Human milk is more than a source of nutrition

Human milk and infant formula

Seven years ago, I wrote an article for Meer in which I described how human milk is the only true superfood ^1-4. So, when the U.S. Food and Drug Administration (FDA) published what it called the largest-ever testing of infant formula for chemical contaminants, ⁵ parents and health professionals around the world had reason to pay close attention, especially because the report also included results from human milk. The report presents the results of 110 human milk samples that were tested for four of those contaminants: lead, mercury, cadmium, and arsenic.

The FDA's headline conclusion about infant formula was reassuring: the vast majority of samples had undetectable or very low levels of contaminants, "affirming the safety of the U.S. infant formula supply." That is good news, and it deserves to be said clearly. But this was also true for human milk, although human milk samples were characterized as contaminated. A careful reading of the same report raises a troubling question: why were the human milk results presented so differently, and in a way that makes human milk appear far more concerning than the underlying data actually warrant?

Two commodities, two standards

Science communication depends on consistency. When you compare two things, you must measure and describe them the same way. The FDA report does not do this, and the asymmetries all point in the same direction.

Statistical detail

For each contaminant, the report provides a rich statistical picture for infant formula: the full range of values detected, the median concentration, and the 95th percentile (the level below which 95% of samples fell). These tools let a reader understand not just whether something was detected, but how much was typically present and how common any given result was. Note that concentrations are expressed in parts per billion (ppb), where one ppb is equivalent to one drop of water in an Olympic-sized swimming pool. None of these statistics—no range, no median, no percentile—were provided for the human milk samples. A reader comparing the two sections of the report has no equivalent basis for understanding the human milk data.

Framing of results

The results of the analysis of infant formulas stressed the proportion of samples in which no contaminants were found, a reassuring frame. The results for human milk, by contrast, emphasized the proportion of samples in which one or more contaminants were detected. Specifically, the report states that 85% of human milk samples had at least one detectable contaminant across all four metals tested. This figure was never computed or reported for infant formula, even though applying the same arithmetic to formula would likely yield a similarly high combined-detection rate, simply because detecting trace levels is what sensitive modern instruments do. The same type of finding was framed as evidence of safety for one product and as cause for concern for another.

The pie chart problem

For each contaminant, the FDA used pie charts to show the distribution of results across concentration tiers, but the tiers used for infant formula and human milk were not the same. For lead, formula results were broken into four tiers: not detected; detected up to 0.2 ppb; 0.3 to 0.5 ppb; and above 0.5 ppb. Human milk results were collapsed into just three tiers with entirely different boundaries: not detected; detected up to 1 ppb; and above 1 ppb. That much wider middle tier for human milk obscures exactly the distributional detail that the formula chart provides clearly.

When the human milk data are reconstructed using the same tier boundaries applied to formula (which is possible using the raw data the FDA published in its own appendix), the picture changes considerably. For lead, 70% of human milk samples were below the limit of detection, 7% fell in the detected-to-0.2-ppb range, 14% in the 0.3-to-0.5-ppb range, and only 9% above 0.5 ppb. That distribution is not alarming. It was simply never shown.

The missing benchmark

The infant formula section of the report notes prominently that all formula samples were below U.S. Environmental Protection Agency (EPA) drinking water standards for lead, mercury, cadmium, and arsenic, in most cases by a wide margin. Formula lead levels, for instance, were described as at least eight times below the EPA drinking water standard. What the report does not say—but what the data clearly show—is that the human milk results were also below EPA drinking water standards for all four metals. That is, the human milk samples were less contaminated than drinking water.

This context, applied selectively to one commodity, gives a reader the impression that formula has been validated against a meaningful safety benchmark while human milk has not. The benchmark applies equally to both. It was simply not mentioned for one of them.

A caveat applied to only one commodity

The report acknowledges, quietly, that its human milk data came from 110 samples from a single donor source and cannot be generalized to all human milk. This is a legitimate limitation to disclose. But no equivalent acknowledgment appears for the formula data, even though 300 samples of a commercially diverse product is also not a complete census of what is available on U.S. shelves. Flagging limitations for one commodity and not the other is not neutral science communication.

What "detected" actually means

Modern analytical chemistry is extraordinarily sensitive. Laboratory instruments used in food safety testing can detect substances at concentrations far below any level associated with harm—concentrations equivalent to a single drop in an Olympic-sized swimming pool. A positive detection does not mean a sample is unsafe. It means the instruments are working.

This distinction is critical when interpreting the human milk findings. The statement that 85% of human milk samples had "at least one detectable contaminant" sounds alarming. In fact, those contaminants were present at trace levels, in most cases well below the same EPA thresholds used to reassure readers about the formula results. The report applies that reassuring context to the formula. It withholds it from human milk.

Setting aside all presentational asymmetries, what do the data actually show? Both infant formula and human milk contain trace levels of heavy metals. In both cases, the vast majority of samples fall well below EPA drinking water thresholds. For mercury, 95% of formula samples had no detectable levels—and the picture for human milk is similar: 39% were below the detection limit and 56% fell in the lowest detected tier. For lead, 70% of human milk samples were below the limit of detection entirely.

This is not a story about contaminated human milk. It is a story about trace environmental exposures that are present across our food system at levels that regulatory agencies have not identified as harmful. However, these trace levels are only characterized as harmful for one of the two commodities analyzed in the report.

Why this matters: what human milk actually is

As mentioned in my previous article, the World Health Organization (WHO), American Academy of Pediatrics, and Department of Health and Human Services in the USA have all recommended breastfeeding for at least the first six months of life and including it in a mixed diet for at least until the infant is two years old ^2-4.

As mentioned in my previous article, breastfeeding is good for both the mother and her babies. It helps them form a strong, loving bond that can last a lifetime and help during difficult times later in life. Moreover, human milk meets the baby’s nutritional needs. It contains bioactive peptides and proteins, as well as growth factors, hormones, lipids, and carbohydrates ^6-11. It protects infants from neonatal sepsis and stimulates immunity while promoting growth and development ^6-9.

As mentioned in my previous article, lipids are the largest source of energy in human milk ¹. There are also short-chain fatty acids in human milk that are essential for proper maturation of the baby’s gastrointestinal tract (GIT). There are also sphingomyelins that are especially important for forming myelin sheaths for neurons in the central nervous system.

Human milk also contains over 400 different proteins that provide nutrition and have antimicrobial as well as immunomodulatory activities while stimulating the absorption of nutrients. The antibodies found in human milk target the infectious agents encountered by the mother immediately before and after birth, as well as the infectious agents most likely to be encountered by the baby. Unlike infant formula, which has a fixed content, breastmilk adapts its content of nutrients and bioactive components to the changing needs of infant growth at different stages.

As mentioned in my previous article, there are also many complex carbohydrates in milk, as well as the disaccharide lactose and human milk oligosaccharides (HMO) ¹. HMOs encourage the growth of beneficial bacteria in the baby’s GIT ^12-14. This protects the baby from being colonized by pathogenic bacteria. So, HMOs help prevent neonatal diarrhea and respiratory tract infections. HMOs are different than the oligosaccharides that are in other mammals. This is one reason why breastfeeding is better than formula.

HMOs also act as prebiotics to stimulate the growth of healthy bacteria in a baby’s gut. They also prevent the adhesion of harmful bacteria and act as receptor decoys that keep pathogenic bacteria from colonizing mucosal surfaces. They are especially useful in preventing the growth of Streptococcus agalactiae, more often known as Group B Streptococcus (GBS). It’s a common cause of neonatal sepsis and meningitis. HMOs also help keep GBS from forming biofilms that would protect them from the infant’s developing immune system.

HMOs are being manufactured on a large scale for potential uses, including as additives to infant formulas. Eight different HMOs have received numerous market authorizations for use in infant nutrition. HMOs are complex carbohydrates abundant in human milk and distinguished by their characteristic structural features. More than 200 different HMO structures have been identified so far, all of which are built from five distinct monosaccharide building blocks, namely glucose, galactose, N-acetylglucosamine, fucose, and N-acetylneuraminic acid, also known as sialic acid. ^12-14.

HMOs are the third most abundant solid component of breast milk, and 2′-fucosyllactose (2′-FL) is the most abundant HMO in breast milk, accounting for about 20% by weight of the total HMOs. They have several healthy effects, including building a healthy infant gut microbiota. So, adding manufactured HMOs to infant formula brings the composition closer to human milk and may have many health benefits in early life.

Also, 21 clinical studies have established the safety and physiological effects of manufactured HMOs in infants and young children. Since 2′-FL, it was one of the first two HMOs to be manufactured at an industrial scale and to receive regulatory approvals in the USA and the European Union in 2015 and 2016, respectively. In addition, 2′-FL has become the most clinically studied HMO. Further development and clinical testing of blends containing five or more HMOs that come closer to matching the complex composition of HMOs in human milk. In addition, the total HMO dose has also increased.

While some of the early HMO clinical trials tested 0.2–0.25 g/L of a single HMO, two recent clinical trials of five mixtures of HMOs tested total doses of 5.75 g/L. This is in the range of median levels of these HMOs in mature breastmilk (15–90 days). HMOs serve as fermentation substrates for specific gut microbes that can degrade them, especially certain Bifidobacterium species.

Seven clinical studies in infants have evaluated the effects of HMO supplementation on gut microbiota composition. Five randomized controlled trials compared the gut microbiota of infants who received infant formula supplemented with HMOs to respective control groups who received the same infant formula without HMOs or other oligosaccharides. They all reported statistically significant increases in the relative abundance of Bifidobacterium in the HMO groups compared to the controls. Preclinical studies found that some HMOs can protect against infections by acting as decoy receptors for pathogenic viruses and bacteria.

Several HMOs intervention trials in infants have reported on outcomes related to infection and immunity, such as incidence of respiratory tract infections, use of antipyretic and antibiotic medica tions, and levels of immune biomarkers. One trial evaluating 2′-FL noted that there were significantly fewer reports in the adverse events category of infections and infestations in the 2′-FL group compared to the control group.

The physiological effects of HMOs supplementation have also been evaluated in clinical studies in children and adults. The most reported physiological effect of HMO supplementation in children and adults was the modulation of gut microbiota composition. Also, a five-week clinical trial tested whether supplementation with 2 g/d of 2′-FL in combination with 8 g/d of oligofructose or maltodextrin could affect gut microbiota composition and mood in healthy adults with mild to moderate anxiety and depression. This is consistent with the well-known link between the enteric and central nervous systems.

Human breast milk-derived exosomes (HMDEs) also make human milk better than milk from cows, goats, and even infant formula ^15-16. HMDEs are mainly composed of lipids, proteins, RNA, and DNA released from mammary cells. High-throughput analysis has identified up to 1523 microRNAs (miRNAs), 1963 proteins, and 395 lipids in HMDEs. In addition to miRNAs, HMDEs also contain other types of RNA, including messenger RNAs (mRNAs), long-noncoding RNAs (lncRNAs), and circular RNAs (circRNAs), which may have important roles in cellular communication and regulation.

HMDEs carry and exchange biological information between mother and child. They can cross the gastrointestinal barrier, have low immunogenicity, and have good biocompatibility and stability. HMDEs support intestinal barrier integrity in newborns. In addition, they have important properties that are reformable and modifiable. This may make them useful in the prevention and treatment of neonatal intestinal diseases.

The miRNAs in human milk have many health benefits ^17-18. Human milk contains over 1,400 distinct miRNAs. They are packaged in sturdy HMDEs that resist digestion and can cross the blood–brain barrier, contributing toward optimal development, spatial learning, and memory. They leave a lasting molecular signature that optimizes gene expression for immune function, stress, and metabolism, as well as the structure and connectivity of the neuroendocrine immune system.

HMDEs protect the miRNAs within them from enzymes, chemicals, or mechanical degradation. There are immunological miRNAs, including miRNA-148a-3p, miRNA-181a-5p, miRNA-182-5p, miRNA-16-5p, and miRNA-99b-5p. They help to regulate the maturation and differentiation of B and T lymphocytes. Regulatory T cells (Tregs) play an important role in the control of autoimmunity and immune tolerance.

Human milk miRNAs may promote the process of converting self-reactive thymocytes into stable Treg regulatory cells, thereby lowering the level of atopy (a genetic tendency to develop allergic conditions, driven by the overproduction of immunoglobulin E (IgE) antibodies in response to harmless environmental triggers).

MiRNAs enter the systemic circulation and influence the epigenetic programming of several organs. They help control the regulation of insulin secretion, insulin-like growth factor-1, α-synuclein, and forkhead box P3. MiRNA-148a and miRNA-30b stimulate the expression of uncoupling protein 1, a key thermogenesis inducer that transforms brown adipose tissue which helps to prevent obesity. Research is ongoing into the potential of human milk exosomal miRNAs to inhibit the proliferation of cancer cells in vitro and in vivo.

As mentioned in my previous article, in addition, human milk contains a diverse community of healthy bacteria ^1,10-11. The gut microbiome affects the development of the brain. It is especially important for the formation of synapses that connect neurons with the blood-brain barrier, as well as the proper function of microglia that support neurons. Many of the metabolites produced by gut bacteria are important in the optimal development of the young brain. The gut microbiome primes the innate immune system that is connected to both the peripheral and central nervous systems.

Human milk also contains healthy bacteria that colonize her baby’s gut during breastfeeding and work with HMOs to establish a baby’s healthy microbiome. The gut microbiome affects brain development. It is especially important for the formation of synapses connecting neurons, the proper function of microglia that support neurons, and the priming of the innate immune system connected to both the peripheral and central nervous systems.

Many of the metabolites produced by gut bacteria are critical to optimal early brain development. This microbiome foundation, established in part through human milk, contributes to long-term protection against chronic diseases, including obesity, type 2 diabetes, and cardiovascular disease. In contrast, although modern infant formulas are designed to mimic the nutritional composition of human milk and support infant growth, they lack many of these important bioactive components.

When a government report presents data in a way that could cause parents to doubt the safety of breastfeeding, without scientific justification for that concern, the consequences are not abstract. Parents make real decisions based on what they read from authoritative sources. Those decisions affect infant health outcomes.

What should happen next?

The fix is not complicated. It requires only consistency: apply the same statistical summaries, the same concentration tier boundaries, the same regulatory benchmarks, and the same affirming language to both commodities. The FDA's own publicly available appendix data make this straightforward. The data do not need to be reanalyzed. They need to be honestly represented.

We encourage the public and regulatory agencies in other countries to recognize that human milk is not only at least as clean as infant formula but also provides a wealth of healthy bioactive compounds that infant formula, cow's milk, and goat's milk cannot match. Regulatory agencies and international bodies like the WHO and Codex Alimentarius have both the standing and the scientific mandate to request that kind of consistency from the FDA. Independent researchers have the ability to publish their own analyses of the publicly available data. Pediatric and public health professional organizations have the credibility to raise these questions openly.

I join with countless others in encouraging further work and international collaboration. Laboratories and milk banks around the world should work together to compare analytical methods and the results obtained from them, and to analyze human milk samples that are meant for consumption. The scientific and medical communities are also very interested in using AI and biotechnology to identify currently unknown bioactive components of human milk and their significance to infant health. The Human Milk Institute (HMI) holds an annual meeting in La Jolla, California—the next is scheduled for March 9–11, 2027 ¹⁹—and actively encourages exactly this kind of global collaboration.

The science of human milk deserves the same rigorous, honest, and contextually complete presentation that is applied to every other food in the regulatory system. The need to protect our babies and to give parents accurate and unbiased information to make feeding decisions is common to all people around the world. The data to support that already exist. The question is whether the people responsible for communicating with them will choose to present these data without bias and agenda.