Coffee, citrus and genetics: What scientists found inside human breast milk

By Tarun Sai LomteReviewed by Susha Cheriyedath, M.Sc.Oct 14 2026

From coffee consumption to inherited genetic differences, researchers examined the biological and dietary factors behind hundreds of molecular variations in human milk.

*Important notice: bioRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.

Study: Maternal diet and genetics shape the human milk metabolome. Image Credit: FaceStock / Shutterstock

In a recent study posted to the bioRxiv preprint* server, a team of researchers investigated how maternal diet, genetics, clinical characteristics, and milk-cell gene expression were associated with variation in the human milk metabolome.

Background

Human milk supplies nutrients and bioactive compounds that support infant growth, immune development, and microbial colonization during the first 1,000 days of life. How does the mother’s diet affect the molecular composition of her milk? What role do inherited genetic differences play?

Human milk contains numerous metabolites involved in energy production, lipid metabolism, and a variety of cellular processes. Some components are synthesized in the mammary gland, whereas others enter milk from the maternal circulation.

Previous research has linked maternal characteristics to milk composition, but their relative contributions remain incompletely understood. Further research is needed to clarify these relationships across diverse populations and stages of lactation.

About the study

The researchers analyzed data from the Mothers and Infants Linked for Healthy Growth (MILK) study, an observational study involving healthy mothers and full-term infants from Minnesota and Oklahoma, United States. The cohort included 349 participants with human milk samples collected approximately one month after delivery.

Participants intended to exclusively breastfeed and were enrolled during pregnancies described as uncomplicated, though 15% had gestational diabetes. Milk was collected two hours after participants fed their infants, then processed and stored at −80°C until analysis.

The researchers analyzed 458 metabolites using three mass spectrometry platforms: gas chromatography-mass spectrometry (GC-MS), liquid chromatography-mass spectrometry (LC-MS), and liquid chromatography–tandem mass spectrometry (LC-MS/MS). Milk macronutrient, hormone, and cytokine levels were also assessed.

Maternal dietary intake was evaluated using the Diet History Questionnaire (DHQ), with data available for 344 participants. Milk-cell ribonucleic acid (RNA) sequencing provided gene-expression measurements for 299 participants, while genetic data from 197 participants supported genome-wide association analyses.

Statistical analyses included correlation testing, least absolute shrinkage and selection operator (LASSO) regression, and machine-learning approaches to identify relationships between dietary characteristics, gene expression, genetic variants, and metabolite abundance.

The investigators also compared genetic associations of milk metabolites with findings from plasma studies. Models evaluated the relative contributions of dietary, clinical, and genetic features.

Study results

The researchers identified extensive variation in human milk metabolites and grouped them into seven modules of co-occurring compounds. Overall, associations with other milk characteristics were stronger than those with maternal characteristics.

Higher acylcarnitine concentrations, a group of fatty acid esters, correlated with elevated interleukin-6 (IL-6) levels and lower milk glucose. These compounds transport fatty acids into mitochondria for energy production.

Maternal diet showed the strongest relationship with the metabolome among the maternal characteristics evaluated. Acylcarnitines were an exception to the predominantly positive diet–metabolite correlations, suggesting that they may reflect maternal metabolic state rather than direct dietary transfer.

Dietary analyses compared 199 dietary features with 458 metabolites and identified 1,038 significant diet–metabolite correlations. Of these, 920 (88%) were positive.

Coffee and caffeine consumption were strongly associated with milk quinic acid and 1,3-dimethyluric acid, and dietary features explained 34% of quinic acid variation. Dietary features explained 12.5% of variation in proline-betaine, a metabolite associated with citrus consumption.

Higher alcohol intake was associated with lower milk sphingosine-1-phosphate (S1P), a signaling lipid potentially involved in intestinal development and immune function. Overall diet quality, measured using the Healthy Eating Index (HEI), was positively associated with milk riboflavin and N-acetylornithine.

Gene-expression analyses identified relationships between the metabolites present in milk and gene expression in milk cells. Higher expression of quinolinate phosphoribosyltransferase (QPRT) was associated with lower quinolinic acid levels, consistent with the enzyme’s role in metabolizing this compound.

Higher expression of dipeptidase 1 (DPEP1) was also associated with lower cysteine-glycine levels in milk. Instrumental-variable analysis and genetic colocalization supported the latter relationship, but the contribution of DPEP1 activity in other tissues remains uncertain.

Genome-wide association studies (GWAS) revealed nine significant metabolite–locus genetic associations meeting the study-wide threshold. Two notable genetic associations involved variants near PDE6A, linked to purine metabolism, and variants near GNE, associated with free N-acetylneuraminic acid, a form of sialic acid.

The GNE association colocalized with a milk expression quantitative trait locus (eQTL), suggesting a relationship potentially specific to the lactating mammary gland. The associated variant did not show a clear relationship with the 19 milk oligosaccharides measured in this study. Comparisons with plasma studies identified both common and potentially milk-specific genetic associations.

Finally, models combining dietary, clinical, and selected genetic features from 183 participants explained a median of 13.2% of metabolite variation among the 246 metabolites for which predictors were selected. Dietary features accounted for the largest proportion of explained variation in 168 models (68%).

The findings suggest that milk composition reflects multiple maternal influences, with diet contributing most strongly among the measured factors. These models did not estimate total genetic influence, and the authors cautioned that predictive performance could be overestimated.

Conclusions

Maternal diet accounted for the largest share of explained variation among measured maternal factors, but genetic variation and mammary gland biology were associated with specific metabolites. Associations involving coffee-derived compounds, dietary patterns, and genetic loci illustrate the complex relationships shaping milk composition.

These findings provide a foundation for investigating whether milk metabolites can serve as indicators of maternal nutritional status or influence infant development. The study was observational and did not establish effects on infant health. It also assessed milk at approximately one month postpartum, relied on dietary questionnaires, and included a predominantly European-ancestry population and a relatively small genetic sample.

*Important notice: bioRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.