Cardiac Intensivist The Children's Heart Clinic/Children's Minnesota Minneapolis, Minnesota, United States
Abstract: Background Thrombosis is a major complication after cardiac surgery in children with congenital heart disease (CHD), and it significantly increases the risk of mortality and other complications and prolongs hospital stay. The underlying mechanisms for the development of thrombosis after cardiac surgery in children with CHD remain poorly understood. Aim To identify novel circulating metabolites before cardiac surgery that are associated with thrombosis and to develop a prediction model for thrombosis after surgery in children with CHD. Methods Untargeted metabolomic data on 1,115 metabolites were measured in plasma from 203 children (age range: 0 days-18 years) with CHD undergoing cardiac surgery. A total of 776 metabolites met quality control criteria (call rate>75% and Spearman correlation coefficient between blind duplicates>0.5). Logistic regression models were used to examine the associations of individual metabolites with, and elastic net regression models were used to select a prediction model for, thrombosis after surgery in children with CHD. Results In total, 26 children (13%) developed thrombosis. Among the 776 metabolites, 195 were significantly associated with thrombosis (false discovery rate < 0.05). The top three metabolites showing the strongest associations with thrombosis were eicosapentaenoate (EPA; 20:5n3), andro steroid monosulfate C19H28O6S, and formiminoglutamate (false discovery rate = 0.01). Pathway analysis showed that pathways of nicotinate and nicotinamide metabolism (P = 7.25×10-4) and glycerophospholipid metabolism (P = 0.002) were significantly enriched and had significant impact on the development of thrombosis. In elastic net regression analysis, the area under the characteristic operating receiver curve of a prediction model for thrombosis was 0.94 in the train sample (70% of the total sample ) and 0.84 in the testing sample (the remaining 30% of the total sample). Conclusion We have identified promising novel metabolites and metabolic pathways associated with thrombosis and developed a prediction model for the development of thrombosis after surgery in children with CHD.
