Abstract
BACKGROUND: Human milk (HM) contains bioactive constituents that continuously change throughout lactation and that are critical for optimal infant development. However, the temporal variation of the HM metabolome, its maternal nutritional determinants, and its links to infant outcomes remain insufficiently characterized in low-middle-income countries (LMIC) where childhood disease burden and HM exposure are substantial, and maternal nutritional status is poor. OBJECTIVES: We aimed to 1) characterize temporal changes in the HM metabolome, 2) quantify maternal nutritional status-related differences in HM metabolomic patterns, and 3) determine whether HM metabolomic patterns associated with maternal nutritional status are in turn associated with infant growth, in a cohort of rural Bangladeshi maternal-infant dyads. METHODS: In a lactation cohort (n = 99, 6% preterm at 34-36 wk), we characterized metabolites in HM samples at 2 (n = 94) and 5 (n = 88) mo postpartum (PP) using untargeted metabolomics profiling. We performed a paired t-test to identify metabolites that differed between timepoints and principal component analysis to consolidate metabolites measured at each timepoint into "factors," representing metabolite patterns. We used linear regression models to determine associations between maternal nutritional predictors, factor scores at 2 and 5 mo, and infant growth outcomes, and performed metabolite set enrichment analysis (MSEA) to characterize relevant pathways. P values were adjusted for multiple comparisons. RESULTS: Among the 51 metabolites that differed in relative abundance across the 2 time points, 37 (73%) metabolites decreased over time, mainly comprising lipids (sphingolipids and phospholipids), and amino acids (AA). MSEA did not reveal specific metabolic pathway alterations over time. HM from mothers in the highest (compared with middle, ref) body mass index (BMI) tertile PP had significantly lower scores for a metabolomic pattern (factor 5) characterized by AA that enriched in 6 critical metabolic pathways (metabolism of glutathione, glutamate, alanine, glycine and serine, and aspartate; and the ammonia recycling pathway). In turn, lower scores for this profile were associated with greater (β = 0.22; 95% confidence interval: 0.00, 0.43) infant weight-for-age z-score at 3 mo but not 6 mo. Additionally, HM from undernourished mothers had lower scores for a metabolomic pattern (factor 4) characterized mainly by AA and acylcarnitines involved in branched chain AA and fatty acids metabolism, although MSEA were not significant. This pattern was not associated with infant growth. CONCLUSIONS: In this LMIC cohort, the abundance of several lipid and AA metabolites declined over time. Furthermore, HM metabolome differences in mothers with high BMI may underpin early weight gain in their infants. More studies from LMICs are needed to understand maternal nutrition-driven differences in HM composition that may impact infant growth and development to guide interventions in these high-risk breastfeeding populations.