The process of diagnosing a child with autism relies heavily on parents' descriptions of their child's behavior and expert observations, leaving considerable room for human error.
Parental anxiety may bias survey responses, health care providers may be biased and underdiagnose certain groups, and symptoms in children may vary widely depending on culture, gender, and other factors.
The study, published Monday in Nature Microbiology, adds to a growing body of research that suggests a surprising route to a more objective autism diagnosis: the gut microbiome.
The researchers analyzed more than 1,600 stool samples from children aged between 1 and 13 years old and found several distinctive biological “markers” in samples from children with autism. Chi Hsu, a researcher at the Chinese University of Hong Kong and lead author of the study, said the unique signatures of gut bacteria, fungi and viruses could form the basis of diagnostic tools in the future.
Biomarker-based tools could help experts diagnose autism earlier, allowing children to receive more effective treatment at a younger age, he said.
“We rely too heavily on questionnaires,” says Sarkis Mazmanian, a microbiome researcher at the California Institute of Technology. “Anything we can find that we can measure is a big step forward.”
For decades, researchers have combed through people's genomes, medical histories and brain scans looking for reliable indicators of autism spectrum disorder, with limited success. The US Food and Drug Administration has approved two diagnostic tests based on eye-tracking software, but Dr. Su said the tests require significant involvement from psychiatrists.
Over the past 15 years, some researchers have begun to explore whether stool, a window into the trillions of fungi, bacteria, and viruses that live in the gut, might provide clearer answers. Until now, most of these studies have relied on small groups of people and often produced mixed results.
The idea that the gut microbiome might play a role in the development of autism is still a topic of debate among researchers, said microbiologist Gaspard Taroncher Oldenburg, who published a landmark paper on the subject last year.
He called Monday's paper one of the largest of its kind and a “significant milestone” in the wider acceptance of the work. “The winds are shifting,” he said. “People are now acknowledging that the microbiome may not just be one part of it, but that it may be a fundamental piece of the puzzle.”.”
In the new study, researchers used machine learning to identify key biological differences between the stool of children with autism and other samples.
While previous studies have focused primarily on gut bacteria, the researchers broadened their scope to also look at other microbes in the gut, including fungi, archaea, and viruses, and their associated metabolic processes. The scientists identified 31 biological features that distinguish each group.
They then tested an entirely new set of samples to see if these markers could correctly identify which stool samples came from people with autism, and Dr Su said the model got its predictions right almost every time.
But Dr. Sue and Dr. Taroncher Oldenburg cautioned that it's difficult to say whether stool sample tests would work as well in a clinical setting.
More research is needed to convince skeptical scientists that these biomarkers are valid indicators of autism. Mazmanian, who was not involved in the new paper, said he would like to see research that sheds light on how the microbiome relates to autism and whether it plays a key role in causing ASD.
Some researchers argue that the relationship is in the other direction, i.e. that autistic children are more likely to be “picky eaters,” which alters the composition of their microbiome.
Dr Su also said the research model needed to be validated in more diverse samples of children, most of whom were recruited from Hong Kong.
“This study is just the beginning of a long journey,” he said.