AIDS RESEARCH AND HUMAN RETROVIRUSES Volume 27, Number 6, 2011 ª Mary Ann Liebert, Inc. DOI: 10.1089/aid.2010.0206
AIDS Vaccines and Preexposure Prophylaxis: Is Synergy Possible? Jean-Louis Excler,1 Wasima Rida,2 Frances Priddy,1 Jill Gilmour,3 Adrian B. McDermott,1 Anatoli Kamali,4 Omu Anzala,5 Gaudensia Mutua,5 Eduard J. Sanders,6,7 Wayne Koff,1 Seth Berkley,1 and Patricia Fast1
While the long-term goal is to develop highly effective AIDS vaccines, first generation vaccines may be only partially effective. Other HIV prevention modalities such as preexposure prophylaxis with antiretrovirals (PrEP) may have limited efficacy as well. The combined administration of vaccine and PrEP (VAXPREP), however, may have a synergistic effect leading to an overall benefit that is greater than the sum of the individual effects. We propose two test-of-concept trial designs for an AIDS vaccine plus oral or topical ARV. In one design, evidence that PrEP reduces the risk of HIV acquisition is assumed to justify offering it to all participants. A two-arm study comparing PrEP alone to VAXPREP is proposed in which 30 to 60 incident infections are observed to assess the additional benefit of vaccination on risk of infection and setpoint viral load. The demonstrated superiority of VAXPREP does not imply vaccine alone is efficacious. Similarly, the lack of superiority does not imply vaccine alone is ineffective, as antagonism could exist between vaccine and PrEP. In the other design, PrEP is assumed not to be in general use. A 22 factorial design is proposed in which high-risk individuals are randomized to one of four arms: placebo vaccine given with placebo PrEP, placebo vaccine given with PrEP, vaccine given with placebo PrEP, or VAXPREP. Between 60 and 210 infections are required to detect a benefit of vaccination with or without PrEP on risk of HIV acquisition or setpoint viral load, with fewer infections needed when synergy is present.
NAIDS estimates that in 2008 2.7 million people worldwide became newly infected with HIV and 2 million people died from AIDS. The total number of people living with HIV is estimated to be near 33.4 million, with 97% living in low- and mid-income countries and 48% being women.1 Preventing HIV infection and AIDS is a global priority, and a number of behavioral, barrier, antiretroviral drug-based and vaccine approaches have been the subject of intense research.2,3 While antiretroviral treatment (ART) in HIVinfected individuals slows progression to AIDS and death, its benefit at a population level is dependent on the availability of drugs and adherence to treatment regimens. A safe, effective, and durable vaccine is needed to stem the pandemic.
The development of an AIDS vaccine has faced many challenges over the past 25 years. Current AIDS vaccine candidates do not elicit broadly neutralizing antibodies capable of preventing infection by the vast diversity of HIV isolates.4 The first test-of-concept trial of a cell-mediated immunity (CMI)-based vaccine did not show efficacy and raised the concern that the vaccine might enhance the risk of infection for some vaccinees.5 Encouragingly, the recent RV144 Thai trial of a prime-boost vaccine regimen aiming at inducing both CMI and antibody responses showed in a modified intention-to-treat analysis a 31% reduction in the risk of infection.6 Vaccine candidates in current trials are designed to induce HIV-specific CMI responses that could slow HIV replication, destroy HIV-infected cells, preserve immune memory, delay disease progression, and possibly prevent
International AIDS Vaccine Initiative, New York, New York. Biostatistics Consultant, Arlington, Virginia. IAVI Human Immunology Laboratory, Imperial College, London, United Kingdom. 4 Medical Research Council, Uganda Virus Research Institute, Uganda Research Unit on AIDS, Entebbe, Uganda. 5 Kenya AIDS Vaccine Initiative, University of Nairobi, Nairobi, Kenya. 6 Centre for Geographic Medicine Research—Coast, Kenya Medical Research Institute (KEMRI), Kilifi, Kenya. 7 Centre for Clinical Vaccinology and Tropical Medicine, University of Oxford, Headington, United Kingdom. 2 3
670 secondary transmission by reducing the quantity of virus in body fluids (viral load).7–9 Other HIV prevention modalities have shown promise. Prompt postexposure prophylaxis reduces the risk of transmission due to occupational10 or sexual exposure.11,12 Treatment of mother and/or child before and after childbirth has reduced the risk of mother-to-child transmission.13,14 Very recently, the first successful demonstration of protection by tenofovir-containing vaginal gel has provided proof-ofconcept for topical microbicide preexposure prophylaxis (PrEP). Tenofovir gel reduced HIV acquisition by an estimated 39% overall, and by 54% in women with high adherence to the dosing regimen of before and after sexual exposure.15 Animal studies have provided evidence that PrEP and postexposure prophylaxis (PEP) with tenofovir disoproxil fumarate (TDF), alone or in combination with emtricitabine (FTC), can protect against simian immunodeficiency virus (SIV) and chimeric simian HIV (SHIV) infection.16–22 Furthermore, in some experiments, animals that became infected despite receiving TDF as PrEP or PEP showed delayed onset of viremia and delayed seroconversion,18,20,23 demonstrating that even in the case of incomplete protection against infection, PrEP or PEP may lead to attenuated acute infection (perhaps by allowing the immune system to gain some control of the virus). The demonstrated efficacy of TDF in the treatment of HIV infection, the lack of reported serious safety problems associated with its use,24 its long halflife,25–28 and its relatively high threshold to drug resistance29 make this antiretroviral agent an attractive candidate for systemic or topical prophylactic use. Tenofovir and related drugs have high concentrations in genital tissues,30–32 so repeated exposure with locally aborted HIV infection is theoretically possible and, if it occurs, may lead to the development of HIVspecific cellular immune responses. Several clinical efficacy trials using TDF alone or in combination for oral or topical prophylaxis are ongoing (www.prepwatch.org). Other drugs are also being considered for use in such regimens.33 However, more safety and efficacy data are needed, and there are significant practical barriers to widespread prophylactic use of ARV.34 ART for HIV-infected individuals may prove useful in the prevention of secondary transmission. Viral load is the single greatest risk factor for all modes of transmission.35,36 Sexual transmission of HIV has been closely linked to viral load in the blood of the infected host,37–41 which probably serves as a surrogate, albeit imperfect, for HIV concentration in the genital tract.42,43 In addition to lowering plasma viral load to nearly undetectable levels, ART can decrease viral load in genital secretions,44,45 although patients having a detectable semen HIV load may have no detectable virus in their blood plasma, highlighting the residual risk of HIV-1 transmission during unprotected intercourse.46 Observational studies and recent modeling work have triggered considerable interest and concerns regarding the use of ART in HIV-infected individuals to prevent secondary transmission.32,47–54 In the coming years, systemic PrEP or drug-based microbicides (which in this discussion is considered a topical form of PrEP) may be proven to safely confer protection against HIV infection. In that instance, even though the implementation of a ‘‘background’’ intervention would require careful consideration by national governments,55 PrEP might be proposed as a baseline intervention for AIDS vaccine trial
EXCLER ET AL. participants. In effect, by allowing repeated exposures to HIV that do not lead to established infection, PrEP may allow exposure-induced immunity to alter vaccine-induced immunity; on the other hand, vaccination could influence susceptibility, thus altering the efficacy of PrEP. Similarly, ART in HIV-infected individuals to prevent secondary transmission may gain momentum in populations with a high prevalence of HIV infection. Low-dose exposure, presumably more common when the donor is receiving ART, which can reduce virus concentration in both blood and genital secretions, might therefore be countered more easily by preexisting, vaccine-induced immunity in the recipient. These possibilities offer new opportunities for combination biomedical HIV prevention research and raise several questions about potential interactions and/or synergies between PrEP, ART for prevention, and AIDS vaccines. Such potential interactions should be considered when designing intervention trials and when ‘‘translating’’ the results of randomized, controlled trials to predictions of how the interventions, when introduced into broad use, may influence public health. Hypotheses When given in combination, an AIDS vaccine and PrEP could act synergistically to reduce the risk of HIV infection or attenuate the initial infection, thereby slowing disease progression. The overall benefit of the combination could be greater than the sum of the individual benefits of vaccine and PrEP. In this article we describe evidence in animals and humans to support this notion and examine the design of clinical trials exploring simultaneous use of an AIDS vaccine and PrEP. In this section, evidence is presented to support two hypotheses:
PrEP may reduce the infectious exposure sufficiently that a person can withstand it, especially if she or he has preexisting, vaccine-induced immunity. Repeated abortive infections (potentially possible under PrEP) may allow mucosal or systemic immunization that could be enhanced by vaccine priming and/or boosting.
Weaker challenge dose favors vaccine protection In the SIV model, vaccines work better against lower virus concentration and low-dose repeated challenges.56–58 By reducing initial HIV viral replication, PrEP might convert a strong HIV ‘‘challenge’’ (through natural exposure) into a weaker challenge that a modest level of vaccine-induced immunity could overcome, preventing establishment of HIV infection or leading to a lower viral load and slower disease progression in the long term. Additionally, when given around the time of vaccination, PrEP might provide a protective ‘‘cover’’ if there is a transient period of enhanced susceptibility during and shortly after vaccination, followed by a period of immunity. Mucosal exposure in the absence of chronic infection can induce immune responses In some nonhuman primate (NHP) experiments, virusspecific mucosal immune responses have been detected after a single low-dose exposure to SIV by the rectal route in the absence of overt infection.59 Both timing and location may be
AIDS VACCINE AND PrEP SYNERGY
critical determinants of outcome prevention in the earliest stages of infection. When infected founder populations are small and focal, and have not as yet expanded sufficiently to disseminate and establish systemic infection, a relatively small number of effectors at the mucosal site of entry might be at the right place, ‘‘enough and soon enough’’ to clear infection,60–62 and this aborted infection might induce immune responses. Animal experiments suggest that sequential mucosal and systemic (or reverse order) vaccinations can contribute to a better immune response to mucosal challenges.63–65 Systemic vaccination followed by continued mucosal exposures in the presence of PrEP that is sufficient to prevent established infection may allow more effective development of mucosal response. The possibility of a detrimental interaction between ARV for prevention and vaccination has not yet been assessed. The prognostic significance of HIV-specific cytotoxic T lymphocyte (CTL) responses in highly exposed seronegative individuals is unknown. A systematic review of these studies suggests that natural resistance to HIV infection in highly exposed individuals is mediated by multiple mechanisms, conferring on them an ‘‘immunologic advantage’’ that may be related to innate and/or adaptive immune systems.66,67 Although offering only a partial explanation to the natural resistance to HIV infection,68 observations in female sex workers in Nairobi with frequent HIV exposure,69–71 in exposed uninfected infants with strong HIV-1-specific T cell responses,72 and more recently in discordant couples73 suggest that recurrent viral exposure may lead to priming of HIV-specific systemic and mucosal immune responses. These findings suggest that persons receiving PrEP who are exposed to HIV may generate HIV-specific T cell responses, potentially contributing to protection against the establishment of HIV infection. These immune responses could possibly boost or prime vaccine-induced responses. One scenario for vaccine and PrEP synergy would include a short period of limited mucosal infection terminated by PrEP, which would boost existing vaccine-induced responses. In this sense, PrEP would allow mucosal ‘‘vaccination,’’ which has been shown to boost virus-specific immune responses in animal studies. Systemic prime followed by mucosal boost vaccination or mucosal prime followed by systemic boost can lead to stronger systemic responses compared to either separately, although for the systemic-mucosal sequence both groups exhibited equivalent, significant protection and robust post-SIV challenge cellular immunity.30,61–63
with tenofovir in macaques infected with SIV76–81 and SHIV82 has been associated with the stimulation of SIV- or SHIVspecific CD4þ and CD8þ T cell responses and resistance to rechallenge with heterologous virus more than 1 year following the initial challenge.83–85 A possible interpretation of these results is that drug treatment suppressed viral replication in the infected animals sufficiently to allow development of effective immune responses that controlled or eliminated SIV.23,75,86,87 However, in other studies, animals did not develop readily measurable cellular anti-SIV immune responses, and did not resist homologous rechallenge with SIVmac239.88 Rhesus macaques experimentally infected with SHIV while on ART showed long-lasting virus-specific cell-mediated immune responses despite low and/or undetectable viral loads compared to control animals, leading to long-term disease protection.89 The generation of immune responses in PrEPinfected animals could lead to synergy with vaccine-induced immune responses. Nonhuman primate studies support the concept of a significant protective effect against repeated low dose intrarectal SIV and SHIV challenge, after rectal application of tenofovir prior to virus challenge.20 The hypothesis is that extremely low levels of virus replication could occur during PrEP and induce virus-specific immune responses. Macaques protected by PrEP from challenge may harbor occult virus that may be initially contained by cytotoxic T cells, but could emerge later if immunity fades. Interestingly, in the absence of seroconversion, gag-specific interferon (IFN)-gsecreting T cells were detected in the blood of protected animals.74 Other studies using CD8 depletion showed that there was no virus present, suggesting that PrEP may have cleared the virus.90,91 In humans, although anecdotal, drug-susceptible HIV-1 infection despite intermittent fixed-dose combination TDF/ FTC as prophylaxis was associated with low-level viremia, delayed seroconversion, attenuated clinical course, and relative sparing of mucosal CD4þ T cells in the gastrointestinal tract of a patient. Although HIV-specific immune responses were not collected, this observation suggests an associated immune mechanism.92 More immunological data are expected from ongoing studies.55
PrEP may lead to abortive or controlled infection associated with immune responses
Experiments in macaques support the idea that antiretroviral drugs given around the time of infection can suppress viral replication in infected animals sufficiently to allow the development of immune responses.74 Mathematical modeling of ARV treatment during acute SIV infection suggests that transient treatment minimizes immune impairment and allows immune sensitization, leading to high levels of memory CTL for long-term SIV control.75 SIV exposure followed by PEP may activate immune responses to protect against subsequent infection with heterologous SIV challenge virus.17,76 Transient early treatment
Clinical Trial Designs Nonhuman primate studies could test the concept of AIDS vaccine and PrEP synergy to protect against HIV infection or to modulate disease.93,94 This article focuses exclusively on clinical trial designs in humans. Two clinical trial designs are proposed.
The first design assumes that PrEP efficacy is firmly established in human trials; it aims at detecting additional benefit from an AIDS vaccine candidate. The basic AIDS vaccine and PrEP (VAXPREP) trial design is a two-arm, randomized, double-blind study comparing a single AIDS vaccine candidate to vaccine placebo in a setting in which all participants receive open-label PrEP. Entry criteria would be such that the expected HIV incidence rate would be at least 2% per year. After a run-in period to assess compliance with PrEP, individuals who tolerate PrEP and take at least 80% of intended doses would be randomized in a 1:1 ratio to receive either the
672 AIDS vaccine candidate or placebo. The optimal dose and schedule for the vaccine would be based on data from previous Phase I/II trials. Vaccination is assumed to take 6 months after which participants would be tested for HIV infection every 3 months. Participants would be followed for a minimum of 12 months after vaccination is completed. PrEP would be given throughout the trial period, although a brief period of post-PrEP observation might be used at the end of the trial to ensure that there were no infections masked by PrEP or vaccine. HIV infection would be determined by a prespecified diagnostic algorithm including a serological test that can distinguish between vaccination and HIV infection and nucleic acid detection assay. Trial participants who become HIV infected would discontinue PrEP and have their HIV RNA plasma viral load measured at two time points postinfection, approximately 3–6 months after PrEP is stopped, with the geometric mean of the measurements used as their setpoint viral load. The time points correspond to at least 3 and 6 months after the initial positive HIV test, which would exclude acute HIV infection viral load peaks. Infected participants would be offered enrollment into a long-term follow-up study to assess trends in postinfection outcomes such as disease progression, viral load, CD4þ T cell count, and drug resistance. Trial participants found to be HIV infected at the time of the last dose of study vaccine or placebo would have baseline specimens assessed by PCR to rule out a preexisting infection. Those found to be infected at baseline would be excluded from the modified intention-to-treat (mITT) analysis set for the principal evaluation of VAXPREP efficacy. These participants, however, would be included in the safety evaluation of VAXPREP. We postulate that the combination of an AIDS vaccine and PrEP may reduce the risk of HIV infection or reduce the plasma viral load in those who become infected, relative to PrEP alone. In statistical terms, we treat HIV infection and viral load as coprimary endpoints and test a composite null hypothesis that the vaccine neither reduces the risk of HIV infection nor the viral load at setpoint. The composite null hypothesis is expressed as H0: VEs ¼ 0 and DVL ¼ 0, where VEs is the percent reduction in the cumulative HIV incidence due to vaccination and DVL is the difference in average viral load measured on a log10 scale between vaccine recipients and placebo recipients who become infected. Following the approach of the Step trial,95 we would test H0 against the onesided alternative H1: VEs > 0 or DVL > 0. (See the Appendix for detailed statistical considerations.) Table 1 presents the total number of incident HIV infections required to have 80% power to reject the composite null hypothesis for different values of VEs and DVL. In these calculations, we have assumed that PrEP by itself reduces the risk of HIV acquisition by 50% in the face of adherence similar to that seen in PrEP trials and reduces setpoint viral load from an average of 4.5 to 4.0 log10. Individual setpoints are assumed to vary with a standard deviation of 1.0 log10, a value somewhat larger than that in cohorts such as the MACS,95 but in line with at-risk groups in Africa (IAVI, unpublished data). We also assume no postrandomization selection bias for viral load. For example, if vaccination was more likely to prevent infection in vaccinees who would have had higher viral loads had they not been vaccinated, then viral load might appear
EXCLER ET AL. Table 1. Total Number of Incident HIV Infections Required to Have 80% Power to Reject the Composite Null Hypothesis H0: VEs ¼ 0 and DVL ¼ 0 for a 2-Arm VAXPREP vs. PREP Trial DVL VEsa
0% 10% 20% 30% 40% 50% 60%
>100 >100 >100 >100 87 55 36
>100 >100 98 75 55 40 0. For the endpoint of HIV infection, a test statistic Z1 based on the difference in cumulative incidence and its corresponding one-sided p-value p1 would be calculated. For the endpoint of viral load, a test statistic Z2 based on the Wilcoxon rank sum test and its corresponding one-sided p-value p2 would be computed. The two p-values are then combined using Fisher’s method. If the resulting p-value is less than 0.05, preliminary evidence of efficacy is demonstrated and the vaccine candidate is considered for further efficacy testing. Fisher’s test gives equal weight to the infection and viral load endpoints. Therefore, VAXPREP could be declared efficacious over PrEP if either a significant reduction in the risk of HIV infection or a significant decrease in setpoint viral load is observed. If one endpoint is felt to be more important for making a decision to move VAXPREP forward into Phase III trials, a weighted version of Fisher’s test could be employed in which the endpoint of greater importance is given greater weight. Another approach is to split the Type I error probability between the two endpoints. VAXPREP, for example, could be considered efficacious if the observed reduction in HIV incidence is significant at the 0.04 level or if the observed decrease in setpoint viral load is significant at the 0.01 level.
Address correspondence to: Patricia Fast International AIDS Vaccine Initiative 110 William Street, 27th Floor New York, New York 10038 E-mail: pfa[email protected]