Secondary Logo

Journal Logo


Post-Step modifications for research on HIV vaccines

Corey, Lawrence; McElrath, M Juliana; Kublin, James G

Author Information
doi: 10.1097/QAD.0b013e32830e6d6d
  • Free

In November 2007, a large multinational trial called Step, evaluating the lead candidate in what the HIV vaccine field termed T cell-based vaccines, was halted at its first interim analyses because of the vaccine's lack of efficacy [1,2]. The vaccine candidate used in Step included a replication incompetent adenovirus type 5 (Ad5) viral vector, and the objectives of the trial were to investigate whether this candidate vaccine was able to reduce HIV acquisition or to modulate viral load [3]. This vaccine prototype first entered human clinical trials in 2000. The initial clinical studies resulted in several modifications to the vector's promoters as well as an attempt to increase the breadth of the host response by the addition of the pol and nef genes to the original gag construct. The final vaccine candidate contained clade B sequences of gag/pol and nef and underwent an extensive series of clinical trials between 2003 and 2006. These studies demonstrated detectable T cell immunogenicity as measured by the ELISpot assay in about 80% of recipients with a median magnitude in the range 275–300 spot forming units (SFU)/million peripheral blood mononuclear cells (PBMC), a frequency and magnitude greater than any other T cell-based approach at the time of the initiation of the Step trial in 2005 [4]. More detailed analyses of the T cell responses indicated that more than 60% of the persons with detectable ELISpot responses after vaccination exhibited CD8+ T cells to HIV antigens and approximately 40% exhibited CD4+ responses. Durability of the responses was prolonged (>6 months after last vaccination) [5,6]. Thus, the lack of success of the vaccine, especially on controlling postinfection viral load (setpoint viremia) among vaccinated individuals who became infected, has sent reverberations throughout the scientific community for its implications on our ability to develop a globally effective HIV vaccine.

The Step trial involved 3000 healthy, uninfected volunteers who were randomized to receive a placebo or a vaccination. Vaccine recipients in the Step trial were stratified by their prior exposure to adenovirus into those who were considered seronegative to type 5 adenovirus (<18) and those with low (18–200), moderate (200–1000) and high (>1000) antibody titers to Ad5 at enrollment, a stratification based on the above phase 1 clinical trials that suggested prior immunity to the vector reduced the vaccine's immunogenicity [7]. Vaccination was given at 0, 1, and 6 months and all enrollees were followed for HIV infection over time. Those who became HIV infected had their infection confirmed and several measurements of their viral load taken over the following 3–6 months. At the time of the analysis of the trial in November 2007 [8], 84 cases of HIV infection had occurred over the course of the study; 82 of the 84 cases occurred in men, about all of whom acquired HIV through sexual activity with other men. One of the disconcerting and unanticipated findings from the trial was that among men who entered the trial with prior Ad5 infection (Ad5 titers >18), the incidence of HIV acquisition was two-fold higher among vaccinated versus placebo recipients (Table 1). In contrast, the acquisition rate of HIV was identical among men who entered the trial seronegative for Ad5 (relative incidence vaccine to placebo of 1.0). A more recent multivariate analysis has shown that uncircumcised men, especially those who were Ad5 seropositive, had as much as a four-fold higher rate of acquisition of HIV infection [95% confidence interval (CI) 1.3–11], suggesting a statistically increased rate of acquisition in this subgroup, particularly when compared with the lack of increased risk among circumcised men with no preexisting Ad5 immunity [9].

Table 1
Table 1:
Incidence (95% confidence interval) per 100 person-years of HIV infection modified intent to treat population (men).

Recently, the first data from the companion Phambili trial have become available [10]. This trial was conducted in South Africa and used the MRK Ad5 gag/pol/nef vaccine to investigate whether it would demonstrate efficacy in a clade C region. The trial enrolled 801 participants before it was stopped. Of the 11 acquisitions reported to date, nine were among those seropositive to Ad5. HIV acquisition occurred in six who received the vaccine and three who received placebo; the two HIV infections in the Ad5 seronegatives at entry were split one and one in each group; 10 of the 11 HIV infections were in women. These very preliminary results from South Africa continue to provide evidence that preexisting immunity to the Ad5 vector was an independent risk factor in the increased role of HIV acquisition among vaccine recipients. The placebo in these clinical trials was a saline solution, thus limiting the ability to evaluate the impact of the empty Ad5 vector on HIV-1 acquisition. This increased risk for HIV-1 acquisition among persons entering the trial with prior Ad5 immunity was unanticipated and not encountered in any nonhuman primate (NHP) challenge experiments using adenovirus vectors [11–14].

Why is it that such a highly immunogenic vaccine did not control postinfectious viral load? Perhaps more importantly, how did prior Ad5 infection increase the frequency of HIV acquisition, an observation that has implications for other vaccines using adenovirus vectors? Were the immune responses elicited by the vaccine lower in individuals who became infected as compared with those who did not? Was the quantity, quality, or breadth of the immune responses suboptimal? Although definitive answers to these issues are not yet available, several procedures built into the Step trial have started to provide some data of relevance.

ELISpot responses were measured in a random sample of 25% of all recipients at week 8, and four weeks after the second dose of vaccine – the time period at which phase 1 studies had shown near-peak ELISpot responses. These analyses indicate that the frequency of responses as measured by ELISpot in PBMC was similar between vaccinated persons who subsequently developed HIV infection and those who did not. Among Ad5 seronegatives, these responses were 74 versus 76% for gag, 63 versus 73% for pol, and 74 versus 70% for nef for infected versus noninfected vaccine recipients, respectively [9]. The ELISpot geometric mean titers (GMT) were also similar in the two groups (∼350 SFU/million PBMC). Interestingly, ELISpot responses in the Step trial in persons who had preexisting immunity to Ad5 were somewhat lower than those in the earlier phase 1 trials (GMT ELISpot responses ∼170/SFC/million PBMC) [4,8,11]. However, again, no discernible differences were seen between those seropositive vaccinees who developed HIV and those who did not; for example, the ELISpot responses to Gag were 46 and 54% among those who subsequently acquired HIV versus those who remained uninfected, respectively. Notably, over 50% of the HIV-specific CD8+ T cells elicited after vaccination produced interferon (IFN)γ alone and did not exhibit interleukin (IL)-2 or tumour necrosis factor responses, suggesting they exhibited a less polyfunctional phenotype [9,15]. Although these data are still preliminary, they clearly indicate that the MRK gag/pol/nef vaccine was immunogenic; that vaccinated persons who acquired HIV-1 did respond to the vaccine; and that no measurable differences in the overall quality, quantity, or magnitude of response have yet been defined between vaccine recipients who subsequently developed HIV and those who did not. Thus, there appears to be no obvious correlate between immune responses to the vaccination and the acquisition of HIV infection. Perhaps this is not surprising, in that most authorities in the field (including these coauthors) felt there was little likelihood that a vaccine containing just gag/pol and nef genes would reduce HIV acquisition. Until an immunogen that elicits high levels of neutralizing antibodies is developed, most experts feel that the most likely effect of T cell-based vaccines will be on controlling postinfectious viral load and lower mucosal tissue titers of HIV-1.

NHP data have indicated that T cell-based vaccines are more likely to control the postinfection course of HIV infection [13–16]. The analyses evaluating the association between the magnitude or quality of the T cell response and postinfection viral load are just being initiated in Step trial participants. Although it is apparent that on a group basis, there was little difference in setpoint viremia and postacquisition viral load, data analyses are proceeding to determine if a subgroup of persons, for example, those with very high responses to the vaccines had overall lower loads of setpoint viremia. As control of setpoint viremia was hypothesized to be the most likely endpoint for the study, we must await these analyses before determining if the Step trial provides some leads on whether T cell responses after vaccination influenced viral load or CD4+ cell count depletion post-HIV acquisition.

Epitope-specific responses

One of the central hypotheses of the Step trial was that T cell responses to the conserved regions of the HIV gag/pol and nef genes would be elicited and that the immune responses would ‘match’ the strains circulating in the population and result in reduced HIV acquisition or disease progression [17,18]. At present, data that define whether the epitopes elicited by vaccination are commonly found in the infecting strains are not available. In other words, is the lack of vaccine efficacy related to the lack of epitope coverage from the vaccine? Available data from phase I studies indicated an average of one epitope per gene product in most vaccine recipients [16]. This in itself is of interest in that biometric analysis of the sequence data from the gag, pol, and nef genes in the MRK vaccine indicates that there are potentially over 150 epitopes contained in the gag vaccine insert and over 50 in pol and nef[19]. Yet, immunodominance is displayed in nearly all people [20]. The initial data on epitope mapping from the phase I/II trials suggest that many of the gag epitopes elicited from vaccination are present in circulating strains in clade B populations. However, the epitope coverage – that is, the epitopes elicited by vaccination that are frequently found in potentially infectious strains – may be quite low [19,21]. The narrowness of the immune response may be an important factor in the vaccine's overall lack of efficacy. Studies on full-length sequences of infecting isolates from the 82 cases are underway by Drs Francine McCutchen and James Mullins, and the sequences of the infecting isolates will be analyzed with those epitopes that were elicited prior to and after infection. Additional evaluations will focus on the viral load levels among persons who made significant numbers of CD8+ T cells that ‘matched’ the infecting strain of HIV-1 versus those who did not. Designing and developing T cell vaccines that elicit T cell responses to multiple conserved regions of HIV should be a major area of emphasis for subsequent T cell-based vaccines.

In the past 12 months, several biometric proposals to design inserts that are more likely to elicit a diverse set of epitopes to circulating strains have been described [22–24]. It is unclear whether such an approach will overcome the immunodominance one sees with the current viral vector vaccines. What is clear is that strategies must be developed to elicit greater epitope breadth to provide more optimal T cell responses to subdominant responses, which may provide more optimal matching of vaccine-elicited T cell responses to the diverse array of circulating strains of HIV [21]. Immunization with a variety of peptides matching defined CD8 epitopes would be an attractive approach for accomplishing such a goal. However, to date, peptide-based vaccines have not been shown to be immunogenic in humans, either when given as protein or with a viral vector such as a poxvirus [25]. It may be that alternative approaches to achieve breadth, such as protein subunits conjugated to adjuvants capable of eliciting CD8+ T cell responses, are necessary [26]. Recent evidence has shown that continued boosting with homologous vectors changes the T cell response to vaccination and also appears to enhance immunodominance [27]. Strategies to use more complex heterologous vectors with heterologous inserts given at separate sites and times should be investigated. Such approaches require more complex vaccination regimens and combinations of vectors, but data are present that such an approach will likely be required to improve T cell breadth [28].

Assessing the ‘quality’ of vaccine-induced T cell responses

Recently, several groups have stressed that T cells that exhibit a variety of functions (polyfunctional cells) appear to be more effective in controlling experimental viral infections than cells that exhibit a more selective phenotype [29]. Some studies on the clinical course of HIV-1 have associated T cells, especially CD4+ T cells exhibiting both IL-2 and IFN-γ responses, with better control of disease [30,31]. Polyfunctional T cells have been associated with protection from Leishmania infection and have been elicited by vaccinia and yellow fever vaccination [32–34]. The techniques for describing these polyfunctional cells are based on surface marker expression, and current studies define ‘associations’ between such cells and do not provide mechanistic proof of phenotypic function. Unfortunately, an adoptive transfer model to define what type of polyfunctional cells are most likely to control HIV replication has not yet been devised. Thus, quantitative methods to define what types of immune functions are associated with control of HIV replication are needed. Complicating this issue are data to show that innate immune responses influence both the qualitative and quantitative features of adoptive immunity [12]. Hence, even if the required responses are well defined, eliciting such responses may require considerable empiricism. NHP studies may be useful in advancing knowledge about these questions and provide direction for future immunogen design and testing. However, at present, only human clinical trials are likely to provide insight into defining whether a particular type of response is effective in modifying HIV-1 replication.

Increased acquisition of HIV in persons who entered the Step trial with proven Ad5 immunity

The most surprising finding from the Step trial was the two-fold increased risk of HIV acquisition among Ad5 seropositive persons, an effect not seen in any prior NHP studies of adenovirus vectors. As yet, it is unclear whether this risk is biologic or confounded by behavior or other factors, such as geographic clustering of cases or infection with strains of unusual virulence. There were distinct differences in Ad5 seroprevalence between sites; Ad5 seronegative individuals were two times more common in North America than in South America or the Caribbean. However, among placebo recipients, infection rates were higher in Ad5 seronegative than seropositive persons, albeit not statistically different (Table 1).

At present, some degree of uncertainty still exists about whether the increased acquisition in Ad5 seropositive individuals who received the vaccine has an underlying biologic mechanism. However, as shown in Fig. 1, among placebo recipients there was no statistically significant increase in HIV acquisition between Ad5 seropositive and seronegative men – in both univariate and multivariate analyses. No increase in HIV acquisition was seen among Ad5 seronegative men or women who received the vaccine. These latter data indicate that trials of adenovirus vector vaccines for HIV (as well as other diseases in HIV at-risk populations) can be safely conducted in Ad5 seronegative persons.

Fig. 1
Fig. 1:
Kaplan–Meier plot (% failure) for subjects who received placebo: MITT* population (males). Time of estimated HIV infection (midpoint between the first evidence date and last negative date) in weeks relative to randomization. *Modified intent to treat. P value = 0.103 (unadjusted), RR = 1.79 (≤18 versus >18); P value = 0.207 (covariate-adjusted), RR = 1.65 (≤18 versus >18).

Several hypotheses for explaining increased susceptibility to infection after vaccination among Ad5 seropositive versus seronegative persons have been put forth, including the following:

  1. Receipt of the vaccine boosted adenovirus-specific T cells (especially CD4+ T cells and especially at mucosal sites) among persons with previous immunity to the vector. This increase in activated target cells resulted in increased susceptibility to infection following high-risk sexual behavior.
  2. Prior Ad5 immunity ‘skewed’ the immune response to the vector and reduced innate immune response to HIV-1 infection, hence leading to higher acquisition rates in sexually exposed men.
  3. Prior immunity produced some ‘enhancing’ antibodies, which led to increased susceptibility to infection.

Additional explanations are possible, as are combinations of all three presented hypotheses. An intriguing factor about the increased rate of acquisition is that 100% of HIV acquisition in the Step trial occurred after the second dose of vaccine. Thus, most of the adenovirus seronegative persons likely had seroconverted to adenovirus and exhibited evidence of adenovirus-specific CD4+ and CD8+ T cell responses by the time of HIV-1 acquisition. Any mechanistic explanations for the difference in HIV acquisition rate between Ad5 seropositive and Ad5 seronegative individuals must account for this observation. One explanation suggests the difference may be due to ‘imprinting’ of a certain phenotype, for example, homing to mucosal sites from natural mucosal Ad5 infection rather than from intramuscular injection of the replication incompetent vector. Regardless of the eventual explanation, one of the lessons learned from the Step trial is the need to assess mucosal and tissue-specific immune responses to both the vector and HIV vaccine inserts in future clinical trials.

The collaborators in the Step and Phambili trials (HVTN and Merck Research Laboratories) have initiated several modifications to the ongoing trials to attempt to answer the above questions. Biopsies of mucosal tissue sites among vaccinees are underway to evaluate whether the quantity or quality of T cells present in such sites differ between those with naturally acquired Ad5 infection and those with Ad5 immunity induced only by vaccination. As most of the Step recipients had received their last vaccination 6–11 months prior to these mucosal studies, this approach, while necessary, may not be optimal. Ideally, a prospective trial to evaluate such issues should be conducted to provide optimal information regarding the tissue-specific T cell responses after adenovirus vaccination. As defining a potential mechanism for acquisition is essential for the continued development of adenovirus-based vectored vaccines, it is hoped that the availability of the immunogens to perform such investigations will be forthcoming. For the T cell vaccine field, answering these questions are of fundamental importance. As such, the trial sponsors – Merck Research Laboratories, NIAID, and the HVTN – have established a scientific review committee to lead the scientific program being developed to evaluate these issues. Similarly, the outside investigative community has been invited to participate in a process to submit ideas for funding. The application process and review is described on the HVTN website (


The Step trial has been a milestone event in the area of T-cell-based vaccines. It has resulted in several unique scientific contributions to the HIV field, including the use of the test-of-concept trial for defining vaccine efficacy for prototype vaccines; evidence that HIV 89.6P challenge in genetically sensitive animals (MAMU-A*01 or B*08) should not be used to gauge vaccine effectiveness; and the finding that vector-induced immunity should be tightly evaluated in the course of immunogen development. Although the Step trial clearly did not produce an effective vaccine, critical isolates and immunologic assays are just being analyzed. These data will define whether there are correlates between host responses to vaccination and postinfection control of viremia and will be central to resetting the immunological measurements that should be made on future vaccine candidates.


The authors would like to thank Susan Buchbinder (HIV Research Section, San Francisco Department of Public Health), Michael Robertson (Merck Research Laboratories), and Devan Mehrotra (Merck Research Laboratories) for their collaboration.

This study was supported in part by NIH grant AI-46747.


1. Vaccination and enrollment are discontinued in phase II trials of Merck's investigational HIV vaccine candidate. Interim analysis of Step study shows vaccine was not effective. In: Press release from Merck and the HIV Vaccine Trials Network. Whitehouse Station: New Jersey; Seattle, Washington; 21 September 2007.
2. Cohen J. AIDS research. Promising AIDS vaccine's failure leaves field reeling. Science 2007; 318:28–29.
3. Mehrotra DV, Li X, Gilbert PB. A comparison of eight methods for the dual-endpoint evaluation of efficacy in a proof of concept HIV vaccine trial. Biometrics 2006; 62:893–900.
4. Priddy F, Brown D, Kublin J, Monahan K, Wright DP, Lalezari J, et al, Merck V520-016 Study Group. Safety and immunogenicity of a replication-incompetent adenovirus type-5 HIV-1 clade B gag/pol/nef vaccine in healthy adults. Clin Infect Dis 2008; 46:1769–1781.
5. Fu TM, Dubey SA, Mehrotra DV, Freed DC, Trigona WL, Adams-Muhler L, et al. Evaluation of cellular immune responses in subjects chronically infected with HIV type 1. AIDS Res Hum Retroviruses 2007; 23:67–76.
6. Shiver JW, Emini EA. Recent advances in the development of HIV-1 vaccines using replication-incompetent adenovirus vectors. Annu Rev Med 2004; 55:355–372.
7. Isaacs R. Impact of preexisting immunity on the immunogenicity of adenovirus serotype 5-based vaccines [abstract 69]. AIDS Vaccine. Lausanne; 2004.
8. HIV Vaccine Trials Network 2007 Full Group Meeting.; Nov 7. [Accessed 24 April 2008.]
9. Buchbinder S. Efficacy results from the Step study (Merck V520 Protocol 023/HVTN 502). A phase II test-of-concept trial of the MRKAd5 HIV-1 gag/pol/nef trivalent vaccine.15th Conference on Retroviruses and Opportunistic Infections; Boston, Massachusetts; 2008.
10. NIAID Bulletin: an update concerning the HVTN 503/Phambili HIV Vaccine Study.
11. Shiver JW, Fu TM, Chen L, Casimiro DR, Davies ME, Evans RK, et al. Replication-incompetent adenoviral vaccine vector elicits effective antiimmunodeficiency-virus immunity. Nature 2002; 415:331–335.
12. Sekaly RP. The failed HIV Merck vaccine study: a step back or a launching point for future vaccine development? J Exp Med 2008; 205:7–12.
13. Casimiro DR, Wang F, Schleif WA, Liang X, Zhang ZQ, Tobery TW, et al. Attenuation of simian immunodeficiency virus SIVmac239 infection by prophylactic immunization with DNA and recombinant adenoviral vaccine vectors expressing Gag. J Virol 2005; 79:15547–15555.
14. Barouch DH, Yang ZY, Kong WP, Korioth-Schmitz B, Sumida SM, Truitt MD, et al. A human T-cell leukemia virus type 1 regulatory element enhances the immunogenicity of human immunodeficiency virus type 1 DNA vaccines in mice and nonhuman primates. J Virol 2005; 79:8828–8834.
15. Barouch DH, Nabel GJ. Adenovirus vector-based vaccines for human immunodeficiency virus type 1. Hum Gene Ther 2005; 16:149–156.
16. Emini EA. A potential HIV-1 vaccine using a replication-defective adenoviral vaccine vector. Keystone Symposia. Banff, Alberta, Canada; 2003.
17. Johnston MI, Fauci AS. An HIV vaccine – evolving concepts. N Engl J Med 2007; 356:2073–2081.
18. Duerr A, Wasserheit JN, Corey L. HIV vaccines: new frontiers in vaccine development. Clin Infect Dis 2006; 43:500–511.
19. Li F, McKenney DM, Malhotra U, Crimi C, Nolin J, Corey L, et al. Towards prediction of degenerate CTL epitope recognition. Hum Vaccin 2007; 4:115–121.
20. De Rosa SC, Thomas EP, Huang Y, Morgan C, Kalams S, Graham B, et al.Kinetics of T cell responses to vaccination for HIV and heterologous DNA primar-rAd5 boost contrasted with homologous rAd5 prime-boost [abstract 161]. Keystone Symposia. Banff, Alberta, Canada; 2008.
21. Frahm N, Kiepiela P, Adams S, Linde CH, Hewitt HS, Sango K, et al. Control of human immunodeficiency virus replication by cytotoxic T lymphocytes targeting subdominant epitopes. Nat Immunol 2006; 7:173–178.
22. Nickle DC, Rolland M, Jensen MA, Pond SL, Deng W, Seligman M, et al. Coping with viral diversity in HIV vaccine design. PLoS Comput Biol 2007; 3:e75.
23. Li F, Horton H, Gilbert PB, McElrath JM, Corey L, Self SG, et al. HIV-1 CTL-based vaccine immunogen selection: antigen diversity and cellular response features. Curr HIV Res 2007; 5:97–107.
24. Letourneau S, Im EJ, Mashishi T, Brereton C, Bridgeman A, Yang H, et al. Design and preclinical evaluation of a universal HIV-1 vaccine. PLoS ONE 2007; 2:e984.
25. Gorse GJ, Baden LR, Wecker M, Newman MJ, Ferrari G, Weinhold KJ, et al. Safety and immunogenicity of cytotoxic T-lymphocyte poly-epitope, DNA plasmid (EP HIV-1090) vaccine in healthy, human immunodeficiency virus type 1 (HIV-1)-uninfected adults. Vaccine 2008; 26:215–223.
26. Wille-Reece U, Wu CY, Flynn BJ, Kedl RM, Seder RA. Immunization with HIV-1 gag protein conjugated to a TLR7/8 agonist results in the generation of HIV-1 gag-specific TH1 and CD8+ T cell responses. J Immunol 2005; 174:7676–7683.
27. Smith CL, Mirza F, Pasquetto V, Tscharke DC, Palmowski MJ, Dunbar PR, et al. Immunodominance of poxviral-specific CTL in a human trial of recombinant-modified vaccinia Ankara. J Immunol 2005; 175:8431–8437.
28. Koopman G, Mortier D, Hofman S, Niphuis H, Fagrouch Z, Norley S, et al. Vaccine protection from CD4+ T-cell loss caused by simian immunodeficiency virus (SIV) mac251 is afforded by sequential immunization with three unrelated vaccine vectors encoding multiple SIV antigens. J Gen Virol 2004; 85(Pt 10):2915–2924.
29. Seder RA, Darrah PA, Roederer M. T-cell quality in memory and protection: implications for vaccine design. Nat Rev Immunol 2008; 8:247–258.
30. Betts MR, Nason MC, West SM, De Rosa SC, Migueles SA, Abraham J, et al. HIV nonprogressors preferentially maintain highly functional HIV-specific CD8+ T cells. Blood 2006; 107:4781–4789.
31. Harari A, Vallelian F, Meylan PR, Pantaleo G. Functional heterogeneity of memory CD4 T cell responses in different conditions of antigen exposure and persistence. J Immunol 2005; 174:1037–1045.
32. Darrah PA, Patel DT, De Luca PM, Lindsay RW, Davey DF, Flynn BJ, et al. Multifunctional TH1 cells define a correlate of vaccine-mediated protection against Leishmania major. Nat Med 2007; 13:843–850.
33. Pulendran B, Ahmed R. Translating innate immunity into immunological memory: implications for vaccine development. Cell 2006; 124:849–863.
34. Querec T, Bennouna S, Alkan S, Laouar Y, Gorden K, Flavell R, et al. Yellow fever vaccine YF-17D activates multiple dendritic cell subsets via TLR2, 7, 8, and 9 to stimulate polyvalent immunity. J Exp Med 2006; 203:413–424.
© 2009 Lippincott Williams & Wilkins, Inc.