Differences in gut microbiota and faecal metabolomics characteristics in preterm infants with feeding intolerance

Introduction. Feeding intolerance (FI) is one of the most common clinical issues in preterm infants, and there are currently no internationally unified diagnostic criteria.Gap Statement. Screening valuable biomarkers and evaluating their diagnostic value for FI in preterm infants is of great significance.Aim. This study aimed to identify and assess diagnostic biomarkers for feeding intolerance in preterm infants.Methodology. This study included clinical data from 49 preterm infants admitted to a tertiary maternal and child health hospital in Zhejiang's coastal region (January to June 2024). Based on feeding assessments at day 21 postpartum recorded in electronic medical records, infants were divided into feeding-tolerant (FT, n=34) and feeding-intolerant (FI, n=15) groups. Patient data analysis incorporated maternal age, gestational age, parity, antibiotic use, pregnancy complications and neonatal factors (birth weight, Apgar scores, delivery/feeding methods, vomiting, abdominal distension, gastric residuals, kangaroo care and enema use). Faecal samples underwent microbiome and metabolomic profiling to identify diagnostic biomarkers.Results. Baseline data showed no significant differences in maternal-infant characteristics between groups (P>0.05). Dynamic monitoring of feeding tolerance in 21-day-old preterm infants revealed that the incidence of vomiting, abdominal distension, abnormal intestinal morphology and gastric residual volume >30% or >2 ml kg^-1 was significantly higher in the FI group than in the FT group (P<0.001), whereas there was no significant difference in the frequency of nasogastric feeding between the two groups (P>0.05). Microbial analysis revealed enrichment of Escherichia (10.92%) and Klebsiella (6.88%) in FT infants, while FI infants specifically harboured increased Clostridium_P (3.93%), Burkholderia (4.06%) and Limosilactobacillus (4.94%). Metabolomic profiling identified significant pathway differences in ATP-binding cassette transporters (ABC transporters), carbohydrate digestion/absorption and propanoate metabolism. The receiver operating characteristic (ROC) analyses showed that metabolites arginine-proline (Arg-Pro, AUC=0.920), glutamic acid-arginine (Glu-Arg, AUC=0.873), lactaldehyde (AUC=0.900) and genera Clostridium_P (AUC=0.947), Escherichia (AUC=0.765), Staphylococcus (AUC=0.733) and Bifidobacterium (AUC=0.851) exhibited robust predictive value for FI.Conclusion. Our study demonstrates that bacterial genera such as Staphylococcus, Clostridium_P, Bifidobacterium and Escherichia in the gut microbiota, along with metabolites including Arg-Pro, Glu-Arg and lactaldehyde identified in metabolomics, can serve as diagnostic criteria for feeding tolerance in preterm infants. Klebsiella shows a certain degree of diagnostic efficacy but falls into the category of 'low accuracy', requiring comprehensive evaluation considering the research background, sample characteristics and clinical context.