Temporal changes in gastrointestinal fungi and the risk of autoimmunity during early childhood: the TEDDY study

Abstract

Fungal infections are a major health problem that often begin in the gastrointestinal tract. Gut microbe interactions in early childhood are critical for proper immune responses, yet there is little known about the development of the fungal population from infancy into childhood. Here, as part of the TEDDY (The Environmental Determinants of Diabetes in the Young) study, we examine stool samples of 888 children from 3 to 48 months and find considerable differences between fungi and bacteria. The metagenomic relative abundance of fungi was extremely low but increased while weaning from milk and formula. Overall fungal diversity remained constant over time, in contrast with the increase in bacterial diversity. Fungal profiles had high temporal variation, but there was less variation from month-to-month in an individual than among different children of the same age. Fungal composition varied with geography, diet, and the use of probiotics. Multiple Candida spp. were at higher relative abundance in children than adults, while Malassezia and certain food-associated fungi were lower in children. There were only subtle fungal differences associated with the subset of children that developed islet autoimmunity or type 1 diabetes. Having proper fungal exposures may be crucial for children to establish appropriate responses to fungi and limit the risk of infection: the data here suggests those gastrointestinal exposures are limited and variable.

Fig. 1. Fungal exposure increases with age.

a Abundance of fungal reads in TEDDY metagenomic samples (n = 12,262 from 888 children; median 2.6 × 107 total reads/sample). Curves show LOESS fit for samples from Finland (blue; n = 2855 from 238 children), Sweden (yellow; n =  4360 from 302 children), Germany (red; n = 1228 from 86 children), and USA (green; n = 3819 from 262 children). Shaded regions represent 95% confidence intervals. b The fungal percent of metagenomic reads in young children (TEDDY, 3–14 m, n = 6547), older children (TEDDY, 15 48 m, n = 5621), USA adults (HMP, n = 203), and non-Western adults from Fiji, Madagascar, Peru, and Tanzania (n = 315). Displayed are the interquartile range (IQR; boxes), median (line), and 1.5 IQR (whiskers). TEDDY countries left to right are Finland (blue; n = 2835 from 238 children), Sweden (yellow; n = 4331 from 302 children), Germany (red; n = 1220 from 86 children), and USA (green; n = 3784 from 262 children). Samples with zero fungal reads are shown having one read to allow logarithmic scaling. The P values were calculated by Kruskal–Wallis rank-sum tests. Source data are provided in the Source Data file.

Fig. 2. Fungal diversity, similarity, and taxonomic patterns from 3 to 48 months of life, and the corresponding dietary trends.

a Shannon diversity of fungal ITS2 (blue; n = 9547 from 825 children) and bacterial 16S rRNA gene (red; n = 12,616 from 910 children) OTUs clustered at 97% identity and rarefied to 3000 reads/sample. Curves show LOESS fit for the data, and shaded regions represent 95% confidence intervals. b Mean Bray–Curtis similarity of the collection of fungal ITS2 (blue; n =  9547 from 825 children) and bacterial 16S rRNA gene (red; n = 12,616 from 910 children) OTUs clustered at 97% identity and rarefied to 3000 reads/sample, between children and within a child. Error bars display the standard error. c Five most abundant overall fungal taxa from 3 to 48 months as detected by analysis of ITS2 OTUs clustered at 99% identity and rarefied to 3000 reads/sample (n = 9547 from 825 children). Curves show LOESS fit for Saccharomyces cerevisiae (blue), Penicillium paneum (pink), Candida albicans (yellow), Candida parapsilosis (green), and Candida zeylanoides (brown). Shaded regions represent 95% confidence intervals. d Breastmilk, formula, and/or food consumption status of children by age (n = 12,127 from 845 children). Source data are provided in the Source Data file.