Wednesday, 18 March 2015

Does cortisol in a mother’s milk really affect an infant’s temperament?

Having written about lactation for more than 45 years, I have tried to avoid writing about the subject in this blog. However, I do get drawn back from time to time—and this is one of those times.

It has been known for years that biologically active substances, hormones for example, are present in milk. All sorts of claims have been made over the years that they may have effects in the infant. However, the evidence for many such claims and suggestions was not at all convincing and so Peggy Neville and I wrote a Commentary* for the Journal of Endocrinology in 1991 setting out the criteria for establishing an effect of a substance in milk on the suckling. In brief, these criteria—the equivalent of Koch’s famous postulates for the transmission of disease—must be satisfied:

  • An effect in the offspring must be obtained in response to exposure to the substance in milk.
  • The effect of the substance must be abolished by removal of the substance from milk and restored when that specific component is restored.
  • The substance must be shown to be present and active in milk.
  • The substance must be shown to retain its biological activity in the offspring to the point at which it is postulated to act or to be activated by partial digestion within the digestive tract.

I should make it clear that none of these criteria is easy to fulfil in practical terms. However, at the time and soon after, there were two sets of studies that fulfilled these criteria: those by the late Otakar Koldovsky and his colleagues in Arizona on epidermal growth factor (EGF) and those by the late Steve Frawley and his group in South Carolina on the developmental programming of the pituitary by an agent in milk. All these results were obtained in rats.

Work in this field has continued and claims of changes in milk hormone concentrations having short- or long-term effect in the offspring have been made.

A study that made the newspapers recently continued the theme that maternal ‘stress’ affects and programmes the behaviour of the offspring via a change in the concentration of glucocorticoids—the classical ‘stress’ hormones—in milk.

The study widely reported was on a captive colony of Rhesus Macaques. The authors attempted to distinguish between the effects of milk energy intake (the subject of earlier work from the same group) and cortisol (the major glucocorticoid of primates) concentration on the behavioural characteristics of the young. Multiple regression statistical models were used to try to find what was associated with what. In short, mothers who showed a higher concentration of cortisol in their milk had offspring of a more nervous/less confident temperament, an effect particularly seen in daughters. 

Teleologically, I could see the arguments that the authors were making in relation to the temperament of the offspring produced by the different mothers. However, I am yet to be convinced that the differences in temperament that were observed are a maternal effect transmitted by cortisol in milk. This is why:

In general and in all species studied thus far, milk cortisol concentrations reflect maternal plasma cortisol concentrations. I don’t want to spend time explaining the likely kinetics of cortisol movements between the maternal circulation and milk, suffice it to say that recent increases in cortisol in the mother will be seen as an increase in milk. Short-term maternal distress would therefore result in an increase in milk cortisol concentrations. I then looked at how the authors had collected the milk samples: …infants and mothers were captured in their outdoor enclosures between 7:30 and 9:00 AM and were relocated together in temporary housing. To prevent nursing, mothers were placed in mesh jackets and allowed a standardized period of milk accumulation of 3.5-4 h. This allowed infants to remain in contact with their mothers during this period. At peak lactation, the mothers and offspring were separated for the period of milk accumulation and beyond for the behavioural assessments to be made. The procedures used to obtain the milk samples—even the initial stage of capture—are ones that would result in the ‘stress’ production of cortisol. Moreover, those of us instilled with the knowledge that because the rate of milk secretion drops rapidly in ‘stressed’ mothers, great care must be taken in interpreting the quantity of milk and milk constituents produced under such circumstances. I am also not convinced that the method used to estimate milk yield and the composition of ingested milk (in terms of fat concentration) provide a sufficiently reliable measure of energy intake at times other than during the procedure to obtain the milk sample; even then no evidence is presented that stored milk removal was complete.

If the mothers are themselves variable in temperament and respond differently to the ‘stress’ of capture, wearing a mesh jacket and/or separation from their young, then might not the variation in cortisol concentrations seen reflect not the ‘resting’ cortisol concentration but the temperament of the mother in response to these distressing events. A nervous mother could produce a nervous offspring by genetic inheritance and not via a maternal effect of cortisol concentration in milk.

Looking at the data on milk cortisol concentrations in the paper, the variation is very great (19-444 nmol/l at 1 month and 21-622 at 3-4 months) with some mothers showing major changes upwards or downwards between the two samplings; in other words, the sort of pattern that could well be expected from variable maternal responses to the procedure used to get the milk samples.

In conclusion then, I do not consider the evidence sufficient at present to conclude that cortisol in milk affects the temperament of the offspring in Rhesus Macaques nor that milk intake does not. This is not to say that further experimental evidence will not support the proposed maternal effect in the future. However, this research really does illustrate the great practical difficulties in testing hypotheses on putative milk-borne signals to the offspring, just as Peggy and I envisaged when we wrote our commentary nearly 25 years ago.

*Peaker M, Neville MC. 1991. Hormones in milk: chemical signals to the offspring. Journal of Endocrinology 131, 1-3.

†Hinde K, Skibiel AL, Foster AB, Del Rosso L, Mendoza SP, Capitanio JP. 2015. Cortisol in mother’s milk across lactation reflects maternal life history and predicts human temperament. Behavioral Ecology 26, 269-281.

‘Stress’: I have used the word ‘stress’ in inverted commas because the word still means different things to different people—a point to which I shall return at some time in the future.