The STEP Study Provides a Hint That Vaccine Induction of the Right CD8⁺ T Cell Responses Can Facilitate Immune Control of HIV

(See the article by Fitzgerald et al., on pages 765–72)

The STEP study was a large, randomized, placebo-controlled, test-of-concept vaccine trial using an Ad5 vector expressing human immunodeficiency virus (HIV)–1 Gag, Pol, and Nef designed to induce T cell immunity to HIV-1. This trial was stopped and unblinded in 2007 after an interim analysis showed that the vaccine did not achieve efficacy for the 2 primary study end points, HIV-1 acquisition and plasma HIV-1 RNA levels 3 months after diagnosis of HIV-1 infection, and suggested that vaccinated individuals with high preexisting antibody titers against Ad5 might be at a higher risk of acquiring HIV-1 infection [1–2]. These results initiated reconsideration in the HIV-1 research field of the role of T cells in protection from HIV-1 infection and disease progression.

Fitzgerald et al, in this issue of the Journal, now provide follow-up data on the vaccine effect on markers of HIV-1 disease progression in 87 male participants in the STEP trial who became HIV-1-infected by November 2007. After a median of 24 months of postinfection follow-up, no effect of the vaccine on HIV-1 load setpoints, CD4⁺ T cell counts, time to initiation of antiretroviral therapy, and AIDS-free survival was observed. Overall, these data provide further evidence that the HIV-1–specific T cell responses induced by the Ad5 vaccine had no protective effect on HIV-1 disease progression in the majority of individuals.

Since the initial publication of the disappointing data from the STEP trial in 2008, the results of this trial, and their consequences for HIV-1 vaccine research, have been extensively discussed [3–11], and samples collected from the trial have been used to gain additional insights into the immune responses induced by the vaccine and into the consequences of the preinfection Ad5 serostatus on vaccine-induced immunity [1, 12–14]. In general, HIV-1–specific CD8⁺ T cell responses induced by the vaccine were weak and directed against a very limited number of T cell epitopes. Only a median of 2 epitope-specific CD8⁺ T cell responses were detected in vaccine recipients when 15mer peptides were used to detect HIV-1–specific T cell responses in an interferon-γ enzyme-linked immunosorbent spot assay (Frahm, AIDS Vaccine 2009 meeting). In particular, CD8⁺ T cell responses directed against HIV-1 Gag epitopes, which have been shown to be associated with control of viremia in natural HIV- 1 infection in individuals that targeted ≥2 epitopes within Gag [15–16], were only rarely detected in vaccinated individuals. Therefore, the overall weakness of vaccine-induced CD8⁺ T cell responses against HIV-1, and in particular HIV-1 Gag, might have contributed to the failure to mediate protective immunity, and new vaccine concepts aimed at inducing more robust HIV-1-specific CD8⁺ T cell responses are currently being evaluated [17–19].

In a subanalysis presented in the current article by Fitzgerald et al the authors describe that vaccinated individuals expressing human leukocyte antigen (HLA) class I alleles associated with better control of viremia (HLA-B27, B57, and B5801) had significantly lower HIV-1 RNA levels over time than individuals expressing these alleles who only received the placebo. These data are of interest, because they suggest that individuals expressing these protective alleles might have benefited to some extent from the vaccination. Several studies have shown that HIV-1–infected individuals expressing protective HLA class I alleles control HIV-1 replication substantially better and experience a slower progression to AIDS [20]. HIV-1–specific CD8⁺ T cell responses restricted by these alleles contribute over-proportionally to the virus-specific CD8+ T cell response during acute HIV-1 infection, and individuals expressing these alleles appear to control viremia quickly following infection to low levels [21–22]. This protective effect appears to be mediated by HIV-1–specific CD8⁺ T cell responses restricted by these alleles that target highly conserved areas of the virus within Gag, and viral escape from these Gag-specific CD8+ T cell responses is either not possible or occurs at a high cost for viral replicative fitness[23–24]. The enhanced ability of vaccinated individuals expressing HLA-B27, B57, and B5801 to control viral replication after infection, compared with unvaccinated controls expressing these alleles, suggests that priming virus-specific CD8⁺ T cell responses restricted by protective HLA class I molecules prior to infection can be beneficial. Further studies dissecting the epitope-specific CD8⁺ T cell responses induced by the vaccine will be needed to further elucidate whether CD8⁺ T cell responses restricted by these protective alleles were, indeed, induced by the vaccine.

The improved control of viremia, with a mean of 0.86 log₁₀ lower viral load (95% confidence interval, −1.52 to −0.20 log₁₀) in vaccinated individuals expressing protective HLA class I alleles, was statistically significant (P = .03). This is remarkable, given the very small number of study subjects expressing protective HLA alleles (9 in the vaccine arm vs 5 in the placebo arm). Caveats one may need to consider in interpreting these data include the fact that, owing to missing data, as much as 43% of the viral load data in this study were imputed, as opposed to being actual measurements. Furthermore, in the absence of the raw viral load data, it is possible that the B*27/57/5801-positive subjects in the placebo arm, whose viral loads were higher than those in the vaccine arm, may have had viral loads that were unrepresentatively high; in this case, a difference between the 2 would result not from a vaccine effect on subjects with protective alleles, but from chance.

Although these data on viral setpoint are based on a small number of individuals and need to be interpreted with caution, as carefully highlighted by the authors, they might provide some new insights into the potential consequences of vaccine-induced CD8+ T cell responses on the control of HIV-1 viremia. Vaccinated individuals who encoded for HLA class I alleles or allele combinations associated with faster HIV-1 disease progression (B*3502, B*3503, B*3504, B*53, or showed homozygosity in at least 1 HLA locus for HLA alleles other than B*27/57/5801), had a 0.19 log₁₀ higher viral load in the STEP trial than individuals encoding for these alleles who received placebo. This difference in viral load, however, did not reach statistical significance (95% CI: − .46 to .84), but might provide a hint that vaccine-induction of CD8+ T cell responses restricted by unfavorable HLA class I molecules might not be beneficial. Previous studies have suggested that vaccine-induced immunodominance patterns of T cell responses largely mimic the responses induced to the same antigens during natural HIV-1 infection (N. Frahm and J. McElrath, personal communication). While the preservation of these immunodominance patterns for vaccine-induced responses that are protective in natural infection might be beneficial for the “lucky few” possessing protective HLA class I alleles, future vaccine approaches may need to test immunogens designed to break these immunodominance patterns for HLA class I alleles that are not inducing protective responses in natural infection. This can be accomplished by deleting immunodominant epitopes restricted by these nonprotective alleles from the immunogen. For example, if the dominant CD8+ T cell responses in acute infection restricted by unfavorable HLA alleles, such as HLA-B*35 and B*53, are in Nef and Env (which they are [25]), and if these are ineffective responses, then induction of Nef responses by a Nef-containing vaccine could magnify an ineffective response at the expense of other, potentially more effective, non-Nef responses.

One additional discussion point these studies raise is the possibility that ”only” those people expressing protective alleles can be helped by a T cell vaccine. First, it is worth noting that, although some populations might benefit very little if such were the case, such as in Japan, where B*27/57/5801 is expressed in only 1% of the population [26], other populations might benefit substantially, such as is Barbados, where B27/57/5801 is expressed in 27% of the population [27]. In the STEP study, as many as 16% of the infected study participants expressed one of HLA-B*27/57/5801. Second, it is clear that HLA alleles in addition to B*27/57/5801 can also be strongly protective against HIV-1 progression, including HLA-B*1302, B*1516, and B*8101 [28–30], and that some of the highly prevalent HLA alleles, such as B*4201 and B*4403, also have significant, albeit modest, beneficial effects [30]. Thus, there are a number of HLA alleles conferring some degree of protection against HIV-1 disease progression, and a vaccine that could improve on immune control mediated by all these alleles would benefit a substantial proportion of any population.

Taken together, these follow-up data from the STEP study provide some indication that induction of the right CD8⁺ T cell responses by a vaccine can make an important difference to bring about successful immune control of HIV infection. However, the induction of more CD8⁺ T cell responses per se might not result in a successful vaccination strategy, because the majority of HIV-specific CD8⁺ T cell responses are ineffective. This concept includes the possibility that a T cell vaccine might make immune control more difficult than in natural infection by skewing CD8⁺ T cell responses towards the ”wrong” epitopes. It remains of critical importance, therefore, to determine with greater clarity than is currently available what the effective CD8⁺ T cell responses against HIV-1 are and whether they can be distinguished by protein-specificity alone, or by protein region, to facilitate optimal control of HIV-1 viremia in vaccinees who subsequently become infected, irrespective of their HLA class I genotype.

Funding

M.A. is a Distinguished Clinical Scientist of the Doris Duke Charitable Foundation, and P.J.G. is supported by the Wellcome Trust.