Feeding peanut to high-risk infants cut peanut allergy from 13.7% to 1.9%

Deliberately feeding peanut to infants at high risk of peanut allergy prevented most cases of it. In the LEAP trial, among infants who had no measurable reaction on a skin-prick test at entry, 13.7% of those assigned to avoid peanut were allergic to it at 60 months of age, against 1.9% of those assigned to eat it — a difference the trial reported at P<0.001 [s1]. That single result overturned a decade of advice telling parents of at-risk babies to keep peanut out of the house.

It is also the most misread trial in allergy medicine, because a second trial run at the same time, in a different population, did not find the same thing.

What LEAP did

LEAP randomly assigned 640 infants to consume or avoid peanut until 60 months of age [s1]. These were not typical babies. Every one of them had severe eczema, egg allergy, or both — the entry criteria that defined "at risk" [s1]. They were at least 4 months and younger than 11 months at randomisation, and they were split into two separate cohorts on the basis of a skin-prick test done at entry: one cohort with no measurable wheal, the other with a wheal of 1 to 4 mm, meaning some pre-existing sensitivity to peanut extract [s1].

The primary outcome was assessed independently in each cohort. In the 530 infants with a negative skin-prick test, peanut allergy at 60 months was 13.7% with avoidance and 1.9% with consumption [s1]. In the 98 infants who already had a positive test — a group that was, by definition, further along the road to allergy — the figures were 35.3% and 10.6%, at P=0.004 [s1]. There was no significant difference between the groups in serious adverse events [s1].

The immunological signature moved with the clinical result. Peanut-specific IgG4 rose mostly in the consumption group, while a greater share of the avoidance group developed elevated peanut-specific IgE [s1].

The follow-up question everyone asked

The obvious objection to LEAP was that eating peanut every week might simply be suppressing an allergy that would reappear the moment it stopped. LEAP-On tested exactly that: at the end of the primary trial, all participants were told to avoid peanut for 12 months [s2].

Of 628 eligible participants, 556 enrolled, and 550 had complete primary-outcome data [s2]. At 72 months, peanut allergy was 18.6% in the original avoidance group and 4.8% in the original consumption group [s2]. Within the consumption group, a year off peanut moved the rate from 3.6% at 60 months to 4.8% at 72 months, a change that was not statistically significant at P=0.25 [s2]. Three new cases appeared in each group [s2]. The authors' conclusion is carefully bounded: a 12-month period of avoidance was not associated with an increase in peanut allergy, and "longer-term effects are not known" [s2].

The trial that did not replicate it

EAT recruited from the general population rather than a high-risk clinic: 1,303 exclusively breast-fed three-month-olds, randomised either to early introduction of six allergenic foods — peanut, cooked egg, cow's milk, sesame, whitefish and wheat — or to the UK standard of exclusive breastfeeding to about six months [s3].

In the intention-to-treat analysis, which is the analysis a trial lives or dies by, food allergy to one or more of the six foods developed in 7.1% of the standard-introduction group and 5.6% of the early-introduction group, at P=0.32 [s3]. That is a null result. The trial did not show efficacy [s3].

The per-protocol analysis — restricted to families who actually managed the demanding feeding schedule — looked very different: any food allergy 2.4% versus 7.3% (P=0.01), peanut allergy 0% versus 2.5% (P=0.003), egg allergy 1.4% versus 5.5% (P=0.009), with no significant effect for milk, sesame, fish or wheat [s3]. Consuming 2 g per week of peanut or egg-white protein was associated with significantly lower rates of those allergies than consuming less [s3].

Per-protocol analyses are not randomised comparisons. Families who succeed at a hard regimen differ from families who do not, in ways that are not measured. The EAT authors put the finding as a question rather than a conclusion: whether prevention by early introduction of multiple allergenic foods is dose-dependent [s3]. The trial's own summary is that early introduction of all six foods "was not easily achieved but was safe" [s3].

What the guidelines did with this

In the United States, the National Institute of Allergy and Infectious Diseases convened an expert panel with 25 professional organisations, federal agencies and patient advocacy groups, and issued addendum guidelines specifically on preventing peanut allergy [s4]. The addendum sets out three separate guidelines for infants at different levels of risk, and covers the definition of those risk categories, the appropriate use of specific IgE measurement, skin-prick tests and oral food challenges, and the timing and setting for introducing peanut-containing foods — in a clinician's office or at home [s4].

That structure is the honest translation of the evidence. LEAP tested a high-risk population and found a large effect [s1]. EAT tested a general population and did not, in the analysis that counts [s3]. So the guidance is stratified by risk rather than issued as one instruction to everybody, and it builds in testing before introduction for the infants at highest risk [s4].

What is still not known

Nothing in this literature establishes how long the protection lasts. LEAP-On covers one year of avoidance and says so [s2]. Nothing here says that early introduction prevents allergy to foods other than peanut and egg in a randomised comparison; EAT looked for that across six foods and found it for none of them on intention-to-treat [s3]. And nothing here addresses the question that the rise in food allergy actually poses — why prevalence climbed in the first place. LEAP's own background notes that peanut allergy prevalence among children in Western countries doubled in the ten years before the trial, and was becoming apparent in Africa and Asia [s1]. The trial explains what can be done about a high-risk infant. It does not explain the epidemic.

Decisions about when and whether to introduce an allergenic food to a particular infant — especially one with severe eczema or an existing egg allergy — belong with that infant's clinician, working from the risk stratification the guidelines set out [s4]. This article describes what the trials measured, not what any individual family should do.

Sources

• Randomized Trial of Peanut Consumption in Infants at Risk for Peanut Allergy — New England Journal of Medicine, 2015-02-23
• Effect of Avoidance on Peanut Allergy after Early Peanut Consumption — New England Journal of Medicine, 2016-03-05
• Randomized Trial of Introduction of Allergenic Foods in Breast-Fed Infants — New England Journal of Medicine, 2016-03-04
• Addendum guidelines for the prevention of peanut allergy in the United States — Journal of Allergy and Clinical Immunology, 2017-01-06