A Systematic Review and Meta‐Analysis on the Induction of Confirmed Eosinophilic Esophagitis as a Side Effect of Allergen Immunotherapy: An EAACI Task Force Report

ABSTRACT

The European Academy of Allergy and Clinical Immunology (EAACI) established a Task Force to assess the existing data on the relationship between eosinophilic esophagitis (EoE) and allergen immunotherapy (AIT). This systematic review and meta‐analysis aimed to study the incidence of confirmed EoE, developing as a side effect of AIT to food or airborne allergens, following the Preferred Reporting Items for Systematic Reviews and Meta‐Analysis (PRISMA) 2020 guidelines. The literature search was performed in three databases (PubMed, Embase and Scopus). Databases were searched from inception to March 31st, 2023. A total of 17 studies met the criteria for inclusion in the review. Fifteen studies, comprising 3,302 patients, were on food desensitization, and the overall estimate of EoE incidence, combining the results of these individual studies, was 2.31% (95% CI 1.45, 3.36). Registered data reported de novo cases of eosinophilic esophagitis, and its diagnosis was usually made during the maintenance phase of food desensitization. With the adopted searching strategy, only two studies on sublingual immunotherapy with aeroallergens meeting the inclusion criteria were retrieved, comprising 1,436 patients and not reporting cases of EoE. The meta‐analysis showed that the development of EoE is a common adverse effect of oral immunotherapy with food allergens, whereas it is uncommon during sublingual immunotherapy with aeroallergens.

Trial Registration: PROSPERO: CRD42023425917

Short abstract

1. Introduction

Eosinophilic esophagitis (EoE) is an immune‐mediated disease characterized by esophageal dysfunction, typical endoscopic features, and histologically predominant intraepithelial eosinophilic inflammation [1, 2, 3]. Its prevalence is about 0.01% in the general population, while, in high‐risk populations, such as siblings of EoE patients and individuals with food allergies, the prevalence is 1.7%–4.7% [4, 5]. EoE is a chronic disease associated with significant morbidity and impaired quality of life. Despite recent advances, unmet needs regarding its pathophysiology and management remain.

EoE is a multifactorial disease influenced by genetic predisposition and factors causing epithelial barrier dysfunction, such as environmental factors (e.g., allergens), lifestyle, and the microbiome [1, 2, 3, 4, 5, 6, 7]. It shares immunological features with IgE‐mediated allergies, involving eosinophils, mast cells, T‐helper 2 (Th2) cytokines and alarmins like thymic stromal lymphopoietin [6, 7]. Additionally, most EoE patients present atopic comorbidities, such as allergic rhinitis and IgE‐mediated food allergies, with sensitization to airborne and food allergens, respectively [8]. However, EoE is not an IgE‐mediated immune disorder; as a matter of fact, allergy‐test‐driven diets show no advantage over empirical ones, not all EoE patients have allergies or allergic sensitizations, and omalizumab, an anti‐IgE monoclonal antibody, has not been proven effective in clinical studies [9].

The role of food allergens in EoE is reflected in the positive impact of elimination diets on the natural course of this disorder in many patients [10]. Accordingly, the development of EoE in some patients treated with food oral immunotherapy (OIT) further highlights the significance of food allergens in the disease [11, 12]. Moreover, cases of EoE developed during sublingual allergen immunotherapy (SLIT) to airborne allergens have also been reported [12]. OIT, SLIT, as well as other forms of allergen immunotherapy (AIT) represent etiological treatments for IgE‐mediated allergy, aiming to attenuate and possibly stop detrimental immune system reactions to the culprit allergen(s) [13, 14]. For over a century, AIT has been used in various forms and in desensitization protocols to treat respiratory, insect venom and food allergies.

Since exposure of the esophageal mucosa to large amounts of food allergens or to relatively small quantities of airborne allergens has been suggested to lead to EoE [1], it is possible that the same allergens, when administered therapeutically, may induce or worsen EoE [15]. Yet, the relationship between AIT and EoE appears complex and remains elusive. More precisely, it is not known whether AIT may cause the development of de novo cases, worsen or relapse established EoE cases, or unmask a latent condition. At the same time, other forms of AIT that bypass the gastrointestinal tract, including subcutaneous (SCIT) and epicutaneous (EPIT) immunotherapy, may be exploited as a potential treatment for EoE.

Based on these premises, a Task Force (TF) has been established by the European Academy of Allergy and Clinical Immunology (EAACI) to assess the existing data on the complex relationship between EoE and AIT. The TF is planned to focus on three critical aspects: the potential development of EoE as a side effect of AIT, the use of AIT for the treatment of existing EoE, and ‐after outlining the current evidence‐ the establishment of guidelines for physicians who practice AIT and/or treat EoE patients. This paper presents the results of a systematic review and meta‐analysis on the incidence of confirmed EoE developed in patients treated with AIT for airborne and food allergens.

2. Methods

A full review protocol was registered in the International Prospective Register of Systematic Reviews (www.crd.york.ac.uk/PROSPERO) and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses statement (PRISMA) [16, 17]. The detailed methods are reported in the published protocol [18]. A succinct description of the employed methods is given here.

2.1. Search Strategy and Selection Criteria

A systematic strategy for literature search was developed, and validated filters for study design were applied in three electronic bibliographic databases (PubMed, Embase, and Scopus) to identify studies examining the association between EoE and AIT. Databases were searched from inception to March 31st, 2023. The search employed no restriction on the language of publication. In addition, grey literature (e.g., conference abstracts) was also searched and the list of references of full‐text articles was screened to identify further relevant studies. Personal communications were made to detect data on AIT side effects in National/International registries.

2.2. Eligibility Criteria

Studies were included if they provided original data on the occurrence of EoE in patients undergoing any type of AIT, including different protocols of food desensitization and any extract for airborne allergens (See Box 1 for full details). Studies that did not assess EoE or did not examine the occurrence of EoE during AIT were not eligible for the systematic review.

BOX 1.

Population	Allergic patients of any age, sex or ethnicity.
Intervention/exposure	Any type of allergen immunotherapy (AIT) reported to cause, unmask and/or exacerbate histologically (> 15 eosinophils/hpf) and clinically diagnosed eosinophilic esophagitis (EoE) [3].
Comparator(s)	Studies comparing the assessment of EoE before and after AIT and/or studies that histologically/clinically assess AIT‐treated patients to detect the development of EoE.
Study designs	Observational (prospective and retrospective) and interventional studies, including Randomized Controlled Trials (RCT).
Study outcomes	Primary outcomes: To detect the incidence of EoE development after AIT to food or to airborne allergens. Secondary outcomes: The effect of AIT on pre‐existing EoE; the course of EoE after discontinuation of AIT.
Exclusion criteria	Exclusion criteria included reviews and opinion articles, case reports and small case series (including < 10 patients overall), non‐research editorial articles, animal studies and in vitro and ex vivo studies not directly linked to clinical data. Studies using another therapeutic intervention (like omalizumab) in parallel with oral immunotherapy (OIT) were initially planned to be excluded. This was changed considering that the protocols did not affect the development of EoE. The decision was made before any analyses were undertaken.

Abbreviation: hpf, high power field.

2.3. Study Selection

Search results were uploaded into the Mendeley software and underwent de‐duplication. Literature citations were imported into the Rayyan web tool, and two investigators (C.M.R., I.T.) independently scrutinized the eligibility of all identified abstracts [19]. Discrepancies were solved through discussion and a third reviewer (C.P.) was consulted. Full‐text copies were obtained, and their eligibility was independently assessed by two reviewers (C.M.R., M.V.).

2.4. Data Extraction, Analysis and Synthesis

Data were independently extracted from the primary studies by two authors (C.M.R., M.V.) using a standardized detailed data sheet. The extracted data were then re‐assessed by two other reviewers (C.P., K.P.) and any discrepancies were resolved through discussion. The data collected included the first author's last name, publication year, study type, publication period, age, sex, characteristics of food and/or respiratory allergy of the study participant, type of allergen immunotherapy (e.g., oral, gastrointestinal delivery oral immunotherapy), sample size and number of EoE cases.

The meta‐analysis was performed using a random‐effects model (DerSimonian and Laird) [20], allowing for heterogeneity across studies. The outcome measures of the individual studies included cases of EoE and total sample sizes. Freeman‐Tukey‐transformed proportions (F‐T‐tp) were utilized as the effect sizes of each eligible study in the meta‐analysis. The Freeman‐Tukey transformation is a variance‐stabilizing transformation also known as the Freeman‐Tukey double‐arcsine transformation. It guarantees that the back‐transformed confidence intervals (CI) fall within the [0.1] range. The effect sizes in the forest plots present proportions along with the 95% CIs. The inverse Freeman‐Tukey transformation proposed by Miller was used to back‐transform the F‐T‐tp to proportions [21]. The inverse of the variance of the overall effect size was used to back‐transform the overall (meta‐analytic) effect size. The Wilson type was used for the CIs of the proportions of the individual studies presented in the results of the meta‐analysis (including those in the forest plots).

The heterogeneity between studies was evaluated with the I ² metric of inconsistency. Subgroup analyses were also undertaken to examine reasons for heterogeneity due to study design, type of immunotherapy, and type of food allergen. One‐way sensitivity analyses, each time excluding a different study, were also performed. This approach is valuable for evaluating the impact of individual studies on the overall meta‐analysis results and for identifying potential outliers.

The analysis was conducted using the statistical software STATA 15.0 (Stata Corp., College Station, TX, USA).

2.5. Risk of Bias (RoB) Assessment

Risk of bias (RoB) assessment was independently carried out on each eligible study by two reviewers (I.T., C.M.R., K.P. or C.P.). The quality of cohort studies was assessed solely based on whether the paper included comprehensive patient demographic data, the clinical and histopathological diagnostic criteria for EoE, the proportion of patients undergoing AIT and subsequently developing EoE, treatment outcome(s), and study design.

ROBINS‐I was used to assess the RoB of non‐randomized studies [22]. According to ROBINS‐I guidelines, a total score was calculated for each study, based on the fulfilled quality items and excluding studies of interventions with a high risk of RoB [22]. The tool assesses the RoB on several domains: (a) bias due to confounding, (b) bias in the selection of participants into the study, (c) bias in the classification of the interventions, (d) bias due to deviations from intended interventions, (e) bias due to missing data, (f) bias in measurement of outcome, (g) bias in selection of the reported results. The overall RoB of the study is judged to be high if at least one domain has a serious RoB.

Randomized controlled trials (RCTs) were evaluated using the Cochrane RoB tool [23]. The tool assesses the RoB from five domains: (a) arising from the randomization process, (b) due to deviations from the intended intervention, (c) due to missing outcome data, (d) in the measurement of the outcome and (e) in the selection of the reported results. Each domain can be graded as low or high RoB or it can be graded as having some concerns.

3. Results

3.1. Summary of Included Studies

The systematic search yielded 436 single publications, of which 16 studies met the inclusion criteria. During peer review, one additional eligible study that had not been captured in the initial search was identified and subsequently included. The PRISMA flow chart and results were updated accordingly. The PRISMA flow chart (Figure 1), presents the screening and selection process. Table 1 summarizes the studies included in the systematic review, published between 2009 and 2023. Fifteen studies were on food desensitization: 7 RCTs and pooled analysis of RCTs, 2 non‐randomized clinical trials, and 6 observational studies (Table 1) [24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 36, 37, 38, 39, 40]. Two studies focused on SLIT to airborne allergens: one RCT and one pooled analysis of RCTs [34, 35].

FIGURE 1.

Flow diagram of the searching strategies for identifying relevant studies on the development of eosinophilic esophagitis (EoE) after allergen immunotherapy (AIT). Abbreviations: RoB, Risk of Bias; RCT, randomized controlled trial; CT, controlled trial.

TABLE 1.

Characteristics of eligible studies investigating the development of eosinophilic esophagitis (EoE) as a side effect in patients treated with allergen immunotherapy (AIT) to airborne and food allergens.

First author (publication year)	Study design	Type of allergen immunotherapy	Type of extract	Sample size	Cases	Participants age group	Food allergen
Hofmann (2009) [24]	CT	OIT	In house m.	25	1	Children	Peanut
Fuentes‐Aparicio (2013) [25]	RCT	OIT	In house m.	40	1	Children	Egg
Echeverría‐Zudaire (2016) [26]	Observational	OIT	In house m.	128	6	Children	Milk and/or egg
MacGinnitie (2017) [27]	RCT	OIT	In house m.	37	3	Children	Peanut
Virkud (2017) [28]	RCT	OIT	In house m.	108	2	Children	Peanut
Fauquert (2018) [29]	RCT	GIDOIT	In house m.	21	0	Adolescents	Peanut
Nowak‐Wegzryn (2018) [30]	RCT	OIT	In house m.	44	1	Children and adults	Wheat
Yee (2019) [31]	Observational	OIT, GIDOIT	In house m.	13	1	Children	Peanut
Afinogenova (2020) [32]	Observational	OIT	In house m.	783	9	Children and adults	Peanut
García Rodríguez (2020) [33]	Observational	OIT	In house m.	90	3	Children	Egg
Halken (2020) [34]	RCT	SLIT a	Drug	923	0	Children	Timothy grass
Nolte (2020) [35]	RCT	SLIT a	Drug	513	0	Children and adolescents	Ragweed
Nilsson (2021) [36]	RCT	OIT	Drug	1217	12	Children and adults	Peanut
De Vlieger (2022) [37]	RCT	OIT	In house m.	78	0	Children	Egg
Ayats‐Vidal (2022) [38]	Observational	OIT	In house m.	41	1	Children	Milk and/or egg
Morales‐Cabeza (2023) [39]	Observational	OIT	In house m.	607	17	Children	Milk, egg, and/or peanut
Grzeskowiak (2023) [40]	CT	OIT	In house m.	70	1	Children	Peanut

Abbreviations: CT, controlled trial; GIDOIT, gastro‐intestinal delivery oral immunotherapy; In house m.; in house manufactured; OIT, oral immunotherapy; RCT, randomized controlled trial; SLIT, sublingual immunotherapy.

^a For aeroallergen.

Most studies on food desensitization (n = 13) dealt with protocols of oral administration of food allergens; one study evaluated gastro‐intestinal delivery of OIT (GIDOIT), while another one comprised both routes of administration (Table 1). Most food desensitization protocols (n = 14, 93.3%) dealt with in‐house manufactured formulations, while only one study on a commercial product for OIT was retrieved [36]. As to the age groups, 12 (80%) OIT studies had included patients under 18 years of age, while three reported data from both children and adults. A total of 3302 patients following food desensitization protocols were included in the analysis.

In both SLIT‐for‐aeroallergen studies included in our systematic review, EoE was considered a potential side effect. A total of 1436 subjects were enrolled in them, and none developed EoE during SLIT [34, 35].

3.2. Quality Assessment

Overall, the quality of the included non‐randomized studies was high [24, 26, 30, 31, 32, 33, 38, 39, 40]. Only one included study had serious RoB, but its total RoB score was considered appropriate to be included in the meta‐analysis. The RoB ratings of the included non‐randomized studies are shown in Table 2.

TABLE 2.

Quality assessment of the evidence based on ROBINS‐I (“Risk Of Bias In Non‐randomised Studies of Interventions”) in studies included in the systematic review for the development of eosinophilic esophagitis (EoE) after allergen immunotherapy (AIT).

Author, year [reference]	Risk of bias	Domains of bias
Hofmann, 2009 [24]	Moderate	Missing data
Grzekowiak, 2023 [26]	Moderate	Missing data
Echeverria‐Zudaire, 2016 [27]	Moderate	Measurement of outcomes
Yee, 2019 [28]	Low	None
Afinogenova, 2020 [29]	High	Selection of participants, classification of interventions, missing data
Garcia Rodriguez, 2020 [30]	Moderate	Measurement of outcomes
Ayats‐Vidal, 2022 [31]	Moderate	Missing data
Morales‐Cabeza, 2023 [32]	Low	None

Five RCTs were found to have a low RoB, and two RCTs presented some concerns regarding various domains [25, 27, 28, 29, 36, 37]. The “traffic light” plot for domain‐level judgment for each individual result [41] is shown in Figure 2. Both studies on SLIT had a low RoB.

FIGURE 2.

Risk of bias (RoB) domains for the randomized controlled trials (RCT) on the development of eosinophilic esophagitis (EoE) after allergen immunotherapy (AIT).

3.3. Primary Outcome

3.3.1. Incidence of EoE in Patients Undergoing Immunotherapy for Food Allergy

Figure 3 presents the meta‐analysis results for EoE during food desensitization, including 15 treatment arms with a total of 3302 patients and an overall number of 58 EoE cases. The reported incidence of EoE in the individual studies ranged from 0% to 8.11%. The meta‐analytic (overall) estimate of EoE incidence, combining the results of the individual studies, was 2.31% (95% CI 1.45, 3.36). A moderate degree of heterogeneity across studies (I ²: 48.4%, p = 0.019) was found. The diagnosis of EoE was based on both clinical and histological parameters.

FIGURE 3.

Meta‐analysis of studies on the development of eosinophilic esophagitis (EoE) during food desensitization, comparing the frequency among non‐randomized controlled trials (CT), randomized control trials (RCT) and observational studies. CI, confidence interval; DL, DerSimonian and Laird approach.

Figure 4 shows the results of one‐way sensitivity analysis. A potentially more pronounced influence on the overall effect size was observed for the studies of Afigenova et al. and Nilsson et al. [32, 36]. Nevertheless, this impact is not substantial as the CI lines of the sensitivity analyses intersect the reference line based on the meta‐analysis of all eligible studies.

FIGURE 4.

Forest plot of the influence of each individual study (one‐way sensitivity analysis) on the overall (meta‐analytic) estimate of the incidence of eosinophilic esophagitis (EoE) after allergen immunotherapy (AIT). CI, confidence interval.

3.4. Subgroups Analysis

• Study type: There was a trend for a lower incidence of EoE in RCT (n = 7 studies) (1.71%, 95% CI: 0.59, 3.34) than in observational (n = 6) studies (2.78%, 95% CI: 1.41, 4.57) and non‐randomized controlled trials (CT) (n = 2) (2.37%, 95% CI: 0,10, 6.63).

• Age [children versus (vs) adults]: A subgroup analysis based on age was not possible since studies comprising only adults were not retrieved. The retrieved publications reporting food sensitization on adults had also included children or adolescents, without reporting separate data.

• Type of immunotherapy (OIT vs GIDOIT): The analysis indicated that the incidence of EoE in patients undergoing desensitization with oral delivery of (any) food allergen (n = 13 studies) was 2.27 (95% CI: 1.41, 3.33). No incidence was described in the study of GIDOIT‐only delivery (Figure 5). Given this extreme imbalance in both sample size and event occurrence, a meaningful subgroup meta‐analysis could not be performed. Instead, the results are presented descriptively, acknowledging the absence of events in this subgroup. Although a subgroup analysis was initially planned comparing only peanut‐specific OIT with GIDOIT for peanut, the lack of observed events in one subgroup prevented a statistically reliable comparison.

• Type of food (peanut vs. egg vs. wheat vs. milk and/or egg vs. milk, egg and/or peanut): Most studies dealt with peanut (n = 8) or egg (n = 3) immunotherapy. A comparison of peanut vs. egg vs. wheat, showed similar results with an incidence of EoE at 1.68% (95% CI: 0.84, 2.79) for peanut, 1.64% (95% CI: 0.00, 5.21) for egg, and 2.27% (95% CI: 0.40, 11.81) for wheat. A higher incidence of EoE was found in studies (n = 2) reporting milk and/or egg immunotherapy (4.36%, 95% CI: 1.72, 8.04). There was not any statistically significant heterogeneity between groups (p = 0.319) (Figure 6).

• Type of extract (in‐house manufactured vs registered): Only one study adopted a registered product for AIT [36]. There was a trend toward a lower incidence of EoE in the study adopting the registered OIT product (0.99%, 95% CI: 0.56, 1.72) vs. the in‐house manufactured products/extracts of food allergens used for OIT (2.61%, 95% CI: 1.65, 3.77) (p = 0.019).

• Number of food (single vs. multiple): No significant difference in EoE incidence was found across studies including patients receiving immunotherapy for one food (n = 12), as compared to studies (n = 3) including patients receiving more than one food (Figure 7).

FIGURE 5.

Meta‐analysis of studies on eosinophilic esophagitis (EoE) development during food desensitization, comparing the frequency between oral (OIT) and gastrointestinal (GIDOIT) delivery of the food allergens. CI, confidence interval; DL, DerSimonian and Laird approach.

FIGURE 6.

Meta‐analysis of studies on eosinophilic esophagitis (EoE) during food desensitization, comparing the incidence of EoE between different food allergens, and groups of food allergens as reported in the included studies. CI, confidence interval; DL, DerSimonian and Laird approach.

FIGURE 7.

Meta‐analysis of studies on eosinophilic esophagitis (EoE) development during food desensitization comparing the incidence of EoE between studies reporting the delivery of desensitization to a single food versus desensitization to multiple allergens. CI, confidence interval; DL, DerSimonian and Laird approach.

3.4.1. Incidence of EoE in Patients Undergoing Immunotherapy for Respiratory Allergy

Although there are remarkably numerous studies on AIT with airborne extracts in the literature, the use of “eosinophilic esophagitis” as one of the keywords for the search strategy restricted the eligible studies of AIT to airborne allergens to only two of them. These were controlled studies and mentioned the development of EoE as one of the examined side effects [34, 35]. Collectively, they reported an overall of 1436 patients undergoing SLIT for respiratory and ocular allergy, while a total of 1414 individuals received placebo. More specifically, in one of these studies, patients were treated with Ragweed SLIT tablets (513 children/adolescents) [35] and in the other with Timothy grass SLIT tablets (923 children) [34].

Our literature review uncovered several case reports and a case series on EoE induced during SLIT. Yet, these studies were excluded from the analysis since they did not meet the inclusion criteria, but can be found in the Table S1 [42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52].

3.4.2. Secondary Outcomes

In the retrieved studies there were no data reporting preexisting endoscopy and histological examination performed before the start of AIT. On the other hand, the diagnosis of “de novo” EoE was usually made in the maintenance phase (Table S2). The median occurrence time was usually not specified, yet some cases were reported up to 4.5 years after OIT start [33]. In one case EoE was diagnosed 2 years after OIT discontinuation [38]. However, not all patients presenting with gastrointestinal symptoms were referred for endoscopy [28, 36], even in cases of persistent symptoms [28]. Thus, the true incidence of AIT‐related EoE may have been underestimated.

After EoE diagnosis, the most frequent outcome was AIT discontinuation. In the case of persistent disease, the administration of proton pump inhibitors alone or together with topical steroids was described [26]. No cases of worsening of the disease in established EoE patients treated with immunotherapy have been retrieved.

4. Discussion

According to the European Medicine Agency (EMA), the term “common” side effect describes the occurrence of a side effect in 1%–10% of the patients receiving a treatment [53]. Our meta‐analysis confirmed EoE as a common side effect of food desensitization protocols, with an overall incidence of 2.31%, whereas no cases of EoE were observed after immunotherapy for respiratory allergy in the studies retrieved with our searching strategy. Moreover, although the difference did not reach statistical significance, a trend toward a lower estimate was observed in RCTs compared to nonrandomized controlled trials and observational studies. This observation may reflect differences in study design and, hence, duration (RCTs usually have a shorter follow‐up period), which could influence the assessment of EoE development with oral immunotherapy.

The overall incidence estimate of our metanalysis is similar to that of previous work by Lucendo et al. where the overall incidence of EoE after OIT was 2.7% (95% CI: 1.7–4.0). This small difference may be due to differences in methodology, or due to new studies that were conducted in the meanwhile and were included in our updated systematic review [54].

Retrieved food desensitization studies were limited to peanut, milk, egg and wheat, with no statistically significant increased association with EoE found for any of these allergens. Although no milk‐only immunotherapy study was retrieved, studies that included participants treated with milk—either alone or in combination with other allergens—appeared to show a relatively higher incidence of EoE (4.36% in studies of milk and/or egg) compared with studies involving peanut‐only (1.68%) or egg‐only (1.64%) immunotherapy. Although the evidence remains indirect, this pattern may suggest a possible association between milk desensitization protocols and a higher likelihood of EoE development. Methodological limitations should be considered, such as overlapping allergen categories (e.g., egg included in multiple groups) and combined allergens (egg and milk) in one group (Figure 6).

Other food sources have also been used in AIT protocols, such as fruit. More precisely, a Cochrane systematic review has included two OIT studies with fruit; one performed with peach extract obtained from fresh peelings and one with fresh apple [55]. Transient gastric complaints during peach desensitization were registered, but no EoE cases have been reported.

Our systematic review has identified two RCTs regarding the use of SLIT with aeroallergens, with sufficient data for inclusion in the meta‐analysis [34, 35]. Yet, no case of EoE related to SLIT was reported in the retrieved studies. There are several clinical trials on SLIT; however, they were not included in this systematic review, since EoE was not expressly mentioned either in their text or abstract and thus, no relevant record was retrieved with our search strategy. In a Cochrane systematic review on the use of SLIT for asthma, data on 3086 patients who received actual extracts and 1724 who received placebo were reported [56]. No cases of EoE were reported in that workA case of erosive esophagitis was registered during a placebo treatment in a RCT with house dust mites (HDM) SLIT tablets [57]. However, “gastrointestinal problems”, including esophageal symptoms, are frequently reported in studies of SLIT [56]. It is possible that some cases of EoE are included under this nonspecific labeling. However, an endoscopy exam is not always undertaken, especially if symptoms are not bothersome, thus limiting the ability to correctly assess the incidence of EoE in patients undergoing AIT.

According to a recent multicenter phase 3 safety trial on adolescents receiving HDM sublingual immunotherapy with tablets, no cases of EoE were reported; however, the observation time was very short (28 days) [58]. In a recent (published after our search) real‐world study, which was based on administrative data and aimed at evaluating the safety of immunotherapy with a HDM tablet product, EoE occurred only in one adult patient with an overall incidence of 2.3 cases per 1000 patient years (95% CI: 0.3, 16.1) [59].

Based on the case reports and not on evidence from clinical trials, the prescribing information leaflets of some, but not all, SLIT products include EoE among the contraindications [60, 61, 62, 63, 64, 65, 66]. Aside from case reports, another source of information is registries. In a personal communication with the Dutch National Center for side effects, 15 cases of EoE have been registered with the use of HDM, Birch and Grass tablets. Moreover, in an international database of the World Health Organization (WHO), 106 cases of EoE with SLIT tablets have been reported, 25 for Birch tablets, 24 for HDM tablets and 57 for Grass tablets. Despite the relevance of these data, it is not possible to calculate the incidence of EoE due to SLIT tablets without knowing the total number of patients who had received such treatments.

SLIT has also been proposed as a desensitization modality for food‐allergic patients. Studies investigating SLIT with peanut, milk, peach, hazelnut, and kiwi have reported promising outcomes [67]. Current evidence indicates that while SLIT for food allergens is less effective than OIT at inducing desensitization, it has a better safety profile due to the lower doses of allergens required for treatment [67]. In our meta‐analysis, EoE is not reported as a side effect of SLIT with food allergens, since no study was retrieved using our searching strategy. Interestingly, it has been reported only in a single case report, in which SLIT for aeroallergens was also co‐administered [68].

Theoretically, SLIT might further reduce the risk of EoE by minimizing the contact of the allergen with the esophageal mucosa. A similar protective effect can be expected by GIDOIT, which bypasses direct esophageal exposure; however, in our meta‐analysis the available data were insufficient to provide statistical evidence of a lower risk of EoE.

In our meta‐analysis, there was no study assessing the impact of AIT on preexisting EoE. In a study assessing adults before starting peanut OIT, asymptomatic esophageal eosinophilia (> 5 eos/hpf) was observed in 5 of the 21 participants. Three of these with mild endoscopic abnormalities met histological thresholds for EoE but they lacked esophageal symptoms, and they were classified as having asymptomatic esophageal eosinophilia rather than clinical EoE [69]. Furthermore, in a study on children before starting milk or egg OIT, the presence of asymptomatic esophageal eosinophilia and accompanying EoE‐related endoscopic findings was reported in 33.3% of patients [70]. Unfortunately, the authors of these two studies did not provide data after OIT [69, 70]. In two case reports of SLIT administered to patients with anamnesis of preexisting but latent EoE, symptoms flared up after SLIT, but also after SCIT in one of them [44, 47]. In both patients, AIT was discontinued with symptom remission.

The strengths of this meta‐analysis include its robust methodology, the inclusion of only studies with a low risk of bias, and the addition of new data on the incidence of EoE following immunotherapy for respiratory allergens, a topic that had not been previously analyzed.

In terms of limitations, the heterogeneity in defining EoE development after AIT across the included studies makes it difficult to calculate the true incidence of EoE. We decided to calculate the incidence of de novo EoE based on the rate of diagnosed cases among patients undergoing the intervention, since this approach was more feasible and practical, though it may suffer from limitations. An incidence rate based on the comparison with pre‐intervention endoscopical assessment might be more accurate but is unrealistic, since baseline endoscopy was rarely performed in the retrieved studies. In this instance, the identification of noninvasive or minimally invasive diagnostic biomarkers of EoE may be of help in future studies [71, 72].

Moreover, since we focused on confirmed cases of EoE among patients undergoing AIT, we did not include general gastrointestinal symptoms among the terms adopted in the searching strategy; thus we were not able to calculate the rate of EoE cases among the proportion of patients developing gut symptoms after AIT. This may have led to an underestimation of the true incidence of EoE, especially in children who did not undergo endoscopy, given the more heterogeneous presentation of EoE in this age group. Importantly, children with active or resolved food allergy—even in the absence of OIT—may have an increased risk of developing EoE, with one study reporting a prevalence of 4.7% compared to 0.04% in the general population [5]. Furthermore, esophageal symptoms suggestive of EoE, but without histologic confirmation or with normal biopsies, are more common than biopsy‐proven cases of EoE and may still lead to AIT discontinuation. According to one available study, the rate of abdominal pain ranges from 7.9% to 14.3% depending on the food used during OIT and may lead to treatment discontinuation [5]. However, mild or transient symptoms may not lead patients to seek medical attention, and symptoms might resolve if AIT is discontinued, ultimately preventing endoscopic confirmation of EoE. Estimating the incidence of de novo cases of EoE based on symptoms may also suffer from limitations, since the percentage of patients developing persistent gastrointestinal symptoms is often not clearly reported in clinical studies.

Another weakness of our systematic review is that the data on the potential effect of SLIT for aeroallergens on EoE development primarily consisted of case reports, which were therefore excluded from the meta‐analysis, making it difficult to calculate the incidence of EoE associated with SLIT. A further limitation of this review is that we did not have access to individual participant data, which restricted our ability to perform detailed subgroup analysis (e.g., by age and allergen type). Although our statistical analysis was rigorously conducted, the comparison between registered and non‐registered AIT products should be interpreted with caution, as only one study represented the registered product group, limiting the generalizability and statistical power of this subgroup analysis.

We did not focus on the incidence of other primary eosinophilic gastrointestinal disorders (EGID) other than EoE after immunotherapy. Of note, among the retrieved studies, only one study reported cases of non‐EoE EGID associated with OIT [26]. More precisely, in that study, two cases were described and reported: one case of eosinophilic gastroenteritis and one case of eosinophilic colitis, both occurring together with EoE during the maintenance phase. Immunotherapy was discontinued in both patients. To the best of our knowledge, only a few additional cases of non‐EoE EGID temporally associated with immunotherapy, both with OIT, are described in the literature [73, 74].

In conclusion, this meta‐analysis confirmed the results of a previous systematic review and meta‐analysis [54], corroborating the substantial risk of EoE after OIT for IgE‐mediated food allergy. Therefore, this side effect should be clearly mentioned, along with the other adverse reactions of OIT, during the management process with the patient/family members. EoE should be ruled out by means of endoscopy with biopsy in patients presenting with suggestive and severe/refractory symptoms after starting OIT. Moreover, the findings of the present meta‐analysis suggest that EoE is a rare side effect of SLIT for aeroallergens and the possibility of developing EoE should not discourage its therapeutic use. The use of registry data is suggested to assess more precisely the risk of EoE after AIT.

Author Contributions

C.P. and C.M.R. conceptualized and supervised the study. E.A., C.M.R. and C.P. wrote the first draft of the manuscript. C.M.R., I.T., M.V., K.P., C.P., A.C. performed a literature search, extracted data, performed quality assessment and gave a significant contribution to data interpretation. K.P. and G.K.N. performed the statistical analysis and gave a significant contribution to meta‐analytic methodology and data interpretation. A.B., G.N.K., D.A.‐A., E.H., A.A.‐P., M.F.‐R., and O.P. gave a significant contribution to data interpretation and discussion in their field of expertise. All authors critically reviewed the manuscript and approved it for publication.

Funding

This systematic review and meta‐analysis received support from EAACI as part of the Task Force “Eosinophilic esophagitis and allergen immunotherapy” (Budget code 40806; years 2022–2024).

Conflicts of Interest

Ingrid Terreehorst has received personal fees from Sanofi and she is a member of EAACI ExCom. George N. Konstantinou is or recently was a speaker and/or advisorfor and/or has received research funding from AstraZeneca, Chiesi, GSK, Menarini, Novartis, Nutricia, Pfizer, Sanofi, Vianex. Darío Antolín‐Amérigo grants and/or personal fees from SEAIC, ALK‐Abelló, AstraZeneca, Chiesi and Gebro, AstraZeneca, Chiesi, Gebro, GSK, Leti Pharma, Menarini, Mundipharma, Novartis, Roxall, Sanofi. He is co‐chair of the Continuing Medical Education and Specialty Committee of EAACI and a member of the Allergen Immunotherapy Interest Group of EAACI. Enrico Heffler reports grants and/or personal fees and/or travel support from Sanofi, Regeneron, AstraZeneca, Novartis, GlaxoSmithKline, Chiesi, Almirall, Bosch, Lofarma, Stallergenes‐Greer. Oliver Pfaar reports grants and/or personal fees and/or travel support from ALK‐Abelló, Allergopharma, Stallergenes Greer, HAL Allergy Holding B.V./HAL Allergie GmbH, Bencard Allergie GmbH/Allergy Therapeutics, Laboratorios LETI/LETI Pharma, GlaxoSmithKline, ROXALL Medizin, Novartis, Sanofi‐Aventis, Sanofi‐Genzyme, Med Update Europe GmbH, streamedup! GmbH, Pohl‐Boskamp, Inmunotek S.L., John Wiley and Sons/AS, Paul‐Martini‐Stiftung (PMS), Regeneron Pharmaceuticals Inc., RG Aerztefortbildung, Institut für Disease Management, Springer GmbH, AstraZeneca, IQVIA Commercial, Ingress Health, Wort&Bild Verlag, Verlag ME, Procter&Gamble, ALTAMIRA, Meinhardt Congress GmbH, Deutsche Forschungsgemeinschaft, Thieme, Deutsche AllergieLiga e.V., AeDA, Alfried‐Krupp Krankenhaus, Red Maple Trials Inc., Königlich Dänisches Generalkonsulat, Medizinische Hochschule Hannover, ECM Expro&Conference Management, Technical University Dresden, Lilly, Japanese Society of Allergy, Forum für Medizinische Fortbildung, Dustri‐Verlag, Pneumolive, ASIT Biotech, LOFARMA, Almirall, Paul‐Ehrlich‐Institut, outside the submitted work; and he is Vice President of the EAACI and a member of EAACI Excom, a member of the ext. board of directors DGAKI; coordinator, main or co‐author of different position papers and guidelines in rhinology, allergology and allergen‐immunotherapy; he is associate editor (AE) of Allergy and Clinical Translational Allergy. Constantinos Pitsios is a member of EAACI ExCom and an Editorial Board member of Clinical and Molecular Allergy. The other authors declare no conflicts of interest.

Supporting information

Appendix S1: all70183‐sup‐0001‐AppendixS1.docx.

Table S1: all70183‐sup‐0002‐TableS1.docx.

Table S2: all70183‐sup‐0003‐TableS2.docx.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author.