Back grey_arrow_rt.gif
Thai HIV Vaccine Study, Moving Forward in
HIV Vaccine Development - Perspective
Norman L. Letvin
Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, 02115, USA.
Science 27 November 2009:
Vol. 326. no. 5957, pp. 1196 - 1198
DOI: 10.1126/science.1183278
Early in the AIDS epidemic, efforts to develop a vaccine to prevent HIV infection were focused on two vaccine strategies. It was hoped that the HIV envelope glycoprotein (gp120) would generate an antibody response that would block the initiation of HIV infections, and that a recombinant canary pox construct expressing HIV genes would elicit cellular immune responses that would inhibit HIV replication. However, the nature of the immune responses elicited by each of these vaccine candidates in monkeys and human subjects proved disappointing (1–3). Despite these results, the U.S. Department of Defense (DOD) proceeded with plans to initiate the RV144 trial in Thailand to test the efficacy of a vaccine regimen that included both agents. At that time, the U.S. National Institutes of Health assumed responsibility for a major component of the DOD HIV/AIDS program, and provided a substantial proportion of the funding for the Thai trial.
The decision to commit resources to carry out the RV144 trial in the face of the limited immunogenicity of both vaccine prototypes created considerable concern within the HIV vaccine research community (4). Opponents of the trial argued that the resources being dedicated to the vaccine study would be better spent on basic research that might inform HIV vaccine development. As a consequence of discussions among HIV vaccine investigators, a viral load set point-the amount of virus present in the blood during chronic infection-was added as one of the primary endpoints (an outcome that can be measured objectively to determine whether the vaccination was beneficial) in the RV144 trial a year after its initiation. I shared the concerns of my colleagues about the wisdom of proceeding with this clinical trial, but my opposition was tempered by an appreciation for the commitment made by the DOD and the government of Thailand to test this vaccine.
It is unfortunate that there was a partial release of data before the formal announcement and publication of the trial findings, but we can now digest the complete results in the recently published report of the trial (5). The basic message is that the trial results show modest vaccine efficacy at 3 years after initiating vaccination, with a lowered rate of HIV infection by 31.2% (P = 0.04)-51 infections in the 8197 vaccine recipients and 74 infections in the 8198 placebo recipients. This evaluation was done using a modified Intent-to-Treat analysis, in which all uninfected participants enrolled in the trial were considered, including those who missed a shot or did not receive one at the appropriate time. The modest statistical significance of the finding of protection is lost, however, if a different strategy-a Per Protocol analysis-is employed for determining which study volunteers to analyze. In the Per Protocol analysis, only a subset of trial participants are considered-those who had followed the study design precisely and who also did not become infected during the 6-month period of vaccination. Both of these commonly employed analytic strategies were formally designated as the criteria for evaluating the study data before the trial was initiated, and therefore it is appropriate to evaluate the trial results using both analytic approaches. The loss of the modest statistical significance of the findings using the Per Protocol analysis is likely a consequence of the loss of power in the study resulting from the smaller number of subjects that were analyzed.
Interestingly, in those participants who became infected after vaccination, no better control of HIV replication and no better protection against loss of the CD4+ subset of T lymphocytes were observed when compared with the placebo recipients. Critics have pointed out that the absence of a vaccine effect on HIV replication in vaccinees who became infected contrasts with the findings in numerous HIV vaccine trials in nonhuman primates in which virus replication is partially contained in vaccinated animals that become infected.
The findings in the RV144 trial do not bear out the arguments of either the early supporters or opponents of the trial. The supporters argued at the outset that the study would provide an opportunity to test the utility of a "prime-boost" vaccine strategy that elicits both humoral and cellular immune responses (6). However, because the immunogens used in this trial did not generate antibodies that neutralize diverse HIV isolates or potent responses by CD4+ or CD8+ T cell subsets, and because the protection against HIV acquisition was very modest, the results of the trial cannot be viewed as either supporting or arguing against this prime-boost vaccine strategy. The trial opponents argued that the immunogenicity of the vaccine regimen was too marginal to warrant its testing in an efficacy trial (4). Yet, if the results of the Thai vaccine trial are viewed as hypothesis-generating rather than a step toward vaccine licensure, there are indeed some findings that could catalyze new efforts.
Although the number of subjects evaluated does not allow an assessment of the statistical significance of the finding, the vaccine may have been more effective immediately following administration, and this effect may have diminished rapidly as the weak vaccine-elicited immune responses waned. Immunogens that elicit more durable immune responses might therefore generate more durable protection.
Also of interest is the suggestion that the risk for HIV infection among the volunteers may have had a substantial impact on the results of the trial. Although the study was not sufficiently powered to assess vaccine efficacy in subpopulations of volunteers, it is provocative that the vaccine appeared 40% effective in the lower-risk population (limited number of low-risk sex partners and no needle exposure) and conferred only 3.7% protection in the higher-risk population (needle sharing or high-risk sexual activity). We will need to determine whether recombinant gp120 or recombinant canary pox used alone might confer protection against infection in a low-risk human population. A vaccine that elicits more potent responses may be needed to confer protection in high-risk populations.
However, it is also possible that the findings in the RV144 trial may not provide a useful lead for developing an effective immunization strategy for preventing HIV acquisition in high-risk populations. A vaccine that elicits qualitatively different immune responses may be needed to confer protection in high-risk individuals and may ultimately prove the best vaccination approach for a general population.
It will be important going forward to evaluate the immune responses elicited by recombinant protein and recombinant pox virus immunizations to explore potential immune correlates of protection in the RV144 trial. Because a large number of the immunized individuals were likely not exposed to HIV during the course of the vaccine trial, the most useful approach for exploring this problem will be to evaluate vaccinees who became infected and explore what immune responses were not elicited in them. These studies will be most informative if they are done using robust assays with wide dynamic ranges. The DOD has established advisory committees in the areas of B lymphocytes, T lymphocytes, genetics, and animal models to provide guidance on what studies should be done with the limited available blood specimens collected from the subjects in this study. There are very limited quantities of cells and no mucosal specimens available from the vaccine trial recipients (7). It will therefore be necessary to initiate new, focused immunogenicity studies in the near future to evaluate the cellular and humoral immune responses in systemic and mucosal compartments induced by the vaccine regimen administered in the RV144 trial. Based on our current understanding of the immune responses that are elicited by the immunogens used in this trial, early thinking is focusing on possible contributions to protection by antibodies that bind to but do not neutralize the virus and by CD4+ T lymphocyte populations that augment vaccine-induced antibody responses to the HIV envelope glycoprotein (gp120).
The findings in the Thai trial suggest that different criteria should be considered in the future for determining what vaccine strategies warrant evaluation in advanced-phase clinical testing. Until now it was assumed that blocking HIV acquisition through vaccine-elicited immunity might not be achievable. Therefore, advanced-phase clinical studies have been initiated to determine whether vaccine-elicited immunity can contribute to decreasing the viral load after HIV infection has occurred. Because evidence from the Thai trial suggests that vaccination might result in protection against HIV acquisition, at least in low-risk subjects, clinical efficacy trials of vaccine candidates should in the future be initiated only if there is an expectation that protection from infection might be achieved. Therefore, HIV vaccine candidates being considered for advanced-phase human trials should be evaluated in nonhuman primates to determine whether they confer protection against HIV acquisition. This represents a change in the way nonhuman primate vaccine trials have often been done in the past. In those earlier studies, virus replication was monitored after high-dose, intravenous challenge. A more accurate prediction of vaccine efficacy will require the development of nonhuman primate models in which vaccine protection against the acquisition of HIV infection can be reliably evaluated. These models would ideally be based on repeated, low-dose mucosal exposure of nonhuman primates to a virus. Useful studies can be done using pathogenic simian immunodeficiency viruses (SIVs) for these challenges. Because HIV envelope protein vaccines must be evaluated, it will also be important to develop consistently pathogenic chimeric simian-human immunodeficiency viruses-laboratory-constructed viruses that express the HIV envelope on an SIV "backbone" of structural and enzymatic proteins-that bind to CCR5, the human HIV co-receptor.
In retrospect, modifications in the design of the RV144 trial would have elucidated many of the findings with which we are now grappling. The study of larger numbers of individuals would have clarified the significance of the differences in HIV acquisition observed between the vaccine and placebo recipients. The delivery of the gp120 glycoprotein with a more potent adjuvant, such as one incorporating an agonist of Toll-like receptors (when activated, these receptors trigger immune cell responses), might have potentiated the protein component of vaccine. Similarly, the use of a less attenuated pox virus–based vector such as the NYVAC or MVA strains of vaccinia virus might have boosted immune responses. Also, it would have simplified the interpretation of the data if a second placebo arm in the trial had received the canary pox vector without an HIV gene insert, as this would allow a determination of the possible contribution of vector-induced innate immune responses to protection. Including such a second placebo arm in the trial could only be done if the U.S. Food and Drug Administration determined that the vaccine vector had a sufficiently benign safety profile.
While the pursuit of basic laboratory research will continue to be important in the development of an HIV vaccine, the results of the Thai trial underscore the extraordinary importance of also performing focused human clinical trials of vaccine strategies. Just as the nature of the recent failure of a recombinant adenovirus vaccine to confer protection against HIV infection in the STEP trial could not have clearly been predicted based on the preclinical experiments that had been carried out (8), the findings in the Thai trial were not expected based on preclinical studies and human immunogenicity data. We have learned to expect the unexpected in our efforts to generate an effective HIV vaccine.
* 1. J. Mascola et al., J. Infect. Dis. 173, 340 (1996).
* 2. N. M. Flynn et al., J. Infect. Dis. 191, 654 (2005).
* 3. N. D. Russell et al., J. Acquir. Immune Defic. Syndr. 44, 203 (2007).
* 4. D. R. Burton et al., Science 303, 316 (2004). * 5. S. Rerks-Ngarm et al., N. Engl. J. Med. (10.1056/NEJMoa0908492 (2009).
* 6. R. Belshe et al., Science 305, 177 (2004).
* 7. N. Michael, personal communication
* 8. S. Buchbinder et al., Lancet 372, 1881 (2008).
  icon paper stack View Older Articles   Back to Top