Send Orders for Reprints to
[email protected] Current HIV Research, 2013, 11, 450-463
450
HIV Vaccine Efficacy and Immune Correlates of Risk Robert J. O'Connell1 and Jean-Louis Excler*,2 1
Department of Retrovirology, US Army Medical Component, Armed Forces Institute of Medical Sciences (AFRIMS), 315/6 Rajvithi Road, Bangkok 10400, Thailand; 2The Henry M Jackson Foundation for the Advancement of Military Medicine, Vaccine Clinical Development, US Military HIV Research Program (MHRP), 6720-A Rockledge Drive, Suite 400, Bethesda, MD 20817, USA Abstract: Although immune correlates of protection for HIV vaccines have remained an intractable question, RV144 provided the first evidence that an HIV vaccine could provide protective efficacy against HIV acquisition. The study of correlates of risk has opened large and unforeseen avenues of exploration and hope for the most exciting time of HIV vaccine development. Several elements in the RV144 post-hoc analysis and recent macaque challenge studies suggest that antibodies directed against the V2 loop of gp120 are functional and may have played a protective role against virus acquisition. Several protective mechanisms against sexual transmission of HIV are evoked including blocking the gp120α4β7 interaction and ADCC although possibly mitigated by high levels of Env-specific IgA, both mechanisms contributing at least partially to the protective effect. Several questions remain unanswered that will deserve intensive assessments, in particular, IgG and IgA Env antibodies in mucosal secretions, Env-specific IgG subclasses, cross-reaction of V2 antibodies, role of T-follicular helper cells, and B-cell memory. Whether RV144 correlates of risk are universal and apply at least partially to other populations at higher risk for HIV acquisition and other modes of transmission (rectal, injecting drug users) is unknown and remains to be explored. Future efficacy trials using the same vaccine concept tested in high-risk heterosexual populations and in men having sex with men may answer this question. In addition, the determination of early events in the pathogenesis among HIV-infected vaccine recipients based on current correlates knowledge would offer unprecedented information about correlates biomarkers in the peripheral blood and gut mucosa during early acute HIV infection.
Keywords: Efficacy trial, immune correlates, HIV, prime-boost, V2 antibodies, vaccine. INTRODUCTION At the end of 2011, an estimated 34 million people were living with HIV worldwide, the number of new infections being 20% lower than in 2001 [1]. There is an urgent need to strengthen and scale-up existing and new prevention methods such as HIV testing and counseling, behavioral interventions, condom use, treatment of sexually transmitted diseases, harm reduction, male circumcision [2], and antiretroviral drugs for prevention [3]. New prevention strategies to control the epidemic and prevent new infections, including pre-exposure prophylaxis [4], New prevention strategies to control the epidemic and prevent new infections, including pre-exposure prophylaxis [5], topical microbicides [6], prevention of mother-to-child transmission [7], and HIV preventive vaccines must be explored and their access ensured. The development of a preventive vaccine against HIV-1 remains among the best hopes for controlling the HIV/AIDS pandemic [8]. Experimental preventive HIV-1 vaccines have been administered to over 44,000 human volunteers in over 187 separate trials since 1987, tested mostly in Phase I and II clinical trials. Different HIV vaccine approaches along with their scientific and programmatic challenges and lessons learned have been amply reviewed [9-13]. Identifying *Address correspondence to this author at the US Military HIV Research Program (MHRP), 6720-A Rockledge Drive, Suite 400, Bethesda, MD 20817, USA; Tel: 63 947 893 7459; E-mail:
[email protected] 1873-4251/13 $58.00+.00
correlates of protection remains one of the major hurdles to guide and accelerate HIV vaccine development. While animal models [14, 15] and studies of the immune responses to acute HIV infection [16] are providing important leads, testing vaccine efficacy in humans remains key to understand what immune responses are relevant to protection against HIV acquisition and/or progression to disease. To date only six HIV vaccine regimens have reached Phase IIB or Phase III efficacy testing (Table 1). This review intends to revisit the notion of correlates of protection, put the efficacy and immunogenicity results of HIV vaccine trials in perspective along with recent findings on immune correlates of risk in humans and in non-human primate models, and address unanswered questions and path forward. DEFINING IMMUNE CORRELATES FOR VACCINE PROTECTION As evidenced by several recent attempts to clarify nomenclature about immune correlates for vaccine protection [17-19], a clear definition has been elusive. For immunogen design, demonstration of precise mechanistic relationship to protection is required, but for a regulatory agency seeking to evaluate a vaccine against a pathogen that causes widely dispersed epidemics, clear association without knowledge of mechanistic linkage may be sufficient. The US Food and Drug Administration defines a correlate of protection in vaccine efficacy trials as “generally a laboratory parameter that has been shown from adequate and well-controlled trials to be associated with protection from © 2013 Bentham Science Publishers
HIV Vaccine Efficacy and Immune Correlates of Risk
Table 1.
Current HIV Research, 2013, Vol. 11, No. 6
451
Completed and Ongoing Phase IIb and III Human HIV-1 Vaccine Trials.
Study Protocol
Candidate Vaccine
Phase
Volunteers
Location
Results
RV144
ALVAC-HIV vCP1521 and AIDSVAX B/E (MN and CRF01_AE CM244) rgp120 in alum
III
16,403
Thailand
31.2% efficacy against HIV acquisition. No effect on plasma viral load
HVTN 505
DNA (VRC-HIVDNA016-00-VP) and rAd5 (VRCHIVADV014-00-VP) (A, B, and C)
II b
2,494 in MITT analysis
US
Stopped for futility; no efficacy on HIV acquisition and plasma viral load
HVTN 502/Merck 023
MRKAd5 HIV-1 gag/pol/nef B
II b
3,000
US
No efficacy; transient infection risk
HVTN 503 Phambili trial
MRKAd5 HIV-1 gag/pol/nef B
II b
3,000
RSA
No efficacy; increased HIV infection rate in vaccinees
Vax003
AIDSVAX B/E gp120 (MN and CRF01_AE CM244) gp120 in alum
III
2,500
Thailand
No efficacy
Vax004
AIDSVAX B/B gp120 (MN and GNE8) gp120 in alum
III
5,400
US
No efficacy
Step trial
MITT: modified intent-to-treat analysis. Pox: Recombinant Poxvirus-vectored vaccine. ALVAC-HIV (vCP1521): recombinant canarypox vector expressing Gag and Protease subtype B (LAI) and env gp120 CRF01_AE (TH023) linked to the transmembrane-anchoring portion of subtype B gp41 (LAI) genes. Ad5: Replication-defective recombinant Adenovirus 5-vectored vaccine. VRC-HIVDNA-016-00-VP: DNA plasmids expressing Gag, Pol, and Nef subtype B (strains HXB2, NL4-3, NY5/BRU, respectively) and HIV-1 Env subtype A (strain 92rw020), B (strains HXB2/BaL), and C (strain 97ZA012). VRC-HIVADV014-00-VP: mixture of four rAd5 vectors encoding the HIV-1 Gag-Pol polyprotein subtype B (strains HXB2-NL4-3) and HIV-1 Env A, B and C matching the DNA Env components. US: United States. RSA: Republic of South Africa.
clinical disease,” and states that an immunological correlate of protection “is most useful if clear qualitative and quantitative relationships can be determined, e.g., a certain type and level of antibody correlate with protection [20].” Plotkin and Gilbert have recently brought authoritative clarity by defining a correlate of protection (CoP) as reflecting a statistical relation between an immune marker and protection with two mutually exclusive sub-categories: 1) mechanistic correlates of protection where the CoP is mechanistically responsible for protection, and 2) nonmechanistic correlates of protection, where the CoP is a predictor of protection but has not been proven to mechanistically underlie protection [21]. CORRELATES OF PROTECTION FOR LICENSED VACCINES Fundamentally, preventive vaccines aim to prevent morbidity and mortality associated with infectious diseases. Major licensed vaccines prevent disease accomplish this goal through a variety of mechanisms including inhibition of pathogen bloodstream circulation, mucosal replication, mucosal or skin invasion, toxin production, reactivation in neurons, or replication in macrophages [18]. Immune mechanisms of protection may not be known, and indeed, vaccination strategies to prevent smallpox, rabies, and anthrax all preceded and led to rudimentary immune discoveries such as antibodies [22]. Modern assessments have revealed that the majority of successfully licensed vaccines protect through elicitation of protective antibodies [22, 23]. Most of these prevent disease from acute cytopathic childhood infections [24]. Similar to
passive protection afforded to an infant from maternal antibodies, these vaccines induce immunoglobulins that either provide direct pathogen/toxin neutralization, (classic examples include smallpox and yellow fever) or provide a link to other effector mechanisms such as antibody dependent cell-mediated cytotoxicity (ADCC), or opsonophagocytosis. The strongest evidence demonstrating antibody correlates has come from passive infusion studies, where such data exist for smallpox, diphtheria, tetanus, pertussis, Haemophilus influenzae type b, pneumococcus, hepatitis A, hepatitis B, varicella, measles, rubella, poliomyelitis, and rabies [18]. Antibodies protect against viruses despite the obligate intracellular replication because an extracellular phase exists for such pathogens, providing transient susceptibility to antibody effects [23]. Pathogens that cause persistent infections tend to elicit partially effective but pathogenic chronic immune responses. Controlling such infections without collateral pathogenesis has generally not been possible using existing vaccine approaches. Two possible exceptions are the bacilli Calmette-Guerin (BCG) vaccine against tuberculosis, and zoster vaccine, for which no true correlates are known but for which cellular responses appear particularly important [25, 26]. In spite of the importance of cellular responses, no available vaccine protects exclusively via T-cell immunity [24]. Two obvious elements required to study potential vaccine-induced correlates of protection are documented protection against infection and/or progression to disease and measurement of immune parameters to be analyzed. These two conditions were met with the RV144 HIV Vaccine trial that triggered considerable interest and research including the revisit of older HIV vaccine efficacy trials.
452
Current HIV Research, 2013, Vol. 11, No. 6
RV144 VACCINE EFFICACY AND IMMUNOGENICITY The Thai “Phase III” trial, RV144, provided the first evidence that an HIV vaccine could provide protective efficacy against HIV acquisition [27]. The prime-boost vaccine regimen consisted of a non-replicating recombinant canarypox vector, ALVAC-HIV prime (vCP1521, expressing gag, protease subtype B (LAI) and env gp120 CRF01_AE with a gp41 subtype B (LAI) transmembrane anchor) administered at 0, 1, 3, and 6 months and a bivalent AIDSVAX® gp120 B/E MN and CRF01_AE boost given at months 3 and 6. AIDSVAX® gp120 B/E comported a deletion of 11 amino acids (aa) at the N terminal of the C1 domain of gp120 replaced by an HSV-2 leader and a 27-aa HSV-2 gD protein tag. This was a community-based trial initiated in 2003 in Rayong and Chon Buri provinces of Thailand. Roughly 90% of incident infections in RV144 were CRF01_AE infections, also known as subtype E, which is the predominant circulating strain in Thailand and much of South East Asia. The modified intent-to-treat analysis (excluding those randomized, but HIV-infected at the first vaccination visit) was conducted on 16,395 volunteers. No data were collected on the status of male circumcision or on serologic analyses for adenovirus type 5 or herpes simplex virus type 2. The analysis showed 31.2% efficacy after 42 months of follow-up. There was no effect on early postinfection HIV-1 RNA viral load or CD4+ T-cell count. In addition, in a post-hoc analysis with Bayesian statistics, which was not pre-specified in the study’s statistical analysis plan, the likelihood that the result was correct varied depending on the Bayesian “priors”, and a scenario factoring in the negative VaxGen trials and data from pooled analysis of 5 other prime-boost trials gave a 71% probability that RV144 showed efficacy [28]. Although not included in the pre-specified analysis plan, vaccine efficacy appeared to be higher (60%) at 12 months post vaccination, suggesting an early, but nondurable, vaccine effect. The authors pointed out that future HIV vaccine trials should recognize potential interactions between challenge intensity and risk heterogeneity in both population and treatment effects [29]. One can speculate that the RV144 regimen was able to protect because the number of sexual contacts were limited in time and could be countered by a marginally efficacious vaccine, which might or might not hold true with communities at higher risk of sexual transmission and increased number of exposures such as MSM and female sex workers. Interestingly, a simple, combined analysis of previous phase I and II ALVAC-HIV and gp120 prime-boost studies showed a rate of HIV-1 infection of 0.59 per 100 person-years in the vaccine group and 1.2 per 100 personyears in the placebo group, for a vaccine efficacy (VE) of 50%, although not statistically significant; the results also showed no effect on viral load [30]. Vaccination did not affect the clinical course of HIV disease after infection [31]. A further analysis of the effect of vaccination on disease progression after infection showed a weak evidence of lower plasma viral load and higher CD4+ count in the vaccine group. Interestingly, a lower mucosal VL was observed in semen of vaccine recipients. This corroborates the observation in rhesus macaques where a DNA/Ad5 vaccine also yielded lower seminal fluid viral load after SIV challenge [32].
O'Connell and Excler
The vaccine regimen was safe and generally well tolerated [33]. Immune responses were assessed in a subset of volunteers. Vaccination induced an HIV-specific response, as measured by IFN-γ ELISPOT assay to either Env or Gag antigen, in 19.7% of volunteers 6 months after the final dose of vaccine was administered. Response rates for CD4+ Env-specific ICS were higher in the vaccine group than in the placebo group (32% vs 2%). Rates of positivity in the gp120 and p24 binding-antibody assays and the lymphoproliferation assay were similar to those in the phase II study. Binding antibody against Env was nearly uniformly present to MN and A244 strains, whereas p24 responses were less frequent. The stimulation index was significantly higher in vaccine recipients compared to baseline and placebo recipients. IFN-γ ELISPOT positive responses were measured in 25 (41%) vaccinees and were predominantly CD4+ T cellmediated. Responses were targeted within the HIV Env region, with 60% of vaccinees recognizing peptides derived from the V2 region of HIV-1 Env, which includes the α4β7 integrin binding site. Intracellular cytokine staining confirmed that Env responses predominated (63% of vaccine recipients) and were mediated by polyfunctional effector memory CD4+ T cells. Proliferation assays revealed that HIV antigen-specific T cells were CD4+, with the majority (80%) expressing CD107a. HIV-specific T-cell lines obtained from vaccine recipients confirmed V2 specificity, polyfunctionality, and functional cytolytic capacity [34]. Neutralization was assessed with tier 1 and tier 2 strains of virus in TZM-bl and A3R5 cell assays. Neutralization of several tier 1 viruses was detected in both RV144 and Vax003. ALVAC-HIV (vCP1521) priming followed by two boosts with gp120 protein was superior to two gp120 protein administrations alone, confirming a priming effect for ALVAC-HIV. Sporadic weak neutralization of tier 2 viruses was detected only in Vax003 using the A3R5 cell assay [35]. Non-response to vaccine was associated with DRB1*11 and DRB1*16:02 alleles. Vaccine recipients with HLA-DQ heterodimers encoded by DQA1*05:01 and DQB1*03:01 alleles, were less likely to produce neutralizing antibodies (NAb). These data suggest that the lack of response to a vaccine designed to induce clade-specific HIV NAb is associated with the presence of certain HLA class II alleles and heterodimers in some Southeast Asians [36]. RV144 CASE CONTROL CORRELATES OF RISK The efficacy observed in the RV144 trial provided the first opportunity to study immune correlations with vaccine efficacy against HIV. A diverse consortium of investigators systematically evaluated assays that detect antibody, innate, and cellular immune responses [37]. Initial pilot studies evaluated 32 assays of 17 types on the basis of reproducibility, non-redundancy, low false positive rate, and large dynamic range using 50-100 samples from uninfected RV144 volunteers collected at baseline and peak immunogenicity (80% of whom were vaccine and 20% placebo recipients). In order to optimize statistical power to show a correlation of risk between vaccinated persons who acquired versus those that did not acquire HIV-1 infection, 6 primary variables were identified as the focus of a casecontrol analysis using peak immunogenicity samples from
HIV Vaccine Efficacy and Immune Correlates of Risk
Table 2.
Current HIV Research, 2013, Vol. 11, No. 6
453
Assays Employed in the RV144 Case-Control Correlates Analysis. Variable
Assay Description
Institutions
IgA antibodies binding to ENV
ENV IgA binding score [14 ENVs]*
Duke University
Avidity of IgG antibodies for ENV
IgG Avidity to AE Δ11 A244gD-
Duke University
ADCC
ADCC AE.92TH023-infected CD4 T cells
Harvard University
Neutralizing antibodies
Neutralizing antibody panel score of both TZMbl + A3R5 neutralizing assays [n=6]
Duke University; AFRIMS, Siriraj Hospital, Bangkok
V1V2 binding antibodies
IgG binding antibody to scaffolded V1V2 protein**
New York University
CD4+ T cell responses
ENV-specific CD4 T cells (IFN-γ, IL-2, TNF-α, CD154)
FHCRC
*Envelopes included A244 gD- gp120, AE.92TH023 gD- gp120, A.00MSA gp140CF AE.97CNGX 140CF, A1.CON03 140CF, B.CON.03140CF, C.CON.03 140CF, G.CON.03 gp140CF, AE.CON.03gp140CF, G.DRCBL gp140, B.JRFLgp140, US1 gp140 SIV CPZ, B. **Target used was a fusion protein consisting of the V1V2 domain from a subtype B strain, A2, attached to gp70 from murine leukemia virus [138], a construct known to bind to monoclonal antibodies that only recognize conformational epitopes. AFRIMS- Armed Forces Research Institute for Medical Research, Bangkok, Thailand FHCRC- Fred Hutchinson Cancer Research Center, Seattle, Washington.
vaccine recipients who acquired HIV-1 infection and controls who did not acquire infection. The assays are shown in Table 2. Two of these variables correlated significantly with HIV1 infection risk: plasma IgG binding antibody to scaffolded gp70 V1V2 envelope proteins correlated inversely with risk, while ENV plasma IgA (monomeric) binding score correlated directly with risk, raising the hypothesis that IgA responses against ENV and IgG responses directed against V1V2 may be mechanistically associated with RV144 vaccine regimen-mediated protection. Neither low levels of V1V2 antibodies nor high levels of Env-specific IgA antibodies were associated with higher rates of infection than were found in the placebo group, suggesting there was no evidence for enhancement of infection risk in the overall study or associated with any of the case-control assay variables. In vaccinees with low levels of Env-specific IgA antibodies, four of the other five primary variables, IgG avidity, ADCC, neutralizing antibodies, and Env-specific CD4+ T cells, were inversely correlated with infection, while vaccinees with IgA antibodies to the first conserved region C1 of gp120 had a higher risk of infection than vaccinees without these antibodies. The reasons for negative association between high levels of plasma IgA and protection are unclear. It has been hypothesized that IgA may block the action of IgG [38], in particular ADCC [39] and phagocytosis [40]. ADCC responses were predominantly directed to the C1 conformational region of gp120 [41-43] although other epitope specificities (i.e., V2) also contributed to the overall response [44]. Another hypothesis is that C1 region Env-specific IgA could block C1-specific IgG effector function due to their ability to bind to different Fc receptors on effector cells. It was recently demonstrated that IgA antibodies elicited by RV144 could block C1 regionspecific IgG-mediated ADCC (via natural killer cells) [45]. This suggests that the study of Fc receptor-mediated antibody function may be important in the evaluation of HIV-1 vaccines. The protective role of dimeric mucosal IgA could not be assessed in RV144 and should deserve further studies in future trials. In an earlier Phase I trial, the RV144 regimen was shown to induce potent ADCC activity [46]. The combination of low plasma anti-HIV-1 Env IgA antibodies and high levels
of ADCC inversely correlate with infection risk. One hypothesis is that the observed protection in RV144 may be partially due to ADCC-mediating antibodies. The majority of a representative group of vaccinees displayed plasma ADCC activity, usually blocked by competition with the C1 regionspecific A32 Fab fragment. Using memory B-cell cultures and antigen-specific B-cell sorting, 23 ADCC-mediating non-clonally related antibodies were isolated from 6 vaccine recipients. These antibodies targeted A32-blockable conformational epitopes, a non A32-blockable conformational epitope, and the gp120 Env variable loops. Fourteen antibodies mediated cross-clade target cell killing. ADCC-mediating antibodies displayed modest levels of Vheavy (VH) chain somatic mutation (0.5-1.5%) and also displayed a disproportionate usage of VH1 family genes (74%), a phenomenon recently described for CD4-binding site broadly neutralizing antibodies [bNAbs]. Maximal ADCC activity of VH1 antibodies correlated with mutation frequency. The polyclonality and low mutation frequency of these VH1 antibodies reveal fundamental differences in the regulation and maturation of these ADCC-mediating responses compared to VH1 bNAbs [42]. The HIV-1 Env gp120 monoclonal antibody (MAb) A32 binds to the surface of transmitted/founder HIV-1-infected CD4+ T cells earlier in the course of in vitro infection than the gp120 Env MAbs 17b and 2G12. MAb A32 was able to mediate ADCC activity that was 4- to 6-fold higher than that of the other two anti-gp120 MAbs when either gp120-coated or HIV-1infected target CD4+ cells were used. Antibodies that are blocked by A32 Fab comprise a majority of CD4-inducible ADCC-mediating antibody responses elicited during the course of HIV-1 infection [41]. A sieve analysis was performed on 1,025 genome equivalents from 121 RV144 participants who became HIVinfected during the trial. Targeted V1/V2-focused analysis identified two signatures of vaccine pressure within the V2 loop corresponding to sites 169 and 181. Intriguingly, VE against viruses matching the vaccine at position 169 was 48% whereas VE against viruses mismatching the vaccine at position 181 was 78%. The explanation of a greater VE associated against mismatched HIV-1 with the sieve effect at site 181 remains unclear. It is speculated that vaccineinduced responses may have hindered HIV-1 infection with
454
Current HIV Research, 2013, Vol. 11, No. 6
181 variants, other explanations including involvement of other unidentified sequences near position 181 or inability of this variant to establish infection due to steric hindrance with vaccine-induced antibodies. Although the full interpretation of these results remains unclear, the sieve analysis provides additional evidence supporting the hypothesis that vaccination-induced immune responses directed against the V2 loop were associated with protection [47]. Supporting this hypothesis, monoclonal antibodies generated from RV144 vaccine recipients targeted a critical residue in V2 (K169), thus providing evidence that vaccine-induced antibodies could potentially mediate a virus sieve effect. These V2-specific antibodies can mediate ADCC, neutralization and low-level virus capture [44, 48]. The assessment of a T-cell based sieve effect in envelope V1/V2 revealed an association between an HLA class I allele ad VE, suggesting that VE was restricted to A*02(+) participants and that iGA-C1 antibodies inhibited protective effects of other responses in A*02(-) participants, highlighting the importance of host genetics assessment on VE in future HIV vaccine trials [49]. Two studies have characterized binding antibody responses against the V2 region. First, using peptide microarray, surface plasmon resonance, and ELISA, 97% of 32 studied plasma samples from RV144 vaccine recipients 2 weeks post last vaccination contained antibodies that recognize V2 region synthetic peptides. Percent responders fell to 19% at 28 weeks post last vaccination, and V2 responses were significantly more frequent in vaccine recipients as compared to individuals naturally infected with HIV-1 [50]. In addition, Zolla-Pazner et al. described results from the 13 primary and exploratory assays associated with the correlates analysis [51]. All V2-related odds ratios were
