir–emtricitabine, HPTN 069 (NEXT-PrEP: Novel Exploration of Therapeutics for PREP) is assessing the safety and tolerability of maraviroc-containing PrEP regimens [97]. HPTN 069 is enrolling 400 MSM and 200 heterosexual women and will compare 48-week safety and tolerability of daily maraviroc, maraviroc–emtricitabine, maraviroc–tenofovir, or tenofovir–emtricitabine. Secondary objectives will include evaluation of electronically monitored adherence, pharmacokinetics in systemic and genital tract compartments, and efficacy of ex-vivo HIV challenge in tissue biopsy explants.

Alternative drugs and delivery systems more appropriate to PrEP are also being pursued. Vaginal rings for women containing antiretrovirals would be similar in concept to vaginal rings currently used for contraception and hormone replacement therapy [98]. These rings could maintain long-term, sustained antiretroviral release for local efficacy. Due to the long-term drug release, rings can be used in a coitally independent manner and inserted monthly, which could have an adherence benefit over gels or pills.

Dapivirine [TMC120; Janssen R&D Ireland (previously Tibotec)] has been formulated into a ring and is currently undergoing phase III studies [99]. Two phase I studies evaluating dapivirine 25 or 200 mg delivered from a vaginal ring over 7 days in 25 healthy women found the ring to be well tolerated, with adverse effects similar to placebo. Mucosal fluid sampled at up to 7 days after insertion in all women from the introitus, cervix, and ring area had mean drug concentrations 1000 times the 50% effective concentration (EC₅₀) against wild-type HIV-1 [100,101]. A combination dapivirine and maraviroc ring is currently undergoing a phase 1 safety and pharmaco-kinetic study.

An alternate formulation for better adherence is a long-acting injectable product that could be administered every 30–90 days. A rilpivirine (RPV) (TMC278 LA) nanosuspension is in early phase of development for this purpose [102,103]. In an exploratory study among 32 HIV-negative participants, a single intramuscular injection with either 300, 600, or 1200 mg of RPV showed varying, prolonged plasma, genital tract, and male rectum concentrations over 84 days [103]. Optimal dosage, differences in drug concentrations in various biological compartments, and long-term safety of multiple doses all require further study.

Although a potentially strong addition to the biomedical technologies available for HIV prevention, PrEP has given rise to a number of concerns [104]. First, HIV resistance is an important consideration while using oral antiviral agents. Among the 10 individuals enrolled in iPrEx with unrecognized HIV infection at baseline (and therefore subsequent unwitting exposure to ART directed at PrEP rather than treatment), three developed FTC-resistant mutations; this resistant mutation would be expected from the double-drug therapy employed [105]. However, among 36 men who became HIV infected during the trial in the FTC-TDF group, no ART resistance was observed. These results have sometimes been interpreted to indicate that PrEP does not threaten the utility of the ART agents used [106]. However, given the low adherence rates recorded in iPrEx, these results could also suggest that the study individuals used no ART product during the time after HIV was acquired and diagnosed. Detection of ART resistance markers represents a surrogate for failed PrEP usage, and ART resistance might compromise future management of people using PrEP who acquire HIV [105].

Second, the long-term biological impacts of FTC-TDF taken for PrEP purposes (as compared with HIV treatment purposes) will require additional study. Specifically, tenofovir has been linked with renal injury [89,107] and loss of bone mineral density (BMD) [108] when used for HIV treatment. Individuals with preexisting renal conditions have been excluded from PrEP studies [109,110]. In the iPrEx study, a nonsignificant trend toward elevated creatinine levels was found among the intervention arm [81], and a substudy showed a small but significant (up to 1%) loss in BMD [111].

An additional concern is that the use of ART as PrEP could affect sexual behavior [112]. In the iPrEx study, the investigators reported no increase in sexual risk behaviors. However, an individual's behavior within an unproven medical trial as compared with behavior of an individual under the belief of effective PrEP in a real-world setting must also be considered. ‘Risk compensation’, wherein individuals alter their behaviors in response to perceptions of risk, has been documented in relation to the availability of antiretroviral drugs and other biomedical HIV-prevention approaches [113–115]. Furthermore, even within a controlled trial setting such as the iPrEx study, self-reported adherence to ART was less than perfect, and other STDs were detected in both groups, challenging the veracity of the behavioral data collected. Finally, PrEP has generated discussions about resource distribution and how to balance the current HIV treatment coverage gap with expansion of preexposure prevention measures [116].

Taking antiretroviral treatment to scale: modeling, observation, and empirical data

Although investigators have been evaluating the potential utility of antiviral agents for prevention, a virtual parallel universe of researchers have been making the case that the benefits of ART are both inevitable, and already visible. This work has been conducted through evaluation of observational and ecologic data and mathematical modeling.

Mathematical modelers take the best available data and make assumptions about biology and behaviors to provide predictions of the future that are often provocative. A large number of models have focused on the usage of ART by HIV-infected persons to reduce the spread of HIV (Table 5) [117–125], as described in numerous review articles [2,126–128]. The results of several such modeling studies have suggested that expanded ART use could result in substantial reductions in HIV incidence under certain, optimistic conditions [120,129]. However, modeling studies have also shown that the population-level transmission prevention benefits of ART could be severely compromised by such factors as increased risk behavior [130–132], ongoing transmission during acute and early HIV infection [133], antiretroviral drug resistance [124], concentration of risk in population subgroups [122], and sub-optimal ART coverage, effectiveness, or adherence [125,134] (Table 5). Considerable uncertainty surrounds the fundamental assumptions and parameter values used in these models; more empirical data about risk behavior patterns, STD co-transmission, ART uptake and adherence, and effects of ART on infectivity (especially for anal contact and parenteral transmission) will facilitate more reliable model projections.

Table 5.

Selected modeling studies examining the impact of treatment as prevention on the HIV epidemic.

Source	Model setting	Model risk group(s)	Main predictions
Blower et al. [117]	San Francisco	MSM	Increased ART use could substantially reduce epidemic, but if risk behavior increases as a direct result of increased ART use, net benefit could be ~0
Gray et al. [118]	Uganda	Heterosexuals	ART alone cannot control mature HIV epidemics; ART in combination with a low-efficacy vaccine and in the absence of behavioral disinhibition could control the epidemic
Baggaley et al. [119]	Southern Africa	Heterosexuals	Increased treatment is unlikely to control the epidemic, regardless of extent of ART roll-out; counseling to reduce risk behaviors is essential
Granich et al. [120]	South Africa	Heterosexuals	An annual test-and-treat strategy could eliminate the HIV epidemic within 10 years
Bezemer et al. [121]	The Netherlands	MSM	Decreasing risk behavior will have the greatest impact on the HIV epidemic, but earlier diagnosis and treatment can also prevent substantial numbers of infections
Dodd et al. [122]	Theoretical	Heterosexuals	Test-and-treat interventions can substantially reduce HIV transmission in some contexts, but inadequate intervention coverage and uneven distribution of risk behaviors could dramatically compromise effectiveness
Long et al. [123]	USA	MSM, IDUs, heterosexuals	Substantial reductions in risk behavior will be required to markedly affect the epidemic, even with substantial expansion of HIV screening and treatment
Smith et al. [124]	San Francisco	MSM	Transmission of antiretroviral-resistant strains could substantially compromise HIV treatment programs
Walensky et al. [125]	Washington, DC	Not specified	A test-and-treat intervention with annual testing and realistic rates of intervention uptake is unlikely to halt the HIV epidemic

ART, antiretroviral therapy; IDUs, intravenous drug users.

One area of special interest is the potential balance between PrEP and treatment, as well as the targeted use of antiretroviral-based prevention strategies within discordant couples. Using available data to look at management of HIV-discordant couples, Hallett et al. [135] reported that use of PrEP by the uninfected partner could be at least as cost-effective as earlier ART initiation by the infected partner, provided that the annual cost of PrEP is less than 40% the cost of ART and the effectiveness of PrEP exceeds 70%. At the population level, Pretorius et al. [136] used a mathematical model to examine the impact and cost-effectiveness of PrEP relative to ART in South Africa. These investigators concluded that PrEP use would be most cost-effective if utilized before ART reaches 65% of HIV-positive persons; as ART coverage increased beyond this level, the cost-effectiveness of PrEP was predicted to decrease rapidly. El-Sadr et al. [137] evaluated treatment of HIV-infected partners within serodiscordant couples as a strategy for reducing HIV incidence at the population level and found that the predicted effectiveness in a given setting depended on HIV prevalence and the degree of HIV discordancy. In general, these results do not support a focus on HIV-discordant couples for public health purposes. As with the modeling studies of treatment as prevention described above, these studies indicate that PrEP effectiveness will depend on many factors and that the choice of intervention must take into account the epidemiological context in a given setting.

Combination prevention: the way forward

Implementation of ART as prevention faces substantial challenges, including logistic limitations [73,75], potential changes in risk-taking behaviors, and cost. Indeed, ART usage will need to be part of a combination strategy [138]. In a remarkable 2011 speech at the NIH, Secretary of State Hillary Clinton expressed hope for an ‘AIDS Free Generation’ [139]. On World AIDS Day 2011, President Barack Obama pledged an additional US$ 35 million for state AIDS drug assistance programs based on the HPTN 052 findings [140]. PEPFAR and the WHO have recommended the use of ART to prevent HIV transmission among heterosexual partners [46,141], and UNAIDS is looking into similar guidelines [68]. But attention to broader use of ART cannot ignore other parts of a prevention package [138]. Secretary Clinton concluded that broader prevention of mother-to-child transmission, more circumcision, and optimal and broader use of ART will point us in the right direction [139]. Additionally, combination prevention strategies [142] will need the continued efforts of behavioral interventions to increase condom use, reduce high-risk behaviors, and address suboptimal antiretroviral adherence and risk compensation. The community-based clinical trials described in this article, though focused on ART, all embrace a similar combination prevention strategy. And all of these trials and advances must also recognize the important discoveries in the field of HIV vaccine [143,144]. It seems reasonable to expect that – after 30 years of work – the tools now available in the HIV prevention toolbox and those that will become available from ongoing research can be expected to control the spread of HIV.