No, autism is not caused by the gut microbiome

Last week, the New York Times published a paper claiming that we may be able to use the gut microbiome to diagnose autism. Times The story was based on a recently published scientific paper that claimed the same thing.

This report set off all my skeptical alarm bells. My first reaction was, “Oh no, more bad science about autism.” For one thing, as most scientists who study autism know, the modern anti-vaccine movement began with a scientific paper in 1998 that falsely claimed that childhood vaccines caused autism. That paper The Lancet The article was revealed to be fraudulent and was eventually retracted, but not before enormous damage was done. Its lead author, Andrew Wakefield, became a hero of the anti-vaccine movement and continues to promote anti-vaccine misinformation to this day.

The new article (extract from the journal Microbiology of nature) does not make outrageous statements like that, and he did not New York Times. However, anyone who claims that autism is caused by microbes in the gut should know that the famous Lancet The study was based on the “leaky gut” hypothesis, a hypothesis that was long ago discredited. (I don’t want to give it any credence, but that hypothesis held that viral particles from some vaccines somehow “leaked” out of the gut and into the brain. It was absurd then and still is.) This is one reason why the idea that gut microbes might cause autism (or even be used to diagnose it) raises so much alarm.

I have now looked at the study and, frankly, I don’t believe a word of it. I want to be clear, though: I am not trying to scientifically prove that the study is wrong, which would take months of effort and far more detail than I can possibly express in a column anyway. Fortunately, there is an earlier study that did that work for me, and I will get to that later.

However, the science behind this study is closely related to my own work, so I feel quite comfortable giving my opinion. So what did the authors do?

As the new study explains, they collected feces (known as “stool samples”) from 1,627 children, some of whom had been diagnosed with autism and some of whom had not, and sequenced the DNA in the feces. They then looked for bacteria, viruses, and other microbes in the DNA sequence data.

That’s right: The “gut microbiome” is actually a polite term for the bacteria that live in the intestines and colon, some of which are excreted in stool. Sure, some bacteria in stool can come from the food a person has eaten, but these are mostly so-called gut bacteria.

I have participated in many studies like this myself, so I have seen that these experiments produce results. hundreds of different species from each sample. The data sets are very complex and a common problem in the field is that these data are often misinterpreted. Microbiology of nature In this paper, the authors took these very complex datasets and fed them to a machine learning program, and voila! The AI ​​program was able to do a pretty good job (far from perfect, I should point out) of identifying autistic children, based on the mix of microbes present in their feces.

That’s right. I don’t believe in all this, as I wrote above. Why not? Well, first of all, machine learning programs are really good at distinguishing between two groups of subjects (like autistic and non-autistic children) if you give them enough data. It sometimes turns out that machine learning programs focus on irrelevant features that scientists didn’t anticipate.

For example, this 2021 paper looked at over 400 studies that used machine learning to predict Covid-19, all of which had claimed some success, and found that all of the studies were essentially useless “due to methodological flaws and/or underlying biases.” Of course, the gut microbiome study wasn’t one of them, and some machine learning experiments do work, but we should be very skeptical.

Another reason for skepticism is that the new paper doesn’t even try to tell us what the machine learning models actually learned — it simply treats the programs as a “black box” that we should trust.

Perhaps the biggest flaw in the study of autism and the gut microbiome in children is this: autistic children tend to be picky eaters, and their parents try all sorts of diets in the hopes that they can at least alleviate the symptoms of autism through food. There are countless websites—many of which are unfortunately scams—that claim that special diets can help these children. Why is this important? Because a special diet will change your gut microbiome, sometimes quite significantly.

So even if the machine-learning models in the new study are correct, the causality almost certainly goes the other way: autistic children might have a different microbiome because they eat different foods. So it’s autism indirectly affecting the microbiome. Unfortunately, both New York Times and the scientific article suggested otherwise.

Now let’s move on to the scientific paper I mentioned above. It turns out that 3 years ago, a group of Australian researchers published a major study in the journal Cell This study addressed precisely the problem I just highlighted. In this study, scientists collected and sequenced stool from 247 autistic and non-autistic children. They found “negligible direct associations between ASD (autism spectrum disorder) and the gut microbiome.”

On the contrary, the authors warned: “Microbiome differences in ASD may reflect dietary preferences…and we caution against claims that the microbiome plays a driving role in ASD.”

In other words, three years ago, a study published in a major scientific journal found that there was no connection between autism and gut microbiota content. They then cautioned that if you see differences in the gut microbiota in autistic children, these are caused by their diet, so don’t claim that the microbiota causes autism. The authors of the new study and the journalists at the New York Timesapparently decided otherwise.

So no, the gut microbiome cannot be used to diagnose autism.