Pasteurization and lactoferrin: measured losses by heat treatment

Pasteurization names more than one process, and the answer splits along that line.

Holder pasteurization, the 62.5 °C half hour that milk banks apply to donor human milk, removes most of the lactoferrin. Four studies put the loss between 75 and 87.5 percent. High-temperature short-time treatment, 72 °C for 15 seconds, is the hotter of the two and takes consistently less, though the published figures for it range from nothing measurable to 83.5 percent and do not agree. In cow's milk on a retail shelf, the protein survives pasteurization well enough that pasteurized samples group with raw samples rather than with ultra-heat-treated ones.

Heat load, not peak temperature, is what the measurements track. Thirty minutes at 62.5 °C is a larger thermal dose than fifteen seconds at 72 °C, and lactoferrin registers the difference.

Holder pasteurization: four studies, one direction

Milk banks run this program more often than any other, which makes it the best-measured case on the page.

The largest matched design is Davis and colleagues, who took 50 pairs of raw and pasteurized milk from a single US milk bank, each pair from a different approved donor, and measured 29 nutrients and bioactive factors across both. Copper, calcium, thiamine and nonprotein nitrogen lost under 15 percent. Immunoglobulin A lost 35 percent. Lactoferrin lost more than 80 percent. The authors' own reading is that Holder pasteurization leaves energy and the essential nutrients of donor milk broadly intact and that the antimicrobial proteins are the casualty.

Tanaka and colleagues ran the same program on three machines and found that the hardware matters. Donor milk went untreated, through one water pasteurizer and through two dry pasteurizers. Macronutrients, minerals and oligosaccharides came through unaffected or nearly so. Lactoferrin fell 75 percent in the water device and 86 percent in each of the dry devices, all three at P below 0.001, and secretory immunoglobulin A fell 23, 42 and 44 percent on the same three machines. The dry pasteurizers also ran a plateau temperature more than 0.5 °C above the water pasteurizer and a peak more than 1 °C above it. The abstract reports the temperature gap and the eleven-point loss gap side by side without formally testing the link, and concludes that the dry method is unsuitable for donor milk. Either way, a nominally identical program delivered on different hardware moved the number by eleven points, which is the kind of margin a single published figure for "Holder pasteurization" quietly absorbs.

Aceti and colleagues, working from the Bologna milk bank, reported 87.5 percent. Kontopodi and colleagues reported a 50 to 98 percent band across four bioactive proteins without separating lactoferrin out, so their Holder figure is not directly comparable and is left out of the range above.

The short-time figures contradict each other

Three studies ran the same 72 °C, 15 second regime and reported losses of 42.3, 52 and 83.5 percent. That is a wider spread than the Holder rows show, and it has to be stated rather than averaged away, because the gap between 42 and 84 percent is the difference between a promising alternative and a pointless one.

Two of the three were done on the same samples as a Holder arm, which makes their internal comparisons more trustworthy than their absolute numbers. Aceti found short-time treatment barely better than Holder on lactoferrin, 83.5 against 87.5 percent. Immunoglobulin A came through it intact in the same study, and of the four arms only high-pressure processing reduced that protein significantly, by 38.8 percent. Kontopodi found short-time treatment clearly better, 52 percent against a band reaching 98, and also measured function rather than concentration alone: bile salt-stimulated lipase activity was still 81 percent down after the gentler treatment. A protein can be present and not working, and only one of these designs could have caught that.

The third study isolated a variable the other two did not vary. Martysiak-Żurowska and colleagues heated milk to the same temperature for the same time by microwave and by convection. Lactoferrin fell 42.3 percent under convective heating and did not change significantly under the microwave. Vitamin C fell by about half under both. So the stated regime does not fix the outcome: how the milk reaches 72 °C, and how long it spends on the way, changes what arrives at the other end.

Franco and colleagues, reviewing heat, drying, pressure and pulsed electric field treatments, report that the protein's heat stability depends on how iron-saturated it is and on the medium it is heated in. That is a narrative review rather than a measurement, so it carries less weight than the rows above, but it names two variables none of the human milk studies controlled and offers a mechanism for the scatter.

Cow's milk is a different question with a different answer

The milk in a supermarket carton is a separate case, and the evidence there points the other way.

Brick and colleagues took three pools of raw cow's milk and put each through six treatments: homogenization, fat separation, pasteurization, ultra-heat treatment, extended shelf-life processing and ordinary boiling. Mass spectrometry identified 364 serum proteins, and the 140 detected in at least two thirds of samples went into a clustering analysis. The samples sorted into two groups, and the split was not treatment by treatment: high heat (boiling, UHT, extended shelf-life) on one side, no or low heat (raw, skimmed, pasteurized) on the other. Retail pasteurized milk grouped with raw milk. Comparing the two groups, 23 proteins came out significantly lower after heating, lactoferrin among them at a log2 fold change of -0.37 with p = 0.004, which is an abundance about 23 percent lower.

Two things follow. That 23 percent belongs to the contrast between the heat groups, not to pasteurization on its own, because the pasteurized samples sat inside the low-heat group. And the treatments that did measurably cost lactoferrin are UHT, extended shelf-life processing and ordinary boiling, so bringing pasteurized milk to the boil in a pan at home is a harsher treatment than the dairy applied to it.

What these measurements do not establish

None of these studies followed an infant. They measured what comes out of a machine, which is a different question from whether the loss changes any outcome in a child fed the milk. The nearest trial evidence runs the other way round, putting lactoferrin back in rather than measuring what processing took out, and it is set out under lactoferrin for preterm infants.

The assays also cut across each other. An immunoassay reports whatever its antibody still binds, mass spectrometry reports peptides, and neither says whether the molecule still folds into a shape that holds iron. Kontopodi's activity assays are the only measurements here that address function directly. A concentration figure is the floor of this question, not the whole of it.

This page carries no evidence grade, for the same reason the lactoferrin concentration table does not. The grading scale on this site rates claims about effects in people against the human trial evidence behind them. What heat does to a protein in a bottle is a laboratory measurement, and running it through a clinical scale would say nothing true about either.

Readers arriving from the general account of lactoferrin will find the untreated figures by species and stage of lactation in the concentration table, and the rest of the processing picture under lactoferrin sources.