Seven milk molecules offer new clues to infant growth and maternal nutrition

Studies suggest breast milk composition reflects a complex blend of maternal diet, genetics and mammary-gland biology, but researchers caution that the findings are not yet a diagnostic tool.

Breast milk has long been treated as something close to a biological mystery: responsive, protective and deeply individual, yet difficult to pin down in scientific terms. Two recent lines of research are giving that mystery sharper edges. One large international study published in Science examined 1,543 milk samples from Canada, Pakistan and Burkina Faso and identified seven molecules that tracked most consistently with maternal nutrition and infant growth. A separate preprint from the United States suggested that everyday diet, including coffee and citrus, leaves measurable traces in milk, while genetics and mammary-gland biology help shape its composition as well.

The Science study focused on a broad sweep of milk chemistry, combining metabolites, proteins, vitamins, carbohydrates and human milk oligosaccharides with infant growth data. Among the compounds that most strongly matched patterns linked to maternal nutritional status and early growth were four short-chain acylcarnitines, known as C2, C3, C4 and C5, along with lysine, 6′ sialyllactose and selenium. Researchers described these as candidate biomarkers rather than a clinical test, but the consistency of the pattern across three countries and different nutritional settings is what makes the work notable.

One compound in particular, C5, drew close attention. The researchers found it was associated with signals connected to lower milk synthesis, tissue remodelling and immune activity, although those links do not prove that the molecule itself causes poorer milk production or slower infant growth. They have proposed a possible explanation involving pantothenic acid, also known as vitamin B5, because acylcarnitine metabolism depends on coenzyme A, which the body makes from that vitamin. The team behind the study is now investigating that pathway more directly, including how it may operate during colostrum and transitional milk stages.

The U.S. preprint adds a different layer to the picture. Analysing milk from 349 women in Minnesota and Oklahoma about a month after delivery, researchers measured 458 metabolites and compared them with diet questionnaires, gene expression data and genetic information from subsets of participants. Coffee intake was strongly associated with quinic acid and 1,3-dimethyluric acid, while citrus intake was linked to proline betaine. The study found that diet was often the largest measured contributor when predictors were selected, but it also identified genetic variants near genes such as GNE and PDE6A that were tied to specific metabolites, suggesting that milk reflects more than just what a mother eats.

That is an important caution for parents and clinicians alike: these findings do not mean that coffee or citrus should be treated as universally harmful or beneficial in breastfeeding mothers. The preprint is preliminary, has not been peer reviewed and cannot establish cause and effect. Its sample was largely of European ancestry, the genetic subgroup was relatively small and milk was collected at a single point in lactation. The Science paper, although published, still stops short of offering a validated screening tool. Its molecules are promising clues, not a ready-made diagnostic panel.

Taken together, the studies support a more nuanced view of human milk. It is shaped partly by nutrition, partly by maternal biology and partly by processes within the breast itself. That complexity matters. As earlier research has also suggested, including studies linking breast milk composition with infant growth and maternal health, the make-up of milk can influence development in ways that extend beyond calories alone. The newer work does not settle the question of how best to change milk composition, but it does move scientists closer to understanding which signals matter, and why.