| Literature DB >> 30207476 |
Mengjie Yu1, Shuilin Sun1, Jiangqing Yu2,3, Fen Du1, Shouhua Zhang4, Wenlong Yang1, Juhua Xiao5, Baogang Xie1.
Abstract
To identify and screen serum biomarkers to determine pediatric patients with congenital heart diseases (PCH) from healthy control children (NC), a total of 614 clinically diagnosed subjects from three hospitals, including 491 PCH and 234 NC, were enrolled for nontargeted proton nuclear magnetic resonance spectroscopy (1H NMR)-based and targeted ultra-high-performance liquid chromatography-tandem mass spectroscopy (UPLC-MS/MS)-based metabolomics studies. Nineteen serum metabolites distinguishing PCH from NC were identified by 1H NMR-based metabolomic analysis. The amino acid and choline metabolic pathways were considered to be closely related to PCH. The serum levels of 13 metabolites in these two pathways were further determined by UPLC-MS/MS and observed to be altered significantly in PCH. Taurine, glutamine, and glutamate presented considerable diagnostic value for the diagnosis of PCH (AUROC > 0.80). Logistic regression analysis showed that a combination of four variables, namely, betaine, taurine, glutamine, and phenylalanine, yields a high diagnostic value (AUROC = 0.949) and prediction accuracy (89.1%) for differentiating PCH from the NC, and the sensitivity and specificity were 93.9 and 95.2%, respectively. Further double-blind sample prediction showed that the accuracy of the model was 83.8% for 80 unknown samples. Our results showed that the serum amino acid and choline metabolite levels in PCH were changed considerably. The combination of four metabolites, namely, betaine, taurine, glutamine, and phenylalanine, can be used as potential serum biomarkers in PCH diagnosis, which contributes to the early PCH screening.Entities:
Keywords: diagnostic biomarkers; metabolomics; pediatric patients with congenital heart diseases; predictive modeling
Mesh:
Substances:
Year: 2018 PMID: 30207476 DOI: 10.1021/acs.jproteome.8b00466
Source DB: PubMed Journal: J Proteome Res ISSN: 1535-3893 Impact factor: 4.466