(P27-082-25) Network Analysis Reveals Associations Between Maternal-Infant Multi-Omics and Infant Anthropometry at Birth in the Women First Maternal Periconception Nutrition Trial

Objectives: Underlying molecular links between pre-conception nutrition and infant growth are largely unknown. Using a multi-omic network analysis, we aimed to identify whether relationships between maternal and infant multi-omic predictors and birth anthropometrics differ by timing of maternal nutrition supplementation.

Methods: This ancillary analysis included 138 Guatemalan mother-infant dyads from the Women First Trial. Mothers received small-quantity lipid-based nutrient supplement (sqLNS) pre-conception (Arm 1, n = 46), at 12 weeks gestation (Arm 2, n = 47), or no sqLNS (Arm 3, n = 45). Circulating maternal targeted one-carbon metabolites at pre-conception and at 34 weeks of gestation (wk 34); wk 34 maternal gut microbiome via 16S rRNA gene sequencing; and DNA methylation (DNAme) from amnion tissue via Roche NimbleGen SeqCap Epi CpGiant were measured. Infant length-for-age Z-score (LAZ) and weight-for-age Z-score (WAZ) were assessed at birth. With the Visualization of Linear Regression Elements (VOLARE) method, we conducted 32,605 pairwise regressions (m1 ~ arm + m2 + arm x m2, where m1 and m2 are biomarkers from different assays or birth outcomes). We created a network using the 25 best models per assay pair with p-value < 0.025 comparing slopes of Arms 1 and 2 (n=104). Effect sizes are reported as the change in the estimated outcome when the predictor changes from the 25^th to the 75^th percentile (i.e., interquartile range).

Results: The association between wk 34 methionine and LAZ was negative in Arm 1 (-0.08) and positive in Arm 2 (0.45), p = 0.013. Prevotella and LAZ were associated positively in Arm 1 (0.45) and negatively in Arm 2 (-0.50), p = 0.003. Wk 34 methionine (-0.05), alanine (-0.07), and histidine (-0.08) were negatively associated with WAZ in Arm 1, while positively associated in Arm 2 (methionine: 0.46, alanine: 0.53, histidine: 0.42). Slopes for these metabolites differed between Arms 1 and 2 (methionine: p = 0.009, alanine: p = 0.005, histidine: p = 0.021). Differences between Arms 1 and 2 were seen in 17 DNAme sites linked to wk 34 arginine (p ≤ 0.012) and 10 linked to Ruminococcaceae (p ≤ 0.005).

Conclusions: Multi-omic biomarkers were associated with infant growth, and these associations differed by timing of sqLNS suggesting molecular biomarkers may link maternal nutrition and fetal growth.

Funding Sources: Bill & Melinda Gates Foundation, NIH, ODS