What's in Glaxo's HIV Vaccine?

"The first human trial with GlaxoSmithKline Biologicals' (GSK Biologicals) candidate HIV vaccine will test whether the vaccine, which prevented AIDS in rhesus monkeys, is safe and immunogenic in humans."

-- GlaxoSmithKline Biologicals Press Release, Jan 31, 2002

On January 31, GlaxoSmithKline (GSK) garnered a good deal of press by announcing the launch of the first human trial of their HIV vaccine candidate. The company trumpeted data from macaque experiments as the rationale for moving forward with their "novel" candidate, but details of these animal studies were sparse. So, what are the data? And what exactly is in this "unique vaccine candidate"? The answers are perhaps not as thrilling as GSK's press release might lead you to expect. The macaque experiments were conducted under the aegis of SmithKline Beecham (SKB), and the data has recently been presented at several meetings by researcher Gerald Voss, with the public debut occurring at the Vaccines & Immunotherapy Meeting that took place in Puerto Rico in May of 2001.

The vaccine itself is a mix of three recombinant clade B HIV proteins: gp120 and a nef/tat "fusion protein." These proteins are administered in a proprietary adjuvant (immune stimulator) called ASO2A, which is a mix of oil/water emulsion and two experimental compounds called 3D-MPL (derived from an immune-stimulating part of the salmonella bacteria) and QS21 (derived from the soapbark tree). In an SKB-sponsored phase I human study using AS02A (then known as SBAS-2) with just the gp120 protein from HIV-1, the adjuvant improved T-helper and antibody responses when compared to a traditional adjuvant called alum, but adverse events were also reported to be more frequent and more severe (see Vaccine 2000;18(13):1166-77). In a larger, recently published study of GSK's experimental malaria vaccine that includes AS02A (see Lancet 2001;358:1927-34) the most common adverse events reported were pain, headache and malaise, and the investigators described the adjuvant as "well tolerated."

At the Puerto Rico meeting, Gerald Voss presented two separate macaque experiments conducted with the HIV vaccine. The first study utilized six groups of four rhesus macaques each:

2gp120/nef/tat/SIV nef/AS02A
3nef/tat/SIV nef/AS02A
4gp120/nef/tat/SIV nef/AS06*
5nef/tat/SIV nef/AS06
6controls (AS02A alone)
*AS06 was another experimental adjuvant, now discontinued.

Somewhat confusingly, due to the differences between HIV's nef and the analogous protein from SIV, any animal receiving nef/tat was also administered a homologous SIV nef protein (for the other vaccine antigens, Voss stated that the gp120 was approximately 20% different from the protein in the challenge virus while the tat protein differed by about 10%). Immunizations were given at months 0, 1 and 3. Thirty days after the final shot all animals were challenged with an SIV/HIV hybrid virus called SHIV89.6P.

Voss reported that, while no animals were protected from infection, all four animals that received gp120 and nef/tat in ASO2A recovered CD4 counts after an initial dip and appeared to have good control of viral replication out to 18 months of follow-up. Three of four control animals rapidly lost CD4 cells, developed symptoms of simian AIDS and were euthanized, consistent with the rapid disease course that can be induced by SHIV89.6P. Groups receiving gp120 alone or nef/tat alone did no better than controls. The group receiving all three antigens in the alternative adjuvant ASO6 remained healthy but 3/4 had persistently high viral loads. The only statistically significant differences in terms of CD4 counts and viral load were between the group receiving all three antigens in ASO2A and the controls. The observation that all three proteins were required to see any effect was mysterious to the researchers, with Voss stating in Puerto Rico "we don't have any solid explanation for this." Voss did delineate the rationale for including nef and tat, which was based on their early expression in the viral life cycle and relative conservation among different HIV strains. However, he did not offer an explanation for the absence of other commonly targeted HIV proteins, such as gag, from the vaccine construct.

In a second "confirmatory" study, Jonathan Heeney's group in Leuven immunized and challenged on a similar schedule. There were four groups of six animals each. Again, animals receiving nef/tat were also administered a homologous SIV nef protein. Five of six animals receiving the full vaccine showed relative preservation of CD4 cells but one animal experienced a clear decline. Control of viral load was variable and not as robust as in the first experiment. None of the control animals experienced the characteristic "crash" in CD4 cells that is associated with SHIV89.6P infection and 4/6 ultimately controlled their viral load. It also appeared from Voss's graphs that a group that received tat alone also controlled viral load well. There were no statistically significant differences between vaccinees and controls in this second study. Voss was unable to explain the differences between the first and second studies of the vaccine, although he stated that Heeney used a slightly different (but unspecified) subspecies of rhesus macaque than that employed in the first experiment.

The only immune response data Voss presented were CD4 T-helper cell proliferation and production of interferon-gamma or IL-5 (using an ELISpot assay) in response to vaccine antigens. While responses could be detected in most animals after immunization, they were variable and poorly maintained. Most disturbingly, both proliferative and ELISpot responses declined post-challenge, and loss of virus-specific T-cell immunity is associated with eventual progression in both SIV-infected macaques and HIV-infected humans (as a contrast, Harriet Robinson's recent studies of a DNA/MVA vaccine showed almost complete preservation of proliferative responses after challenge -- see Science 2001;292[5514]:69-74). No cytotoxic T-lymphocyte (CTL) data are available from GSK's macaque studies. Voss reported that neutralizing antibodies were rarely detected and "were not important for the protection seen in this model." It actually remains an open question as to whether the vaccine can induce CTL at all, since the earlier human study using the same vaccine with gp120 alone was unable to detect any CTL responses. Voss stated that GSK was planning new macaque challenge experiments with more detailed analyses of cell-mediated immune responses to the vaccine, but as yet there have been no reports of such studies being conducted. The two studies covered here also remain unpublished.

Despite the recent promotional activity, it appears that GSK's HIV vaccine is an outlier amidst the new generation of candidates that can induce both T-helper and CTL immunity against HIV antigens. At the very least, more data is needed before its potential can be rationally assessed. Rumor has it that SKB had shelved the product due to the dubious animal results and, if so, it is unclear what might have prompted GSK to dust it off now for new human trials.

Related articles: