We selected 33 children with GHD diagnosed at Longgang District Maternity and Child Health Hospital as the disease group, and 24 healthy children of the same age as the control group. The ages of children in the two groups ranged between 5 and 14 years, with no statistical difference seen in the comparative analysis (P > 0.05) (Table 1). All the children with GHD were diagnosed at the Department of Growth and Development, Shenzhen Longgang District Maternity and Child Health Hospital. The disease group met the diagnostic criteria for GHD in Chinese children (15): ① Below the third percentile of the height of normal healthy children of the same age and sex (−1.88 standard deviations [−1.88 SD] or minus 2 standard deviations [−2 SD]); ② Annual growth rate <5 cm/year; ③ Symmetrical dwarfism and childish face; ④ Normal intelligence development; ⑤ Bone age lagging behind actual age; ⑥ Peak values of two GH drug provocation tests of <10 µg/L; ⑦ Lower than normal level of serum IGF-1. The exclusion criteria for children in the two groups included: ① Severe liver or gastrointestinal disorders; ② Severe infection; ③ Treatment with antibiotics or probiotic preparations within one month before the test. All children provided informed consent from their guardians before enrollment.

The ade4 package in R (v3.3.3) software was used to perform principal component analysis (PCA) based on the composition and relative abundance of bacteria in all samples at the genus level. The overall distribution of the microbiota compositions in the two groups was plotted. Bacteria were classified to the phylum and genus levels, and different species between the two groups were investigated by the Wilcoxon method where P < 0.05 indicated a significant difference. The 16S rDNA sequencing data were used to evaluate differences in bacterial functions between the two groups of children based on the functional analysis performed by the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. SPSS 22.0 software was used for general data analysis. The age, weight, height, and IGF-1 values were compared by χ² tests or two-group independent sample t-tests.

The GM of two groups of children showed significant differences in α-diversity (P = 0.033) (Figure 1). We used PCA to reduce the dimensionality of the GM data of the two groups, finding that there were marked differences in the GM between the two groups. The genera that contributed most to this difference included Prevotella (P < 0.001), Megamonas (P = 0.01), Bacteroides (P = 0.765), Bifidobacterium (P = 0.011), and Faecalibacterium (P = 0.094) (Figure 2).

The top five dominant bacterial phyla differed between the groups with a significant increase in the abundance of Bacteroides in the disease group (P = 0.000) together with a significant reduction in the abundance of Firmicutes (P = 0.000). In addition, there was also a significant difference between the two groups in the abundance of Fusobacteria and Actinomycetes (P < 0.05) (Table 2 and Figure 3).

We selected the top 15 dominant bacterial genera in the two groups for comparison. The results showed that the abundance of Prevotella, Fusobacterium, Klebsiella, and Alistipes was significantly increased in the disease group (P < 0.05) while that of Lachnospiracea incertae sedis, Megamonas, Blautia, Clostridium XlVa, Bifidobacterium, and Eubacterium was significantly decreased (P < 0.05) (Table 3 and Figure 4).

In comparison with the healthy children, the GHD patients showed significant changes in GM functions, including the decreased “Membrane transport” (P < 0.001, FDR < 0.001), “Lipid metabolism” (P = 0.025, FDR = 0.042), and “Transcription” (P < 0.001, FDR < 0.001, Figure 5), which indicated the. In contrast, the functional categories, such as “Energy metabolism” (P < 0.001, FDR < 0.001), “Metabolism of cofactors and vitamins” (P < 0.001, FDR < 0.001), “Nucleotide metabolism” (P = 0.008, FDR = 0.016), “Glycan biosynthesis and metabolism” (P < 0.001, FDR < 0.001), and “Folding sorting and degradation” (P < 0.001, FDR < 0.001) were enriched in the GHD patients (Figure 5). These elevated GM metabolic activities in the GHD patients, especially the “Glycan biosynthesis and metabolism” function, affect the neuro-regulations in hosts and is probably related to the occurrence of GHD.

Spearman's correlation analysis was used to investigate associations between the GM of children with GHD and eight endocrine hormones. Our results showed that Bacteroides were positively correlated and Prevotella was negatively correlated with insulin, while Alistipes and Haemophilus showed a negative correlation with GH. A positive correlation was also reported between Fusicatenibacter, Fusobacterium, and Sutterella, whereas Veillonella was negatively correlated with prolactin. Faecaliterium and FSH were positively correlated (Figure 6).

Compared with the healthy controls, children in the disease group showed reduced α-diversity in the GM, consistent with results reported in malnourished children (16). The phylum Bacteroides was more abundant in children from the disease group than in those from the control group while the opposite trend was observed for Firmicutes, in contrast to findings on obese and diabetic patients (17). The abundance of Prevotella in the disease group was also significantly higher than that in the control group. Prevotella can degrade broad-spectrum plant polysaccharides (18), and carbohydrate-based diets tend to form a Prevotella-dominated “gut type.” Increased abundance of Prevotella abundance has been shown to reduce blood sugar and insulin levels, thus affecting energy absorption and promoting weight loss (19). In the disease group, the abundance of Fusobacterium, Klebsiella, Alistipes, and other genera was found to be significantly increased. Fusobacterium is present in the normal oral flora and can inhibit the immune response as well as promote the transformation of inflammation to malignancy (20). An increase in the abundance of both Klebsiella and Alistipes has been shown to be associated with intestinal inflammation (21, 22); therefore, the increase in the population of these genera can promote chronic inflammation in the intestine and disrupt the function of the intestinal barrier. This can lead to a cellular biochemical imbalance, reduced absorption capacity, and increased susceptibility to enteric pathogen infections, and consequently affect energy metabolism and nutrient absorption (23). In addition, Klebsiella and Alistipes are both associated with neurological diseases (24) and can produce neurotransmitter-related metabolites such as serotonin, dopamine, and histamine (25). These neurotransmitters enter the brain through the gut-brain axis to regulate the energy balance and function of the hypothalamus (26, 27). The hypothalamus is the highest regulatory center of thehypothalamic–pituitary–growth axis (HPA) and can reduce appetite and cause weight loss (28, 29). Lachnospiracea incertae sedis, Megamonas, Blautia, Clostridium XlVa, and Bifidobacterium were found to be significantly reduced in the intestines of the disease group, which could lead to reduced concentrations of SCFAs such as butyric acid produced by these beneficial bacteria (30). Decreased abundance of Lachnospiracea incertae sedis might also affect protein synthesis (31), disturb the intestinal energy supply, and retard growth and development. Jensen et al. (5) reported that increased Prevotella abundance together with reduced numbers of Bifidobacterium can reduce the levels of growth hormone-releasing peptide (GHRP) and leptin, thus reducing the release of GH.

The enriched functional categories in the GHD group included “Replication and repair, Energy metabolism, Poorly characterized, Metabolism of cofactors and vitamins, Nucleotide metabolism, Cellular processes and signaling, Nucleotide metabolism, Glycan biosynthesis and metabolism, Transcription, Folding sorting and degradation”. Children with GHD showed dysregulation in energy metabolism, vitamin and related factor metabolism, and polysaccharide metabolism and biosynthesis. Considering that Prevotella significantly increases the catabolism of carbohydrates and that the abundance of butyric acid and other bacteria such as Lachnospiracea incertae sedis is significantly reduced in children with GHD, the GM imbalance in this population may affect the functions of the flora. This phenomenon may lead to chronic inflammation of the intestine and poor intake and absorption of nutrients such as fats and proteins, affecting both growth and development.

We conducted a correlation analysis of the GM and endocrine hormones and found that Prevotella abundance was negatively correlated with insulin. Significantly higher abundance of Prevotella can affect insulin secretion, which can not only regulate food intake (19, 32) but also modulate blood glucose levels through signaling pathways essential for maintaining energy storage, glucose metabolism, sugar production, adipogenesis, cell growth, survival, and reproduction (33). We speculate that this significant increase in Prevotella abundance may be detrimental to growth and development. We also found a variety of other intestinal bacteria related to endocrine hormones and confirmed the interaction between the GM and endocrine hormones. Maintaining the stability of the GM is conducive to the promotion of growth and development.

There are many factors that cause insufficiency in GH secretion in children with GHD, and GM imbalance may be one of the major factors. On the one hand, GM imbalance leads to the abnormal secretion of endocrine hormones as well as an abnormal production of microbial metabolites, especially neurotransmitters, that can influence the HPA through the gut–brain axis. The sample size in the present study was small, consisting of only 33 children with GHD; hence, large-sample, multi-center research is needed to verify the associations between the GM and GHD. Studies combined with metabolomics could better clarify the mechanism of action of the GM and its metabolites in growth and development.