Deviated and early unsustainable stunted development of gut microbiota in children with autism spectrum disorder

Abstract

Objective: Recent studies have provided insights into the gut microbiota in autism spectrum disorder (ASD); however, these studies were restricted owing to limited sampling at the unitary stage of childhood. Herein, we aimed to reveal developmental characteristics of gut microbiota in a large cohort of subjects with ASD combined with interindividual factors impacting gut microbiota.

Design: A large cohort of 773 subjects with ASD (aged 16 months to 19 years), 429 neurotypical (NT) development subjects (aged 11 months to 15 years) were emolyed to determine the dynamics change of gut microbiota across different ages using 16S rRNA sequencing.

Result: In subjects with ASD, we observed a distinct but progressive deviation in the development of gut microbiota characterised by persistently decreased alpha diversity, early unsustainable immature microbiota, altered aboudance of 20 operational taxonomic units (OTUs), decreased taxon detection rate and 325 deregulated microbial metabolic functions with age-dependent patterns. We further revealed microbial relationships that have changed extensively in ASD before 3 years of age, which were associated with the severity of behaviour, sleep and GI symptoms in the ASD group. This analysis demonstrated that a signature of the combination of 2 OTUs, Veillonella and Enterobacteriaceae, and 17 microbial metabolic functions efficiently discriminated ASD from NT subjects in both the discovery (area under the curve (AUC)=0.86), and validation 1 (AUC=0.78), 2 (AUC=0.82) and 3 (AUC=0.67) sets.

Conclusion: Our large cohort combined with clinical symptom analysis highlights the key regulator of gut microbiota in the pathogenesis of ASD and emphasises the importance of monitoring and targeting the gut microbiome in future clinical applications of ASD.

Keywords: Autism spectrum disorder; Gut microbiota; Neurotypical.

Figure 1

General characteristics of gut microbiota and clinical information of cohorts. (A) Geographical features of residence of the studied cohort. Subjects with ASD (n=773) were from 25 provinces of China, while NT subjects (n=449, 20 adults included) were from 14 provinces of China. (B) Histogram showing the summative distributions of grouped subjects according to age and gender. (C) Unweighted PCA at OTU level (for PC1, PC2 and PC3) showed that the gut microbial composition of subjects with ASD was separated from that of NT and healthy adults. The p values between each group were tested using mutational multivariate analysis of variance (Adonis). (D) Phylum-level distribution of gut microbiota in ASD, NT and healthy adults. (E) The Shannon diversity index of each group or age category. The mean values±SEM are plotted. One-way analysis of variance, ***p<0.0001. (F) Diverging bar chart of absolute microbial abundance changes by Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC) between NT and ASD. (G) Horizontal bars indicate the impact (R²) of each host factor on gut microbiota variations. Subjects were subdivided into two groups (group 1: age ≤3 years or group 2: age >3 years), and the effect of each host factor was determined by EnvFit (vegan). Factors were roughly classified according to metadata categories, and the factors with significant effects are indicaed with an asterisk (FDR adjusted p value, *p<0.05 and **p<0.01). (H) The severity of ASD showed a significant correlation with severity of GI (Wilcoxon signed-rank, p=8.274e-06), sleep disorder (Wilcoxon signed-rank test, p=0.0001537) and allergy (Wilcoxon signed-rank test, p=0.03008). ASD, autism spectrum disorder; NT, neurotypical; ns, not significant; OTU, operational taxonomic unit; PC, principal component; PCA, principal component analysis.

Figure 2

Deviated developmental spectrum of gut microbiota in children with ASD. (A) Three-dimensional diagram of unweighted PCA based on OTU-level Bray-Curtis dissimilarity. Plots of each sample were dyed gradients according to their physiological age. Arrows with gradient colours showed the developmental trends of the gut microbial community in ASD (red) and NT (blue) from young to old. (B) Heat map showed the mean relative abundance changes (10-based logarithm) of 30 age-discriminatory bacterial taxa across the physiological ages of subjects. (C) Predictions of microbiota age in both ASD, NT and adult subjects (above). Each circle represents an individual faecal sample, and the curves are a smoothed linear fit between the microbiota age and physiological age. The values of physiological age minus (−) predicted microbiota age of each group and the microbiota-for-age Z score (MAZ) of the subjects with ASD are shown in the Figure 2C chart below. Mean values±SEM are shown. (D) Shannon diversity index with age. (E) The taxon detection rate difference between NT and ASD remained constant with age. The detection rate curves of Bifidobacterium, Veillonella, Faecalibacterium, Lachnospira and Blautia are highlighted. Arrows indicated the time points of a specific bacteria with an abnormally fluctuating detection rate. ASD, autism spectrum disorder; MAZ, microbiota-for-age Z score; NT, neurotypical; OTU, operational taxonomic unit; PC, principal component; PCA, principal component analysis.

Figure 3

Deviated development in the microbial relationship in children with ASD. (A) Altered microbial community network between NT and ASD before 3 years of age. (B) Altered microbial community network between NT and ASD after 3 years of age. (C) Box plot of profile monitoring (PM) scores between NT and ASD at different ages. The PM score, that is, the microbial relationship alteration, is significantly reduced in NT subjects and children with ASD after 3 years of age when compared with that for children under 3 years of age (Wilcoxon signed-rank test, p<2.2e-16). (D, E) The isometric log-ratio transformed the abundance scatter plot of Fenollaria and Chloroplast before and after 3 years of age, respectively. The relationship beween these two taxa is significantly altered between NT and ASD groups under 3 years of age. This difference disappeared in NT subjects and children with ASD after 3 years of age. (F) Expanded alteration in the microbial community relationship with increasing ASD score. The edge width is proportional to the linear slope in the regression of the PM score to ASD symptom severity. (G) The identified PM score for 54 altered genera relationships is shown in figure 3F, which increased with ASD severity. (H–K) The isometric log-ratio transformed abundance scatter plot of (Ruminococcus)_gauvreauii_group and Coprobacillus in different ASD symptom severity groups. The microbial community alteration networks in A, B and F were derived using PM2RA. The edge width represented the interaction of the PM score. The node size represents the relative abundance change, as well as the label of the nodes specified taxonomic affiliation. The red node represents the increase of taxon abundance in ASD, and the green nodes represent the decrease. p<0.05 *, p<0.01 **, p<0.001 *** and p<0.0001 ****. ASD, autism spectrum disorder; NT, neurotypical.

Figure 4

Age-specific taxonomic and microbial-metabolic signatures in the ASD group. (A) Column chart illustrating the average relative abundance of the 20 taxa with significant abundance changes across different age brackets between ASD and NT (20 adults included). Only columns with significant abundance changes at a specific age are circled on their top surface (based on the Kruskal-Wallis test, and p values were detailed in online supplemental table S13. (B) Heat map showing the significant changes in microbial metabolic functions across age. Functions annotated as GBM (left) or a member of the biocycle (METACYC, right) were exhibited (based on the Kruskal-Wallis test, and p values are shown in online supplemental table S14 and S15). (C–E) Relationship between microbial taxa/function and clinical phenotypes. Correlations with p<0.05 were visualised based on Spearman’s correlation coefficient. The circle size represents both the grouping schemes used to calculate the correlation and the degree of significance (only in the two bigger circles). From left to right, the two bigger circles represent the grouping schemes ≤3 years or >3 years, and the small circles represent the age brackets according to the age axis of figure 3A. Correlations between the significantly altered microbial genera and predicted microbiota age were obtained from all subjects (C). To show the relationships between GBM and phenotype more intuitively, functions with significant correlations between phenotypes (all four ASD-related phenotypes) among subjects aged ≤3 years or >3 years or ≥4 individual subdivided age brackets are visualised (D). Correlations between METACYC and phenotypes are shown as circles (E). The values of rho and the p value of each taxa/GMB/METACYC for each phenotype are shown in online supplemental table S16–S18. ASD, autism spectrum disorder; GBM, gut-brain modules; NT, neurotypical.

Figure 5

Diagnostic potential of gut microbiota in current and validation sets. (A) Microbial taxonomic and metabolic markers for detecting subjects with ASD were identified from random-forest classifiers based on the genus, GBM and METACYC. The x-axis represents the mean SHAP value (average impact on model output magnitude) of the features to the model prediction in each test (see ‘Methods’ section). The length of the column represents the total SHAP values of a specific marker by summing the SHAP value of ASD (red) and NT (blue). (B) A decision tree heat map for predicting whether a subject is diagnosed with ASD (purple) or NT (yellow). The heat map colours indicate the value of a sample relative to the rest of the group for each feature. (C) Performance of the classifiers using AUCs in both current (solid red line) and validation (solid blue line) sets. (D) AUC values in different prediction groups. (E) AUC values of all features and top 20 features in the current cohort across age brackets. AUC, area under the curve; ASD, autism spectrum disorder; GBM, gut-brain modules; NT, neurotypical; OTU, operational taxonomic unit.