Eosinophilic Esophagitis without Eosinophils: Do You Want to Mock Me? Open Access

Eosinophilic esophagitis (EoE) is a clinico-histological diagnosis that is based clinically on the presence of symptoms of esophageal dysfunction (mainly dysphagia) and histologically on the presence of an eosinophil-predominant infiltration of the esophageal epithelium [1, 2]. Having long been considered a rare disease since its first description in the early 1990s, EoE is now known to be the most common cause for solid-food dysphagia in young adults [3, 4]. Incidence and prevalence have been steadily increasing in both the USA and Europe [5, 6]. It is currently estimated that one out of 2,000 suffers from the disease, although this might only be the tip of the iceberg. Many patients still remain undiagnosed or are misdiagnosed as having gastroesophageal reflux disease (GERD). A substantial proportion of EoE patients presents with a food bolus impaction necessitating endoscopic removal, but follow-up of these patients (including esophageal biopsy sampling) remains insufficient [7]. However, such follow-up is key for establishing the diagnosis and initiating the relatively simple, yet highly efficacious treatment.

When symptoms of esophageal dysfunction are present, the diagnosis of EoE can be established by the identification of an eosinophil-predominant infiltration of the esophageal epithelium, which is defined by a threshold of at least 15 eosinophils per high-power field (hpf) [1, 2]. As EoE can be a patchy disease, esophageal biopsies need to be taken from the distal and the proximal part, with 5 biopsies having a diagnostic sensitivity of almost 100% [8]. It is important to rule out other diseases potentially being associated with esophageal eosinophilia, particularly GERD, but also inflammatory bowel disease (IBD, such as Crohn’s disease), non-EoE eosinophilic gastrointestinal diseases (EGIDs) that affect the gastrointestinal tract beyond the esophagus, or connective tissue disorders [1, 2]. If there are no clinical or endoscopic signs for IBD, EGIDs, or connective tissue diseases, the most important differential diagnosis is GERD. The recent Lyon 2.0 consensus therefore states the GERD requires conclusive evidence of reflux-related pathology on endoscopy and/or abnormal reflux monitoring, in the presence of compatible troublesome symptoms [9]. In clinical practice, GERD is usually excluded by the following [10]: clinically by lack of typical symptoms (acid reflux, heartburn) that respond to proton-pump inhibitor (PPI) treatment, endoscopically by absence of signs of reflux esophagitis, and/or by negative pH monitoring. However, PPI response per se does not indicate reflux disease as a subset of EoE patients respond clinically and histologically to a PPI trial [11]. Therefore, the entity of PPI-responsive EoE has been removed from current guidelines with PPI now being accepted as a first-line treatment option for EoE [12, 13]. Still, the potential overlap of GERD and EoE, as both are common diseases, has to be considered.

With the last 2 decades of intensive research in the field, data on disease presentation of EoE have become more granular, from a clinical, endoscopic, and histological perspective. Despite its most frequent clinical presentation with dysphagia, several non-dysphagia symptoms have been appreciated, among which are odynophagia, thoracic pain not related to swallowing, heartburn, abdominal pain, nausea vomiting, and failure to thrive (particularly in young children) [14]. Recently, a novel syndrome associated with EoE has been described that is characterized by the acute onset of a burning sensation within the esophagus, which is precipitated by the intake of different food allergens (so-called food-induced immediate response of the esophagus) [15]. Endoscopically, several endoscopic changes have been identified as being specific for EoE, which has led to the implementation of the EoE endoscopic reference score (EREFS), that is now widely used both in clinical practice and for research purposes [16]. The EREFS comprises the following endoscopic features: edema, rings, white exudates, furrows, strictures, and eventually crepe paper-like mucosa [16]. Finally, from a histological perspective, several changes of the mucosa that go beyond simple eosinophilic infiltration have been identified and have led to a novel histological scoring system, the EoE-HSS [17]. These changes include basal zone hyperplasia, dilated intercellular spaces (spongiosis), surface epithelial alteration, dyskeratotic epithelial cells, and lamina propria fibrosis [17].

The pathogenesis of EoE is at best partially understood [18]. As of yet, EoE is considered a chronic immune-mediated food-allergen-induced inflammatory disease of the esophagus [19]. A disrupted epithelial barrier enables crossing of food allergens, which then initiate and perpetuate an inflammatory process, mainly mediated through T cells and basophils, which recruit mast cells and eosinophils to the site of inflammation [19‒21]. Fibrosis then happens through activation of tissue-resident myofibroblasts [19, 21]. Key cytokines involved in this cascade are the Th2 markers IL4, IL5, and IL13 but also IL-33 and eotaxin-3 [18]. Chronic inflammation itself results in epithelial leakage enabling further exposure with food allergens [22]. Lately, the esophageal microbiome has also been considered as a potential contributor to the inflammatory cascade [23, 24].

In contrast to the complexity of disease presentation and pathogenesis, current therapeutic options appear simple and consist of the 3 Ds: drugs, diet, and dilation [25]. Drugs and diet both tackle underlying inflammation, while endoscopic dilation is an add-on treatment for severe fibrostenotic disease. As EoE is a food-allergen-mediated disease, elimination diet is a very effective treatment option, although the long propagated 6-food elimination diet (elimination of dairy products, soy, eggs, nuts, wheat/gluten, and fish/seafood) can be cumbersome [26, 27]. However, newer studies suggest that a step up 2-4-6 food elimination strategy or even a single milk elimination diet might be almost equally effective [28, 29]. With regard to drugs, PPI and swallowed topical corticosteroids are considered first-line treatments, although data are more rigorous for topical steroids [30]. These are highly effective in both short and long term, with only few side effects (mainly esophageal candida infection) [31, 32]. More recently, the approval of dupilumab, an anti IL4/IL13 antibody, has heralded the era of biologics in EoE [33].

Eosinophils and their role in EoE have been studied for years. In normal conditions, the esophageal mucosa does not contain any eosinophils. Therefore, their existence itself suggests a pathogenic role. Eosinophils are late-phase effector cells that release toxic granule proteins, which cause damage to the adjacent tissue [34]. Eosinophil-released mediators are involved in inflammation, immunoregulation, tissue remodeling, and repair [34]. Among the most important mediators are IL4, IL5, IL9, IL13, the cytotoxic EPX. Eosinophils are recruited to the esophageal epithelium by inflammatory signaling molecules that trigger a Th2 inflammatory response (e.g., IL13) resulting in the release of eosinophil chemotactic factors (e.g., eotaxin 3) [35]. Eosinophils have been associated with epithelial barrier maintenance, repair, and immune tolerance – all potentially beneficial effects [34]. However, in EoE, eosinophils exhibit pathological activities promoting sustained inflammation by production of eosinophil survival factors (e.g., GM-CSF and IL5) [36]. By secreting IL9, eosinophils recruit mast cells to the site of inflammation [37]. Eosinophil granule proteins can damage the epithelium perpetuating the inflammatory process [34]. Eosinophil-derived IL13 together with MBP causes extensive tissue remodeling resulting in fibrosis [34]. Another way how eosinophils cause damage in EoE is by the formation of so-called eosinophil extracellular traps, a complex of DNDA fibers and cytotoxic granule proteins [38, 39].

In recent years, several non-eosinophil-related pathomechanisms have been identified, among which is the involvement of different inflammatory cell types such as T cells, B cells, and mast cells. T cells are increasingly recognized to contribute to EoE, particularly at its early stages. EoE patients have higher numbers of T cells in esophageal biopsies when compared to controls [40]. Single-cell RNA sequencing revealed that indeed only subpopulations of T cells are upregulated, such as putative T-regulatory cells and effector Th2-like cells [41]. B cells are also increased in active disease [42, 43], probably associated with the involvement of a pro-angiogenic phenotype [42]. Increased angiogenesis is known to contribute to tissue remodeling (and fibrosis) in EoE [44]. Mast cells – tissue-resident immune cells – are increased in EoE, although their exact role is not clear [45, 46]. Nevertheless, mast cells counts respond to treatment and correlate with disease recurrence [45]. Activated mast cells might contribute to ongoing disease activity in otherwise remission status [47‒49] as it is known from other disorders where mast cells can contribute to gastrointestinal pain [50].

As data on EoE have become more granular, the role of eosinophils is increasingly questioned. They might be particularly implicated in the formation of subepithelial fibrosis, an important (but late) step in EoE. In early stages of the disease, however, other cells appear to contribute to EoE pathophysiology, mainly T cells and basophils. Several findings underline the inferior role of eosinophils in EoE. First, it has long been known that symptom severity does at best only moderately correlate with esophageal eosinophilia [51]. Symptom scores were able to detect histological remission with an area under the curve of 0.6–0.68 [51]. In other words, a subset of patients remains symptomatic despite achieving histological remission. In the long-term follow-up, this proportion might increase up to 65% [52]. The mismatch between symptoms and eosinophilic infiltration has also been shown in several prospective studies evaluating anti-eosinophil medications. Despite their sometimes impressive potential to lower or even deplete eosinophil infiltration, no effects on clinical activity were seen. Several studies failed with regard to the clinical activity outcome [53‒55]: the anti-IL5 antibodies mepolizumab and reslizumab showed promising results when looking at eosinophil reduction but did not result in significantly improved clinical disease activity [53‒56]. The latest setback was the failure of the phase III randomized placebo-controlled MESSINA study evaluating the anti-IL5Ra antibody benralizumab. Finally, novel forms have been described where patients present with typical EoE symptoms but absent or only low-grade esophageal eosinophilia [57, 58]. In these patients, EoE typical histological changes beyond eosinophilia (such as basal zone hyperplasia or spongiosis) can be found along with an impaired epithelial barrier function [58]. These findings all suggest that factors other than pure eosinophil infiltration contribute to symptoms and disease activity per se in EoE.

Given these uncertainties with regard to the role of eosinophils in EoE, the research focus has shifted away from esophageal eosinophilia toward other aspects of the disease. This is mirrored in the recently released summary from an EoE expert meeting discussing endpoints in EoE studies [59]. Here, the use of peak eosinophil counts as histological readout in clinical practice and research studies is questioned given the abovementioned findings [59]. In fact, the use of peak eosinophil counts is discouraged to determine efficacy of therapeutic interventions given the fact that they reflect only 1 cell type in the myriad of inflammatory cells present in EoE that include lymphocytes (ILC2, Th2), antigen presenting cells, basophils, mast cells, and fibroblasts [59]. Due to these drawbacks, Collins and colleagues developed the abovementioned EoE-HSS, which appears to outperform peak eosinophil counts and better explains failure of eosinophil-depleting, targeted therapeutics [17].

Despite the consensus regarding these uncertainties, the optimal histologic outcome in EoE remains to be discussed in the future. Nonetheless, it has become increasingly clear that EoE is much more than just eosinophil infiltration.

Straumann and colleagues first coined the term EoE-like disease using this description for patients presenting like having EoE but not fulfilling the diagnostic criteria with regard to an esophageal eosinophilia of at least 15 eosinophils per hpf [57]. To avoid any gray zone, they first reported on patients without any eosinophils in the esophageal epithelium. Five patients from four EoE families were rigorously assessed from a clinical, endoscopic, histological, and molecular perspective [57]. Of note, offspring of these EoE-like patients were diagnosed with EoE in 33% suggesting a strong and common inheritance of EoE and EoE-like disease [57]. Clinically, EoE-like disease presented similarly to conventional EoE with dysphagia being the leading symptom (1 patient even required endoscopic food bolus removal) [57]. In contrast, EoE-like disease patients did not show typical endoscopic features of EoE captured by the EREFS [57]. Functional abnormalities were excluded by high-resolution manometry and 24-h pH testing [57]. Despite the absence of eosinophil infiltration, some considerable overlaps with EoE were found [57]. Most EoE-like disease patients showed increased T-cell and mast-cells infiltration [57]. However, in contrast to EoE, these T cells were CRTH2 negative, and mast-cell counts were considerably lower than those seen in EoE [57]. Cytokine expression was lower in EoE-like disease compared to EoE (TSLP, TNFa, and eotaxin-3) [57]. These findings were all pointing toward a possible defect in the eosinophil recruitment pathway in these patients resulting in a non-eosinophilic inflammatory pattern. Strikingly – despite these differences – RNA analyses revealed considerable overlaps with regard to gene expression between EoE and EoE-like disease, particularly in EoE typical genes, which are known to be dysregulated in IL13-stimulated esophageal epithelial cells [57]. Based on these findings – although small in population size – it was speculated that EoE-like disease and EoE share a uniform underlying pathogenesis and that the two entities are phenotypes or nuances of a broader inflammatory dysphagia syndrome, similarly to other spectrum disorders [57].

The concept of EoE-like disease has been shaped in more detail in the last few years. An international multicenter consortium was formed to study a cohort of EoE-like disease patients [58]. To further account for the complexity of chronic inflammatory processes in the esophagus, to avoid confusion, and to highlight potential nuances, the term EoE-like disease was replaced by the term EoE variants. EoE variants were further subtyped into EoE-like esophagitis, non-specific esophagitis, and lymphocytic esophagitis. The following definitions shown in Table 1 were used.

Patients were excluded if they suffered from any other disease potentially being associated with eosinophilic infiltration of the esophageal epithelium such as GERD, non-EoE EGIDs, or IBD, if they had undergone esophageal surgery, had congenital disorders of the esophagus (e.g., esophageal atresia), or suffered from other non-IgE-mediated diseases including lichen planus, celiac disease, connective tissue disorders, or drug hypersensitivity [58]. Most importantly, all biopsies were re-examined by two reference pathologists to avoid a missed diagnosis of EoE [58] Furthermore, all patients were off any anti-eosinophil treatment at the time of diagnosis (including anti-eosinophil treatment for other indications such as asthma or nasal polyps), and given the potential patchy nature of EoE, at least 6 biopsies had to be available for re-analysis [58].

The cohort has first been described in a cross-sectional analysis including 69 patients (36 EoE-like esophagitis, 14 lymphocytic esophagitis, 19 nonspecific esophagitis) [58]. Patients with an EoE variant had clinically and histologically active disease, despite the absence of a significant eosinophil infiltration [58]. In addition, in a considerable proportion (>50%) of patients, endoscopic EoE features (captured by the EREFS) were found questioning the specificity of this endoscopic scoring system and its availability to discriminate between EoE and EoE variants. Except for lymphocytic esophagitis, no inflammatory cell infiltrates were found, but considerable structural changes were detected based on histology and protein expression [58]. All patients showed signs of an epithelial barrier dysfunction. Bulk RNA sequencing revealed distinct molecular fingerprints of each variant disease with – however – some considerable overlaps with conventional EoE [58]. Of note, clustering analyses confirmed the presence of three different clusters, corresponding to an EoE-like, a lymphocytic, and a nonspecific variant cluster [58]. The cohort was recently described in a second analysis including long-term follow-up data (for 54 of the 69 patients) [61]. Patients with EoE variants showed high response rates to topical steroids (particularly patients diagnosed with EoE-like esophagitis and nonspecific esophagitis) [58]. Intriguingly, transition from one variant to another variant was seen in 35.2% [61]. In 8 patients, progression to EoE was observed after a median of 14 months [61]. Sequential RNA sequencing revealed that only few genes are upregulated during this progression, among which were TSG6 and ALOX15 [61]. Of note, this progression was associated with a switch from a Th1 gene signature to a Th2 gene signature, which was confirmed by an increased number of GATA3-positive cells [61].

These analyses from a unique cohort of patients underscore the presence of EoE variants, the potential flux between different variants (including lymphocytic esophagitis), and the potential progression to EoE over time. All these findings point toward the presence of a spectrum disorder, where EoE represents only the most extreme phenotype (Fig. 1).

Lymphocytic esophagitis has first been described by Rubio and colleagues in 2006, as an entity distinct from EoE [62, 63]. It has been defined exclusively based on histological criteria, namely, the presence of a lymphocyte-predominant inflammation with at least 30 lymphocytes per hpf, gathered mainly in the peripapillary region, together with peripapillary spongiosis, and in the absence of intraepithelial granulocytes [60]. It still remains an enigmatic disorder with only few studies analyzing its disease presentation and course. It might be more frequent in older women, thereby contrasting EoE’s epidemiology [58], while the clinical presentation is similar with dysphagia being the most common symptom, although odynophagia appears to be more frequent [64, 65]. Lymphocytic esophagitis has anecdotally been treated with PPI, steroids (topically and systemically), and repetitive dilations [66]. However, no treatment has been evaluated in a prospective fashion. Retrospective data suggest a response rate to topical steroids that is lower than that seen in EoE or other EoE variants [58].

Despite the first description in 2006 and emerging (although few) data about this disease, the abovementioned EoE variant cohort represents the most-detailed description of this entity using RNA sequencing data and comparing it with EoE, GERD, and healthy controls [58]. While lymphocytic esophagitis is clearly distinct from reflux disease and esophagus healthy individuals, considerable overlaps with other EoE variants and EoE can be seen with regard to upstream pathway regulators [58]. Intriguingly, axonal guidance signaling – the top affected pathway in lymphocytic esophagitis – was also among the top affected pathways in EoE, suggesting a strong overlap [58]. Further tertiary analyses revealed the involvement of Th2 activation, which is a key aspect of EoE [58].

With the description of EoE-like disease and more recently the three EoE variants EoE-like esophagitis, lymphocytic esophagitis, and nonspecific esophagitis, several questions arise. First, long-term outcome beyond 1–2 years remains unknown. It might be speculated that all patients eventually end up with an EoE diagnosis, at least those with the EoE variants EoE-like esophagitis and nonspecific esophagitis. Whether or not lymphocytic esophagitis can progress to EoE has not been answered yet, although the transition from lymphocytic esophagitis to the two other variants suggests such a flux. Some of the EoE variants could be burned-out EoE, where eosinophilic inflammation is not active anymore. However – while fibrosis pathways were found in the RNA sequencing analyses – all patients in the EoE variant cohort showed signs of inflammation (either histologically or molecularly); thus, a burned-out status is rather unlikely. The most obvious looming question is the one about pathogenesis and potential pathogenic overlaps with EoE. Whether or not EoE variants are also food-allergen mediated needs to be confirmed by elimination diet therapeutic approaches or antigen challenges of T cells. So far, none of the patients in the abovementioned international cohort have been treated with a dietary approach. Moreover, EoE variants need to be distinguished from other esophageal diseases such as GERD and motility disorders. RNA sequencing data already show a clearly distinct transcriptomic profile compared to GERD [58]. In addition, motility disorders have been rigorously excluded at least in the case series by Straumann et al. [57]. However, there are also emerging data that the overlap between eosinophilic diseases and motility disorders (such as achalasia) appears to be more considerable than previously thought. Finally, it can be assumed that the 69 patients from the international cohort are only a small tip of the iceberg and that the frequency of EoE variant is much higher. Thus, this rather rare phenomenon might become more common over time – as we have seen with EoE in the past 3 decades.

Lymphocytic esophagitis has been included as an EoE variant although it has been known for over a decade now. So, its position within this group of entities remains to be discussed. However, several findings support that lymphocytic esophagitis is indeed part of the EoE variants: (i) transition from lymphocytic esophagitis to the two other EoE variants; (ii) pathogenic overlaps with the two other variants as well as with EoE; (iii) response to topical steroids. Nevertheless, lymphocytic esophagitis patients appear to be considerably different from other EoE variant patients (older, more females), have a considerably inferior response rates to steroids, and so far, no progression to EoE has been described.

Nonspecific esophagitis needs in-depth characterization as its definition – for now – relies on the distinction from EoE-like esophagitis and lymphocytic esophagitis. Thus, it is a diagnosis of exclusion and not based on specific diagnostic criteria. However, preliminary RNA sequencing data suggest that this variant is much more specific than its name suggests and that it is clearly distinct from the other variants, from EoE, healthy controls, and GERD in particular.

To make it even more complex, Dellon and colleagues recently coined the term mast-cell esophagitis that has been described in 87 patients [67]. Mast-cell esophagitis was defined by an infiltration with mast cells of at least 15 mast cells per hpf (identified through tryptase staining). All patients had normal endoscopy and otherwise normal histology but presented with symptoms of esophageal dysfunction, thus somewhat similar to what was seen in the EoE variant cohort. In fact, in terms of age, atopic comorbidities, and female preponderance, these mast-cell esophagitis patients appear to be comparable to EoE variant patients [58]. However, in the EoE variant cohort, mast cells were not significantly increased [58]. Thus, mast-cell esophagitis appears to be a distinct entity. Its relation to the other EoE variants needs to be examined in future studies and molecular data are needed to describe this novel entity in more detail. Of note, no correlation between mast-cell counts and clinical features were found, which at least questions the pathogenic relevance of this cell type.

The pathogenic view of EoE has shifted from a single cell disease (eosinophils) toward a Th2-mediated disorder with a myriad of inflammatory processes being involved. In fact, the role of eosinophils has been questioned by various findings and there is increasing agreement among experts that eosinophils are not the contributing cell and that other histological readouts are needed to define histological response in clinical practice and clinical trials. There is growing evidence that EoE variants or nuances exist that resemble EoE but lack the typical eosinophil infiltration. Future studies, particularly mechanistic in nature, will help to characterize these variants in more detail and to answer the looming question whether or not an EoE spectrum disorder exists, where EoE is only the most extreme phenotype.

So, eosinophilic esophagitis without eosinophils, do you want to mock me? Not really. It is probably time to embrace the fading role of eosinophils in EoE. This is more fact than a fiction.

Thomas Greuter has consulting contracts with Sanofi-Regeneron, Janssen, BMS, Takeda, AbbVie, and Falk Pharma GmbH and received travel grants from Falk Pharma GmbH, AbbVie, Takeda, and Vifor, speaker’s fee from Norgine, and an unrestricted research grant from Novartis. Elena Gonzalez has no conflicts of interest.

This work was supported by grants from the Swiss National Science Foundation to T.G. (Grant No. P2ZHP3_168561). This work was further supported by the Swiss EoE Foundation, a young investigator award from the Swiss Society of Gastroenterology to T.G., a research grant from the Novartis Foundation for Medical-Biological Research to T.G., a research award from the Swiss IBDnet to T.G.

Elena González de Béthencourt: drafting manuscript. Thomas Greuter: drafting manuscript and critical review and responsible investigator.