Field of Science

On autism, sAPPa, immunity, and subsets

via Wikimedia Commons
A new paper is making the rounds, purporting to link autism, a protein called sAPPa, and immune function. The paper, published in the journal FASEB**, isn't about autism, however. It's about a potential link between elevations in a protein called sAPPa and its effects on immune factors. Lead author A.R. Bailey and the other authors produced mice that make a whole lot of sAPPa in specific organs and found that these mice showed compromised production of specific immune cells and molecules compared to animals not overexpressing the protein. Period. They did no work specific to autistic behaviors and did not assess the mice overexpressing sAPPa for such behaviors.

Scientists use the Introduction of a scientific paper to lay down their rationale for the work they're doing. In the Introduction to this paper, the researchers in this work first sought to link autism and immune function. To do so, they cite two papers. One is a 2010 review of existing literature that refers to autism as a "devastating disease," apocalyptic terminology that does little good for autistic people or their families.

That 2010 review (full text here) covers a great deal of ground, with abundant citations of studies linking autism with altered levels of various molecules associated with inflammation responses. What remains unclear is whether these changes are causative in autism or the result of the stress and anxiety that often accompany autism. The review authors describe new "exciting" research findings showing that immune molecules called antibodies from mothers of children with autism appear to target the fetal brain, but these findings apply to a "subset of mothers of children with autism." You'll see that word subset more and more as researchers parse the likely multiple variants of the developmental difference we generally call "autism." 

The other paper the FASEB authors cite to support the immune link to autism is from 2004. That's the groundwork they lay for an immune connection. 

In the next paragraph, the Bailey team offers their rationale for focusing on sAPPa. The background consists primarily of this study. It is quite similar to this study, published in PLoS ONE, with both papers reporting that in a subset--there's that word again--of patients with severe autism and aggression, sAPPa levels are increased. The sAPPa protein is related to APP, a molecule that has been central to research into the causes of Alzheimer's and associated with other brain derailments. In this second paragraph, the FASEB authors also link sAPPa to immunity, noting that immune cells pump it out when they're stimulated, although the studies they cite focus on Alzheimer's, which the authors seem to suggest is, molecularly speaking, the opposite of autism. At the end of this second paragraph, they say that sAPPa is being considered as an autism biomarker, a molecule that would identify a person as having or not having autism. 

The issue with that concept is that one of the two studies (full text here) cited to support it focused on a subset--there's that word again--of autistic people. The subset consists of children with severe autism and aggression, as in the PLoS ONE and Sokol studies. The other cited study compared sAPPa levels in 25 autistic children and 25 non-autistic children and found relatively elevated levels in the autistic group. The authors of that last study also identified sAPPa in umbilical cord blood, but as they established no link between that and autism in the owners of said umbilical cords, the relevance remains unclear.

The authors of the PLoS ONE study, which involved children whose average ages were 6 to 8 years,  note that their sAPPa findings "do not extend to children with mild or moderate autism."

Finally, the FASEB authors work to thread it all together, suggesting that because immune changes are linked to autism, sAPPa is linked to autism, and sAPPa is linked to immune effects, then a sAPPa influence on immunity might be a cause of autism. To test that idea, they created a mouse that made a lot more sAPPa than normal in the brain, thymus, and spleen and found that more sAPPa in these tissues resulted in notable changes in various immune measures. 

This work addresses a potential link among these factors. But then the FASEB authors reach and overstate. The final statement of an Introduction sometimes summarizes the authors' take-home message, and that applies in this case. They write:

We conclude that sAPP- is instrumental in modifying T-cell development and function, and therefore may be directly involved in the pathophysiology of autism.
Except for a couple of problems. Nowhere in this work do they establish that elevated sAPPa in these mice results in behaviors that characterize autism, which is diagnosed on behavioral characteristics. All of the work in the paper focused on looking at the immune outcomes. Nor has there been a previously established unequivocal connection between autism and T-cell development and function. Finally, sAPPa has been linked to severe autism with aggression, not to autism in general, which encompasses a minutely variegated spectrum of manifestations.

The Discussion of a scientific paper is a place for authors to note the shortcomings of their work. The Discussion of this paper includes no such assessment. It does, however, include some more overstatement of the findings.

Our findings of impairment in the T-cell memory function as a result of high levels of sAPP- may explain why children with autism are more susceptible to infection (57, 58).
The two references they provide don't support the assertion that autistic children are "more susceptible to infection." The first study, from 2006, based on the abstract, actually found that while autistic children had more ear infections, non-autistic children had more "illness-related fevers." The other paper they cite, a Danish cohort study from 2010 (full text here), noted that the associations they identified did "not suggest causality" and that the association was not "specific for infection or for autism spectrum disorders." In their paper, the cohort study authors note again the increased incidence of ear infections, and offer three non-immune-related possible explanations for why autistic children might present more often at the hospital.

Finally, the authors of the sAPPa paper close with the following overstatement:

These findings are important because they substantiate the strengthening hypothesis that sAPP- is involved in the initiation of the T-cell response, and they provide a theory for the possible association between two phenomena observed in patients with autism: elevated levels of sAPP- and aberrant T-cell immunity. Future studies investigating this potential connection are imperative and will lead to the necessary development of individualized treatments for the unique subset of patients with autism for whom these studies are relevant.
The authors do not define the unique subset in question, but as the papers they cite illustrate, the "unique subset" is children with severe autism and aggression. They're the population with elevated sAPPa levels in the small studies done to date. As for abberrant T-cell immunity, as I noted above, that's not an unequivocal association with autism. I think the authors missed a very large, important boat in these conclusions, but I'll keep what that is in my pocket until the end of this post.

And then...

There is this news release associated with the FASEB paper. If you agree with me that the paper authors reached quite a bit in stating their conclusions, get a load of what the news release says.

First, there's the headline: "Autism may be linked to abnormal immune system characteristics and novel protein fragment." Except that nowhere in this paper do the authors make this link.

Then there's the dek. "University of South Florida researchers made the discoveries using mouse models of autism." Nope. They used a mouse model of overexpression of sAAPa. Using a mouse model of autism would not have helped them address the research question they posed unless they reversed the parameters.

Nowhere does this news release mention a "subset" of autism. In fact, the opening salvo refers to "immune system abnormalities that mimic those seen with autism spectrum disorders." Plural. The implication is that there are clearly defined immune abnormalities in autism that can be mimicked. But that's not the case. There are hints. Signs. Signposts. Outcomes that in some cases neatly dovetail with the FASEB paper's findings...and some that do not.

The release states that "the USF researchers concluded that the protein fragment might be both a biomarker for autism..." Not at all. They didn't conclude that in the paper, noting only that sAPPa was already under consideration as a biomarker for autism. Specifically severe autism with aggression. 

Then things get rather personal. The chair of the department where the team did the work is quoted as saying, "The prognosis for autism is poor." What does that mean? Do people die from it? Will the autistic children whom I know today never mature, develop, move forward? That statement alone propagates an image of autism akin to malignant melanoma, and as with "devastating disease" does no good for autistic people. The chair is also quoted as saying that this work "may lead to earlier diagnosis and more effective treatments." That's beyond speculative, and at best, it would be diagnosis of the subgroup that shows elevated levels of sAPPa. That's important, but I'm still keeping that one in my pocket for last.

The release goes on to say that the research team inserted the gene for sAPPa into the genome of the mice. Not really. The gene was turned on only in spleen, brain, and thymus. This work does not indicate what the outcome would be were the gene expressed organismally. 

In what ultimately is the best description of this work, the release quotes one of the study authors as saying that they used molecular biology and immunochemistry techniques to characterize immune cell development in the thymus and spleen function in the genetically altered mice and compared them to normal mice. That is exactly what the authors of this paper did. There is no characterization of autism in this work. 

That doesn't stop the news release from closing with a quote from study author Jun Tan, saying, "Our work suggests that the negative effects of elevated sAPPa on the adaptive immune system is a novel mechanism underlying certain forms of autism." Another overstatement that's sure to set the interwebz aflame with fingerpointing at vaccines as triggers for autism. But then he offers the real conclusion: "The findings also add support to the role of sAPPa in the T-cell response." This conclusion finds support in the paper results. Anything having to do with autism does not.

What would I have liked to have seen from this paper and the accompanying news release? More precision and accuracy. A clear emphasis on the real findings of the paper, the effect of sAPPa on immune endpoints. If autism had to be mentioned, the fact that sAPPa is related only to a "unique subset" of autistic people should have been made quite clear in both the paper and the news release. Finally, the news release should not--in the headline, quotes, or other text--have used the word "linked" in association with autism, as no aspect of autism itself was investigated in this work. Doing so simply leads to flame headlines, inflaming controversies, misapplication of science, and further muddling of people's perception of autism.  

The work itself is cool stuff. Had the authors and the news release noted the closer relevance to the "unique subset" of autism, the marketing of the work would have been more accurate. Frankly, the findings would have been of greater interest had they specified the subset, given the growing recognition that there are likely many "autisms," rather than a monolith that some--including in these publications--have been calling it. Any work teasing out these different strands of the autism monolith is forward looking and worth emphasizing.

Without ever really acknowledging or emphasizing it, the authors here may have added to a growing body of work that could lead to stratification--identification of subsets--of autism into what are clearly different forms. How immensely helpful that would be in terms of people with severe autism and aggression. What a useful contribution to testing and treatment for severe autism and aggression. Yet somehow, that important aspect, that target audience, went overlooked in favor of big, splashy references to what remains an eyeball magnet: Autism. Just...autism.

Aberrant T-lymphocyte development and function in mice overexpressing human secreted amyloid precursor protein-α: implications for autism; A. Bailey, H. Hou, D. Obregon, J. Tian, Y. Zhu, Q. Zou, W. Nikolic, M. Bengston, T. Mori, T. Murphy, J. Tan; The FASEB Journal; published online Nov. 15, 2011


  1. I accept your reservations about the misrepresentation of the study. I am a little concerned that you find the phrase "this devastating disease" so reprehensible. It may be that you pick it out for particular reprobation because it is easily understood by lay people. The autistic spectrum certainly encompasses people for whom "devastating" would be an accurate description. "Disease" probably has a definition which may be problematic in some cases. Maybe "what can be a devastating condition" would be acceptable to you? I know that for many autism is not at all devastating and adds quirkiness to an entirely acceptable and successful personality. I fully accept that and have no wish to challenge it. I simply think it unkind when alluding to the non-devastating presentations of autism to imply that that is all there are. Some people's lives are devastated and alluding to that should perhaps not be condemned as "terminology that does little good to autistic people and their families". The devastated are the most vulnerable of the autistic community. Remembering them does no harm.

  2. My concern with fatalistic language about autism is linked to (a) what I hear from parents of children with intense autism who find that kind of terminology harmful to them and from autistic people with intense autism who strongly resent it, and (b) that it can drive a level of concern that itself can range from being obstructive to progress to having catastrophic and devastating effects, e.g.,

    That is why I noted that this terminology does no good to autistic people and their families. How can it help to promulgate a perception of autistic people--especially severely autistic people--as "devastating" and "diseased"? Is it not so much more complex than that? Are they not capable of development, advancement, growth, and maturity? "Devastation" implies hopelessness. It is exactly the most severely affected of whom I think, and I think that this catastrophic language *does* harm them; it makes them seem like hopeless, near-fatal cases, so why bother?


Comments for this blog are closed.

Note: Only a member of this blog may post a comment.