Evaluating the efficacy of therapeutic HIV vaccines through analytical treatment interruptions

Introduction The development of an effective therapeutic HIV vaccine that induces immunologic control of viral replication, thereby eliminating or reducing the need for antiretroviral therapy (ART), would be of great value. Besides the obvious challenges of developing a therapeutic vaccine that would generate effective, sustained anti-HIV immunity in infected individuals is the issue of how to best assess the efficacy of vaccine candidates. Discussion This review discusses the various outcome measures assessed in therapeutic HIV vaccine clinical trials involving individuals receiving suppressive ART, with a particular focus on the role of analytical treatment interruption (ATI) as a way to assess the virologic control induced by an immunotherapy. This strategy is critical given that there are otherwise no readily available measures to determine the ability of a vaccine-induced immune response to effectively control HIV replication. The various outcome measures that have been used to assess vaccine efficacy in published therapeutic HIV vaccine clinical trials will also be discussed. Outcome measures have included the kinetics of viral rebound, the new viral set point and changes in the size of the viral reservoir. Clinically relevant outcomes such as the CD4 decline, the time to resume therapy or the time to meet the criterion to resume therapy, the proportion of participants who resume therapy and/or the development of clinical symptoms such as acute retroviral syndrome are also measures of vaccine efficacy. Conclusions Given the lack of consistency between therapeutic HIV vaccine trials in how efficacy is assessed, comparing vaccines has been difficult. It would, therefore, be beneficial to determine the most clinically relevant measure for use in future studies. Other recommendations for future clinical trials also include studying compartments in addition to blood and replacing ATIs with single-copy assays in situations in which the use of an ATI is not ideal.


Introduction
The idea that HIV-positive individuals might benefit from therapeutic immunization was first proposed by Jonas Salk in 1987 [1]. The discovery since then of long-term nonprogressors and elite controllers whose immune systems naturally control HIV infection without the need for antiretroviral therapy (ART) provides evidence for effective hostmediated anti-HIV immunity, thus providing a rationale for the development of therapeutic vaccines (reviewed in Refs. [2Á4]).
The development of an HIV therapeutic vaccine capable of inducing control of HIV replication such that ART could be eliminated is a major focus of HIV research [5Á7]. While ART has transformed HIV infection into a chronic, manageable disease for most individuals who have access to it [8,9], ART is associated with a number of disadvantages and limitations. In addition to being a lifelong therapy [7,8,10], ART can be toxic [8,9], is potentially associated with the development of HIV drug resistance [9] and does not eliminate latent HIV in viral reservoirs [6Á9]. Finally, the high cost of ART makes it unavailable to the majority of the world's HIV-positive individuals who live in resource-limited countries [8Á11]. A therapeutic vaccine would, therefore, circumvent many of the limitations associated with ART.
Besides the obvious challenges of developing a therapeutic vaccine that would induce effective, sustained anti-HIV immunity in infected individuals is the issue of how to best assess the efficacy of vaccine candidates [12]. In many clinical trials of therapeutic HIV vaccines (Tables 1AÁ1E), assessing efficacy involves comparing various outcome measures before and after an analytical treatment interruption (ATI), which is used to assess vaccine-induced, immune-mediated viral control [2,5]. While therapeutic HIV vaccine clinical trials typically include the CD4 count as a safety/clinical event, virologic outcome measures vary from trial to trial, making it challenging to compare the results of different vaccine studies. [137] Long-term observation (1.5 years) after immunization in Ref. [136] Percentage of participants who resumed ART Immunogenicity CD4 T cell count pVL Participants with the greatest DTH responses following immunization were less likely to require ART resumption compared to low responders. [138] Observation period four years after enrolling in Ref. [136] Time until ART resumption

Discussion
The current state of non-HIV therapeutic vaccines Only a few therapeutic vaccines are currently licensed worldwide and most of them are used to treat cancer [13]. In 2010, the US Food and Drug Administration (FDA) approved sipuleucel-T (Provenge † ) to treat hormone-refractory prostate cancer [14]. A therapeutic vaccine for ovarian cancer has been approved in Dubai [13], while another one was recently given fast track designation by the FDA [15]. Two different therapeutic vaccines for renal cell carcinoma have been approved, one each in Russia and South Korea [13]. Zostavax † is a therapeutic vaccine that reduces the frequency and severity of shingles, which is caused by the reactivation of the varicella zoster virus that causes chickenpox [39]. Zostavax is the first example of a vaccine with clinical efficacy against an established infection [40]. The success of       this vaccine provides hope that it might be possible to induce clinically beneficial immunity against other viruses that establish chronic infections.

The current state of therapeutic HIV vaccines assessed in clinical trials
The features that might make a therapeutic vaccine effective and the inherent challenges of making such a vaccine have been recently described in a number of excellent reviews [3,6,98Á102]. Minimally, a therapeutic vaccine should improve the benefits of existing ART regimens, simplify these regimens or allow for periodic ART interruption [6,98Á100]. Ideally, a therapeutic vaccine would eliminate the need for ART, either by eradicating virus (a sterilizing cure) or by inducing an immune response capable of controlling virus replication (a functional cure) [6,98Á100,102].
Therapeutic vaccination would be of particular value for HIV-positive individuals residing in resource-limited countries in which access to ART is limited [98]. In these settings, the HIV epidemic is fuelled by the higher rate of new infections relative to the rate at which newly infected individuals receive ART [98,99]. An effective therapeutic vaccination could, therefore, help control the epidemic. Therapeutic vaccines would also be invaluable to those who struggle with daily, lifelong ART compliance [98].
Over the course of more than two decades, more than four dozen therapeutic HIV vaccine candidates have been evaluated in clinical trials for safety, immunogenicity and, in some cases, for efficacy. The results of these trials have shown limited success (reviewed in Refs. [3,6,99Á102]) with respect to their ability to control HIV replication or maintain CD4 T cell counts in the absence of ART [99,102,103]. While the majority of these trials involved therapeutic vaccination of individuals who initiated ART during chronic HIV infection, vaccination of individuals who initiated ART during acute or early infection was also ineffective [104Á108]. In one of these studies, the dynamics of viral rebound following vaccination and ATI were similar to those observed in studies of chronically infected individuals who discontinued ART [105].
In a trial assessing the effects of receiving two vaccines, ALVAC-HIV vCP1433 and Lipo6-T, followed by IL-2 administration, Levy et al. [109] observed that a significantly greater proportion of vaccinated HIV-positive participants had a lower viral set point 12 weeks after stopping ART, compared to unvaccinated controls. The times to viral rebound and to resume therapy were also significantly delayed in the vaccinated participants.
Garcia et al. [110] observed that therapeutic vaccination of HIV-positive participants with an autologous dendritic cell vaccine loaded with autologous, inactivated HIV-1 resulted in a decrease in the viral load set point following ATI. Unfortunately, the decrease in the viral load induced by vaccination was transient. Furthermore, vaccination did not prevent the CD4 T cell count decline after interruption of ART. It was subsequently reported that, although no change was observed in the size of the viral reservoir during the vaccination period, vaccine-induced T cell responses transiently delayed the replenishment of the viral reservoir after ATI [112].
Pollard et al. [111] observed that the Vacc-4x vaccine, which contains a mixture of conserved Gag peptide domains, was able to significantly reduce the viral load following ATI, resulting in a new viral load set point. However, vaccination did not affect the change in the CD4 T cell count following ATI, nor did it affect the proportion of participants who resumed therapy or the time until therapy was resumed.
The role of ATIs in assessing the efficacy of therapeutic HIV vaccines ART may be interrupted as part of a structured treatment interruption (STI) or as part of an ATI. The main goals of the STI are to reduce ART-associated burden (reviewed in Refs. [4,113Á115]) and/or to induce HIV ''autoimmunization'' (reviewed in Refs. [3,4,114,116]), whereas the purpose of the ATI is to assess the efficacy of an experimental therapeutic candidate [12]. STIs and ATIs are discussed in further detail below.
STIs have been used in the past to address the ARTassociated issues of toxicity, cost and resistance (reviewed in Refs. [4,113Á115]). Another goal of the STI was to allow for viral rebound, resulting in ''autoimmunization'' with increased exposure to HIV antigens (reviewed in Refs. [3,4,114,116]). It was hypothesized that the resulting viremia would boost the anti-HIV immune response sufficiently to induce viral control, thus avoiding ART resumption. Unfortunately, the various clinical trials that assessed the immunological benefits of STIs failed to show any benefits (reviewed in Refs. [4,114]), while the SMART study showed that treatment interruptions can increase morbidity and mortality [117].
When ART is interrupted, plasma HIV RNA levels typically first become detectable within days or weeks [118Á120], reach a peak and then decrease to a steady state level, or viral set point [121]. Exceptions to the occurrence of viral rebound following therapy interruption do occur and may be more frequent in those who are treated during acute primary infection [122], although the exact immune mechanisms responsible for this degree of viral control are currently unknown.
The ATI is an intentional interruption of ART that is included in controlled clinical trials of therapeutic vaccines (reviewed in Refs. [2,4,5,115,123]). The ATI is a frequently used strategy for assessing HIV therapeutic vaccine efficacy [12]; it is considered by some to be the ''gold standard'' [5]. This strategy, which is used to assess the virologic control induced by an immunotherapy given while the individual is still taking suppressive ART [2,115,123] is necessary because there are currently no laboratory assays that measure the ability of the immune system to effectively control HIV replication [2,99,123]. In addition to assessing the kinetics of viral rebound [2,99], ATIs also allow for the assessment of a potentially new viral set point as well as CD4 T cell dynamics following treatment interruption [115,123].
The SMART study revealed that HIV-positive participants who interrupted ART experienced an increased risk of developing AIDS and non-AIDS events compared to participants who continued therapy [117]. However, it also showed that individuals having high CD4 counts (!500 cells/ml), high CD4 nadir (!200 cells/ml) and undetectable virus levels ( B50 copies/ml) can safely undergo treatment interruptions in carefully monitored clinical trials without increasing their risk of death and non-AIDS events or developing viral resistance [115,124Á126]. It was recently shown that chronically infected individuals having undetectable virus levels and preserved CD4 counts, including those with low CD4 nadir, can also safely undergo treatment interruptions if the interruptions are short, that is if treatment is reinitiated upon detection of viral rebound [127].
Despite the safety of ATIs, the increased viral load that occurs following treatment interruption can occasionally be associated with the development of an acute retroviral syndrome [128,129] or thrombocytopenia [130], as well as with an increased risk of HIV transmission by individuals involved in high-risk activities [131].

Alternatives to ATIs
In studies that include an ATI, therapy is typically reintroduced either at the end of a fixed period of treatment interruption (e.g. 16 weeks), during which time a new viral set point is usually achieved, or when specific virologic, immunologic or clinical outcomes are met. A new, alternative treatment interruption strategy in clinical trials of HIV immunotherapies is the monitored antiretroviral pause (MAP), which reintroduces ART as soon as viral rebound occurs [123]. The advantage of the MAP is that, by reintroducing ART as soon as viral rebound occurs, the risk is reduced compared to the risk associated with an ATI. However, since the MAP does not allow a new viral set point to be established, this strategy cannot be used to determine whether the immunotherapy being tested improved virologic control by the immune system. Thus, whether an ATI or a MAP should be used in a clinical trial of an HIV immunotherapy depends on the scientific question being asked, with the MAP lending itself to assess therapies designed to measure the time to viral rebound, which may be a surrogate measure of the size of the viral reservoir, while the ATI should be used to assess therapies designed to improve immune control of HIV. It should be noted, however, that it has not yet been established whether the time to viral rebound following ATI is, in fact, a surrogate measure of the size of the viral reservoir [132].
Recently, single-copy reverse transcriptase (RT)-qPCR assays with single-copy sensitivity (i.e. the single-copy assay (SCA) for HIV-1 RNA) were used to detect virus in the plasma of individuals who had undergone myeloablation and autologous stem cell transplantation for the treatment of lymphoma [133]. Since these patients had undetectable plasma viremia by standard methods, it was hypothesized that their lymphoma treatment had resulted in HIV eradication; the results of the SCA, however, proved otherwise. Therefore, in this setting, SCA was used to guide the decision regarding whether to interrupt ART; because virus was detected using this assay, ART was not interrupted and the viral rebound that would have otherwise occurred was avoided. However, had the SCA failed to detect virus, then an ATI would have been warranted. The use of these highly sensitive assays has been suggested as an additional approach to the assessment of therapeutic vaccine efficacy [5]. The inclusion of such assays into future clinical trials of HIV immunotherapeutics could expedite these trials if only subjects with undetectable viral load by SCA proceeded on to ATI. However, such an alternative approach would need to be validated first by concurrent analysis in clinical trials in which it can be determined that SCA results predict virologic rebound following ATI [5]. In the interim, or until some other strategy is validated, treatment interruptions, the current gold standard for assessing therapeutic vaccine efficacy [5], will continue to play a crucial role in the evaluation of HIV therapeutic vaccines and should only be replaced with some other strategy if treatment interruption must be avoided.
It was recently shown that the ex vivo antiviral capacity of CD8' T cells [134] predicts the rate of CD4 T cell loss in early HIV infection and is inversely correlated with viral load set point [135]. It has been suggested, therefore, that this assay [134] be included as a read-out in clinical trials of therapeutic vaccines. However, whether this accurately measures vaccineinduced immunologic control of viral replication remains to be established [135].

Read-outs of therapeutic HIV vaccine studies that incorporate ATIs
More than four dozen therapeutic HIV vaccine candidates have been evaluated in clinical trials for safety, immunogenicity and, in some cases, for efficacy. Tables 1AÁ1E summarizes the outcome measures of efficacy that have been assessed in published vaccine trials that include ATI.
Since the correlates of viral suppression/immunological response that should be used to assess the therapeutic benefits of vaccines are not well defined [7,99,135,164Á166], the surrogate measure(s) used as the primary end point(s) to assess the clinical benefits of therapeutic vaccine candidates vary from trial to trial.
The virologic outcome measures assessed following vaccination and ATI may include the time to detectable viremia, the peak level of viremia, the new viral set point, the time to reach the new viral set point, the time to reach a certain viral load threshold, the viral load at a predefined time following ATI and changes in the size of the viral reservoir.
Of these read-outs, it has been suggested that the new viral set point is the most relevant clinical assessment of the antiviral efficacy of a therapeutic vaccine (reviewed in Ref. [5]). Whereas the new viral set point, the peak level of viremia and the time to rebound are all affected by the strength of the host's anti-HIV immune response, the peak level of viremia and the time to rebound may also be affected by the number of susceptible target cells and the size of the viral reservoir, respectively. It has also been suggested that the new viral set point established after immunotherapy and ATI should be the primary end point of clinical trials for assessing the effectiveness of anti-HIV immunotherapies; a difference of at least 0.5 log copies/ml between the experimental and control arms of a study is probably clinically significant, as determined by the results obtained in studies of antiretrovirals (reviewed in Ref. [2]). One disadvantage of using the new viral set point as a primary end point, however, is that it may miss important virologic and immunologic events that occur early in the immune response to the vaccine [165]. Another disadvantage is that the establishment of a new viral set point may be delayed, thus extending the length of treatment interruption and its associated risks.
Another primary end point that is commonly used in HIV therapeutic vaccine clinical trials is the time to detectable virus or the time to viral rebound (i.e. the time to achieve a viral load !50 copies/ml) following an ATI. However, an accurate assessment of this requires frequent viral load monitoring [165]. While this outcome would seem to be clinically relevant, its value is unknown since this measure does not appear to correlate with other virologic outcome measures. In a therapeutic vaccine trial of ALVAC-HIV vCP1452 by Jacobson et al., the time until viral rebound did not correlate with any of the other virologic measures assessed, such as the new viral load set point [150]. As a result, it has been suggested that the time to viral rebound is probably not an appropriate outcome measure for assessing the effectiveness of HIV therapeutic vaccines [2]. Similarly, our own study of ALVAC-HIV vCP1452 with or without Remune † failed to find a correlation between the time to viral rebound and the new viral set point, nor with the magnitude of the viral rebound [120]. A correlation was observed, however, between the time to viral rebound and the time to restart therapy, as well as the time to meet the criteria to do so. In the one other trial in which vaccination was found to delay the time to viral rebound (this trial involved ALVAC-HIV vCP1433 and Lipo-6T), no assessments were made for correlations with other virologic outcome measures [109].
The CD4 count, which is routinely used to determine the risk of opportunistic infection [167], is typically included in trials of HIV therapeutic vaccines. In addition to monitoring the change in the absolute CD4 count, changes in the percentage of CD4 T cells, the CD4:CD8 ratio, the time to decline to a predefined level or the change in the slope of the CD4 count have also been assessed in clinical trials of HIV therapeutic vaccines. However, this is not an ideal primary outcome as it requires waiting for a decline in the CD4 count.
In addition to virologic and immunologic outcomes, some studies of therapeutic HIV vaccines include the assessment of clinical outcomes. These outcomes include the development of clinical events, including symptoms of acute retroviral syndrome after ATI [128,129] or the time until either ART is resumed or the criteria for ART resumption is met. In addition, the proportion of participants who do or do not resume ART may also be determined.
Therapeutic HIV vaccines and their potential role in an HIV cure strategy One of the research priorities recently identified by the International AIDS Society (IAS) Global Scientific Strategy ''Towards an HIV Cure'' working group is the development of a therapeutic HIV vaccine capable of boosting the immune system of the infected host to control HIV replication in the absence of ART, thus producing a functional cure [7] similar to that experienced naturally by long-term non-progressors and elite controllers (reviewed in Refs. [7,99,166]). According to this IAS working group, a therapeutic vaccine should be directed to conserved HIV epitopes and either 1) elicit a humoral response consisting of neutralizing anti-HIV antibodies that would a) prevent cell-to-cell transmission or b) recognize virus-producing cells for destruction by antibodydependent cellular cytotoxicity; or 2) induce a strong cytotoxic cellular response for the destruction of cells producing virus before virus progeny is released [7]. These strategies should lead to a sustained reduction in the number of cells actively transcribing virus and induce an immune selective pressure that would lead to loss of viral fitness and replicative potential.
The persistence of the viral reservoir is considered to be the major obstacle to curing HIV infection [3,6,7,10]. In fact, when ART is interrupted, the viral rebound that occurs within days or weeks is the result of reseeding from viral reservoirs [11]. Furthermore, high levels of HIV DNA, a surrogate marker of the size of the viral reservoir, are correlated with quicker viral rebound following ART interruption [126]. The few trials of therapeutic vaccines that have assessed the change in the size of the viral reservoir did not observe any significant effect [106,112,149,154,159,168Á172]. Five of these studies assessed whether vaccination induced any changes in the size of the viral reservoir by measuring proviral DNA, either by co-culture assay [154,168,171] or by PCR [149,169]. One of the advantages of assessing changes in the size of the viral reservoir as a read-out of therapeutic vaccine efficacy is that this outcome measure can be made in trials that do not include an ATI [168], thus minimizing the risks that may be associated with treatment interruptions. Disadvantages of using this outcome measure, however, include the fact that no single assay accurately measures the size of the viral reservoir [173,174] and the lack of strong correlations between assays [174].

Conclusions and future directions
The development of a therapeutic HIV vaccine would be a valuable alternative to the use of expensive, toxic, lifelong ART regimens. The results of dozens of clinical trials performed over more than two decades to assess the safety, immunogenicity and, in many cases, the efficacy of various HIV therapeutic vaccines have been published, and more trials are underway. Besides the obvious challenges of developing a successful therapeutic vaccine is the issue of how to best assess the efficacy of vaccine candidates [12]. Currently, the inconsistent assessment of different outcome measures in different trials makes it difficult to compare the relative efficacies of the various vaccine candidates.
In its ''Towards an HIV Cure'' recommendations, the IAS recommends that future clinical trials of therapeutic HIV include studying compartments in addition to peripheral blood, such as the gastrointestinal tract and lymph nodes, both sites of HIV infection and immune responses [7]. Other suggestions for future trials include measuring immune control of viral replication by using highly sensitive SCA in situations in which the use of ATI is not ideal.
Given that immune activation predicts HIV disease progression independently of the CD4 count and viral load, it has also been recommended that, when assessing the efficacy of an HIV immunotherapy such as therapeutic vaccination, concurrent changes in immune activation markers, vaccinespecific responses and viral replication should be assessed during treatment interruption [175].
It is apparent that despite significant efforts made, the therapeutic vaccine candidates studied to date have been associated with limited clinical benefit [3,6,99]. Continued efforts will be required, therefore, to develop and test a safe and effective therapeutic HIV vaccine that will help end the global HIV epidemic. Future work may be influenced by promising prophylactic simian immunodeficiency virus (SIV) vaccines. In one study, virus levels became undetectable following initial viremia in half of the macaques vaccinated prior to SIV challenge [176], while in another SIV/macaque study, one-third of the monkeys that became infected following SIV challenge ultimately became elite controllers [177]. Thus, despite being designed as preventative SIV vaccines, both appeared to induce therapeutic benefits. These simian vaccines may, therefore, provide some valuable insight into the design of effective therapeutic HIV vaccines.
Finally, while the characteristics of a successful therapeutic HIV vaccine are currently unknown, standardizing which outcome measures should be used in future clinical trials to evaluate vaccine efficacy would certainly be beneficial. Competing interests JBA has done contract research for Argos Therapeutics, Inc., Sanofi Pasteur Ltd. and Immune Response Corp. GMG has no competing interests to declare.
Authors' contributions GMG and JBA contributed equally to the preparation of the manuscript.

Author information
JBA is the Co-leader of the Vaccines and Immunotherapies (VIT) Core of the CTN (CIHR Canadian HIV Trials Network). GMG is the VIT Core Research Associate.