Microbiome Resource Center

Microbiome Resource Center

Disclosures: The authors report no relevant financial disclosures.
July 21, 2020
2 min read

Machine learning analysis reveals less diverse microbiome in children with type 1 diabetes

Disclosures: The authors report no relevant financial disclosures.
You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

Children with newly diagnosed type 1 diabetes have more abundant gram-negative bacteria in their gut microbiome associated with the onset of disease compared with healthy children, according to an analysis using machine learning algorithms.

“The study of gut microbiota is a new field of research regarding the environmental factors associated with type 1 — autoimmune diabetes — and machine learning algorithms for analysis now provide a substantial contribution for the validity of results,” Giuseppe d’Annunzio, MD, assistant professor in the Regional Center for Pediatric Diabetes at Giannina Gaslini Pediatric Hospital in Genoa, Italy, told Healio. “Our results suggest that gut microbiota composition impairment is among the environmental factors associated with the development of type 1 diabetes, although this research requires more studies with larger cohorts.”

Using fecal samples, D’Annunzio and colleagues conducted machine learning analyses and metagenome functional analysis to identify significant taxa and their metabolic pathways in the microbiome of 31 white children with type 1 diabetes (20 boys; mean age, 10 years) admitted to the ED between January 2016 and May 2018, as well as 25 sex-matched, healthy children who served as controls. At time of diabetes diagnosis, mean glucose level was 462 mg/dL, mean HbA1c was 10.75% and mean C peptide level was 0.48 ng/mL. DNA was extracted from fecal samples taken from children with type 1 diabetes within 1 week after recovery from metabolic decompensation and during daily insulin therapy. Researchers used two different machine learning algorithms for biomarker classification and identification; study of fecal microbiota was performed using next generation sequencing.

“In our approach, we used the sequencing of seven polymorphic regions to avoid as much as possible potential false negative taxonomic classifications and any bias in the identification of microorganisms in a complex microbial population,” the researchers wrote.

Compared with healthy controls, children with type 1 diabetes had a higher relative abundance of the most important taxa: B.stercoris, B.fragilis, B.intestinalis, B.bifidum, Gammaproteobacteria and its descendants, Holdemania, and Synergistetes and its descendants. Children with type 1 diabetes also had a lower relative abundance of B.vulgatus, Deltaproteobacteria and its descendants, Parasutterella and Lactobacillus, and Turicibacter genera compared with healthy controls.

In cluster analyses, researchers found that BMI-SDS, insulin autoantibodies, HbA1c, Tanner stage and age at diabetes onset were the most positively or negatively correlated with specific clusters of taxa.

The researchers noted that it is not possible to clearly state whether gut microbiota diversity represents a cause or a consequence of autoimmunity in type 1 diabetes.


“Gut microbiota composition deserves attention as a new topic of research in the development of several diseases,” d'Annunzio said in a press release.

For more information:

Giuseppe d’Annunzio, MD, can be reached at the Pediatric Clinic, Regional Center for Pediatric Diabetes, IRCCS Instituto Giannina Gaslini, Via Gaslini 5, 16147, Genoa, Italy; email: giuseppedannunzio@gaslini.org; Twitter: @OspedaleGaslini.