doi: 10.1093/nsr/nwae352.
eCollection 2024 Oct.
Cord blood ceramides facilitate early risk identification into childhood metabolic health
Affiliations
- PMID: 39474264
- PMCID: PMC11519047
- DOI: 10.1093/nsr/nwae352
Item in Clipboard
Cord blood ceramides facilitate early risk identification into childhood metabolic health
Natl Sci Rev.
.
No abstract available
Figures
Association between cord blood Cer and offspring metabolic health. (A) Cord blood Cer were associated with infant birth weights. Forest plot illustrates differential lipids obtained using the limma model for LBW relative to NBW (LBW-NBW) and HBW relative to NBW (HBW-NBW). Log2 fold change (95% CI), P value, and false discovery rate (FDR, 0.05; Benjamini–Hochberg method) are presented. Blue rectangle contains Cer with V-pattern of changes from LBW through NBW to HBW. Red dots indicate Cer significantly upregulated relative to NBW, while grey dots denote non-significant changes. Lipid species are indicated on the y axis. (B) V-pattern of changes in medium-to-long chain cord blood Cer across different birth weight groups illustrated as line plots drawn using z-scores. (C) Correlation between cord blood Cer and clinical indices. Bubble plots illustrate the Spearman correlations between cord blood lipids and maternal indices and other indicators relevant to cord blood biochemistry in LBW, NBW and HBW groups. Only statistically significant correlations are shown. Size of the bubble represents magnitude of P values from Spearman correlations, with positive correlations in red and negative correlations in blue. Color intensity denotes the strength of correlation. (D) Trans-omics integration revealed molecular pathways associated with changes in cord blood very long-chain (C20-C26) Cer. Over-representation analysis (ORA) of pathways using the Reactome database based on placental proteins correlated (P <0.05, |r|>0.7) with very long-chain (C20-C26) Cer in LBW + NBW, and HBW + NBW, respectively. Significant pathways (P <0.05) were ranked based on density of networks. Pathways boxed in red denote common pathways associated with cord blood Cer levels for both LBW + NBW and HBW + NBW analyses, while the pathway boxed in blue was distinct to LBW + NBW analysis. (E) Correlation networks of proteins and lipids categorized under the Reactome pathway R-HSA-2408522 seleno amino acid metabolism for the LBW + NBW and HBW + NBW groups, respectively. SEPSECS, a key enzyme in the biosynthesis of selenoproteins, was negatively correlated to several very long-chain Cer for both LBW and NBW. Correlation networks of proteins and lipids categorized under the Reactome pathway R-HSA-2132295 MHC class II antigen for the LBW + NBW group. Red and blue numerals on edges indicate positive and negative correlation coefficients, respectively; red nodes denote lipids and blue nodes denote proteins. (F) Correlation analysis of umbilical cord blood Cer with clinical parameters. Spearman correlation coefficients were calculated and shown in a heatmap plot. Significant correlations were indicated by color, with positive correlations in red and negative correlations in blue. (G) Baseline levels of cord blood Cer at birth predict childhood lipid metabolism at the 7th year. LDL-Cho at the 7th year were divided into tertiles, labeled as groups 1, 2 and 3. (H) ROC curves on model performance of lasso-selected variables in discriminating different tertiles of LDL-Cho at the 7th year. The cohort was randomly divided into training (90%) and validation (10%) sets. Ten-fold cross validation was conducted and ROC curves were created using the combined training and validation samples. (I) Corresponding changes in AUC of ROC curves distinguishing participants in the 3rd from the 2nd tertiles based on validation set when removing variables one by one. (J) Boxplots showed changes in cord blood Cer d18:1/20:0 and Cer d18:0/20:0 across the three tertiles. P values were from ANOVA test for linear regression analysis. NBW, normal birth weight; LBW, low birth weight; HBW, high birth weight; Cer, ceramides; APN, adiponectin; IGF-1, insulin-like growth factor 1; BMI, body mass index; SF: skinfold thickness; WC: waist circumference; Glu0: fasting glucose; HOMA-IR, homeostatic model assessment for insulin resistance; TC, total cholesterol; HDL-Cho, high density lipoprotein cholesterol; TG, triglycerides; LDL-Cho, low density lipoprotein cholesterol; INS_index, insulinogenic index; QUICKI, quantitative insulin sensitivity check index; ODI: oral disposition index; AUC, area under the curve; ROC, receiver operating characteristic curve.