When It’s Too Good to be True…..

Affiris‘ announcement that one of the two placebos they used in their AD02 antibody trial outperformed everything–including AD02–in stabilizing cognition and hippocampal volume takes Alzheimer’s drug development to a new level of BizarroWorld surprise. Just to recap: They reported that 47% of the 60 or so (300pts, 5 groups) patients receiving the nonclear (“obscure”) placebo were stabilized over eighteen months, compared with 17% of those receiving the ‘clear’ placebo (which is visually discriminable from the AD02 drug product), and 31% of those receiving the highest dose of AD02 (there was a dose-response relationship in the reported results). Affiris is happily puzzled (actually the CEO did not look as happy as one would expect given his claim of a “breakthrough”, he looked like he was testifying in front of a Congressional committee run by the party he does not belong to) is disclosing very little about the ‘obscure’ placebo, and may be exploring ways to nail down some kind of use patent for it.

There was understandable chuckling amongst some observers about the value of ‘sugar-water’, and indeed we’ve had a few inquiries as to why we had not yet chimed in. The first reason was that having this kind of completely counterintuitive–indeed counterlogical–finding risks reinforcing the not-uncommon belief that CNS R&D is so riddled with such incongruities that its pursuit is economically and strategically unwise. Frankly, we would have been happier had Affiris simply had the AD02 data in hand, reporting a smaller magnitude of effect than hoped for, but dose-related, raising the possibility that going up in the dose of antibody might give them clinically and statistically significant results (we presume that if the high dose AD02 impact in 31% of patients hit p=.05, we would have heard about it).

The second reason is that the placebo now known as AD04 was likely an immuno-active adjuvant, such as is commonly utilized in vaccine trials. Such adjuvants enhance the effects of the co-administered antibody, generally believed to be via stimulating immune cell recruitment via a proinflammatory effect. This was not an ‘inert’ placebo, but it was not expected to have an effect without the presence of an antibody to ‘magnify.’ We wanted to look into the adjuvant issue more carefully. So far as we can tell, there are four classes of adjuvants used in vaccine work: Aluminum salts/gels; squalene adjuvants; a saponin-derivative (QS-21); and virosomes. The one interesting and useful comment that was made during Affiris’ 27 minute press conference was that AD04 is an adjuvant that has not previously been used in Alzheimer’s trials, but has been used in influenza vaccine development. That rules out virosomes (used in testing Novartis/Cytos‘ CAD106) and QS-21 (used with bapineuzumab and solanezumab). Which leaves us with aluminum and squalene. The putative relationship between aluminum and autism has been the crux of actress Jenny McCarthy’s fleeting brush with the world of science, while squalene has been linked to the rare (but age-related, more likely in younger subjects) emergence of narcolepsy.

The bottom line is that neither of these adjuvant classes is inert, they are co-administered with antibodies for an immunological reason. But with all of the work done vis-a-vis inflammation in Alzheimer’s, one would think that if a very nonspecific proinflammatory chemical could have a very salient impact on AD neurodegeneration, that would have already been established.

Perhaps it could be argued that it is the duration of exposure/re-exposure  and monitoring that might be the differentiating factor here: The only intelligent question to be heard from the audience during the aforementioned press conference was in reference to the apparent year it took for AD04 to begin to have a stabilizing effect on cognition or hippocampal volume. For the first year, while the curve was slightly higher than those of the other groups, the slope of deterioration was essentially the same. At twelve months, hippocampal volume stabilized, and the composite cognition score actually went up. This just adds to the mystery.

These are small cohorts of about 60 patients each, which increases the possibility of flukiness. But the usual suspects when it comes to patient populations do not show up here, neither Russia nor India supplied patients, it was a European trial.

When a clinical finding is too good to be true, it generally isn’t (hello, Dimebon). And while we would like to believe that a nonspecific proinflammatory chemical, given long enough, can somehow provide neuroprotection in mild Alzheimer’s, our guess is that in the long run, there will be some other explanation for the performance of the ‘obscure placebo’ in this Phase II trial, rather than clinical efficacy. The sad thing is that this may so erode Affiris’ credibility that they will never get a chance to see if higher dosing of AD02–the actual antibody–might provide a larger magnitude of effect.

All in all, it’s more tragic than comical.




This entry was posted in Uncategorized. Bookmark the permalink.

One Response to When It’s Too Good to be True…..

  1. Steve says:

    Very nice anaylsis and I agree with most of the comments and conclusions. Remains just to clarify that AD02 is NOT an antibody but a vaccine!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s