A novel approach to finding the compositional differences and biomarkers in gut microbiota in type 2 diabetic patients via meta-analysis, data-mining, and multivariate analysis

Background/Purpose of the study

Type 2 diabetes mellitus (T2DM)—one of the fastest globally spreading diseases—is a chronic metabolic disorder characterized by elevated blood glucose levels. It has been suggested that the composition of gut microbiota plays key roles in the prevalence of T2DM. In this study, a novel approach of large-scale data mining and multivariate analysis of the gut microbiome of T2DM patients and healthy controls was conducted to find the key compositional differences in their microbiota and potential biomarkers of the disease.

Methods

First, suitable datasets were identified (9 in total with 946 samples), analyzed, and their operational taxonomic units (OTUs) were computed by identical parameters to increase accuracy. The following OTUs were merged and compared based on their health status, and compositional differences detected. For biomarker identification, the OTUs were subjected to 9 different attribute weighting models. Additionally, OTUs were independently analyzed by multivariate algorithms (LEfSe test) to verify the realized biomarkers.

Results

Overall, 23 genera and 4 phyla were identified as possible biomarkers. At genus level, the decrease of Bacteroides, Methanobrevibacter, Paraprevotella, and [Eubacterium] hallii group in T2DM and the increase of Prevotella, Megamonas, Megasphaera, Ligilactobacillus, and Lachnoclostridium were selected as biomarkers; and at phylum level, the increase of Synergistota and the decrease of Euryarchaeota, Desulfobacterota (Thermodesulfobacteriota), and Ptescibacteria.

Conclusion

This is the first study ever conducted to find the microbial compositional differences and biomarkers in T2DM using data mining models applied on a widespread metagenome dataset and verified by multivariate analysis.