Assessment of Disease Activity in Eosinophilic Esophagitis: Is It Clinically Relevant or Simply an Amusement for Experts?

Eosinophilic esophagitis (EoE) is a chronic, local, immune-mediated esophageal disease, characterized, clinically, by symptoms related to esophageal dysfunction and, histologically, by eosinophil-predominant inflammation [1]. First described by Stephen Attwood (USA) and Alex Straumann (Switzerland) in 1993 and 1994, respectively, EoE’s current incidence ranges from 5 to 10 cases per 100,000, and current prevalence estimates range from 0.5 to 1 case per 1,000 [2‒5]. Concomitant allergies to various food antigens or aero-allergens are found in the majority of EoE patients [1]. In affected children, food refusal, gastroesophageal reflux disease-like symptoms, vomiting, and abdominal pain are the leading symptoms, while adolescents and adults mostly present with dysphagia for solids or food impaction [1, 6]. Adult EoE patients frequently adapt to living with dysphagia by practicing various behaviors, such as food avoidance, food modification, or they take a longer time for ingestion of certain foods compared to persons without EoE eating that same food [7]. Histologic diagnosis is based on ≥15 esophageal eosinophils/high-power field (hpf) and exclusion of other conditions associated with esophageal eosinophilia [1]. In the majority of patients, untreated inflammation leads to formation of strictures with the inherent risk of food bolus impactions and associated esophageal perforations [8]. In fact, once anti-eosinophil treatments are stopped, histologic or clinical relapse rates exceed 90% over a 1-year period [9]. Treatment options range from pharmacological and dietary therapies to esophageal dilation [10‒12]. The optimal long-term therapeutic approach still needs to be defined.

Standardized assessment of EoE activity is important for several reasons. First, it allows to define strict inclusion and exclusion criteria within a clinical trial and, thereby, sets the ground for evaluating the efficacy of therapeutic interventions in appropriate populations of patients. Second, it allows the comparison of results among different studies. The conceptual model for EoE activity assessment is shown in Figure 1. Based on the guidance from regulatory authorities, a strict separation between patient-reported outcomes (PRO) like symptoms, quality of life, anxiety and hypervigilance, and clinician-reported outcomes like histology and endoscopy is maintained to avoid introduction of health care professional-inherent bias to patients’ reporting of symptoms and other concepts related to their disease and well-being [13]. The choice of endpoints and instruments to be used strongly depends on the clinical setting and can be stratified into frameworks of randomized clinical trials (RCTs), observational studies, and daily clinical practice [13]. It is evident that requirements for particular outcome tools to be used in RCTs are more strict than those when the tools are used in observational studies and daily clinical practice [13]. Table 1 provides an overview of instruments used to assess EoE activity in adults on the level of symptoms, histology and endoscopy in the framework of RCT, observational studies, and daily clinical practice. It is, among others, the advances in standardization of outcome assessment that paved the way for the regulatory approval of the first few therapies including budesonide orodispersible tablets (Jorveza®) in Europe, Canada, and Australia, dupilumab (Dupixent^®) in USA and Europe, and oral viscous budesonide (BOS, Eohilia^®) in the USA [14‒16].

Assessment of EoE activity was initially based on capturing symptoms of esophageal dysfunction and esophageal peak eosinophil counts using unvalidated, often developed ad hoc or borrowed from other conditions instruments, as validated instruments for assessment of EoE activity were not yet available. Between 2010 and 2020, several validated instruments were developed, notably the dysphagia symptom questionnaire (DSQ) and conceptually similar to that instrument the Dysphagia Symptom Diary (DSD) as well as the EoE activity index (EEsAI) PRO for symptom assessment, the Adult Eosinophilic Oesophagitis Quality of Life (EoE-QoL-A) instrument to assess EoE-specific quality of life, the EREFS (Exudates, Rings, Edema, Furrows, Stricture) grading and classification system to assess endoscopic inflammatory and fibrotic alterations, and the EoE histologic scoring system (EoEHSS) that takes into account the severity and extent of 8 distinct features observed in active EoE patients (Table 1) [7, 17‒22].

Description of these instruments formed the basis for several landmark papers published in the field of EoE outcome assessment in the last decade. The significance of these studies, as well as implications that these have for the field at large, are further discussed.

Symptom assessment in EoE can be carried out using multiple instruments, all of which vary not only with respect to the recall periods but also with respect to the language used to query dysphagia and in the way they capture behavioral adaptations for living with EoE. While the change in symptom scores is a most commonly used symptom-based endpoint, dysphagia-free days and remission definitions should also be assessed, as these are easier to interpret for both clinicians and patients.

The question of association between biologic findings and symptoms has been a subject of much research. In a well-characterized cohort of adults with EoE, Safroneeva et al. [23] showed that EoE-related symptoms correlated in only a modest way with histologic and endoscopic disease activity. One third of EoE patients in clinical remission (defined as EEsAI PRO <20) had ongoing histologic or endoscopic activity, which poses a risk factor for subsequent stricture formation and food bolus impactions. Not only did that study derive the first symptom-based remission definition, but it also showed that EoE patients should be followed not based on symptoms alone but rather undergo regular esophagogastroduodenoscopy even in the absence of symptoms. This allows to, among others, adjust treatment by modifying the dosage of anti-inflammatory drugs or esophageal caliber in case of strictures. Subsequent analysis of The Consortium of Eosinophilic Gastrointestinal Disease Researchers (CEGIR) OMEGA data as well as that of the fluticasone versus budesonide studies clearly showed that the association between biologic findings and symptoms is moderate in never dilated patients, but it’s lacking in patients that underwent esophageal dilation in the last 12 months [24]. That relationship takes several months to recover, although this likely depends on the dilation procedure that vastly varies from one provider to the other. With that being said, even in never dilated patients, it is likely that symptoms are not going to be accurate enough to predict histologic remission. Lastly, the history of dilation may impact the actual change in symptom response to treatment, as improvement in symptoms has been observed in never dilated but not in previously dilated patients. This has been observed in randomized controlled studies of oral viscous budesonide and etrasimod, but not dupilumab, all using the dysphagia symptom questionnaire, and it is likely that variability in dilation practices may partly be responsible for these differences [25, 26]. Therefore, symptoms are not accurate in predicting histologic remission, especially in dilated patients. Dilation leads to a dissociation between biologic findings and symptoms and may impact symptom response to treatment. The effects of dilation may last a long time depending on the dilation practices, which likely vary. These are important consideration for studies and clinical practice alike.

The global assessment of EoE-associated endoscopic findings varies greatly among endoscopists [20]. Therefore, Hirano et al. [19] revolutionized the field of assessment of EoE endoscopic findings by validating the EREFS classification and grading system. The change in EREFS score has been used extensively as endpoint in randomized controlled trials of many therapies. Although in the randomized controlled clinical trials the EREFS is calculated separately for proximal and distal esophagus, it has been shown convincingly that a simple EREFS scoring from 0 to 8 already explains 90% variation in endoscopist global assessment and was most responsive to treatment, with more complex scoring or scoring by segment providing only modest gains, if any [20]. This observation was confirmed by Ma et al. [27], who showed that the EREFS and its modifications were not more reliably scored by segment and also not more responsive when proximal and distal EREFS were summed. Therefore, scoring endoscopy simply on the worst affected area is likely the best with respect to both reliability and responsiveness; this would not only facilitate scoring of the EREFS but would harmonize its assessment across various studies and clinical practice.

Cotton et al. [28] developed an EREFS-based remission definition of ≤2; this definition best reflected histologic and symptomatic measures as well as posttreatment gene score [29]. Although developed for EREFS scoring from 0 to 9, it is likely will hold true for separate scoring of proximal and distal findings, as inflammation mostly impacts either proximal and distal esophagus together or distal esophagus alone [30]. In EoE studies, symptom- and endoscopy-based remission endpoints are underused when compared to those in inflammatory bowel disease. Given a relatively long diagnostic delay typically associated with EoE, it is only reasonable to expect that certain level of findings rather than their complete absence should be accepted as EoE remission. There are now enough data to support the use of EREFS-based remission definition along with histologic remission definition of <15 eos/hpf in studies and clinical practice [28, 29].

Like in inflammatory bowel disease, making endoscopic findings or endoscopy-based remission definition a part of the primary endpoint may become important for several reasons: (1) endoscopy better represents the pan-esophageal inflammatory and fibrotic alterations than histology that provides a peak eosinophil count that only captures one small fraction of the esophageal surface (less than 0.0001% of the esophageal surface); (2) using standard biopsy forceps, subepithelial alterations, such as fibrosis, are captured in only half of the patients, thereby limiting the contribution of histology to evaluate esophageal remodeling [19, 20, 31]. With that being said, stricture assessment has been shown to be least reliable and contributing most to variation in endoscopist global assessment, highlighting the need to better standardize the stricture assessment in EoE.

In 2017, Collins et al. [21] published data on development and validation of the EoEHSS and, in so doing, provided the pivotal evidence for assessing pathologic features beyond esophageal eosinophilia. In a later paper, Warners et al. [32] formally evaluated the reliability of this instrument. Although the authors found the instrument to be reliable overall, inter-rater reliability of features like dyskeratotic epithelial cells was near 0 both for both grade and stage indicating that RCT data should be evaluated to assess the performance of this feature. Nevertheless, until now only esophageal eosinophil count was assessed as part of primary endpoint in trials, with EoEHSS, like EREFS, being used as a secondary endpoint.

However, the data of two recently published randomized controlled trials questioned the use of esophageal eosinophil count without other biologic features of EoE as primary endpoint, a consistent feature of the EoE trials conducted over the last three decades. The KRYPTOS trial, results of which are not yet published in a peer-reviewed journal, evaluated the histologic and clinical efficacy of lirentelimab, a monoclonal antibody against Siglec-8, which is present on eosinophils and mast cells, in adults with histologically and clinically active EoE [33, 34]. Group 1 (n = 93) was treated with a monthly injection of 1 mg/kg of lirentelimab for 6 months, group 2 (n = 91) received a monthly injection of 3 mg/kg of lirentelimab, and group 3 (n = 92) was treated with placebo. After 24 weeks, a histologic remission (≤6 eosinophils/hpf) was achieved in 92.5% (group 1), 87.9% (group 2), and 10.9% (placebo), respectively. However, change in DSQ was not different in lirentelimab-treated patients when compared to placebo [34]. The double-blind placebo-controlled MESSINA trial evaluated the efficacy and safety of benralizumab, a monoclonal antibody against the IL-5 receptor, in adolescent and adult patients with active EoE. The proportion of patients with a histologic remission, defined as a peak eosinophil count ≤6/hpf, and mean changes from baseline in DSQ was evaluated at week 24 posttreatment. The endpoint of histologic remission at week 24 was achieved, but like in the KRYPTOS trial, there was no difference in DSQ change when compared to placebo [35]. Not only these trials showed that selective depletion of eosinophils and mast cells is not sufficient to provide symptomatic improvement pointing to the importance of other T2 cell types, but they also demonstrated that the eosinophil count alone is likely not an appropriate endpoint in randomized controlled clinical trials in EoE. It seems that drugs that act upstream, such as budesonide orodispersible tablets or dupilumab, are able to downregulate the active T2-cell compartment, which seems associated with symptom improvement. In summary, stakeholders recognized that the endpoint focused on symptoms and peak eosinophil counts alone represents an oversimplification of global EoE activity in randomized controlled trials.

The international COREOS group recently published a consensus on the set of outcomes to be used for the purpose or randomized controlled trials and observational studies [36]. The COS consists of four outcome domains for controlled and observational studies (histopathology, endoscopy, patient-reported symptoms, and EoE-specific quality of life). A total of 69 stakeholders (response rate 95.8%) prioritized 42 outcomes in a 2-round Delphi process, and the final ratification meeting generated consensus on 33 outcome measures. These included measurement of the peak eosinophil count, EoEHSS, EREFS, and patient-reported measures of dysphagia, and quality of life. This COS can be applied to adult and pediatric studies of pharmacologic and dietary therapies for EoE and will facilitate meaningful treatment comparisons and improve the quality of data synthesis. At the time when COREOS was conducted, the evidence for using some of the outcomes was lacking, and the recommendations were based on vast clinical experience of the team. The new data suggest that the simple EREFS scoring at the worst affected site and EREFS-based remission definition should consistently be used in randomized controlled and observational studies alike to continue the efforts to harmonize outcome assessment in EoE. Novel endpoints that reflect overall function of the esophagus should be examined for the purposes of the next COS exercise in EoE.

Given that treatment success in EoE should not be assessed by relying on a combination of symptom scores and peak eosinophil count alone, several new outcome measures and endpoints that better reflect overall function of the esophagus and are more easily interpreted were proposed during ASCENT (Assessment of Clinical Endpoints in Eosinophilic Esophagitis for Novel Therapeutics) meeting at the Digestive Diseases Week congress in Chicago in May 2023 [37]. While discussing all the endpoints and outcomes falls outside the scope of this review, several of these are described below.

Instead of change in symptom score, counting the number of dysphagia days over a certain period (e.g., 2 weeks as proposed in the phase III randomized controlled trial evaluating cendakimab for patients with active EoE) has the advantage of being more readily interpretable [38]. However, most symptom-based PRO instruments, like DSQ and DSD, do not include the entire spectrum of language used by patients to describe dysphagia. This means that depending on instrument, achieving an endpoint of certain number of dysphagia-free days may be harder with instrument like PROSE that incorporates the broader spectrum of dysphagia language than with instrument like DSQ that includes a narrower dysphagia definition. In addition, the number of dysphagia days does not account for the severity of dysphagia episodes and behavioral adaptations, like food avoidance, modification, or slow eating. Results of the cendakimab trial, in which dysphagia days was one of the primary endpoints, are eagerly awaited, as these will help inform our understanding of utility of dysphagia-free days as primary endpoint [38].

Taft et al. [39] recently published on the dysphagia stress test (DST) in adults with EoE, gastroesophageal reflux disease, achalasia, or with dysphagia causes of which were not specified. Control patients with non-esophageal diagnoses and healthy esophagus controls were also recruited. A total of 132 participants completed the DST with water, applesauce, rice, bread, barium tablet and rated their swallowing difficulty and pain. A study clinician observed and documented water use and refusal of any challenges. About 90% of patients were able to attempt or pass each of the bolus challenges, suggesting high acceptability of the DST. Applesauce, rice, and bread demonstrated collinearity; thus, the DST was reduced to three challenges. As an advantage, the DST takes into account adaptive behaviors, food avoidance, and modification. It can be regarded as a clinically relevant outcome measure. With respect to shortcomings, the DST needs further evaluation to support its use. Furthermore, there is a risk of food impaction during the test.

The Index Severity for Eosinophilic Esophagitis (I-SEE) is a multidimensional instrument aiming to assess clinical EoE activity by measuring symptom features, complications, and inflammatory and fibrostenotic features on both endoscopic and histologic examination. The I-SEE can be used in children and adults with EoE [40]. The I-SEE has shown responsiveness to therapy in children and adults with EoE [41, 42]. The I-SEE is straightforward to calculate and may develop as an appropriate instrument for clinical practice. However, I-SEE requires further validation.

Impedance planimetry (EndoFLIP) is regarded as an objective biomarker of esophageal remodeling consequences [43]. EndoFLIP allows to measure the esophageal caliper and wall distensibility. The measured distensibility plateau is associated with food impaction. Furthermore, responsiveness of impedance planimetry to therapy was demonstrated in small studies. As a limitation, the costs for the acquisition of the technical setup as well as the costs of the single-use balloons might be limiting factors to a widespread application of this technology. Furthermore, the current metrics are restricted to measurement of the distal esophagus. In addition, muscle layer contractions at the lower esophageal sphincter can serve as a confounder for strictures at the esophagogastric junction. In addition, the examination is operator-dependent [44].

Over the last years, the data from trials targeting eosinophils provided pivotal evidence that assessing EoE activity using a combination of symptoms and peak eosinophil count represents an oversimplification. Therefore, similar to inflammatory bowel disease, more weight should be given to the assessment of endoscopic activity using EREFS grading system, as it allows, in contrast to histology, the assessment of the pan-esophageal inflammatory and fibrotic alterations. Efforts should also be made to develop and validate the endpoints that better reflect overall function of the esophagus. Impedance planimetry is increasingly assessed as an endpoint to evaluate esophageal remodeling processes. Novel and innovative outcomes such as the dysphagia stress test (DST), number of dysphagia days, and the I-SEE are currently undergoing further evaluation. Continued concerted efforts among clinicians, scientists, industry, patient advocacy groups, and regulatory authorities are crucial to provide our EoE patients with adequate outcome measures that will facilitate the broadening of the still restricted therapeutic landscape.

A.M.S. has received consulting fees and/or speaker fees and/or research funding from Adare/Ellodi, AstraZeneca, Dr. Falk Pharma, Gossamer Bio, GSK, Celgene/Receptos/BMS, and Regeneron-Sanofi. E.S. has received consulting fees and/or speaker fees and/or research funding from Dr. Falk Pharma, GSK, Celgene/Receptos/BMS, and Regeneron-Sanofi. She is an employee of Tillotts Pharma AG.

This work was supported by a grant from the Swiss National Science Foundation (32003B_204751/1 to A.M.S.) and the Swiss EoE foundation (to A.M.S.).

A.M.S. and E.S. contributed equally to drafting of the article, literature search, writing, and final approval.