Individualized glycemic responses and gut microbiome changes to dietary intervention with fructooligosaccharides and galactooligosaccharides

Background: The gut microbiota has been implicated in glucose intolerance and its progression towards type-2 diabetes mellitus (T2DM). Relevant randomized interventional clinical trial targeting gut microbiome using high-throughput sequencing technologies is still inadequate. We sought to evaluate the impact of two different types of prebiotics fructooligosaccharides (FOS) and galactooligosaccharides (GOS) on glycemia during oral glucose tolerance test (OGTT) and the intestinal microbiota in healthy and overweight adults. Methods: A randomized double-blind self-controlled cross-over study was performed with 35 adults (18 healthy, 17 overweight) treated daily with FOS and GOS for 9 days (16 g/day). Faeces sampling, OGTT and impedancemetry were performed before and after intervention. The gut microbial composition was profiled via surveys of 16S rRNA-encoding gene sequence from Ion Torrent platform and bioinformatics analysis using QIIME. Results: The intake of both kinds of prebiotics for 9 days led to deterioration in glucose metabolism as demonstrated by OGTT (P < 0.05). A significant increase in the relative abundance of Bifidobacterium was observed respectively in FOS treatment (from 1.45% to 2.20%) and GOS treatment (from 0.88% to 3.17%), while the butyrate-producing bacteria like Phascolarctobacterium was decreased in FOS group (from 1.21% to 0.91%) and Ruminococcus was decreased in GOS group (from 1.14% to 0.56%). Within each group, high inter-individual and intra-individual variability in glucose metabolism on OGTT and gut microbiome was observed. A random forest model using the initial microbiota was developed to predict OGTT levels after prebiotic intervention with relative success (R = 0.72).Conclusions: FOS and GOS increased Bifidobacterium, but reduced some of butyrate-producing microbes like Ruminococcus and Phascolarctobacterium. These community changes occurred with deteriorated glucose metabolism as demonstrated by OGTT. Our random forest modeling result indicated the potential of personalized prediction of glucose level on OGTT using gut microbial community data.