HIV-1 CTL Vaccine: $%@#!

When HIV-1 was first discovered, scientists were optimistic about a vaccine. Small pox, polio, measles, meh– HIV-1 is a virus. We can handle viruses. Give us 6 months. No prob.

Well, >25 years later and we still dont have a damn vaccine. Strategies that worked against other viruses? HIV-1 doesnt bat an eyelash.

So weve been trying to think of new approaches to vaccine design. One of these ideas was to paste HIV-1 genes into adenovirus vectors. Adenoviruses are double-stranded DNA viruses, normally give you a cold-like illness, or make you a little poopy. Not real pathogenic. The idea is, if you can get an adenovirus to make HIV-1 proteins (like gag and pol), you can get a cytotoxic T-cell response to HIV-1, without infecting anyone with HIV-1! Used to be, the only way you could get a CTL response is through live attenuated virus!

And apparently, the time we have to create a CTL vaccine is limited… this is a pretty sweet idea!

Except it wasnt. Scientists put the HIV-1 genes in adenovirus serotype-5. Because adenoviruses are rather ubiquitous, some people who got the vaccine already had Ad5 antibodies. Well, no big whoop… except it turns out people who already had those Ad5 antibodies were 2.3 times more likely to get infected with HIV-1 than people who didnt get the vaccine.

* In volunteers who could be considered as having no pre-existing immunity to Ad5 at enrollment (baseline Ad5 titer less than 18 units), 20 cases of HIV infection were seen in the 382 volunteers in the vaccine group and 20 cases were seen in the 394 volunteers in the placebo group.
* In volunteers with immunity to Ad5 between 18 to 200 units, eight HIV infections were observed in the 140 volunteers in the vaccine group compared to four infections in the 142 volunteers in the placebo group.
* In volunteers with immunity to Ad5 between 200 to 1000 units, 14 cases were seen in the 229 volunteers in the vaccine group and seven cases were seen in the 229 volunteers in the placebo group.
* In volunteers with very high levels of immunity to Ad5, defined as greater than 1000 units, seven cases of HIV infection were observed in the 163 volunteers in the vaccine group and two cases were observed among the 157 in the placebo group.

SHIT!

This was a really good idea, so what went wrong??? Some scientists hypothesized that the vaccine activated a bunch of Ad5 CD4+ T-cells, thus created more nommynommy targets for HIV-1 once the people were exposed. This could be fixable, maybe. Make a collection of vaccines with different adenoviral vectors. Serotype everyone– they get the vaccine with the vector theyve never been exposed to.

But turns out that this probably wasnt the problem with the vaccine. Retrospective analysis of the Merck STEP data indicates that there were not more Ad5-reactive CD4+ T-cells in either population 8 weeks post-vaccination.

Adenovirus-specific immunity after immunization with an Ad5 HIV-1 vaccine candidate in humans– The immunologic basis for the potential enhanced HIV-1 acquisition in adenovirus serotype 5 (Ad5)-seropositive individuals who received the Merck recombinant Ad5 HIV-1 vaccine in the STEP study remains unclear. Here we show that baseline Ad5-specific neutralizing antibodies are not correlated with Ad5-specific T lymphocyte responses and that Ad5-seropositive subjects do not develop higher vector-specific cellular immune responses as compared with Ad5-seronegative subjects after vaccination. These findings challenge the hypothesis that activated Ad5-specific T lymphocytes were the cause of the potential enhanced HIV-1 susceptibility in the STEP study.

Baseline Ad5 serostatus does not predict Ad5 HIV vaccine-induced expansion of adenovirus-specific CD4+ T cells– The mechanisms underlying possible increased HIV-1 acquisition in adenovirus 5 (Ad5)-seropositive subjects vaccinated with Ad5-HIV-1 vectors in the Merck STEP trial remain unclear. We find that Ad5 serostatus does not predict Ad5-specific CD4+ T cell frequency, and we did not observe durable significant differences in Ad5-specific CD4+ T cells between Ad5-seropositive and Ad5-seronegative subjects after vaccination. These findings indicate no causative role for Ad5-specific CD4+ T cells in increasing HIV-1 susceptibility in the STEP trial.

Vaccine making more HIV-1 targets doesnt appear to be the right answer to what went wrong in the Merck STEP trial. We still dont know what happened.

God DAMMIT!

At this point, they dont even think the original vaccine worked in the non-Ad5 people. As time goes by, infections in the ‘vaccine’ group are catching up with the ‘placebo’. But researchers working on similar vaccines want… need to know what went wrong in this study so they can avoid it.

Comments

  1. #1 rrt
    July 24, 2009

    Wow. WTF?

    Okay, I’m sure this is wrong and all, me knowing not a lot about immunology, but the first thing that pops into my head is: Could the immune system be associating the virus with the proteins?

    So I’m an immune system, and I’ve seen AD5. Been there, done that, I know it’s no big deal and I know I’m better able to pounce on it when it does show up. Now I see adenovirus making HIV proteins and I (somehow) decide this means I shouldn’t worry when I see those. So when real HIV shows up, I don’t respond right away, or respond inappropriately. Is that even feasible?

  2. #2 dreikin
    July 24, 2009

    Those are some pretty low numbers for the infections. I wonder if there was actually no effect, and the highs are just a statistical fluke.

    Looking at the placebo groups, the first one has 20/394 infected, and the third has 7/229 infected. Do some math (7/229*394), and the third group has an equivalent to 12/394 infected. The difference between the first and third placebo groups is greater than the difference in each vaccine-placebo pairing.

    Not nice that it failed to provide a vaccination, but I don’t see a reason to believe it had any actual effect at all.

  3. #3 D. C. Sessions
    July 24, 2009

    Someday (I won’t speculate on when that is) this will all be “obvious” in retrospect. Some people would be comforted by that thought.

    I wouldn’t. Rather the opposite. I doubt that I’m alone in that.

  4. #4 rrt
    July 24, 2009

    I’m a little fuzzy on proper statistical analysis, but doesn’t the fact that each of three separate experimental groups showed similar results make it a lot less likely to be a fluke?

  5. #5 dreikin
    July 24, 2009

    rrt:
    So am I, and yes that would probably decrease the likelihood of a fluke – if humans were perfectly random. One of the press releases I looked at identified the infected as men (except 1) who mostly had sex with other men – so I’d at least like to know the distribution of gay+bi vs. straight between the vaccination and placebo groups.

    There could be other factors as well – were genotypes that are more susceptible to HIV infection more predominant in the vaccine group? What was the sexual activity (mates/year or some other metric) in each group and subgroup? What was the distribution of condom usage? Etc.

    ========
    On another note, I found this paragraph from the same page interesting (bold mine):

    Viral load levels were also evaluated across the total population, and there was no obvious correlation between levels of Ad5 immunity and viral load. In a post-hoc analysis of those male volunteers with high levels of immunity to Ad5 who developed HIV infection, the geometric means of HIV RNA levels were observed to be approximately 19,000 copies/mL in the vaccine group (based on 21 infections) and 90,000 copies/mL in the placebo group (based on nine infections observed). In the total study population, including male volunteers with both low and high levels of immunity to Ad5, the geometric means of HIV RNA levels were approximately 29,000 copies/mL for the vaccine group (based on 46 infections) and approximately 38,000 copies/mL in the placebo group (based on 30 infections).

    I have reservations about the actual significance of that as well, for the same/similar reasons, but it looks like the vaccine may1 have inhibited viral reproduction, even if it didn’t prevent infection.

    1: w00t! Power to the house went out at this point in my typing, but firefox + session manager saved the data. Praise be to open sores! Er, source..

  6. #6 rrt
    July 24, 2009

    Yeah…but I don’t think that changes anything. Of course there could be many confounding factors. But assuming the selection process was random and unbiased, multiple trials would show a problem like that. That’s one of the reasons you do them. If I’m understanding the design correctly and these actually were independent groups with their own controls, then the odds are lower that all three groups just happened to have the same confounder, or a mix of confounders, producing the same proportional effect. I think. Maybe.

  7. #7 Joe
    July 25, 2009

    So, this popped a mildly related question into my mind.

    Given how wildly successful (from the viewpoint of the virus) HIV is, why aren’t there more viruses like it? It seems to have a winning strategy for avoiding the immune system, why isn’t there a wide range of viruses all trying to do the same thing?

  8. #8 Stephen Wells
    July 25, 2009

    @7: there are other retroviruses out there. But to an extent it’s a bit like asking why everything isn’t a lion since lions are such good predators. It’s a matter of niches.

  9. #9 Bob O'H
    July 25, 2009

    Those are some pretty low numbers for the infections. I wonder if there was actually no effect, and the highs are just a statistical fluke

    Quite. I’ve just bunged it through an analysis, and there’s no evidence for an interaction between dose and treatment, and there’s only small evidence for an effect of vaccination: the odds ratio is 0.43 (i.e. vaccination increases the chance of infection by about 1.5 times), but with a standard error of 0.23. So there might be an effect, but it’s not well estimated: the study would have been as likely to have shown that vaccination reduces infection, as to have shown it increased infection 2.4 times.

    In reality, there is probably less evidence than this. Chance is lumpy, so there is usually more variation than a simple binomial assumption assumes. From the data ERV provides, it’s not possible to look at this in any further detail, although I guess the original researchers looked at this.

    Finally, another reason to have some healthy scepticism, from the report:

    Post-hoc analyses of the total STEP population, including those with higher Ad5 levels
    Extensive additional analyses have been and continue to be done to better understand the results from STEP. These are post-hoc exploratory analyses and are subject to the statistical limitations inherent in such analyses.
    (emphasis in the original)

    Basically, they’re saying that they were fishing in the data to see what might be happening, and found this, but they probably looked at a variety of other things. Again, chance is lumpy so you’re bound to find some pattern if you look hard enough.

  10. #10 J Todd DeShong
    July 29, 2009

    OK, I got my hissy fit out of the way from above…I read my blogs in batches and go backwards. Anyway, this post is great and is exactly why I read your blog. Very insightful. Well articulated. And a nice bit of humor thrown in. So, thanks.
    Also, have you seen this about an HIV Vaccine specific for Africa started trials with the first person injected July 20.
    http://www.scidev.net/en/news/trial-of-africa-s-first-hiv-vaccines-underway.html
    JTD

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.