66% of kids grow out cow’s milk allergy, even if they completely avoid it. But rate rises to nearly 90% if baked milk introduced. And less restrictive diet good for everyone, of course.

Salmivesi RCT from Finland 2012 – n=28, age 6-14. 81% using daily milk 6400mg protein at 12 months. First dose milk 0.06mg – 8 further observed doses (0.12mg day 8; 0.24mg day 15; 2.0mg day 36 [big jump!]; 4.0mg day 38 [2 days later!?]; 8.0mg day 42; 40mg [big jump again!] day 57; 80mg day 64; and 160mg on day 78) with other dose increases done at home. Monitored for 2hr in clinic. Final dose of 6400mg (=200ml) was given at home (!) on day 162.

Oral immunotherapy (OIT) for severe milk allergy (IgE >85 or low eliciting dose):  at 1yr 36% tolerated 150ml, 54% 5-150ml (good enough for accidental exposure). 10% could not complete protocol. [reference?]

DRACMA update on milk immunotherapy (2021) mostly fresh milk used in research. Randomized trials looked at kids aged 2-14, mean 8. Non randomized had wider range. 67% tolerated 150ml (cf 2.2% of controls), 78% tolerated 5-150ml (=swig protection) (cf 3.3% of controls). Anaphylaxis rate of 5.5 per person-years (although not always defined, and not always usual definition). 63% used IM adrenaline!?

6.9% developed EOE but rarely biopsy confirmed. No deaths.

2–8 weeks after discontinuation of successful OIT, tolerance of cow’s milk persisted in only 36% (20%–91%). So you have to persevere life long, probably.

Quality of life badly reported – only 5 studies. 1 study found parents reported better QOL in 37%, worse in 26%, and unchanged in 38%! 2 conference abstracts reported improved QOL or at least reduced anxiety in general and reduced fear of unexpected reaction. Some studies found worse QOL! Certainty of evidence overall v low…

Only 1 trial and 2 non randomized studies of baked milk OIT! In the 1 RCT of baked milk, 73% achieved tolerance at 12 months (cf none of control group) – other studies did not find evidence of effectiveness however (meta-analysis of Anagnostou. Tolerance to heated or baked milk may be temporary and diminish within a few years. Dantzer. 20% required adrenaline! Only some children assessed for QOL, with evidence of benefit – of the parents studied, no improvement in QOL! [Dantzer and Dunlop 2021]. Low desensitization rate for unheated milk means persisting concern about exposure in real life despite less frequent adverse reactions (8%–33%).

Other studies did not find this relationship, which is consistent with the meta-analysis results of Anagnostou et al. (41). Tolerance to heated or baked milk may be temporary and diminish within a few years. Dantzer et al. found that a protocol involving gradually increasing doses of baked milk was effective in promoting desensitization (73% achieved end dose of 4000mg cf 0% of controls).  Mean age 11. “This suggests that the initial dose [and then building] may influence the efficacy of desensitization.” [Yan Wang, China – Front. Immunol., 04 June 2025]

4% rate of biopsy proven EOE – note EoE typically emerges approximately 2.8 years after initiating the maintenance phase so many studies will miss…

Longer duration better. None of the studies reported on sustained unresponsiveness.

Only Dantzer study reported quality of life! 

Highlights “a dire need for a standardization in outcome assessment and reporting,” and that successful completion of OFC needs to be validated as a predictor of “real world” success. [Natasha trial will hopefully clarify further…]

Separate article on OIT recommendations – apart from balancing risks and benefits, recommends using omalizumab (monoclonal anti-IgE) in advance and in initial stages of oral immunotherapy with unheated cow’s milk. Inferring from limited evidence (use in other food allergies, mostly) that risk of anaphylaxis reduced (RR0.34 but not significant!). Not really any downside? (But costs at least £256 per dose, maybe double? Lasts a month?). Plus, does not recommend baked milk OIT as very low certainty of benefit. More evidence on effect, risks, QOL benefit obviously required.  

Sublingual/epicutaneous immunotherapy (SLIT) for milk? Not much evidence – low rate of success and high rate of relapse.

[EGPHAN 2023, Frontiers in Pediatrics 2019; BMJ cmpa article sept 2013]

Rush study from Japan used microwaved milk – microwaved for 100s at 550W (!). Dosing was 2–4 times a day in hospital for several days (frequent adverse events) aiming for 200 mL dose (total daily, I assume). If not achieved, dose increases changed to 1ml per day. At 200ml, dosing changed to once a day. After 2 months of maintenance, length of time in the microwave gradually reduced.

Another Japanese study looked at using heated milk powder (3 mins at 125degC) vs unheated milk. Rush protocol (5 days in hospital) up to maintenance of just 3ml daily. At 1 year, 35% and 18% in the HM group and 50% and 31% in UM group passed the 3 and 25 mL OFCs, respectively, so not great efficacy. Rates of moderate or severe symptoms significantly lower (halved) in the heated milk group however.

Natasha Trial is looking at Peanut or cow’s milk OIT in UK – groups are Peanut age 6-23yrs, Peanut age 2-3yrs, Cow’s milk age 3-23yrs. Using everyday food products. Final results expected 2027.

Outcome

In small Korean study of kids aged 3-10, 83% has sustained unresponsiveness (stopped cow’s milk for 4 weeks, following 12 months of maintenance.

In a study of OIT plus omalizumab, about half had sustained unresponsiveness.

Duration of treatment appears to be key – in Japanese study, 2 years treatment had significantly more SU cf 1 yr treatment. Review here.

Gold standard for diagnosis of a food allergy or intolerance. A challenge is only considered complete once a normal age-appropriate portion of the food has been consumed.

In certain cases, food challenge is potentially more dangerous, and should only be done after assessment and discussion of likely risks and benefits. If there is no reaction (“negative challenge”) then of course there will be less fear and more food options [at least if food is consumed regularly]. If there is a reaction (“positive challenge” is the preferred term, not “failed”) then there is no longer any doubt, and the child/family will experience a reaction (useful if child does not remember) and see how it is managed. Lots of families say that using their adrenaline autoinjector for the first time (if it comes to that) takes away a lot of the anxiety about using it in future, which is very useful. So always check patient/family prepared to administer if required.

Objective measures of symptoms/signs preferred. Eczema, wheeze, congestion, diarrhoea should all have resolved (or be optimal) prior to doing the challenge, to avoid potential confusion. The challenge can be blinded if there is still doubt.

2024 PRACTALL update (Hugh Sampson)

Discussion about reliability of skin prick tests etc. Table of PPVs based on Riggioni metanalysis.

In terms of safety, severe reactions possible even with milligram level doses (and even in children with tolerance to baked versions!). Test results do not predict, other than possibly 2S albumins and tree allergy.

Outpatient clinic rooms are included in potential setting. Other considerations are delayed reactions; asthma; previous reactions to trace amounts; whether on elimination diet (?).

IV access should be considered if prior “severe anaphylaxis”, severe asthma, or likely difficulties getting access if required.

Lots of levels useful if trying to establish threshold, but makes it harder to achieve “typical” dose within time available – and potential for “rush desensitisation” effect! Some reports that multiple dosing can cause reactions where single dose wouldn’t… So some regimens combine logarithmic and semi-logarithmic increases.

Target dose generally 2-3g protein. 5% false negative rate for 875mg (cumulative 3500mg) top dose. Shellfish particularly needs a high target dose, fish similarly. But depends on size of child too…

Suggests 1 egg, 140-200ml milk, 28g cheese, 2g peanut, 3g tree nut.

Lip challenges not recommended as little data, all suggesting it is unhelpful.

Dosing intervals typically 15-20 mins but not very logical, given many reactions take longer eg milk, peanut, cooked egg. So now recommends 20-30 mins…

For FPIES, have IV fluids and ondansetron available. Check IgE/SPT negative! Consider methylprednisolone (but no data!). Taking baseline and 4-6hr neutrophil counts useful (>1500 cells/ml rise diagnostic where symptoms subjective). 0.06–0.3 g of food protein per kilogram body weight (maximum 3 g protein), administered as a single serving or as 2–3 servings every 15–30 min, followed by 4 h observation.

Else, 1/3 of the food portion for age is done under physician observation followed by a home titration to a full dose (very low rate of mild (mostly diarrhea) delayed reactions later during the day of the OFC or within the first few days of home dosing) – better than risk of causing severe symptoms during a single feeding with a higher dose of food. Some reports of FPIES recurrence after negative challenge…

Extensive use of omalizumab for asthma, FDA approved since 2003. Anaphylaxis rate is 0.1-0.2% so initial injections must be supervised (and adrenaline provided for home). Use in under 5s only approved since 2024, however.

Various small studies showed benefit in food allergy. 2024 OUtMATCH study of 180 patients (mostly children) with peanut allergy and at least 2 other food allergies found that after 16 weeks treatment, 67% of those treated with omalizumab could tolerate a 600mg single dose (1044mg cumulative) of peanut plus 1000mg challenge of the 2 other foods (cashew, egg, milk, walnut, hazelnut, wheat all included).

Dosing is as done for asthma, based on weight and total IgE.

1 study found efficacy with only 3-8 weeks of treatment. Duration of regimens varies from 16-34 weeks.

Sustained unresponsiveness (“cure”) has not yet been documented. In fact, protection seems to drop off within 8-12 weeks, so important to continue OIT.

Skin testing and sIgE levels are unreliable once omalizumab is started, so the only way to monitor response to treatment is with food challenges.

Study of omalizumab as monotherapy, but also as adjunct to multi-food oral immunotherapy. 180 children (age 1+) and 3 adults.

Lots more studies underway, now that Omalizumab approved for treatment of food allergy.

All have peanut allergy, but at least 2 other food allergies (milk, egg, wheat, cashew, hazelnut, walnut).

“Standard” dosing based on weight and total IgE levels, with omalizumab (ranging from 75 mg to 600 mg) administered every 2–4 weeks for a duration of 16–20 weeks.

After 16 weeks Omalizumab, 66.7% achieved challenge dose of 600mg peanut AND 1000mg for the other 2 foods. Success associated with higher total IgE (?), 2 week (vs 4 week) dosing (!), smaller SPT size, higher tolerated baseline threshold dose.

No change in quality of life!!! Only difference in adverse events was skin reactions to injections!

Threshold maintained or increased during open label extension (24 weeks total). Most people with milk, egg or wheat managed to continue foods in diet after negative challenge for 12 months (61-70%), only 38-56% managed to continue peanut or tree nuts… Consumption generally declined over time, apparently due to symptoms and patient preference. Higher screening challenge threshold predicted success, not surprisingly. 2 cases of EOE.

Stage 2 is 8 weeks of omalizumab for everyone, then either omalizumab for another 8 weeks PLUS OIT, or else omalizumab alone (placebo OIT) for 52 weeks total OIT.

Stage 3 (n=80) randomised to continued omalizumab or OIT. Initial Dose escalation was 3, 30, 60, 125, 250, 250, 375mg of each food, every 15 minutes.

Showed more than 60% of patients achieved endpoint after 12 months, whether doing omalizumab alone or doing OIT. Safety similar – 2 of OIT group had serious adverse events during real world transition. 1 case of EOE after omalizumab.

Thomas Casale review says don’t exclude high threshold patients, as still likely to have meaningful benefit. Dosing can include total IgE <30 or weight <10kg, just use nearest. Aim to suppress total IgE to below 10 IU/ml – calculate 0.005mg/kg per week for each 1IU/ml – see table here.

Of the 33% who did not respond, an extra 34% did seem to respond with extension – the rest did not (or actually worsened). Given this treatment is expensive, consider at least a single dose challenge at 20 weeks. Quality of life jumped significantly after challenge, once families were aware of response to therapy (since not ingesting!).

For milk/egg/wheat in young children, discontinue and do challenge at 4 months (5 half lives) after to see if resolved.

Good that no obvious safety problem with children under 6 (not previously studied). Potential concern around herpesviruses, bacterial infections, visual disorders. Congenital absence of IgE (!) possibly associated with asthma, otitis media, sinusitis, autoimmune disorders and cancer…

Similar review by Aikaterini Anagnostou from AAAI.