Alterations in Gut Microbiota Composition Are Associated with Changes in Emotional Distress in Children with Obstructive Sleep Apnea

Abstract

Emerging evidence underscores the pivotal role of the gut microbiota in regulating emotional and behavioral responses via the microbiota-gut-brain axis. This study explores associations between pediatric obstructive sleep apnea (OSA), emotional distress (ED), and gut microbiome alterations before and after OSA treatment. Sixty-six children diagnosed with OSA via polysomnography participated, undergoing adenotonsillectomy alongside routine educational sessions. ED was assessed using the OSA-18 questionnaire, categorizing participants into high ED (scores ≥ 11, 52%) and low ED (scores < 11, 48%) groups. Gut microbiome analysis revealed significant diversity differences, with high ED linked to a reduced Shannon index (p = 0.03) and increased beta diversity (p = 0.01). Three months post-treatment, significant improvements were observed in OSA symptoms, ED scores, and gut microbiome alpha diversity metrics among 55 participants (all p < 0.04). Moreover, changes in the relative abundances of Veillonella, Bifidobacterium, Flavonifractor, and Agathobacter, as well as ultra-low frequency power and low frequency power of sleep heart rate variability, were independently associated with ED score alterations. These findings underscore the gut microbiome's critical role in the emotional and behavioral symptoms associated with pediatric OSA, suggesting that microbiome-targeted interventions could complement traditional treatments for ED reduction and emphasizing the need for further research.

Keywords: adenotonsillectomy; children; dietary profile; educational counseling; emotional distress; gut microbiome; obstructive sleep apnea; sleep heart rate.

Figure 1

Case flow diagram. Abbreviations: HRV, heart rate variability; OSA, obstructive sleep apnea.

Figure 2

Distributions of the gut microbiota amplicon sequence variants (ASVs) between high emotional distress (ED) and low ED groups in pediatric obstructive sleep apnea. (a) The UpSet plot illustrates the unique and shared ASVs between the low ED group (3485 ASVs) and the high ED group (3258 ASVs) at the pre-treatment stage. (b) The UpSet plot illustrates the unique and shared ASVs between the low ED group (3238 ASVs) and the high ED group (3077 ASVs) at the post-treatment stage.

Figure 3

Gut microbial distributions of the top 30 genera between the high and low emotional distress (ED) groups in pediatric obstructive sleep apnea. (a) The heatmap illustrates significant differences in the microbial distribution of the top 30 genera between the high and low ED groups at the pre-treatment stage, with statistical significance confirmed by permutational multivariate analysis of variance with Benjamini–Hochberg correction (p = 0.01). (b) The heatmap illustrates a similar microbial distribution of the top 30 genera between the high and low ED groups at the post-treatment stage (p = 0.33).

Figure 3

Gut microbial distributions of the top 30 genera between the high and low emotional distress (ED) groups in pediatric obstructive sleep apnea. (a) The heatmap illustrates significant differences in the microbial distribution of the top 30 genera between the high and low ED groups at the pre-treatment stage, with statistical significance confirmed by permutational multivariate analysis of variance with Benjamini–Hochberg correction (p = 0.01). (b) The heatmap illustrates a similar microbial distribution of the top 30 genera between the high and low ED groups at the post-treatment stage (p = 0.33).

Figure 4

Gut microbiome stratification between high and low emotional distress (ED) groups in pediatric obstructive sleep apnea. (a) The scatter boxplot illustrates a significantly lower Shannon index in the high ED group compared to the low ED group at the pre-treatment stage (p = 0.03). (b) Principal coordinates analysis (PCoA) highlights significant differences in Bray–Curtis distances between the two ED groups at the pre-treatment stage. (c) The scatter boxplot shows significantly greater microbial community variability in the high ED group compared to the low ED group at the pre-treatment stage (p = 0.01). (d) The scatter boxplot indicates comparable Shannon indices between the two ED groups at the post-treatment stage (p = 0.87). (e) PCoA highlights differences in Bray–Curtis distances between the two ED groups at the post-treatment stage. (f) The scatter boxplot shows a significant increase in microbial community variability within the high ED group compared to the low ED group at the post-treatment stage (p = 0.01). Shannon indices were compared using the independent-samples t-test, and Bray–Curtis distances were compared using permutational multivariate analysis of variance with Benjamini–Hochberg correction.

Figure 5

Heatmap of significant associations (−log(q value) × sign(coefficient)) between the top 30 genera and variables of interest. (a) Heatmap showing significant associations across five genera, emotional distress (ED) score, and male sex. (b) Heatmap depicting significant associations across five genera, semi-skimmed milk, whole-meal bread, and rice consumption. (c) Heatmap illustrating significant associations across three genera, very low frequency (VLF) power, and ultra-low frequency (ULF) power. Positive signs indicate enrichment, and negative signs indicate depletion of the corresponding genera relative to increasing variables of interest. The analysis was conducted using the Maaslin2 R package (v1.15.1).