Maternal lipidomic signatures in relation to spontaneous preterm birth and large-for-gestational age neonates

Abstract

Lipidome-wide metabolites may be useful biomarkers of pregnancy outcomes. We sought to characterize maternal lipidomic signatures associated with preterm birth and neonatal anthropometric parameters. Plasma samples were collected 24-28 weeks gestation, and lipidomic profiling was quantified using high-performance liquid chromatography tandem mass spectrometry. Lipid metabolites were analyzed individually and as whole lipid classes and subgroups based on degree of hydrocarbon chain saturation. Associations were estimated using linear and logistic regression. After false discovery adjustment (q < 0.15), four plasmenyl-phosphatidylethanolamines and three free fatty acids associated with increased risk for spontaneous preterm birth. Five phosphatidylinositols, two phosphatidylglycerols, and one phosphatidic acid were associated with large for gestational age neonates. The saturated plasmenyl-phosphatidylethanolamines held the association with increased risk for spontaneous preterm birth. Both the mono- and poly-unsaturated free fatty acids held the association for increased risk for spontaneous preterm birth. Mono- and poly-unsaturated phosphatidylinositols were associated with large for gestational age neonates. Whole lipid classes (plasmenyl-phophatidylcholines and plasmenyl-phosphatidylethanolamines) were associated with increased risk for large for gestational age at delivery. This study provides evidence that finer omics-scale analysis of the maternal lipidome may be more informative biomarkers of pregnancy outcomes compared to whole class level lipid analysis.

Figure 1

Correlation matrix among lipid hydrocarbon saturation subgroups.

Figure 2

Volcano plots of associations [-log₁₀(p-values)] between lipidome-wide metabolites and pregnancy outcomes: (A) gestational age at delivery; (B) preterm birth (n_cases = 31,n_controls = 69); (C) spontaneous preterm birth (n_cases = 19,n_controls = 69); (D) birthweight z-score; (E) small for gestational age [SGA] (n_cases = 16,n_controls = 75); (F) large for gestational age [LGA] (n_cases = 9,n_controls = 75). Point estimates labeled in red are those that were adjusted for false discovery rate (q-value < 0.15). Models adjusting for maternal age, maternal education, fetal sex, pre-pregnancy BMI, and weight gain.

Figure 3

Forest plots of effect estimates of associations between hydrocarbon chain subgroups of lipid classes and pregnancy outcomes: (A) gestational age at delivery; (B) preterm birth (n_cases = 31,n_controls = 69); (C) spontaneous preterm birth (n_cases = 19,n_controls = 69); (D) birthweight z-score; (E) small for gestational age [SGA] (n_cases = 16,n_controls = 75); (F) large for gestational age [LGA] (n_cases = 9,n_controls = 75). Models adjusting for maternal age, maternal education, fetal sex, pre-pregnancy BMI, and weight gain. Associations highlighted in red have p-values < 0.05, but all associations had corresponding q-values exceeding 0.2.

Figure 4

Forest plot of effect estimates of associations between whole lipid classes and pregnancy outcomes: (A) gestational age at delivery; (B) preterm birth (n_cases = 31,n_controls = 69); (C) spontaneous preterm birth (n_cases = 19,n_controls = 69); (D) birthweight z-score; (E) small for gestational age [SGA] (n_cases = 16,n_controls = 75); (F) large for gestational age [LGA] (n_cases = 9,n_controls = 75). Models adjusting for maternal age, maternal education, fetal sex, pre-pregnancy BMI, and weight gain. Associations highlighted in red have p-values < 0.05, but all associations had corresponding q-values exceeding 0.2.