From Diet to Dysbiosis: The Microbiome’s Relationship to Eosinophilic Esophagitis

Abstract

Purpose of the Review

Eosinophilic esophagitis is a chronic, type 2 inflammatory disease of the esophagus. Diet elimination trials suggest that EoE is triggered by food antigens, however, incomplete response rates present gaps in our understanding. This review aims to synthesize current evidence surrounding the diet-microbiome-immune axis in EoE and propose a conceptual framework for how dietary interventions may influence disease beyond antigen elimination.

Recent Findings

Many studies on the EoE microbiome highlight descriptive differences in cross-sectional cohorts, although findings remain inconsistent and lack a unifying disease signature. We highlight evidence from broader mucosal and gastrointestinal microbiome research and discuss the importance of microbial metabolic function and their role in regulating epithelial barrier integrity and immune responses. We present the case that in EoE, elimination diets may unintentionally alter microbial substrates and metabolic activity, leading to heterogeneity in treatment response independent of dietary antigen removal. We review early and limited experimental data that suggest microbial metabolites such as short-chain fatty acids may influence esophageal barrier function and identify gaps in understanding mechanistic and longitudinal implications for patients.

Summary

The interactions and implications of the diet-microbiome-immune axis is ripe for exploration. We explore reframing elimination diets to think beyond the antigens being removed, and probe what microbially active foods may be introduced (or not) and thus have clinically relevant microbiome-modulating properties. Future work should prioritize longitudinal, diet-controlled, and multi-omic approaches to elucidate causal relationships and support the development of microbiome-informed, precision dietary strategies in EoE.

Introduction

Eosinophilic esophagitis (EoE) is an increasingly common chronic disease of the esophagus characterized by type 2 inflammation, affecting roughly 1 in 700 in the USA [1]. EoE is characterized by symptoms of esophageal dysfunction, including heartburn and dysphagia in adolescents and adults, with pediatric or atypical symptoms being food aversion, nausea or abdominal pain [2, 3]. Histologic inflammatory hallmarks of EoE include intraepithelial eosinophilic infiltration, dilated intercellular spaces, and basal cell hyperplasia. Although the pathophysiology of the disease is not fully understood, EoE displays a predominant type 2 immune response while recent evidence suggests interferon-γ may contribute to pathology [3–5]. Diet elimination and reintroduction studies along with case reports of allergy immunotherapy suggest EoE pathophysiology is triggered by environmental factors such as food or in some cases aeroallergens [6]. However, whether or not other environmental factors such as microbes play a role in pathogenesis or maintaining health remains to be determined.

EoE can be managed pharmacologically or with diet therapy. Diet elimination remains a mainstay in the care and management of EoE with success rates varying across populations, specific diet type, and over time spanning from approximately 40–70% efficacy [7–9]. Elimination diets can be broad removing dairy, wheat, soy, egg, peanuts/nuts and fish/shellfish (6FED), though recent data points toward improved quality of life and similar efficacy rates with one (dairy) or two-food (dairy and wheat) elimination diets [10]. Stepwise removal and reintroduction of trigger foods provide robust indirect evidence that EoE is triggered by food antigens. However, and interestingly, even the broadest elimination diet, elemental feeding, still leaves 10–30% of patients with active EoE [11]. Additionally, more direct methods of detecting food antigen specificity such as immunoglobulins or antigen-specific T cells have to date been unsuccessful for the masses; though these and aeroallergen-triggered approaches are still avidly being explored with early intriguing findings [10, 12, 13]. These knowledge gaps along with studies showing previous infections, antibiotic exposure, and birth via cesarean section as strong risk factors for EoE suggest microbial perturbation may be relevant in EoE [14–16]. Further linking the microbiota to EoE, diet is one of the strongest sculptors of microbial members and function [17–19].

The diet-microbiome-immune axis is an emerging frontier in EoE. While diet is well established to have significant influence on microbial metabolism and epithelial barrier function with downstream effects on immune responses generally, these effects are largely unknown in EoE. Host-diet-microbial interactions could result from gaining pathogenic interactions, or loss of beneficial ones. Because of this, a more comprehensive characterization of the diet-immune relationship and its interplay in EoE may identify novel biomarkers of treatment response and uncover mechanistic targets for more precise, less restrictive therapeutic strategies or prevention.

What We Know (and Don’t) About the Microbiota in EoE

Interest in the microbiome in EoE has expanded over the past decade, although primary literature largely remains limited to cross-sectional and descriptive investigations as a natural starting point. We will briefly summarize many of the findings thus far in the field while also identifying gaps in our understanding to date.

Esophageal and Upper Gastrointestinal Microbial Communities and Characteristics

Studies to date largely utilize 16S rRNA sequencing to identify bacterial community composition with varied approaches in sample acquisition, processing, alignment pipelines, and analyses. In general, the esophagus is recognized to harbor a resident microbial community that is compositionally distinct from stool, which makes ecological sense; while more similar environments like the oral microbiome share many of the same microbes as the esophagus [20, 21]. Spatial heterogeneity of microbes within the esophagus was examined using mouse models where investigators found Bifidobacteriales Order dominating the proximal esophagus while Lactobacillales Order resided distally [22].

At the compositional level, some studies report altered bacterial relative abundance in individuals with EoE compared to healthy adult controls identifying phyla-level differences in the esophagus, with increased Proteobacteria (now known as Pseudomonadota) and decreased Firmicutes (now known as Bacillota) [23]. Other readouts such as bacterial load has been shown to be increased in adults and children with EoE and gastroesophageal reflux disease regardless of treatment status or degree of mucosal eosinophilia compared to healthy controls [24]. Genera such as Haemophilus, Prevotella, and Streptococcus have been shown in varied studies to be increased in patients with untreated or active EoE and many of these organisms are also reflected in saliva [21, 24–26]. However, findings across studies have not always been uniform. Johnson et al. reported that the esophageal microbiome in patients with EoE could not be reliably distinguished from controls, suggesting the absence of a consistent disease-defining signature microbial makeup and/or small sample sizes [27].

Studies have also inquired whether therapy type may influence microbial composition. Laserna-Mendieta et al. found the esophageal microbiome varies according to therapy, suggesting that proton pump inhibitors (PPI), dietary elimination, and topical corticosteroids may independently shape microbial communities [28]. It is well established that PPIs change the intestinal microbial landscape though the significance is not known; topical steroids and dietary modulation of the esophageal microbiome is intriguing where glucocorticoids in other contexts are associated with microbial changes such as Candida blooms, the context in EoE along with dietary changes is relatively unexplored. Collectively, these findings suggest caution with interpretation of varied cross-sectional comparisons and underscores the need for treatment-stratified analyses.

One study to date of 14 pediatric EoE patients (7 responders to PPI and topical steroids, 7 refractory) utilized metatranscriptomics from bulk RNA sequencing data of formalin fixed esophageal biopsies [29]. Investigators described Fusobacterium and Acinetobacter species being abundant in EoE patients refractory to PPI and topical steroids, which to our knowledge may be the first description of Acinetobacter as a potential commensal of the esophagus. Interestingly, Acinetobacter has mainly been described as a detected organism using RNA isolation techniques as opposed to 16S rRNA sequencing. While Acinetobacter species are known as a rare oral microbe that can colonize oral biofilms and plaques, permitting opportunity for aspiration pneumonia or periodontitis, whether it has a role in EoE pathogenesis remains to be determined.

Fecal Microbiota Composition in EoE

Evidence examining the distal gut microbiome in EoE is comparatively sparse. In a small study of 12 adult patients with EoE compared to 12 non-EoE controls, investigators performed 16S rRNA sequencing on stool samples and showed that patients with EoE had lower alpha-diversity by Shannon index and beta diversity based on weighted Uni-Frac [30]. More specifically, there was a significant decrease in Clostridia and Clostridiales in patients with EoE. The only other study to collect stool was multi-center and did so from nearly 70 EoE patients and compared to 13 people without EoE; they did not find differences in diversity, abundance or taxa using 16S rRNA sequencing [21]. Stool-based studies in EoE remain limited in number and sample size, and there is a significant need for more longitudinal research and detailed cohort data on diet, medication use, and lifestyle differences.

Probiotic Studies in EoE

Interventional microbe studies in EoE are rare, though some murine studies have been explored. Holvoet et al. examined the beneficial effects of two organisms, Lactococcus lactis NCC2287 and Bifidobacterium lactis NCC 2818, both previously shown to be regulatory in pre-clinical food allergy studies [31]. Investigators administered the organisms ad lib to mice via their drinking water either before (preventative) or during (treatment) induction of EoE using the Aspergillus epicutaneous sensitization mouse model. L. lactis was found to reduce bronchoalveolar and esophageal eosinophilic infiltration when administered as a treatment following onset of EoE. More recently, in the same study that showed spatial heterogeneity in microbes also demonstrated that early exposure to broad-spectrum antibiotics diminished Lactobacillales, and these mice had worsened type 2 inflammatory responses using the Aspergillus model of EoE [22].

A brief summary of the high-level EoE and microbiome findings are summarized in Fig. 1.

Fig. 1.

Overview of human and murine studies investigating the microbiome in eosinophilic esophagitis. A Human studies have utilized varied biospecimens and sequencing methods, with high-level microbial findings summarized here. B Murine investigations utilized the Aspergillus model of EoE and describe regional heterogeneity in microbial abundance as well as probiotic administration with Lactococcus lactis showing decreased eosinophil inflammation. Figure was created with BioRender

Gaps and Opportunities in the Microbial EoE World

The microbiome science field is in the beginning stages of uncovering microbial composition and how it differs throughout the GI tract in states of health. The small bowel is a particularly challenging area to access during homeostasis, thus Shalon et al. developed a novel device, Cap-Scan, that sampled the small bowel lumen of 15 healthy, living participants during normal digestion [32]. They also obtained fecal samples. Investigators compared these sites and identified substantial differences between bacterial communities, host proteins, and bile-acid profiles across intestinal regions, and metabolomic analysis of the same cohort further demonstrated spatial variation, with microbial metabolites becoming increasingly present in the distal gut [33]. A later study sampled over 50 different sites from the skin and GI tract from 33 recently deceased individuals and performed 16S rRNA sequencing revealing that esophageal and stomach microbiome were most similar, while the oral cavity and large intestine were most different. The ecological environments between these sites likely largely explains the variability observed [34]. Together, these findings indicate that foregut and hindgut microbial communities are connected but remain regionally distinct. Understanding the relationship of regional microbiomes in healthy populations can help define an important baseline for distinguishing EoE disease-specific microbial variations.

Additionally, as with many immune-mediated diseases, human studies and microbiome research in EoE have focused predominantly on taxonomic compositional differences in diseased versus healthy populations. These studies are largely derived from 16S rRNA sequencing and snapshots in time between patient cohorts and usually of the esophageal microbiome. These approaches describe relative abundance patterns but leave much room for discovery into the assessment of microbial mechanisms including functional and/or metabolic activity and its relation to disease pathophysiology. The only intervention study directly comparing therapies demonstrated that food elimination diets (FED) significantly reduced esophageal microbial alpha diversity, whereas PPI and swallowed topical corticosteroids (STC) achieved comparable histologic remission without altering microbial diversity [28]. However, these methods infer potential function rather than directly measuring microbial activity or metabolite production with an intervention, and no studies to date have utilized metagenomics or directly measured longitudinal microbial outputs following nutritional changes. To further elucidate the role of the microbiome on disease onset and therapy response, longitudinal studies and functional analyses are needed. Understanding how diet shapes these microbial communities, and how microbial metabolites may influence epithelial barrier function and type 2-driven inflammation, represents a critical next step in translating associative findings into clinically meaningful insights.

Diet as Regulator: Mechanisms of Diet-Microbiome-Immune Crosstalk in Health and Disease

While EoE-specific data on diet-microbiome-immune interactions remain limited, we can draw on broader mucosal immunology and microbiome literature, in particular from intestinal models, to examine a framework within which these interactions may play a role in EoE.

The gut microbiome plays an important role in mucosal health. In certain contexts, perhaps even more important than the type of bacteria present is the behavior of given microbes in response to the diet. Functional metabolic outputs of the microbiome are termed microbial metabolites. Examples include short-chain fatty acids (SCFAs), tryptophan-derived indole metabolites, secondary bile acids, and sulfur compounds, all of which can influence epithelial barrier and immune function. Understanding how specific dietary components shape these outputs provides a framework for interpreting dietary interventions in inflammatory mucosal diseases, including EoE.

Fermentable Fiber and SCFA Production

One of the best characterized diet-microbiome relationships is the fermentation of dietary fiber to SCFAs, with butyrate, propionate, and acetate being the most well studied. The amount of SCFA produced depends on the type of fiber consumed, how microbially accessible it is, and the composition of the microbiome. In a controlled dietary intervention of 174 adults, resistant starch from potatoes produced the greatest increases in total fecal SCFAs and butyrate, while resistant starch from maize and inulin had more modest effects [35]. It’s worth noting that the ability for a given microbe to be a “SCFA producer” is not exclusive or constant; however genetic machinery and functional outputs for SCFAs have been described in several Clostridial clusters including IV, IX, and XIVa corresponding to Faecalibacterium prausnitzii, Roseburia spp., Eubacterium rectale, Anaerostipes spp., and Coprococcus spp. among others such as Bifidobacteria [36–39]. However, one of the most frequently seen organisms unique to the mucosal layer is Akkermansia mucinophilia, which is known for its unique ecological niche in the mucus lining the colon epithelia, and the ability to generate biologically significant SCFAs by digesting mucin [40]. Collectively, these organisms target specific fiber or carbohydrate structures and do not readily substitute one substrate for another, which partly explains why not all fiber supplements produce equivalent SCFA (or other microbial metabolite) responses even at matched doses.

Another fascinating factor in the microbiota is the importance of the ecological community. Microbes can be metabolically dependent on each other in addition to diet. For example, butyrate production can be dependent on interactions with non-butyrate-producing bacteria. Primary fiber degraders, such as Bifidobacterium species or Ruminococcus bromii, can break down complex polysaccharides and release acetate and monosaccharides that butyrate producers then consume [35]. In co-culture experiments, Bifidobacterium adolescentis degrades inulin and resistant starch to release acetate, which Faecalibacterium prausnitzii then converts to butyrate [41]. These interdependencies suggest that butyrate production reflects the output of a microbial consortium rather than any single organism, and that dietary strategies supporting a diverse fiber-fermenting community are more likely to produce robust SCFA output than those targeting a single taxon.

SCFA production also varies by baseline microbiome composition. Human microbiotas with Bacteroidaceae-Ruminococcaceae-dominated microbiota increase butyrate by up to 96% with fiber supplementation, while those dominated by Prevotellaceae-Ruminococcaceae preferentially generate propionate [42]. This inter-individual variation is a likely factor that contributes to the heterogeneous clinical responses observed with dietary interventions.

Once produced, SCFAs exert what are generally thought to be beneficial effects to hosts across multiple pathways. SCFAs can carry out their actions as histone deacetylase inhibitors (HDACi) or by binding to G protein-coupled receptors GPR41, GPR42, or GPR43. The anti-inflammatory and beneficial effects of SCFAs are also pleiotropic. For example, butyrate can serve as a primary energy source for colonocytes, stabilize hypoxia-inducible factor to maintain the anaerobic colonic environment, and maintain barrier integrity through regulation of claudin-1 and synaptopodin expression [36]. Butyrate also suppresses pro-inflammatory cytokines including IL-6 and IL-12, and inhibits oncogenic signaling via Wnt, and TGF-β pathways [36, 43, 44]. Conversely, acetate can activate GPR43 to regulate insulin sensitivity and inflammatory responses, while propionate can activate PPARα to modulate lipid metabolism [45, 46]. We are only beginning to uncover the myriad effects of SCFAs in health and disease.

Dietary Protein and the Protein-to-Fiber Ratio

Dietary protein influences microbial metabolic output differently than fiber. High-protein diets shift bacterial metabolism toward amino acid degradation, increasing production of branched-chain fatty acids, ammonia, phenolic compounds, and sulfur metabolites. Animal protein combined with minimal fermentable fiber generates the highest ammonia levels, while plant protein consumed alongside fermentable vegetable fiber produces lower ammonia and higher butyrate. The protein-to-fiber (P/F) ratio appears to be an important metric: each unit decrease in the P/F ratio is associated with a 3% increase in microbial gene richness, a composite marker of microbiome functional capacity [47]. High P/F ratios promote proteolytic communities dominated by Bacteroides spp. and Clostridium cluster IX, while lower ratios favor saccharolytic communities enriched in Roseburia spp., Coprococcus eutactus, and Parabacteroides. These findings suggest that protein and fiber are microbially influential and that their ratio may confer even more value.

Dietary Fat and Microbial Metabolism

Dietary fat can also influence microbial membership and metabolism and vice versa through multiple mechanisms. In a recent study using mice, the spore-forming bacteria Turicibacter was found to be metabolically protective when gavaged regularly throughout a high-fat diet, combating weight gain and metabolic dysregulation [48]. Remarkably, Turicibacter treated obesity and metabolic dysfunction by producing a competing lipid of its own. Regular gavage of Turicibacter was critical for metabolic health, as high-fat diet itself was toxic to the bacteria. In the same study, investigators found Turicibacter was notably decreased in humans with metabolic dysfunction such as obesity.

Microbial metabolism is also relevant in bile acid production. A 6-month randomized controlled feeding trial in humans showed that higher-fat diet (40% of energy intake) increased total bile acids, unconjugated bile acids, and secondary bile acids (like deoxycholic acid and taurodeoxycholic acid) compared to a lower-fat diet (20% of energy intake) [49]. These changes correlated with increased abundance of stool bile salt hydrolase (BSH)-expressing bacteria, including Bacteroides, Clostridium, Bifidobacterium, and Lactobacillus. Secondary bile acids activate the receptors FXR and TGR5, which regulate glucose metabolism, lipid homeostasis, energy expenditure, and inflammatory tone. Dietary fiber modulates this axis by directly binding and sequestering bile acids, altering BSH-expressing communities, and reducing intestinal transit time, thereby limiting the window available for secondary bile acid transformation. There is still much to be discovered regarding composition and individual bile acids and their significance in health and disease states.

Potential Relevance for EoE

Taken together, these diet-microbiome metabolic relationships support several principles potentially relevant to EoE. We note that microbial metabolic outputs are shaped more by complex nutrient patterns and ratios than by individual nutrients or taxa in isolation. The P/F ratio influences whether saccharolytic or proteolytic fermentation predominates, and particular lipid consumption can sculpt the microbiota. When EoE patients are prescribed food elimination diets, patients may turn to eating diets high in animal protein or increase consumption of plants. These unmeasured changes very likely change the dietary substrate landscape that shapes and feeds the gut microbiome, with downstream consequences for microbial metabolite production and mucosal signaling, outside the mechanism of direct food antigenicity. These ideas are further explored below.

Rethinking Elimination Diets Through a Microbiome Lens

Elimination diets are a therapeutic mainstay for patients with various food-related reactions, including EoE. We see differences in efficacy of elimination diets in EoE vary based on breadth of elimination, age groups included, and study design (e.g. single vs. multi-site) among other factors. The strictest elimination diet is an elemental diet removing all macromolecules. In a study of elemental diet over 4 weeks, investigators found histologic remission (defined as eosinophils < 10) in 72% of adult EoE patients, suggesting despite removing all dietary macromolecules about 30% of patients remained active [11]. The success of six- and four-food elimination diets has varied from ~ 40–70+% in recent years, and less restrictive diets like one food (dairy) elimination can be even 50% effective in pediatric populations [7, 50, 51]. It is suspected that elimination diets influence the microbiome throughout the gastrointestinal tract, including the esophagus. However, the significance of these changes in EoE is a process not well understood.

The Intended and Unintended Consequences of Elimination Diets

While we have little data on the effects of diet on the esophageal microbiome, one prospective observational study by Nobel et al. found that differences in dietary fiber intake are associated with higher abundances of esophageal Firmicutes (now Bacillota) compared to gram negative bacteria like Proteobacteria (now known as Pseudomonadota) or Bacteroidites (now Bacteroidota) [52]. This study assessed dietary composition via food frequency questionnaire of the previous 4 weeks just prior to a standard of care endoscopy, thus all patients had indications such as gastroesophageal reflux disease or Barrett’s esophagus for endoscopy. Another study utilized an in vitro model of human esophageal epithelial cells and found specific products of microbial fermentation of fiber, SCFAs butyrate and propionate, improved esophageal barrier integrity induced by IL-13 [53].

At the patient level, the field has yet to collect relevant longitudinal data and samples from EoE patients on diet elimination to probe which foods are chosen to be consumed. While elimination diets are designed to remove proposed antigenic triggers, they simultaneously have the potential to reshape the nutritional landscape that in turn feeds the gut microbiome. When removing dairy and/or wheat products, other nutrients like fermentable fibers, resistant starches, and phytochemicals, which are critical substrates for microbial metabolism, may also change. Most dietary counseling in diet elimination focuses on avoidance rather than substitution. As a result, patients may diverge substantially in how they replace eliminated foods. It is unknown at this time if this divergence is significant to diet elimination success or not. For example, some individuals on 2FED may increase intake of whole fruits, vegetables, legumes inadvertently enriching microbial diversity and short-chain fatty acid production. Others may rely more heavily on ultra-processed, allergen-free products that are low in fiber and high in additives, with potential downstream consequences including reduced microbial diversity and altered metabolic output. These divergent dietary patterns highlight that elimination diets are not a uniform intervention, but rather a spectrum of nutritional exposures with variable microbiome effects.

This variability underscores a critical gap and opportunity in current dietary strategies: the absence of a microbiome-informed framework for guiding food replacement during elimination. Rather than focusing solely on the removal of triggers, there is growing rationale to emphasize what should be maintained or enhanced to support microbial resilience. Dietary fiber, plant diversity, and minimally processed foods are key drivers of microbial composition and function, including the production of short-chain fatty acids that influence epithelial barrier integrity and immune signaling. Whether these microbial metabolite factors can have an impact in EoE is yet to be determined. Investigation is needed to better understand these complex host-diet-microbiota relationships to eventually arrive at a precision nutrition approach that integrates both immunologic triggers and microbiome-supportive dietary components. This may help explain why a subset of patients fail even the most restrictive diets and offers a path toward improving therapeutic efficacy. In this context, elimination diets could be reimagined not simply as subtractive interventions, but as opportunities to intentionally shape the microbiome while managing disease.

Conclusion: A Roadmap for Diet-Microbiome Research in EoE

In summary, the role of the diet-microbiome-immune axis in EoE remains incompletely understood, and several lines of evidence suggest this axis may be important and underappreciated. Diet elimination therapies provide strong evidence for a component of food antigens driving pathology, however, their imperfect efficacy even under elemental diet conditions, suggests that factors beyond antigen exposure may be relevant. Importantly, the diet is one of the most powerful modulators of microbial composition and function, capable of shaping metabolic outputs that influence the epithelium, immune cell education and signaling. These observations and data begin to support a model in which diet-mediated microbial changes may contribute to both therapeutic response and/or treatment failure in EoE.

While current microbiome work in EoE has taken important steps toward defining the composition and potential differences in the esophageal microbiome, steps to move the field forward should involve longitudinal, diet-controlled studies that capture not only which foods are eliminated, but what replaces them, paired with multi-omic approaches with a microbial metabolic mindset. Interventional studies designed to modulate microbial function through dietary composition, fiber type and diversity, or targeted microbial therapies will be essential to establish more powerful understandings.

Shifting our perspective to reframe current thinking about elimination diets through a microbiome lens offers an opportunity to move beyond purely subtractive approaches toward a more precise, additive strategy that supports both immune and microbial health. In this context, future therapeutic paradigms may incorporate not only trigger avoidance, but also intentional modulation of the microbiome through diet, engineered probiotics, or other targeted approaches. Advancing this integrated perspective has the potential to improve treatment efficacy, reduce dietary burden, and ultimately refine how we approach nutrition-based therapy in EoE.

Data Availability

No datasets were generated or analysed during the current study.