Fecal sphingolipids predict parenteral nutrition-associated cholestasis in the neonatal intensive care unit

The use of machine learning in paediatric nutrition

PURPOSE OF REVIEW

In recent years, there has been a burgeoning interest in using machine learning methods. This has been accompanied by an expansion in the availability and ease of use of machine learning tools and an increase in the number of large, complex datasets which are suited to machine learning approaches. This review summarizes recent work in the field and sets expectations for its impact in the future.

RECENT FINDINGS

Much work has focused on establishing good practices and ethical frameworks to guide the use of machine learning in research. Machine learning has an established role in identifying features in 'omics' research and is emerging as a tool to generate predictive models to identify people at risk of disease and patients at risk of complications. They have been used to identify risks for malnutrition and obesity. Machine learning techniques have also been used to develop smartphone apps to track behaviour and provide healthcare advice.

SUMMARY

Machine learning techniques are reaching maturity and their impact on observational data analysis and behaviour change will come to fruition in the next 5 years. A set of standards and best practices are emerging and should be implemented by researchers and publishers.

Multi-Omics Unravels Metabolic Alterations in the Ileal Mucosa of Neonatal Piglets Receiving Total Parenteral Nutrition

Total parenteral nutrition (TPN) is life-saving therapy for the pediatric patients with intestinal failure (IF) who cannot tolerate enteral nutrition (EN). However, TPN-induced metabolic alterations are also a critical issue for the maintenance of intestinal homeostasis, and thus the global metabolomic signatures need to be addressed. In this study, ileal mucosal biopsies were collected from 12 neonatal Bama piglets receiving either EN or TPN for 14 days, and changes in the intestinal metabolism were examined by multi-omics (HM350 Metabolomics + Tandem Mass Tag (TMT)-based proteomics). As a result, a total of 240 compounds were identified by metabolomics, including 56 down-regulated and 9 up-regulated metabolites. Notably, tissue levels of fatty acyl-carnitines (decreased by 35-85%) and succinate (decreased by 89%) dramatically decreased in the TPN group, suggestive of disrupted processes of fatty acid oxidation (FAO) and the citrate cycle, respectively. Interestingly, however, no differences were found in the production of adenosine 5'-triphosphate (ATP) between groups, suggesting that these dysregulated metabolites may have mainly led to the loss of bioactive compounds rather than energy deficit. Additionally, 4813 proteins were identified by proteomics in total, including 179 down-regulated and 329 up-regulated proteins. The analysis of protein-protein interactions (PPI) indicated that most of the differentially expressed proteins were clustered into "lipid metabolism" and "innate immune responses". In summary, this work provided new findings in TPN-induced intestinal metabolic alterations, which would be useful to the improvement of nutritional management for IF patients.

Untargeted Metabolomics Reveal Parenteral Nutrition-Associated Alterations in Pediatric Patients with Short Bowel Syndrome

Short bowel syndrome (SBS) is a major cause of intestinal failure (IF) that may require long-term parenteral nutrition (PN) support. However, long-term PN is accompanied by severe complications such as catheter-related blood stream infection (CRBSI) and intestinal failure-associated liver disease (IFALD), and it is associated with high healthcare costs. In this study, we characterized the plasma metabolomic profile and investigated the role of metabolism in predicting long-term PN in pediatric patients with SBS. Untargeted metabolomics was performed in plasma samples from 20 SBS patients with PN support: 6 patients had IFALD and 14 patients had no liver disease. As controls, 18 subjects without liver or intestinal diseases were included for the analysis. SBS patients had distinct plasma metabolomic signatures compared to controls, and several pathways associated with amino acid metabolism and cell death were significantly changed. The presence of IFALD in SBS was associated with alterations of metabolites mainly classified as “amino acids, peptides, and analogues” and “benzene and derivatives”. Serum direct bilirubin levels were negatively correlated with levels of uridine, skatole, and glabrol. Importantly, SBS patients with long-term PN showed significantly increased levels of glutamine compared to those in the short-term PN group. Finally, using multivariate logistic regression analysis, we developed a prediction model including glutamine and creatinine to identify pediatric SBS patients who need long-term PN support. These findings underscore the potential key role of the metabolome in SBS with IF and suggest that metabolomic profiles could be used in long-term PN assessment.