Dynamics and Crosstalk between Gut Microbiota, Metabolome, and Fecal Calprotectin in Very Preterm Infants: Insights into Feeding Intolerance

Abstract

Background: Feeding intolerance (FI) is a significant concern in the care of preterm infants, impacting their growth and development. We previously reported that FI is linked to lower fecal calprotectin (FC) levels. This study aims to explore the postnatal dynamics and interplay between microbiota, metabolic profiles, and host immunity in preterm infants with and without FI.

Methods: Infants with gestational age <32 weeks or birth weight <1500 g were enrolled at the Children's Hospital of Fudan University between January 2018 and October 2020. Weekly fecal samples were analyzed for bacterial profiling, metabolome, and calprotectin levels, exploring their longitudinal development and interrelationships.

Results: Of the 118 very preterm infants studied, 48 showed FI. These infants experienced an interrupted microbial-immune trajectory, particularly at 3-4 weeks of age, marked by a reduced bacterial abundance, alpha diversity, and FC levels. Metabolic changes in FI were pronounced between 3 and 6 weeks. Pantothenic acid and two polyamine metabolites were closely associated with bacterial abundance and FC levels and negatively correlated with the duration to attain full enteral feeding.

Conclusions: FI infants demonstrated compromised microbiome-immune interactions, potentially influenced by specific metabolites. This research underscored the importance of early microbial and metabolic development in the pathogenesis of FI in very preterm infants.

Keywords: calprotectin; feeding intolerance; microbiome; pantothenic acid; polyamine metabolites; preterm infants.

Figure 1

Study design, sample characteristics, and clinical differences in infants with and without feeding intolerance. (A) Flow chart of study design. (B) Fecal samples were collected at a weekly routine of all enrolled infants during hospitalization. X axis and Y axis represent different days of postnatal age and different infants, respectively. Fecal samples were all analyzed for calprotectin level (red color). Fecal samples in blue and yellow color were further analyzed for microbiome (blue color) or multi-omics (microbiome and metabolome, yellow color), respectively. (C,D) Differences in time to reach full enteral feeding (C) and days of hospitalizations (D) between infants with and without feeding intolerance, respectively. Yellow and blue color for infants with and without FI, respectively. (E) Correlations between days to full enteral feeding and days to discharge using Pearson’s method. Yellow and blue color for infants with and without FI, respectively.

Figure 2

Postnatal dynamics and clinical associations of gut microbiome and metabolome in preterm infants in NICU. (A,B) Longitudinal postnatal development of gut microbiome at phylum level through compositional (A) and quantitative (B) perspectives, respectively (n = 294). (C) Effects of different clinical factors on absolute bacterial abundance, estimated absolute abundance of Firmicutes, and relative abundance of Firmicutes, respectively. Error bars indicate 95% confidence intervals. (D) Principal coordinate analysis (PCA) based on the Euclidean distance of the metabolic signatures of all samples. Dots in different grayscale images represent samples collected at different postnatal ages. (E) Multi-variate PERMANOVA analyses revealed the effects of different clinical factors on fecal metabolome profiles (n = 225). Percentages shown are the percentage of the variation explained by the corresponding factors, adjusted for postnatal age. Bars in gray color represent factors that did not have a significant impact on the variance of metabolome profiles (p > 0.05).

Figure 3

Compromised development of the gut microbiome is associated with altered calprotectin level in feeding intolerance. (A) Longitudinal change of microbiome profiles in infants with and without feeding intolerance (FI) using principal coordinate analysis (PCoA) based on the unweighted UniFrac dissimilarity at the genus level (n = 294). Percentages shown are the percentage of the variation explained by the corresponding principal coordinate, and the first two are displayed. (B–D) Longitudinal changes of Shannon index (B), OTU richness (C), and absolute bacterial abundance (D) in infants with and without FI, respectively. (E) The estimated bacterial abundance of all phyla and major genera of samples at 3–4 weeks of age that were significantly altered in infants with and without FI. (F) Longitudinal changes of fecal calprotectin (FC) levels in infants with or without FI, respectively. (G) Correlations between FC level at 3–4 weeks of age and time to reach full enteral feeding using Spearman’s method (n = 110). (H,I) Correlations between FC level and microbial OTU richness (H) and absolute abundance (I) in samples collected at and larger than 2 weeks of age using Spearman’s method (n = 235). Group-wise comparisons are shown using at-risk. Asterisks represent p values: ns: p > 0.05, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Yellow and blue colors indicate infants with and without FI, respectively.

Figure 4

Potential roles of metabolites in the early development and crosstalk between microbiota and gut immunity. (A) Longitudinal change of metabolome profiles in infants with and without FI using principal components analysis (n = 225). Percentages shown are the percentage of the variation explained by the corresponding principal components, and the first two are displayed. (B) Correlation between the longitudinal development of microbiome and metabolome profiles using Procrustes analysis (n = 170). (C) Longitudinal change of eigen value of the ‘pink’ module in infants with and without feeding intolerance, respectively. (D) Top five enriched metabolic pathways from Small Molecule Pathway Database (SMPDB) in the ‘pink’ module using MSEA. The enrichment ratio represents the ratio of the Q-statistic for each pathway to the expected statistic. (E) Correlations between the eigen values of different metabolite modules and sample conditions (age and feeding volume at the time of collection), fecal calprotectin level, and microbial characteristics (total bacterial load and alpha diversity) using Spearman’s method. Only significant correlations are shown (p < 0.05). (F) Scatter plots reveal the correlations between intensities of metabolites in the ‘pink’ module and calprotectin level (X axis) and between metabolites intensities and total bacterial load (Y axis), respectively (n = 152). (G) Mediation linkages among the fecal bacterial abundance, metabolite intensity, and fecal calprotectin level. pACEM and pADE were estimated via the bidirectional mediation analysis. (H) Correlations between fecal metabolite intensities and days to reach full enteral feeding using Spearman’s method (n = 70). Group-wise comparisons are shown using at-risk. Asterisks represent p values: ns: p > 0.05, * p < 0.05, ** p < 0.01, *** p < 0.001. Yellow and blue colors indicate infants with and without feeding intolerance, respectively.