Sunday, September 2, 2012

An analysis of the sources supplied for the NYT autism and inflammation op-ed

Moises Velasquez-Manoff has published his list of sources related to the NYT Op-Ed he wrote in which he suggested that parasitic worm therapy [would fix "maternal dysregulation," which] would “most likely” prevent autism. He promises also to “counter would-be rebuttals” of his piece soon, as well. Unfortunately, offering up citations and sources after the horse has left the barn isn’t that helpful for readers of the original piece. My comments below related to each study are not meant to diminish their importance or to critique design but to emphasize that the vast majority are correlational in nature. I focus my analysis on the context of an immune- or inflammatory-related cause of autism. Correlations are signposts only. What I like about many of these studies is that the authors use just the kind of qualifying language that would have been appropriate in Velasquez-Manoff’s original piece.

The tl;dr version of the below? Very, very little cited below points to a mechanism of immune dysregulation causing autism, but some of it points to immune dysregulation/response as a secondary manifestation of having autism in a subset of cases. This conflation of correlation and causation was a main problem of the original op-ed, which was not only my opinion. The other main problems, not including the overstatements and lack of sourcing, were the cherry-picking of results to support a pre-existing narrative (a common issue in popular science writing, at least of late), the absence of emphasis on subsets, and the lack of necessary qualification and nuance in discussing any of these findings. Those problems persist in the list provided, based on the comments that accompany each source. Given the limited relevance or repetition of some, I get the feeling some links were provided just to add bulk.

There is not a neat, tidy picture here of immunity and autism--even the findings within the field can contradict each other or point in different directions--and to suggest that such a tidy picture exists is inaccurate, at best. What we have here is a few parts to build the frame for such a picture. That’s all we have for a lot of research related to autism. It’s unfortunate for people who really want answers to the mystery that autism seems to be, but that’s the reality of autism research. Hell, that’s the reality of research into any complex condition.
Finally, I want to make clear that I am not even remotely trying to dismiss a link between autism and immune dysfunction. Indeed, our family history of autoimmune disorders (although not my son’s apparently hardy immune system and lack of allergies) makes such a thing anecdotally compelling for me. My issues with the original op-ed are unrelated to any absence of conviction on my part regarding a link. In a subset.

I should note that I was originally alerted to this posting by a commenter on my earlier critique who goes by the handle “Helminthic Therapy.” Who found it so important, s/he posted it twice.

The Annotation
Note: I have included in red boldface italics Velasquez-Manoff’s commentary that accompanied each link he provided.

SOURCE LIST (Velasquez-Manoff’s original subhead)
Comment: Full, open access paper here. This type of inflammation is, according to the authors, related to mitochondria and oxidative stress. Furthermore, of the 15 autistic brains, causes of death included seizures (3), heart failure, bruising of the brain, and sarcoma. Of interest, given the assertion of inflammation involvement here, the cause of death in four of the control group brains was asthma. [ETA, 4:45 EST: I point this out because Velasquez-Manoff suggests in his source list/annotation an "uncanny similarity" between prenatal inflammation in asthma and autism. Conflicts like this litter the source list and annotation, e.g., Th1 vs Th2 and autism, allergies and autism vs infection and autism, etc.]

*Another measurement of inflammation in brain tissue by different group
Comment: Full text of that study is here. This very small group of samples came from individuals who had died, among other things, from seizures and cancer. I’m not arguing that the information isn’t worth pursuing, but when the authors write, “Small sample sizes precluded valid formal between-group (i.e., ASC vs. control) statistical tests. Nevertheless, in many cases there was no overlap in the values obtained for each group,” that tends to signal “preliminary,” at best, which given the dearth of tissue samples is understandable. Overstating it in interpretation? Not so much.

Great review on ongoing inflammation in autism. Note that severity of symptoms often correlates with markers of inflammation.
Comment: And correlation doesn’t indicate causation. As with the CRP story I described here, we can’t tell from a correlation whether the condition results in the factor or the factor results in the condition.

Here’s a review on severity of symptoms correlating with markers of inflammation.
Comment: Full text is available here. This is not a review; it is a research paper from the same senior author as that cited above. The authors themselves state, “A potential role for immune dysfunction has been suggested in ASD.” I agree completely, and this kind of wording is exactly the sort that is needed to qualify just about anything we say about autism. The authors also note the relative dearth of studies looking at blood inflammation markers in autism and the various factors that have confounded the studies that do exist.

Also of interest from this study: “Moreover, when the ASD group was separated based on the onset of symptoms, it was noted that the increased cytokine levels were predominantly in ASD children who had a regressive form of ASD.” This represents a subset of autism. In fact, the authors found different inflammatory molecule profiles within the autism population itself.
* Another recent study showing that elevated levels of the pro-inflammatory IL-17 (which is implicated in many an autoimmune disease) correlates with symptom severity in autism.
Comment: Correlational again. And this study also notes that the majority of the different in levels of this cytokine were attributable to a subset of the autistic population in the study, those with what the authors describe as “more severe” autism. And again, as with all correlations: Does the intensity of the autism result in raised levels, or do raised levels cause the intensity of the autism? For example, autism is associated with a great deal of anxiety, and pro-inflammatory molecules are elevated in anxiety, including IL-17. Indeed, I could write an entire book building a narrative of autism-->anxiety-->inflammation.

* And another, this one focused on a different measure of immune activation
Comment: This paper is paywalled, but it involves a small group of adults (yes, there are adults) with autism spectrum disorder in Italy (22 autistic individuals) and describes a correlation. They seem to have done some pretty fancy statistical analysis for having had only 22 autistic people and 28 controls.

* One showing less of an anti-inflammatory factor called proganulin in autism
Comment: Full text of that paper is here. The authors use exactly the kind of language that we should all use when talking about autism: “However, these data should be treated with caution until further investigations are performed, with a larger subject population, to determine whether the decrease of plasma progranulin levels is a mere consequence of autism or has a pathogenic role in the disease.” Unfortunately, this particular molecule is linked to decreased neural connectivity in the brain, which is the opposite of what many have found to be the case in autism. It is, however, a very complex molecule with complex processing. That’s because it is now a Law that nothing potentially related to autism can be straightforward.

*And another free paper showing less TGF-beta, an important anti-inflammatory signalling molecule, correlates with symptom severity in autism.
Comment: Full text of this is available here. These authors state in their introduction, “However, it must be noted that it is not currently known whether these antibodies are a cause of autism or generated as a result of inflammation in the brain and CNS.” Of interest but generally not discussed in the paper except for a mention in the abstract and the methods, the levels of this protein did not differ (based on the abstract) between the autistic children and their siblings without autism; for some reason, the authors neither report this in the actual results nor elaborate on it. Also, for those of you into that sort of thing, this figure of the actual distribution of values among the autistic, developmental disorder, and control groups is worth a look.

But autism is complicated (it’s a Law). This paper discussing TGF-beta in autism seems to suggest that overexpression of the molecule just postnatally triggered what researchers like to call “autistic-like” behaviors in mice. And while the earlier cited paper found lower levels of TGF-beta in an autistic versus control group, they found no association with severity.

* Activation of both th1 and th1 arms of the immune system in autism.
Comment: I think what was meant here was Th1 and Th2. This paper is paywalled. But see, this is where it gets confusing. Th1 drives many autoimmune diseases--and autoimmune disease increase and in mothers forms one of the pillars of Velasquez-Manoff’s  thesis. But this paper found a “predominance” of Th2 (associated with allergies) in the autistic population. Meanwhile, another paper found a predominance of Th1. If you dig into the autism literature enough, you will find this situation of “that paper’s findings counter this paper’s findings.” The vaccine literature appears to be the only exception. That’s why “Autism is complicated” is a Law. Fun fact about Th1 vs Th2: A shift to Th2 appears to be involved in supporting a successful pregnancy (full text). Pregnancy, you see, triggers immune changes all on its own.

NEUROGLIAL ACTIVATION (Velasquez-Manoff’s original subhead)
Here’s the seminal paper noting neuroglial activation in “autistic” brain tissue
Comment: I mentioned this paper in my critique (I provided a link to the full text, which is here). As I noted, more than half of the 11 brains used in the comparative tissue study were from people who also had epilepsy. For the CSF study, all samples were from people who had “regressive” autism. The study is interesting, but again, because of the inherent limitations of sample size and the confounder of epilepsy and involvement of an autism “subset,” it was far from definitive. In fact, epilepsy and seizure disorder are common in these cohorts, and they also are associated with microglial activation and neuroinflammation.

Comment: This paper is paywalled. It is a paper looking at the morphology of microglia, brain-specific immune cells that activate in the presence of a challenge, of which there are many possibilities. It is not, as far as I can tell from the abstract, a replication of the above. Abstract says, "Microglia appeared markedly activated in 5 of 13 cases with autism, including 2 of 3 under age 6, and marginally activated in an additional 4 of 13 cases.” Again, a subset. And I can’t let this study go unnoticed: Did you know that “social defeat” (in mice) increases anxiety-like behavior and is associated with microglial activation?

* And again from a slightly different angle
Comment: Paywalled, but you can find most results in the form of tables and figures here. This is also not a replication or an “again” of anything previously offered. These authors looked at how microglia are organized in a part of the brain that’s known for neuronal overgrowth in autism and found more microglia associated with neurons than they found in non-autistic brains. Again, as is necessary with these post-mortem studies, the groups were small, and people had in life taken a long list of medications that have varying effects on microglial activation; for example, Risperdal seems to inhibit it, but drugs that inhibit serotonin transport appear to activate microglia and which often are used to treat anxiety.
*Evidence of disturbed neuronal migration in autistic brain tissue (also replication of glial activation).
Comment: Paywalled, but you can see first page of PDF here. Elevated levels of this protein are also associated with seizures.

GENETIC EXPRESSION IN AUTISTIC BRAIN TISSUE (Velasquez-Manoff’s original subhead)
Evidence of immune dysregulation at the transcriptome level in “autistic” brain tissue.
Comment: This is a cool paper (sorry, paywalled) that I’ve cited before. Neuroskeptic did a fine analysis of this paper here. A quote from that analysis:
Another cluster, "M16", was overexpressed; it contains genes involved in immune and microglial function (microglia are specialized immune system cells inside the brain). However, M16 did not contain overrepresentation of suspected asd genes. 
 Also, whatever the abnormality in autism is, it seems to trigger a secondary change in the brain which is immune and/or glial related. By "secondary" I don't mean that it's less important. It might be what causes the symptoms of autism. But it's not the root cause (because if it were, mutations in this network would cause autism, and they don't seem to.) 
This study raises many more questions than it answers, but in a good way. It certainly doesn't explain autism, but it's pointed the way towards more focussed research in the future - gene cluster M12.
*And again, with a focus on over-expression of inflammation-related pathways
Comment: Open access (yay!) available here. Perhaps nothing illustrates The Law of Autism Research better than this statement from that paper: “This underscores the importance of our findings on ASD-implicated genes, as both our approach and whole-transcriptomics studies implicate immune signaling pathways, even though most ASD-implicated genes we profiled are not dysregulated in ASD brain tissue.” Oddly enough, their results didn’t implicated immune-related processes in schizophrenia (!).

Comment: Open access. This would be one of the reasons why I’m surprised the previous paper got the results they did for schizophrenia.

CYTOKINE PROFILES IN AUTISM (Velasquez-Manoff’s original subhead)
Comment: I cited this paper above; full text is here. It’s actually showing that these molecules are skewed toward the Th1 (autoimmune) pathway (as opposed to Th2). I encourage anyone with sufficient interest to take a look at the figures and results text. I can’t figure out why the Th1:Th2 ratios are so low in all groups (?).

*More on pro-inflammatory markers in autistic spinal fluid (the central nervous system)
Comment: Again, correlation, not causation, in 10 children, and it’s one marker--tumor necrosis factor-alpha, a Th1-related molecule. Oddly enough, a previous analysis of the spinal fluid from 12 autistic children found elevated levels of a receptor for TNFa, but not of TNFa itself. Also, spinal fluid can’t be autistic.

Discussion of possibly treating the immune dysfunction in autism
Comment: I don’t see anything about worms in the abstract for this one. Several attempts have been made to target various aspects of putative immune dysfunction in autism. So far, little success has been reported. Also, we have cautionary tales from autoimmune diseases themselves (e.g., multiple sclerosis, rheumatoid arthritis, and TNFa therapy) about making generalized assumptions about immunity, correlations, and therapies, and in the therapeutic context in particular, we must tread very cautiously.

MATERNAL INFECTION IN AUTISM (Velasquez-Manoff’s original subhead)
Large Danish study linking maternal infections to autism
Comment: I mentioned this study in my critique; it was not referenced in the original article, but I inferred that this was the study in question. This is a large study, which makes it even more important to note that it focused on hospitalization for infections, not just having an infection. Indeed, the authors note that they found no association between maternal infection and an autism diagnosis in the child resulting from that pregnancy when looking at the entire period of pregnancy. Another study found an association between fever that went untreated in pregnancy and autism OR developmental delays, with a higher odds ratio for the latter. Oddly, they found no association between having had influenza and having a child with autism.

Two good reviews by Paul Patterson on, among other things, maternal infection and autism. Includes refs to his and others’ experimental work on this. The first one’s free. One
Comment: Full text is here. As Patterson notes in this review, “It is also clear that there is considerable heterogeneity among the autism samples, as might be expected from the extreme disparities in behavioral symptoms among ASD subjects.” I have no issues with this review, as Patterson appropriately uses modifiers and qualified language throughout and notes where there are gaps and discrepancies. It’s interesting to me that he couples autism and schizophrenia generally in the review, given that another recent study did the same, associating a father’s age--and accumulated mutations--with both. That does not seem to be related to the womb or maternal immunity status, although I could come up with a great “if this, then that” series of cards from which to build that house.

Comment: This review by Patterson is older than the above review and paywalled, so not much of an addition to the above.

Recent article in NEJM that covers changes in burden of disease in the US – i.e. the transition from infectious to degenerative disease.
Comment: The inclusion of this paper baffles me. It covers the change in what clinicians have seen in terms of patients and disease states, not in the context of a hygiene hypothesis or immunity changes, but in the context of, Gee, you just don’t see cannonball injuries in the clinic any more. Interestingly enough, in referencing what the NEJM addressed in its earliest issues 200 years ago, the authors mention both asthma and diabetes. The commentary does not reference immunity, immune, inflammation, autism, or the hygiene hypothesis, either directly or indirectly, except to mention that immunizations wiped out smallpox. It does not talk about “the transition from infections to degenerative disease.” It does mention that chronic conditions like heart disease and cancer have become more prominent as causes of death, although influenza makes the list as a top killer today.

MATERNAL AUTOIMMUNITY IN AUTISM (Velasquez-Manoff’s original subhead)
Large Danish study on autoimmunity in the family and especially in the mother, and the risk of ASD in offspring. More refs to come on this. Numerous smaller studies find the same, although sometimes with different autoimmune diseases. But this is largest to date.
Comment: Full text of that study is here. In the paper, the authors state, “We add to previous knowledge by suggesting that this increased risk is possibly limited to maternal history of RA.” Last week, I participated in a Web chat hosted by an autism-related organization with Paul Patterson and asked if he had any insight into the apparent mutually opposing natures of multiple sclerosis and rheumatoid arthritis (one rarely happens with the other** they can show an inverse correlation and therapies for one, based on tumor necrosis alpha, are contradindicated for the other). I did not receive a response to my question, but this finding with RA might be partially one.

The authors also report an association between maternal and paternal history of type 1 diabetes and “infantile autism,” and between maternal celiac disease and autism. Of importance, I note their comment:
We found no general association between all autoimmune diseases combined and autism spectum disorders or infantile autism. However, this was not a surprising result, because different pathogenesis and etiologies apply for different autoimmune disorders.
In other words, we can’t just sweep a hand at “autoimmune disorders are increasing” and then point to autism. As always, these findings are correlations and not indicative of causation or the nature of the relationship. 
Comment: Full text linked here.

And for HLA*DR4 in the mother especially
Comment: As someone who is an HLA increased risk variant queen, I am particularly interested in studies like this. But why doesn’t the mother, having inherited this from *her* mother, for example, have autism? Perhaps because autism is underdiagnosed in females.

And alleles associated with less production of regulatory (anti-inflammatory) cytokine IL10 in mothers who tend to produce those antibodies that bind to fetal brain proteins.
Comment: I’ve seen this in a few studies. 

*Also, interesting study on a link between a gene variant that makes macrophages, critical cells of the innate immune system, more aggressive, and autism. Note that the more circulating macrophage inhibitory factor a subject had (the more activated this aspect of innate immunity, essentially), the worse the symptoms.
Comment: Full text of this study here. It’s a pretty cool study, although four years old now. Unfortunately, a PubMed search on autism and MIF turned up no other hits. I also searched autism and macrophage inhibitory factor and found an ancient paper from 1982 suggesting that the autistic people showed inhibited macrophage response. Invoking the Law here.

FETAL BRAIN-DIRECTED ANTIBODIES (Velasquez-Manoff’s original subhead)
On those antibodies directed at fetal brain proteins showing up in mothers of autistic children
Comment: This study found differences in subsets and emphasizes, “The fact that these bands were not found in all mothers of children with autism further emphasizes the heterogeneity that is widely reported in autism, and the variety of etiologic mechanisms that likely exist.”

Comment: Paul Patterson notes in the review cited above that this study did not focus on the appropriate trimester of pregnancy; he writes, “The choice of 3rd trimester does not, however, fit with what is known about the windows of vulnerability for development of schizophrenia or autism.”

OTHER MATERNAL RISK FACTORS (Velasquez-Manoff’s original subhead)
On maternal allergy and asthma diagnosed during pregnancy associated with autism in offspring.
Psoriasis, an autoimmune disease, is also a risk factor in this study.
Comment: Collectively, correlational.

Comment: Another question I posed to Paul Patterson during the web chat was related to this and another study of the Danish amniotic fluid. I’m inferring--and I could be wrong--that these amniotic fluid samples are likely the same as those that Simon Baron-Cohen used in his analysis of androgens in amniotic fluid from Danish samples. He reported these findings at the IMFAR autism research conference, and I wrote about that here. My question to Dr. Patterson was whether or not these elevated androgens might have an influence on the inflammatory profile in the womb in autism. Given that androgens do influence levels of these molecules, and that it may be that these samples show evidence of elevations in both, I thought it was an interesting question. Dr. Patterson agreed, saying, “Yes, the potential interplay between androgens and inflammatory molecules would be a really important area for study in animals, given the male bias in ASD. Interestingly, we do see sex biases in some of our animal results in the maternal immune activation model. Thx for a key question. PHP

The remainder of Moises  Velasquez-Manoff’s annotation is related to “immunoregulation in modernity” and focused on non-autism expert Parker’s discussions about using helminth worms. I’ve said what I’ve got to say on that subject and want to focus here only on autism. One paper he cites that relates directly to autism is this one. It was published in Medical Hypotheses in 2007. I don’t really think I’ve got anything more to say about that.

**I've modified this to avoid the appearance of saying the two never occur together, as I know of a couple of people who do have both. 


  1. I completely wandered away from one of the autism-related references, having to do with autism in the developing world, which in this case appears to include the thousands-of-years-old Persian and Chinese cultures. At any rate, the cited paper full-text is here:

    The authors say the following about the reasons for the lower prevalence in the developing world. These explanatory factors suggest a heavy cultural and diagnostic influence on autism prevalences and prevalence differences globally, particularly those factors that would suppress actual prevalences. Cambodia is not mentioned.

    "Various factors might explain the lower prevalence of autism in less developed countries such as Iran. In Iranian culture, a diagnosis of disability is likely to be seen as stigmatizing [23]. As parents are keen for their child to attend schools for ordinary developing students rather than being referred to special schools, they may underreport the child’s difficulties to assessors even though they are aware of them. The screening tools that are used in Iran (as well as other countries) rely heavily on parental reports with limited time and opportunity for assessors to observe and interact with the child and for them to make consensus decisions [22]. By contrast, in Western countries parents may be eager to obtain a diagnosis for their child’s difficulties as this enables them to access additional services, whereas these are not readily available in Iran. On the other hand, Iranian parents will voluntarily seek out a diagnosis when they recognise the severity of their child’s problems [24].

    A second reason for lower prevalence rates is that children who have associated conditions such as intellectual disabilities and epilepsy may have been diverted from educational services at an earlier age and, therefore, are not included in the screening for pupils enrolling in elementary schools. In addition, childhood mortality for more severely affected children could be greater, especially in poorer areas [25]. All these factors would reduce the prevalence of ASD compared to Western countries with their more developed health and education services.

    It is possible that different child-rearing practices, adult tolerance, and expectations around children’s behaviours could be other reasons for the difference in the prevalence rates of autism in Iran. Iranian culture and families may be more tolerant of behaviours in children that in western societies would be seen as “abnormal”, although those children presenting for clinical assessments tended to show the same symptoms as documented in other studies [24]. de Giacomo and Fombonne [26] found that the most common parental concerns were for delay in speech and language development, followed by abnormal signs of socioemotional behaviour and medical problems or delay in reaching milestones. In contrast, Daley [2] reported that Indian parents rated social difficulties such as lack of interest in people, poor eye contact, and showing no interest in playing with other children as their primary concern."

  2. Hi Emily -

    Especially nice job linking to a lot of the papers on MVMs list!

    It appears that the overwhelming majority (all of) of the human analysis we have available in this regard is correlational. As I think about this a little bit, I found myself wondering what other type of data we could possible hope to have in humans? Wouldn't a prospective study, intentionally invoking the maternal immune response and then observing the neurodevelopmental outcomes of the offspring be wildly unethical in humans? What would a human study that captures non-correlational data look like?

    Though MVM didn't referene too many animal studies, I'm curious on how you think these studies can inform us on the correlation versus causation problem? In these instances, the direction of causality is not ambiguous, the treatment group shows altered behavioral and immune function while the placebo group does not. I know that there are problems with mouse studies, but we do have primate studies; i.e.,

    Brain enlargement and increased behavioral and cytokine reactivity in infant monkeys following acute prenatal endotoxemia
    Maternal influenza infection during pregnancy impacts postnatal brain development in the rhesus monkey

    Are both in line with the rodent studies. In what ways do you think our animal studies allow us to inform the nature of the correlative data we have in the human population?

    A few of the studies on the MVM list that you didn't reference here (maybe he was still building his list?) would also seem to support the idea that altering the immune environment in cases of MIA has effects; i.e.

    Induction of Toll-like receptor 3-mediated immunity during gestation inhibits cortical neurogenesis and causes behavioral disturbances
    Adult behavioral and pharmacological dysfunctions following disruption of the fetal brain balance between pro-inflammatory and IL-10-mediated anti-inflammatory signaling

    Though that last one tells us the balance is the important part!

    Finally, something that has really been bugging me about the 'correlation != causation' argument, especially in regards to an immune response in the CNS, are some of the new findings on the function of resting microglia. For decades, the consensus was the microglia were in the brain to respond to injury or infection. Part of the problem was measuring the little buggers in vivo, and they were also a lot easier to get a peek at when they are 'activated'. It turns out, however, we are learning that when not 'activated', microglia are responsible for performing some critical operations, including structural optimization of the brain; i.e.,

    Resting microglia directly monitor the functional state of synapses in vivo and determine the fate of ischemic terminals
    Synaptic pruning by microglia is necessary for normal brain development
    Microglia sculpt postnatal neural circuits in an activity and complement-dependent manner
    The role of microglia at synapses in the healthy CNS: novel insights from recent imaging studies
    Functions of microglia in the central nervous system - beyond the immune response

    To my untrained eye, this tells me that any explanation invoking a non-causal relationship between microglia activation in some cases of autism must not only find a way for coincidental findings, but must also argue that the resting state functions of microglia aren't being impaired by a state of chronic activation. While this is possible, I begin to wonder if there is a point at which we ask ourselves why we should ignore established models in favor of possibilities?

    Lastly, from an (even more!) speculative standpoint, perhaps, these previously unkown functions of microglia may help us understand the seeming paradox of an altered neuroimmune state (in some cases!), with a non-degenerative disorder. Maybe it is a function of microglia not doing something, as opposed to doing something. (?)

    - pD

  3. Hi, pD.... thanks for commenting, as always; I was glad to be able to link directly to the full-length text--saves some time.

    About the missing studies you mention: I copied and pasted what I found on the page and then took each link one at a time, so if I missed those, I don't know why. The Toll-like receptor study looked at viral infection and effects in pups, using a knockout model. Definitely a mechanistic look at what viral infection--well, synthetic viral infection--might do to mouse offspring ... although are there really women walking around who didn't have *some* viral infection during pregnancy?

    The second one you cite that I didn't include is here:
    It addresses IL-10, and if I'm recalling correctly, some other studies found too little, while this one found problems with too little AND too much, possibly at specific developmental time points. Balance of some kind is obviously important, but do we all walk the same line or have the same balance to be healthy or "normal"?

    The monkey study you cite regarding influenza is interesting because there's then the Ozonoff study finding no correlation between maternal influenza and autism (!). Typical autism confusion.

    As for mechanistic/animal studies, some of the ones listed here
    target mechanistic evaluations, such as using human maternal antibodies in mice and looking at effects. Not all of those listed here are about autism; e.g., some relate to learning differences, etc. Citations 9-11 look promising but touch the surface of a mechanism, as always with the qualification of "some" cases of autism. Ref 12 is to the monkeys that Patterson said were exposed during the wrong time period. Then that article takes us into a whole new territory: "Confirming an immune mechanism for some proportion of autism cases may help identify at-risk pregnancies, by allowing pregnant women or women planning to become pregnant to be screened for harmful anti-brain antibodies." And what comes of that screening in the absence of a known mechanism? What if a therapy--years down the road as that may be--were available? The rest of that piece gets into territory that is disturbing for me, personally. I may be in a minority, but I have a viscerally negative response to the idea of having done anything to prevent my son from being exactly who he is or, worse, from being here. But my articulation of his importance as a human being and as an autistic human being is commentary for another time.

    But in general, regarding mechanistic studies, I'm thinking of the usual functional in vivo studies, knocking out, overexpression, etc., paralleled with in vitro work. Patterson's work with the induced maternal immune activation in mice is an example of a step in that direction, and I'm extremely intrigued by his reference to a sex bias in some of the findings.

    I like your speculative final statement and the concept of a role for microglial behavior in a non-degenerative disorder. Now, I just need some mice.

    I don't have anything against correlational studies, at all, although it seems like the time has come to dig into mechanisms with this immunity/autism thing. I just don't like seeing them trotted out as incontrovertible proof of ... well, anything ... but in particular, I am tired of seeing people take correlations in autism and run for a touchdown even though they're not still carrying the ball. Sports metaphor!

  4. This comment has been removed by the author.


    great article with references

  6. I received both of the above comments almost simultaneously. I don't dismiss anecdotes as worthless, and they make a great story. I could write about a few of my own. But anecdotes are individual stories, not data. The complexity of human physiology makes it untenable to take one person's anecdote, no matter how convincing, and turn it into something that describes, works for, or applies to all people or people we have grouped together for our clumsy human reasons. Will there come a day when we smack our heads and say, "Worms for autism? What the hell?" or a day when we wonder why anyone thought that was a nutty idea? I can't say. All I can say is that right now, not even the most generous interpretation of the data could form the basis for arguing that *all* autism fits into this idea or that autism could be prevented via maternal treatment with parasitic worms. I restrict myself to the data at hand and am not a seer, nor am I someone who advocates using children or autistic people to test treatments that have a basis in very little mechanistic data. There appears to be a reasonably convincing rationale applying this treatment in inflammatory disorders of the gut, but as far as neurobiological constructs and their "prevention" go? Not convincing at all.

  7. Well, if and when FDA approves TSO and if the side effects are deemed to be relatively harmless and the risk is relatively low, I would love to see a clinical trial conducted with a few hundred autistic kids that have GI symptoms and see of their autistic symptoms improve.

  8. Emily, with regard to "conflation of correlation and causation [as] the main problem of the original op-ed" - Yes, epidemiological findings linking maternal infection and associated markers of infection with autism do not prove causation, as the tobacco industry constantly reminded us about smoking an lung cancer. Nor does the finding of immune dysregulation in the blood and brain in autism prove causation. Nor does the increase in risk for autism when the mother has an autoimmune disease. Causation is usually tackled in animal models, where one can experimentally manipulate the factors in question. That is why it is important that a single day of immune activation (or maternal infection) in a pregnant mouse leads to autism-like behaviors and neuropathology in the offspring. Moreover, immune activation causes immune abnormalities in the offspring, some of which resemble findings in autism. In addition, these offspring display gastrointestinal problems resembling those seen in many autistic kids. Yes, these results were obtained in mice, which are a long way from humans. However, many similar findings are beginning to be made in the maternal infection and maternal immune activation models in monkeys. In my, admittedly biased, opinion, putting the human and animal data together makes a good case for causation. Of course, one cannot state that any individual case of autism is caused by maternal infection any more than one can say that any individual case of lung cancer is caused by smoking. And of course many smokers do not get cancer. I also disagree with your opinion that the NYT writer cherry-picked his sources and data - of course, not every finding fits neatly into a perfect picture, but his sources make the case I just outlined very well. Unfortunately, I don't have the time to go down the list of all the points you raised about all the papers he cited Emily, but I would be happy to do so in a phone call. Cheers, PHP

  9. I agree with everything you say here, having said it myself; however, the issue with the original article was that it generalized and treated correlation *like* causation, and I am not the only scientist who has noted that. The issue with the sources he used to support his careful constructs are that he took only *parts* of the findings that were relevant to his argument and generalized them to all of autism or extended them beyond their immediate implications. That is "cherry picking," and autism scenarios seem particularly susceptible to it. You yourself criticized the monkey study (if we are talking about the same one) as targeting the wrong period of pregnancy, for example, in your own review that he cited in his source list, yet the monkey study (four monkeys in the experimental group, yes?) gets the royal treatment. He's building a story and using specific examples from the literature to support it; in the process, he leaves out the qualifying or equivocal findings that don't fit the story. You know what? If he'd simply focused throughout that op-ed on using qualifiers like "might" or "in a subset" or "some" and left out "most likely prevents," we probably wouldn't even be having this conversation. There'd've been some commentary from various quarters about the tenuous thread he unspools from hygiene hypothesis-->maternal infections/autoimmunity (still can't work out his conflation of Th1 vs Th2 effects, allergy vs autoimmunity)-->autism-->GI effectiveness of helminth worms for *GI* autoimmune diseases-->worms to *prevent* autism, but not about the presentation of his narrative.

    It's interesting that you use the smoking analogy, as though to paint me as someone trying to mask a real, causative link like a tobacco company deliberately lying to people when they had more evidence in-hand than they were letting on. I've seen that before from other quarters regarding pharmaceutical companies and vaccines. I've given my own example of the dangers of inferring causation from even compelling correlations--the CRP-heart disease story--which I think is a more apt analogy, as we are talking about endogenous factors here. Millions of women have infections during pregnancy--I'm very curious how someone would avoid it--without giving birth to autistic children. [Anecdatum: I personally spiked an uncontrolled 104-105 fever for three days with the flu in the first trimester of my pregnancy with my second, *non-autistic* son]. I think it's important to comment on risk and correlation and what they really mean--not because I'm trying to protect the tobacco companies but because women become genuinely concerned and anxious about reports like these, particularly when they are overinterpreted and overstated, and anxiety during pregnancy isn't a terribly good thing, either. I had one very scientifically literate woman I had never met tell me that after reading the NYT piece, she felt as though she'd committed some sort of "moral failing" by not having carried a parasitic worm burden during pregnancy.

    Yes, you likely are biased, as this research path is yours, so why wouldn't you be? I've noticed that you and the op-ed writer are quite complimentary of each other's work/books, too, with frequent cross-citation. Not that it likely matters to you, but I'm deeply intrigued by your research, particularly your reference to a sex bias in the findings, and what I'm writing here is not intended to dismiss it or wave it away. Nevertheless, I have received many emails and messages from other researchers, even some who study immune links to autism, who agreed with my criticisms of the New York Times op-ed on interpretive and scientific grounds. I would be happy to have a phone call with you. My email is ejwillingham (at) g ma il if you would like to get in touch.