In this study we explored several vital questions about food allergy immunotherapy. The first, whether and how long the clinical effects of immunotherapy last when exposure ceases, informs the risk-benefit analysis for these treatments and offers important lessons about how immunotherapy might need to be conducted to be safe. The second, comparing the safety and efficacy of SLIT and OIT methods, underscores the differences and limits of currently available forms of therapy. Finally, our laboratory studies shed light on the mechanism of immunotherapy.
Until now, there have been sparse data about whether mucosally delivered immunotherapy for food allergy leads to a permanent state of tolerance or whether the effects represent temporary desensitization. Two small studies of egg immunotherapy offered different answers to this question in studies of very different lengths. Two of 4 subjects regained reactivity 3 months off therapy in one study
13 compared with no regaining of reactivity after 1 month in the other study.
14 In our current study we found that 6 of the 15 subjects who passed a full milk challenge after 60 weeks of maintenance lost desensitization within 6 weeks. It is reassuring that even after avoidance, the lowest reaction threshold was equivalent to about 2.5 oz of milk, which is well above most accidental exposures, and most did not react until a full cup of milk. In contrast, at the start of therapy, the median food challenge threshold was less than 1 teaspoon. However, it is clear that for most subjects treated with the doses and times in this study, the posttreatment clinical state is far from nonallergic. It is also important to note that 13 other subjects were still reactive after therapy and were therefore not eligible to participate in this test of tolerance. In all likelihood, they would have been much more prone to losing the desensitization that they had achieved with the withdrawal of therapy. Whether a desensitized but not tolerant outcome justifies the risk and effort of immunotherapy needs to be debated.
Two of the subjects regained reactivity within just 1 week, shedding further light on this quandary. The 1-week time point was chosen because we would expect that most patients would avoid an oral allergen for at least 1 week at some point whether because of illness, travel, or aversion to the food. If reactivity can return within this short period, as we demonstrate here, patients will need to maintain strict adherence to inclusionary diets for an indefinite period of time. Doing this might be less difficult for milk and egg for which we expect, but do not conclusively know, that cooked forms will maintain desensitization. For peanut, however, indefinite daily consumption might be impossible. In a study of children with naturally resolved peanut allergy, Fleischer et al
18 found that only 7% ate peanut daily and only 31% ate it at least once per week, likely because of persistence of their learned aversion to peanut. We continue to see this problem in patients with resolved peanut allergy, even when these diets are explicitly recommended. Therefore even after apparently successful immunotherapy, patients might be at high risk for allergic reactions if they do not consume the allergen regularly but feel falsely protected from accidental exposures.
However, it might be that with longer treatment we will find longer-term resolution. For venom and aeroallergen immunotherapy, at least 3 years of treatment is usually required to have long-lasting effects,
19,20 and OIT might have similar require-ments. In a study of peanut OIT, IgE levels continued to decrease at 30 months of treatment.
10 Although cumbersome, we will need longer-term studies to evaluate this hypothesis.
The second novel feature of this study was the direct comparison of SLIT and OIT. Here we sought to minimize the rate of adverse reactions by initiating therapy with SLIT and then to determine whether subsequent treatment with SLIT or OIT might be most promising for future use. At the doses we were able to use, given the limitations of SLIT dosing imposed by both volume and concentration, SLIT followed by OIT was much more effective at desensitization than SLIT alone but was accompanied by an increased risk of systemic side effects. Only 1 of 10 subjects treated with SLIT was able to consume a full serving of milk without symptoms after therapy compared with 14 of 20 subjects treated with SLIT/OIT. However, multisystem reactions were more than 11 times more likely with OIT. The difference in efficacy and safety between the groups is likely related to the dose of allergen given; the cumulative dose that the subjects in the SLIT group received was at least 140-fold lower than the minimum cumulative OIT dose.
Higher SLIT doses might have improved efficacy
4,5 but will require more concentrated forms of therapy. We administered the highest volume we thought practical, 700 μL 3 times, but only achieved 7 mg of CM protein daily. Even this small volume was too much for some subjects: 2 subjects aspirated the liquid and required epinephrine for lower respiratory tract symptoms, a complication of SLIT with liquid extracts, which should be noted. Potentially, multiple doses per day might be more effective.
21 However, given that exercise after dosing is a concern,
15,22 this option might not be feasible for children.
We also examined 2 OIT maintenance doses and did not find significant differences in either the overall rate of reaction or efficacy between these regimens. However, we lacked the power to detect relatively small differences between these regimens, and there was a trend toward improved efficacy with the higher OIT doses. Previous studies have used a wide range of OIT doses
5,6,15,16,23 and protocols, making historical comparisons difficult. Given the suggestion of increased efficacy, it seems reasonable to aim for higher doses with OIT.
The mechanisms of action of specific mucosal immunotherapy remain unclear. Similar to what has been found with SLIT using inhalant allergens,
19,20 we found that CM-specific IgE antibody levels increased early in dose escalation. In the subjects receiving OIT, CM-specific IgE levels subsequently decreased compared with baseline values but remained high throughout the study. Thus deletion of CM-specific plasma cells and decreases in IgE levels are unlikely to be the principal mechanisms of clinical improvement over the interval of this study. The decrease in CM-specific IgE levels was associated with a decrease in SHR, but the clinical significance of this is not clear. IgG
4 antibody has been shown to suppress IgE-mediated mast cell and basophil degranulation by competing with IgE for binding to allergen and by causing coligation of FcεRI and FcγRIIB.
24 CM-specific IgG
4 antibody levels did increase, but these changes did not correlate with clinical outcome (see Results and Fig E1 in this article’s Online Repository). However, we had limited power to examine predictors of tolerance and other clinical outcomes.
Many studies have focused on the basophil as a key cell type underlying clinical desensitization. Decreased expression of basophil activation markers, specifically CD63, which is often used as a surrogate for HR,
25,26 after allergen stimulation has been observed in children undergoing immunotherapy for food allergy.
10 Despite the overall favorable clinical outcomes in this study, we did not observe any significant decrease in allergenor anti-IgE–induced basophil HR at any point, suggesting that immunotherapy did not induce a state of intrinsic basophil suppression. This is consistent with the observed lack of change in expression of Syk, a tyrosine kinase that functions as a key determinant of HR.
27 Prior reports of reduced allergen-induced CD63 expression with immunotherapy used whole-blood assays,
10,28 in contrast to our washed suspensions. We hypothesize these washes might have disrupted binding of IgG or another serum inhibitory factor, which could account for the discrepant results. Interestingly, constitutive expression of CD63 and CD203c did decrease significantly with therapy, potentially suggesting a state of reduced extrinsic basophil activation
in vivo.
CD203c expression on basophils increased very early in therapy. High constitutive CD203c expression has been associated with clinical inflammation, such as that found with asthma exacerbations.
29 We found that subjects who had a greater increase in CD63 and CD203c expression early in therapy had statistically significantly poorer outcomes later in the study, suggesting a potential early marker of clinical response. However, no marker at the time before therapy was withdrawn distinguished those who were tolerant from those who were not. For more details, see Fig E1 and the Results section in this article’s Online Repository.
There are several limitations to this study. In the absence of a placebo, we cannot know how many subjects would have spontaneously improved. However, over the time frame of this study, we would not expect resolution of CM allergy in subjects of this age who reacted at the required baseline challenge dose. A previous placebo-controlled study of CM immunotherapy in subjects drawn from our same clinical population did not find any improvement from baseline.
15 This study was also not designed to fully assess whether starting with SLIT increased safety on transition to OIT. Future studies should compare various initial dose escalation schemes for immunotherapy and different durations of therapy. An additional potential limitation is that we do not know whether subjects who tolerated a full challenge to 8 oz of milk would react at higher doses. Also, we cannot account for day-to-day variation in food challenge threshold.
In summary, in this study we demonstrated that although most patients with even severe milk allergy can be desensitized to milk with OITand SLIT, tolerance is elusive for most after 60 weeks of maintenance. In some subjects desensitization can be lost within 1 week of allergen avoidance, and we lack reliable surrogate markers to predict those who will lose desensitization. Moreover, although most of the adverse events experienced by study participants were mild, there were occasional multisystem reactions, even at doses that had been tolerated for some time without reaction, as has been previously described.
12,15,30,31 Taken to-gether, these findings raise fundamental questions about whether and how immunotherapy should be conducted and emphasize that more work needs to be done before these therapies are ready for the general allergy clinic.
The publisher's final edited version of this article is available at 
