Mode of Neonatal Delivery Influences the Nutrient Composition of Human Milk: Results From a Multicenter European Cohort of Lactating Women

International audience ; BACKGROUND: The effect of the mode of neonatal delivery (cesarean or vaginal) on the nutrient composition of human milk (HM) has rarely been studied. Given the increasing prevalence of cesarean section (C-section) globally, understanding the impact of C-section vs. vaginal delivery on the nutrient composition of HM is fundamental when HM is the preferred source of infant food during the first 4 postnatal months. OBJECTIVE: This study aimed to evaluate the association between mode of delivery and nutrient composition of HM in the first 4 months of life. DESIGN: Milk samples were obtained from 317 healthy lactating mothers as part of an exploratory analyses within a multicenter European longitudinal cohort (ATLAS cohort) to study the HM composition, and its potential association with the mode of delivery. We employed traditional mixed models to study individual nutrient associations adjusted for mother's country, infant birth weight, parity, and gestational age, and complemented it, for the first time, with a multidimensional data analyses approach (non-negative tensor factorization, NTF) to examine holistically how patterns of multiple nutrients and changes over time are associated with the delivery mode. RESULTS: Over the first 4 months, nutrient profiles in the milk of mothers who delivered vaginally (n = 237) showed significantly higher levels of palmitoleic acid (16:1n-7), stearic acid (18:0), oleic acid (18:1n-9), arachidic acid (20:0), alpha-linolenic acid (18:3n-3), eicosapentaenoic acid (20:5n-3), docosahexenoic acid (22:6n-3), erucic acid (22:1n-9), monounsaturated fatty acids (MUFA)%, calcium, and phosphorus, whereas the ratios of arachidonic acid/docosahexaenoic acid (ARA/DHA) and n-6/n-3, as well as polyunsaturated fatty acids (PUFA)% were higher in milk from women who had C-sections, in the unadjusted analyses (p \textless 0.05 for all), but did not retain significance when adjusted for confounders in the mixed models. Using a complementary multidimension data analyses ...