Autism and brain inflammation: A study by Theoharis Theoharides used and abused by antivaxers

With antivaxers, it’s always, first and foremost, all about the vaccines, which in their belief system cause autism, developmental delay, autoimmune diseases, diabetes, sudden infant death syndrome (SIDS), and all manner of other health issues and diseases. In supporting this belief, antivaxers are like the proverbial drunk walking down the street at night, using the lamppost of science not for illumination, but rather for support of his beliefs. Yes, antivaxers are experts at cherry picking and misinterpreting scientific studies. They’re also very good at co-opting any scientific finding that they can spin to give their beliefs even the patina of plausibility. Towards the end of last week, I witnessed that very thing based on the publication of a recent study of autistic brains published last week in the Proceedings of the National Academy of Sciences (PNAS). This led Kate Raines over at The Vaccine Reaction to comment, as well as a fair number of Twitter antivaxers, for example, to start waxing all “science-y” about a couple of cytokines, IL-18 and IL-37 and a study by Theoharis Theoharides and Susan Leeman at Tufts University and Irene Tsilioni, their collaborator at the University of Chieti in Pescara, Italy:

Let’s look at the study. Before I get to meat of the study itself, though, I couldn’t help but note that that one name sounded vaguely familiar to me when I first looked at the paper. So I did what I always do when I see a name that sounds familiar to me. I searched my blogs to see if I had ever written about him before. Guess what? I had! And guess what else? Theoharis Theoharides received a bit of my not-so-Respectful Insolence back in 2013 for peddling an “anti-inflammatory” supplement to treat autism. He was associated with something called AutismFreeBrain (whose website now features cannabis “research” but whose original can still be found at, thanks to the almighty Wayback Machine). Let’s quote a blast from the past, straight from what used to be the AutismFreeBrain website:

Putting an End to Autism by Fighting Brain Immunity Storms™

AutismFreeBrain, Inc. was created to fund innovative research to develop a cure for Autism Spectrum Disorders (ASD). Our studies have identified inflammatory processes in the brain, we called Brain Immunity Storms, that are much like an allergic reaction, releasing surges of molecules that disrupt areas of the brain responsible for emotion and language.

“Fighting brain immunity storms”? As I noted at the time, whenever you see a website like this asserting that neuroinflammation is, in essence, the be-all and end-all of the pathogenesis of autism, you know that there’s likely to be some serious woo involved. It was true six years ago and is still true now that there have been a number studies suggesting an association of markers of inflammation with autism, but the significance of these observations hadn’t been worked out then and still haven’t been worked out. Not that any of this deters antivaccinationists from glomming onto studies like the one I’m about to discuss. The term “inflammation” is a term much beloved and abused by quacks of all stripes, and autism biomeddlers are no exception. This tendency to view autism as an inflammatory condition probably has a lot to do with the general antivaccine leanings of most such practitioners. Basically, inflammation is a convenient means for them to link their two greatest hatreds: vaccines and autism.

This wasn’t Theoharides’ only foray into dubious treatments for autism. He also hawked a supplement called Neuroprotek. It wasn’t even Theoarides’ only foray into cure-all supplements, as he also hawked a supplement he called CystoProtek (to treat interstitial cystitis) and another called Prostaprotek, to treat us old dudes at risk for prostatic hyptertrophy. Indeed, Algonot, Theoharides’ company, was even hit by an FDA warning letter in 2011 for marketing those supplements to treat disease. In any event, as I noted at the time, Theoharides has a very good publication record, nearly 400 publications on a PubMed search that I did yesterday, and many of those publications appear to be respectable. He’s also still the director of the Molecular Immunopharmacology and Drug Discovery Laboratory at Tufts. Unfortunately, he’s lent his name to some dubious supplements and antioxidant concoctions to use to treat autism based on his apparent belief that autism pathogenesis involves neonatal mast cell activation.

So with that background, it is perhaps not so odd to find Theoarides publishing a paper like this in PNAS. I also note that this paper was communicated by his coauthor, Susan E. Leeman. Back in the day, members of the NAS could basically publish anything they wanted to in PNAS by simply communicating it and then picking their peer reviewers, and early in my scientific training PNAS had a reputation as a bit of a dumping ground for NAS members and their friends, for whom they could communicate up to two papers a year. These days, PNAS is much more of a standard peer-reviewed journal, but, even so, I always have a bit of suspicion regarding papers communicated by NAS members for their friends or co-authors.

The paper itself utilized human brain tissue from the NIH NeuroBioBank at the University of Maryland from deceased Caucasian boys aged 3-14 years old, eight with an autism spectrum disorder (ASD) diagnosis and eight neurotypical children. Perusing the methods section, I already saw one big problem:

The only inclusion criteria used were males, 1 to 12 y of age, who had died in car accidents. Unfortunately, there is no available information of how diagnosis of ASD was reached, what the level of cognitive or functional level was before death, or the presence of any comorbidities. Controls were selected without any known brain disease or trauma and were matched to the subjects with ASD to the extent possible as shown in SI Appendix, Table S1.

So right away, we see that this is a very small study and that there’s basically no clinical information about the patients with ASD to tell how severely they were affected, whether they had other comorbidities, such as other developmental delay, seizure disorders, etc. In any event, Theoharides and his investigators did Western blot and isolated RNA for quantitative reverse transcriptase PCR to measure protein and RNA levels, respectively. The authors decided to investigate the levels of two cytokines in particular, interleukin-18 and -37 (IL-18 and IL-37), and they decided to concentrate on the amygdala and prefrontal cortex because of evidence in human and animals connecting the amygdala to social behavior. As for the cytokines studies:

Interleukin-37 (IL-37, previously known as IL-1F7) belongs to the IL-1 family of cytokines (29) and is a natural suppressor of inflammation (30–32). Five isoforms (a–e) have so far been identified with the “b” isoform being the most well studied (33). IL-37 is produced mainly by activated macrophages in response to Toll-like receptor (TLR) activation. An IL-37 precursor (pro– IL-37) is cleaved by caspase-1 into mature IL-37, some of which (∼20%) enters the nucleus and the rest is released along with the pro–IL-37 outside the cells (34) where both are biologically active. Extracellular proteases can then process pro–IL-37 into a much more biologically active form as shown for the recombinant IL-37b with the N terminus Val46 (V46-218) (35).

Although no specific receptor for IL-37 has been identified, a number of studies showed that extracellular IL-37 binds to the alpha chain of the IL-18Rα (36, 37), but with much lower binding affinity than that of IL-18 (38). Moreover, IL-37 binds to an IL-18 binding protein (IL-18BP) (39), and to the decoy receptor 8 (ILR8) (40) via which IL-37 inhibits innate inflammation (35, 41, 42) in vitro and in vivo (42).

To boil it all down, IL-37 is anti-inflammatory, blocking the action of IL-18, which is pro-inflammatory. I can also summarize the findings of the study fairly easily. The authors found higher levels of messenger RNA for IL-37, IL-18, tumor necrosis factor (TNF), and the receptor for IL-18 in the amygdalas of autistic brains than in neurotypical brains. However, if you look at the graph, you’ll see that there is a lot of overlap between the groups:

The same is true of the dorsolateral prefrontal cortex:

The authors also found lower NTR3/sortilin protein levels in amygdala and dorsolateral prefrontal cortex of children with ASD than in non-ASD controls. This is a receptor for neurotensin. To be honest, I wasn’t clear on why the authors even bothered with this other than that because neurotensin stimulates microglial cells, which are a specialized population of macrophages that remove damaged neurons; so supposedly a lower level of microglial activation would allow damaged neurons to remain in the areas of autistic brains studied.

Before I go into the significance of Theoharides’ paper (which isn’t much), I also can’t help but point out some rather poor statistics. In the cell culture experiments in which the authors treated cultured microglial cells with IL-37 in the presence and absence of neurotensin, they appear to have used Student’s t-test to compare groups for statistical significance. That’s a big no-no in an experiment with more than two experimental groups. (There were five, counting a positive and negative control in each of this set of experiments.) That’s the second time yesterday I saw that particular error in a manuscript, an error that is a particular pet peeve of mine, because earlier in the day I had reviewed a manuscript sent to me by a journal and it had the same statistical issue. Seeing such a shoddy statistical mistake reminds me just how badly some biomedical researchers are with statistics. Even I know they should have used ANOVA or another. Moreover, I always wonder: Why didn’t the peer reviewers catch such an obvious mistake?

Be that as it may, the cell culture experiments don’t really add anything to the paper anyway; so let’s go to the authors’ interpretation of their results:

Microglia are responsible for innate immunity of the brain (46, 47). Recent evidence indicates that brains of children with ASD have activated microglia (25, 26, 48, 49). The increased gene expression of TNF, IL-18, and IL-18R reported here supports the presence of inflammation in the amygdala and dorsolateral prefrontal cortex of children with ASD. We also show that the gene expression of IL-37 is increased in these same areas, but the reason for this increase is not clear. A speculation why this might occur is that IL-37 gene expression may be increased in an effort to suppress the inflammation in that part of the brain. Increased IL-37 gene expression was reported in the brain of patients after ischemic stroke and protected them from further inflammatory brain injury (50). Other studies also showed elevated serum IL-37 concentration in patients with sepsis (51) and in ankylosing spondylitis (52).

Overall, Theoharides and colleagues propose a model in which increased expression of IL-18 and its receptor indicates inflammation in the amygdala and dorsolateral prefrontal cortex. Neurotensin released in these areas or entering the brain through the blood brain barrier stimulates microglia primarily in the amygdala through activation of the NTR3/sortilin receptor. (Except that the authors’ results show that NTR2/sortilin receptor is lower in the relevant areas of the autistic brains that they examined, and the only reason to mention neurotensin entering through the blood-brain barrier seems to be to invoke a source of neurotensin from the gut, given that since Wakefield gut inflammation is supposed to contribute to autistic symptoms.) According to the authors’ model, microglia then exhibit abnormal synaptic pruning (the process by which synapses are “pruned” to the proper number) and secrete IL-1β and CXCL8, which contribute to focal inflammation resulting in direct damage to neurons and disrupted connectivity between them that contribute to the symptoms of ASD. IL-37 is increased in these brain areas in an effort to inhibit the release of the proinflammatory molecules thus providing a potential treatment option for ASD. So to boil it all down, the idea is that the pro-inflammatory cytokine IL-18 is elevated, supporting the presence of inflammation, but that IL-37 is also elevated, potentially to try to block the inflammation caused by elevated IL-18, but for whatever reason it’s not elevated enough to block the action of IL-18. Why authors seem to prefer IL-37 as a target for treatment, I don’t know. It’s usually easier to block the action of a protein than to augment it. Why not try to block the action of IL-18 instead?

Basically, it’s all a lot of handwaving and speculation on the part of Theoharides and colleagues.

There is, of course, evidence suggesting a role for inflammation in the pathogenesis of autism and ASD, thus leading antivaxers to “reason” thusly: “Aha! There’s evidence of inflammation in the brain in autism. Vaccines can cause inflammation! Therefore this evidence of inflammation must indict vaccines as a cause of autism!” Unfortunately for them, the problem is that we really don’t know whether the inflammation is an epiphenomenon (i.e., a secondary effect or byproduct that arises from but does not causally influence a process) or part of the actual physiologic process that results in ASD. We do know, however, that there is a large genetic component to autism and autism spectrum disorders. We also know, with about as much certainty as it is possible to have in biomedical science, that vaccination is not associated causally with the development of autism. A small paper with unconfirmed findings mixed with a whole lot of speculation won’t change that.