Vaccine Scientists Struggle to Come to Grips With the Field's "Berlin"
The AIDS Vaccines 2004 Conference was held from August 30-September 1 in Lausanne, Switzerland, the first of these meetings to be held outside the US (the first two were in Philadelphia in 2001 and New York City in 2003). Unlike many recent conferences which have concluded without any clear highlights or obvious signature issues to mark them, the Lausanne conference is likely to be remembered for spotlighting several key themes of broad relevance to the AIDS vaccine field.
Firstly, the disappointing immune responses seen with the International AIDS Vaccine Initiative's DNA-MVA candidate capped a growing concern about the prospects for vaccines based on current DNA and MVA platforms. Secondly, there was widespread agreement that the recently adopted ELISpot assay for assessing T-cell immune responses based solely on production of the cytokine interferon-gamma is not providing a complete picture of vaccine-induced T cells.
Thirdly, reports from human trials suggested that the ability of macaque models to accurately predict vaccine immunogenicity (capacity to trigger immune responses) might be more limited than previously realized. Richard Jefferys prepared this summary for TAGline.
IAVI's DNA-MVA Disappoints
The major news story to emerge from Lausanne is that IAVI is unlikely to move their first vaccine candidates into efficacy testing. The vaccines, a combination of a DNA "prime" and modified vaccinia Ankara strain (MVA) "boost" (produced by a team at Oxford University led by Andrew McMichael and Tom Hanke), induced HIV-specific CD8 T-cell responses in approximately 10-20% of a total of 205 volunteers participating in multiple phase I and II studies in the UK and Kenya, according to presentations in Lausanne. This fell far short of reaching the 60% or greater response rate that IAVI views as necessary to justify launching a phase III efficacy trial.
IAVI released a statement to coincide with the opening of the conference, noting that the remaining studies of the vaccines will be completed and some additional immune responses looked at. But "unless there are new immune response data that are dramatically different, IAVI will not develop the candidates further, and will focus on its other research and development projects."
In the hallways and around the coffee tables of the Beaulieu Conference and Exhibition Centre, people mulled the implications of IAVI's data for the larger vaccine field. In terms of DNA constructs, IAVI's is not the first to show disappointingly poor immunogenicity in humans: Merck, Wyeth Ayerst and others have reported similar findings. Merck also found no advantage to giving its DNA as a prime before subsequently immunizing with a different vaccine (in this case, an adenovirus vector) as a booster, echoing IAVI's experience with DNA/MVA.
A commonly expressed opinion holds that part of the problem with current DNA vaccines is related to dosage; several conference participants estimated that a DNA vaccine dose of at least 8 milligrams is needed to match the dose given to macaque monkeys in pre-clinical studies. Only the NIH's Vaccine Research Center (VRC) has attempted this high a dose in humans. And preliminary results suggest that it did not result in a significant improvement in the immunogenicity of the DNA vaccine compared to a lower 4 mg dose. However, it's also very difficult to inject this much DNA into a human being, and experiments using more than 8 mg are essentially impossible.
Researchers are continuing to explore ways to enhance the response to DNA vaccines at feasible doses, including using separate constructs to encode each vaccine antigen and incorporating cytokines and other potential adjuvants. IAVI's results suggest that these efforts will have to be successful if DNA vaccines are to become viable for human use.
The data on IAVI's MVA-based construct are sobering given that there are over a dozen candidate HIV vaccines utilizing MVA as a vector. The extent to which the poor results reflect problems with the specific MVA candidate as opposed to the MVA platform in general is a matter of debate.
The ability of the vector to express the HIV antigens it is carrying once in the body is influenced by precisely where in the MVA genome the genetic code for the HIV antigens is inserted, and it is believed that other MVA candidates may be able to express higher levels of their antigen payload than IAVI's construct. It remains uncertain, however, whether such improvements can raise the immune response rate to an acceptable level.
Another potential problem with MVA is the large size of the vector; some researchers believe that this skews the immune response toward the vector itself rather than the antigens it contains. A small phase I trial of an MVA-based HIV vaccine produced by Bavarian Nordic recently reported that all participants developed MVA-specific T-cell responses, but only a minority showed evidence of responses to the HIV nef antigen contained in the vaccine.
Analyses of MVA-specific responses induced in the IAVI trials have not yet been reported. While immunogenicity studies of additional MVA-based HIV vaccines are going forward, it is entirely possible that the shortcomings of the approach will eventually lead to its demise, sending a large chunk of the HIV vaccine pipeline down the drain.
Multifunctional T-Cell Testing Comes of Age
The current crop of HIV vaccine candidates primarily aim to induce CD4 and CD8 T-cell responses targeting the virus, due the fact that effective methods for inducing neutralizing antibodies have yet to be discovered. In order to assess the potential of these vaccines, researchers have expended considerable effort on developing and standardizing assays for quantifying vaccine-induced T-cell responses.
The most rigorously evaluated approach is the interferon-gamma based ELISpot, which counts the number of T cells capable of making interferon-gamma after brief stimulation with HIV antigens (this is the assay used in the IAVI studies described above). Over the past few years, new studies have been presented showing that measuring interferon-gamma production alone may underestimate the size of the vaccine-induced T-cell population because some cells produce other cytokines.
In Lausanne this issue came to the fore, and there was widespread agreement that a broader assessment of T-cell responses will be important in future studies. Among the assays under discussion are intracellular cytokine staining, which several groups have used to show that T cells producing IL-2 can make up a substantial proportion of the response to commonly utilized vaccines. Helen Horton, from the HIV Vaccine Trials Network, demonstrated that GSK's HIV vaccine candidate incorporating nef, tat and gp120 proteins induced a CD4 T-cell response mainly comprised of cells capable of producing IL-2 -- not interferon-gamma. Stephen De Rosa from the VRC has previously presented similar findings from a phase I study of a DNA HIV vaccine.
Researchers from Guiseppe Pantaleo's lab also showed data implicating IL-2-producing T cells as important components of the HIV-specific immune response in infected individuals. Extending previous work demonstrating an inverse correlation between the frequency of HIV-specific CD4 T cells making IL-2 and viral load (see the recent review by Pantaleo & Koup, Nature Medicine 10: 806-810, 2004), Simone Zimmerli reported that long-term non-progressors (LTNP) possess significantly more IL-2-producing HIV-specific CD8 T cells than individuals with progressing infection.
Matthias Lichterfeld from Bruce Walker's group in Boston described a study that tracked the ability of HIV-specific CD8 T cells to proliferate both during and after primary HIV infection. Lichterfeld found that HIV-specific CD8 T-cell proliferation was detectable during primary infection but subsequently declined as individuals progressed to chronic infection. The proliferative capacity of HIV-specific CD8 T cells appeared to be linked to the presence of IL-2-producing HIV-specific CD4 T cells, according to Lichterfeld, "providing evidence of a direct functional linkage between these two cellular subsets."
The technique used to measure proliferation in this study involves staining cells with a dye called CFSE prior to stimulation with HIV antigens. (T cells that are able to proliferate lose 50% of the CFSE dye each time they divide, allowing researchers to quantify the degree of CFSE loss as a marker of proliferative capacity using a flow cytometer-based assay.) This new technique is a considerable improvement over previous proliferation tests and is another candidate for use in vaccine studies.
Michael Betts from the Vaccine Research Center debuted data obtained using a newly developed "multi-parameter" assessment of HIV-specific CD8 T-cell function. The assay developed by VRC allows simultaneous assessment of several CD8 T-cell functions, including production of the cytokines IL-2, TNF-alpha and interferon-gamma, the chemokine MIP1-beta and expression of a marker known as CD107a (a surrogate for the cell-killing potential of CD8 T cells).
Betts was able to identify a befuddling 32 different "flavors" of CD8 T cell in humans based on their ability to perform differing combinations of these functions. In a comprehensive analysis of HIV-specific CD8 T cells in infected individuals, Betts found that the spectrum of functions appeared to be connected to the clinical status of the study participant. LTNPs consistently displayed a more "polyfunctional" CD8 T-cell response; e.g., more of their HIV-specific CD8 T cells were capable of elaborating all five of the above-described functions compared to individuals with progressive disease. Betts noted that this difference did not depend solely on differences in viral load among the study participants.
In a closing "report-back" session at the conference, Clive Gray stressed that a take home message was that these types of comprehensive analyses of T-cell function will be important to consider in future vaccine trials. While it's not likely to be feasible to employ every possible test, those that capture responses missed by interferon-gamma ELISpot (such as intracellular cytokine staining for IL-2 and/or TNF-alpha) were cited as candidates for the same rigorous evaluation and standardization that ELISpot has undergone.
Predictive Value of Monkey Models Comes Into Question
Before beginning human studies, the ability of vaccines to induce immune responses is almost always evaluated in monkey models (most commonly rhesus macaques). MVA vectors have generally been reasonably immunogenic in these models, but the IAVI data described above suggests that the picture in humans is rather different. Merck also presented data in Lausanne that adds to this concern. Merck is developing an HIV vaccine based on an adenovirus vector (dubbed Ad5) that will shortly enter an efficacy trial (see TAGline, July 2004 for more information on this study).
One of the problems with Ad5 is that many people have been exposed to the virus (which occurs naturally in the environment and causes severe colds) and therefore possess anti-Ad5 antibody responses that can severely reduce the immunogenicity of Merck's Ad5-based HIV vaccine. Based on monkey studies, it appeared that this problem might be circumvented by giving Aventis Pasteur's ALVAC vaccine as a booster following Ad5 immunization. But Robin Isaacs presented data at the Lausanne conference showing that this approach did not show comparable success in humans: immunizing with Ad5 followed by an ALVAC boost induced similar HIV-specific T-cell responses to those seen when immunizing with Ad5 followed by simply another Ad5 shot -- regardless of the level of pre-existing anti-Ad5 antibodies in the study participants.
These results suggest that immunogenicity data obtained in macaques need to be interpreted very cautiously until they can be confirmed in humans. More encouragingly, Merck was able to report that their Ad5 vaccine containing the gag, pol and nef genes from HIV performed comparably in people to their original test construct that contained only gag. (There had been some concern that including additional genes might reduce the magnitude of the T-cell response -- and/or the % of responders -- to each one, but that didn't seem to happen.) And Isaacs confirmed that this "trivalent" vaccine will be utilized in the upcoming phase IIb efficacy trial that is slated to begin at the end of this year. Merck is also continuing to explore the vaccine potential of alternative types of adenoviruses (such as Ad24) that may be less affected by the problem of pre-existing antibodies than is Ad5.