RESUMO
BACKGROUND: Human milk (HM) composition data are widely used in clinical, regulatory, and public health initiatives. The existing HM profiles in U.S. and Canadian nutrient databanks are outdated and now considered inappropriate to estimate current nutrient intakes. Recent reviews have underscored the limited North American data available to generate a new profile. OBJECTIVE: To describe concentrations and sources of variability of nutrients in HM from a large cohort collected in Canada. METHODS: The Maternal-Infant Research on Environmental Chemicals (MIREC) study recruited participants in the first trimester of pregnancy from 10 Canadian cities between 2008-2011. HM samples (n=559-835, depending on nutrient) were collected 3-10 weeks post-partum and analyzed for minerals (calcium, magnesium, phosphorus, potassium, sodium, manganese, molybdenum, zinc, copper, iodine, selenium), vitamin D (vitamin D3, 25-(OH)D3), folate vitamers (folic acid, 5-methyltetrahydrofolate, total folates), and fatty acids (panel). We examined associations between participant characteristics and log-transformed nutrient concentrations using linear regression. RESULTS: Concentrations of HM components in MIREC samples were within the range observed in literature except for manganese, which was >100 fold lower than the value in the existing Canadian nutrient databank profile (2.43 [SD 2.84] compared to 260 ng/g). In multivariable models, concentrations of folate vitamers, vitamin D and fatty acids demonstrated greater variability with maternal and sample characteristics than minerals. Factors such as relevant supplement use, body mass index (BMI), and for vitamin D, skin color and season, had a larger impact on nutrient concentrations than characteristics typically standardized in HM research, such as maternal or infant health, and method of collection. CONCLUSION: HM mineral concentrations from this study meet the methodological inclusion criteria for updating nutrient databank values and dietary reference intakes. Consideration of factors such as diet, skin colour, and BMI will be important for selecting studies for developing representative reference values based on human milk.