A link between autism and the gut microbiome has been researched for almost a decade, but which directs which? Some studies, albeit inconsistent, have suggested that people on the autism spectrum tend to have a smaller array of bacteria and other microbes in their gut, leading to the theory that their biota drives their behaviour.
An Australian study has turned this around, finding that it's selective eating - more common among children on the spectrum - which is shaping the gut microbiome.
"While it's a popular idea that the microbiome affects behaviour, our findings flip that causality on its head," says Chloe Yap, a researcher at the University of Queensland and Mater Research, and lead author on a paper in Cell.
The link between gut microbiome and autism rests on small-scale studies, and animal-based research - which doesn't necessarily translate well to human behaviour.
Yap says the field was "getting ahead of itself in some ways".
"Within the past 10 years or so, there's been a real explosion in this field of microbiome research. In fact, nowadays, if you think of a trait or condition, there's probably microbiome research being done on it at the moment."
While the causative links between microbiome and autism were fairly weak - and didn't appear across all studies - the ones that did return a positive result got more attention, because they were more exciting.
"Sexy hypotheses are going to sell better than common sense ones," says Yap.
"This hype around the autism-microbiome [link] led to the marketing of these so-called 'therapies' - targeted at parents, often - that claim to help and support their child on the spectrum by modifying the microbiome."
Therapies for sale can include probiotics, diets, and even faecal transplants.
"These are all expensive. Families have to change their lifestyle around some of these diets, and some of them may actually just do more harm than good," says Yap.
The Autism Cooperative Research Centre funded Yap, along with colleagues from UQ and Mater Research, to investigate the link in more detail.
They examined the gut microbiome of 247 children: 99 of whom had been diagnosed with autism, 51 siblings of these children, and 97 unrelated undiagnosed children.
"Ours was a larger study, I'd say probably the largest to date, and it also benefited from having very high-resolution microbiome data [...] that gives us gives us a lot more granularity in terms of accurate taxonomy."
The researchers also had a detailed dataset for each child covering "clinical, lifestyle, dietary, genetic, and other biological data".
"What that allows us to do is really interrogate: what is related to the microbiome, what isn't, what could be confounding?" says Yap.
The researchers ran several types of analysis to discern the link - including looking for behavioural and non-behavioural traits that predicted changes in the microbiome, looking for species of microbe that were more prominent among children on the spectrum, and examining genetic data between siblings and non-siblings.
All these analyses led to dozens of pieces of evidence that mostly dis-established the microbiome-autism hypothesis.
For example, out of over 600 different bacterial species sequenced, only one was closely associated with autism diagnosis. And while restrictive eating is associated with autism and did predict changes in the microbiome, other traits like IQ and sleep patterns did not.
"That allowed us to make this inference that no, it doesn't look like the microbiome causes autism. Instead, it looks like characteristics of autism seem to affect the microbiome rather than the other way around," says Yap.
Trudy Bartlett, an autistic woman from Brisbane, says she has found that "many autistics have gut issues, which I thought may be linked to the fact that many of us - including myself - have restricted diets, so we may not get all the nutrients we should. Wanting to know more about it is like walking through a minefield, trying to filter fact from fiction.
"Having evidence-based research like this study will help members of the autism community to navigate this space and not spend copious amounts of money and time on fads that claim to improve the quality of life for an autistic person."
The researchers plan to back their findings up with more data from other places, as well as examining more of the microbiome.
"Really the gold standard of science is replication," says Yap.
"So getting that evidence in other settings and other countries would be amazing."