Ethics: Nil - Science: Bad - Outcomes: Poor

MacFabe et al’s paper “Neurobiological effects of intraventricular propionic acid in rats: Possible role of short chain fatty acids on the pathogenesis and characteristics of autism spectrum disorders” concluded the following:

“We propose that some types of autism may be partial forms of genetically inherited or acquired disorders involving altered PPA metabolism. Thus, intraventricular administration of PPA in rats may provide a means to model some aspects of human ASD in rats.”

The form of this research is peculiar. The paper cites 190 references. Not one refers to any data with the actual levels of PPA in stool, urine, serum or CSF in autistic children. Instead, the research moves immediately to the deaths of 74 rats and a sophisticated appraisal of their demise. Naturally the exercise involves the gainful employment of a number of psychologists and the acquisition of considerable expertise in histochemistry, brain imaging and other useful techniques. One might speculate that the reason for this omission – an actual and factual basis for the research, might be that a negative or ambivalent result could seriously undermine the rationale of the experiment in the minds of potential funders.

The authors lean heavily on the heterogeneous nature of autism pointing to the all-purpose ‘some children with ASD’ without in any way specifying the characteristics of the sub-group to which their thesis might be relevant. In politics this would be called plausible deniability. In this paper it takes the form of assembling a pastiche of vague resemblances between propionic acidemia and some aspects of ASDs with blurring of the obvious differences especially at points that are crucial to the authors’ thesis.

The most obvious example is the descriptor of ASD as a movement disorder citing an opinion piece which uses gait as the measure of dysfunctional movement. In those affected with inborn errors of PPA metabolism the movement disorder takes a very defined, involuntary and severe form – choreoathetosis, which in no way is equivalent to the ataxia of gait movement in ASD. I refer the reader to Leary and Hill 1996 for the definitive description of ataxia as it affects folks on the autistic spectrum. In the same vein, gastrointestinal disturbances among the autistic population are quite cavalierly equated with the severe vomiting evident among those with propionic acidemia, which has hitherto, not been mentioned as being relevant to the ASD population. Neither are autistics or any subgroup thereof known to have such a complete loss of appetite as to require feeding tubes, a clinical feature of propionic acidemia.

The list of symptoms associated with errors of propionic acid metabolism is as follows:

• Seizures
• Reduced muscle tone
• Impaired immune system
• Skin rash
• Hearing loss
• Hair loss
• Mental retardation
• Vomiting
• High blood ammonia level
• Acidosis
• Propionic acidemia

It takes more than a little academic spandex to see autism in this list. The logical conclusion of such an equivalence would be that all persons with propionic acidemia would be classified as autistic and the most severely autistic at that. The literature makes no mention of autism in relation to this population.

The second pillar of the experimental effort is the use of the behaviour of rats, injected with propionic acid directly into the brain as proof of the link between autism and inborn errors of propionic acid metabolism. The flaws in this argument are legion. The exposure limit for humans to propionic acid, a chemical labelled corrosive, is 10 PPM. This is set to guard against damage to skin and mucosal tissue. It was not set for the ‘direct injection into the brain’ scenario. The low dose the rat got in each injection was around this level and the high dose around five times that level. Most of the paper and its conclusions dwell on the response of the rat to these injections, with much made of the inflammatory response, the seizures and so on. Inject a rat with a corrosive substance, directly into the brain, at a level known to cause damage to human skin and what else could one expect?

There are two cogent reasons to reject the author’s interpretation of the observable movement of the mice following PPA injection. Firstly, who is to say that the movements observed are not characteristic of choreoathetosis in rats? No compelling reason is given. Secondly, the cerebral organization of rats and men differs markedly and not just in size. The authors noted that cell densities are greater in autistic populations but failed to mention that the organizing structure, the mini-column varies from species to species. Given the differing structural arrangements, the validity of rat models for complex multi-factorial behaviours such as social interactions or even the presumably less complex ‘movement behaviour’ is highly questionable. The model is extremely weak and as such, the authors have made the basic error of attempting to validate a hypothesis through the use of a very poor analogue, a logical impossibility. In point of fact, what these authors have attempted to do points to a wider concern in certain sections of brain research.

I have no quarrel with animal studies per se. The 2007 Nobel Prize for Medicine was given to Mario Capecchi, Martin Evans and Oliver Smithies for their pioneering development of knockout gene technology. That technology led to the development of over 500 different rodent models of various disorders having a genetic basis. The expected and indeed valid utility of these models lies in being able to assess the physiological effects of genetic manipulation. But this is not what MacFabe and any number of others are doing. They have conceived the entirely invalid notion of assuming an etiology, that is, assuming the genetic aberration and then using some aspect of animal behaviour to validate their notions. This seems to get the method exactly backwards – using the behaviour to finger the gene(s) and using behaviour from animal populations that, cognitively at any rate, we have little in common. Cohen sums up the problems of this approach as follow:

“To speak of "drug cravings" in animal addicts made physiologically dependent on drugs, to speak of animals forced to drink alcohol as "being high," to consider an "opponent-process theory of acquired motivation" as having anything to do with the etiology of drug addiction, to refer to monkeys taught learned helplessness as being "depressed," to conclude that primates undergoing deprivation are "schizophrenic," and to label animals traumatized in laboratories as "neurotic," is to disdain epistemology, indulge in self-serving fantasy, demean proper scientific exploration, obfuscate the significance of empirical facts, and manipulate, mislead, and delude everyone.”

What applies to other ‘isms’ of psychology applies equally well to autism. Principles of model validation in this area require three things apparently; face validity – does it look like what the model stands for, construct validity – a common etiology really and predictive validity, which is self explanatory. MacFabe’s work fails on all three counts. There is no particular reason to believe that the results of this work are anything other than what would be expected from propionic acidemic rats and no particular reason to believe that the movements, seizures or brain abnormalities of these mice resemble autism.

However, this work was published and will henceforth bear the title ‘peer reviewed’. It might be only this branch of the research community, but I get the strong impression that there is a gentleman’s agreement among the members: I won’t spit on your research if you don’t spit on mine. There is no peer review. There is no chance that this paper could float on the Dead Sea, the holes are so large. As such it is poised to do a lot of damage, to autistic children and by extension to their families. That is to be expected from bad Science, the inevitable result of nothing much in the way of ethics, like reasonable grounds for a host of assumptions.

The treatment of propionic acidemia usually involves large doses of carnitine in an attempt to clear the buildup of toxic metabolites. Carnitine is also used in the diet industry because it raises the metabolic rate. A common side effect is sleep disturbance and for a sizeable number of already sleep disturbed spectrumites, this would not be a good thing. What chance that parents won’t jump on this research and immediately start treatment? Other forms of damage include the waste of material resources, which could be better spent elsewhere and future time and resources wasted on refuting speculative hypotheses.

Spot the Leech.

I think we can expect a large number of research initiatives like this one. Many will want to hitch their research wagon to the rising autism funding star. This paper, a dead loss in many respects, redeems itself slightly by having in one place a large number of the characteristics to be expected in research paradigms grafted haphazardly on to autism.

1. The research is judged ‘preliminary’. True, and there it will remain but in the meantime this is a general disclaimer to shield the authors from any and all repercussions.
2. In much the same vein as [1], the hypothesis will be so general that there is no data the experiment can produce that could falsify it.
3. In the same vein as [1] and [2], the target population will be defined as a subgroup of the autistic population, any subgroup of the autistic population, which will remain undefined.
4. The number of weasel word statements (may, possibly, could be associated with) will outstrip the number of definitive statements by a considerable margin. Occasionally the authors slip and a possibility becomes transformed into a definite at the stroke of a pen.
5. Non sequitur conclusions will be plentiful. In this paper, one example takes the form that some autistics have seizures as do those with propionic acidemia, therefore they are related or could be. What remains missing is the link.
6. Selective citation - debatable interpretations will be strengthened by the omission of reference data, which could suggest a more reasonable alternative. Of, the many examples in this paper, the conflation of ‘gastro-intestinal disturbances’ between the PPA crowd and the autism crowd without in any way pointing out the vast differences in the details is a notable example.
7. References are bountiful but many (most) are not particularly relevant to the actual hypothesis and some will not be mentioned in the body of the text at all. A perusal of this paper’s references reveals that many citations are not referenced in the actual text.
8. Last but not least, there will be great emphasis placed on the future benefits to be derived from this research and if the research is an animal model methodology, equal emphasis on the undeniable fact that this research could not possibly be done on humans.