A conceptual approach to understanding treatment response in eosinophilic esophagitis

Abstract

While the diagnosis and initial treatment of eosinophilic esophagitis (EoE) are becoming more standardized, there are still major gaps in knowledge related to measuring treatment response. One such question centers on how to measure treatment response and what treatment end-points should be. This impacts not only patient care and engagement in decision-making, but also the field of drug development. In addition, studies so far have use a myriad of treatment end-points including over a dozen histology end-point criteria. This review will discuss the various stake-holders involved in assessment of treatment end-points of a complex condition, including patients, practitioners and regulatory agencies, and the myriad care settings in which treatment response is assessed, including routine clinical care, clinical trials, and observational studies. Potential parameters/treatment end-points such as histology, symptoms, patient-reported outcomes, endoscopy, biomarkers are discussed along with associated challenges and opportunities. A framework on how to define treatment outcomes is discussed, and a conceptual approach treatment response is proposed. This takes into account histology, symptoms and endoscopic findings and harnesses existing, validated tools. It includes definitions of non-response, complete normalization, and a graded response category between these two extremes, and also permits flexibility and latitude for modifications as newer knowledge emerges. In addition, ways to position the pediatric population in these endeavors are discussed as are future research directions.

Introduction

Eosinophilic esophagitis (EoE) is a chronic allergic disease characterized by a marked esophageal eosinophilic infiltrate and symptoms of esophageal dysfunction, in the absence of competing causes of eosinophilia.^1–3 The prevalence is currently estimated at ~57/100,000,⁴ with health care costs approaching $1 billion/year.⁵ The natural history of EoE is felt to be a progression from inflammation to fibrosis,^{6, 7} which is reflected by differences in symptoms and presentation across the age spectrum and by length of symptoms prior to diagnosis.^8–11 After diagnosis, treatment modalities can include proton pump inhibitors (PPIs), dietary elimination, swallowed or “topical” corticosteroids, and esophageal dilation when strictures are present.^12–17 With the quickly expanding knowledge of EoE pathogenesis,¹⁸ novel treatment modalities including biologics and small molecules are also under development.^{19, 20}

While the diagnosis and initial treatment of EoE are becoming more standardized, there are still major gaps in knowledge related to measuring treatment response. At its most basic level, it is a simple question – “How do we know when a patient with EoE is better?” – that is complicated to answer. First, there are multiple perspectives (patients, physicians, regulatory agencies) and different treatment settings (clinical and academic practices, observational studies, clinical trials) in which outcomes are assessed, and different stake holders are likely to value different outcomes.²¹ Second, similar to inflammatory bowel disease (IBD),²² EoE is a multifaceted disease with clinical, endoscopic, and histologic markers of disease activity, with a both biologic and patient-reported outcomes.²³ Moreover, there may only be modest correlation between these different outcomes,²⁴ the timeframe of response may vary between outcomes,²⁵ and there are few data linking most biologic outcomes to important long-term clinical benefits. Third, the use of patient-reported outcomes (PROs) is complicated by differences in symptoms in children compared to adults, symptom variability in children, the impact of esophageal dilation on symptoms of dysphagia, and a relative lack of rigorously validated disease-specific metrics. For all of these reasons, and because there is not yet a consensus on a core outcome set (COS) or reporting for EoE clinical trials, there is a substantial heterogeneity in reported outcome measures.²⁶ This heterogeneity has strong parallels to IBD, where a recent systematic review of randomized clinical trials (RCTs) in Crohn’s disease reported significant heterogeneity in definitions of end points.²⁷

The goal of this paper is to review the current understanding of treatment outcomes in EoE and present possible common approaches for considering treatment response in both clinical practice and research. We will discuss the range of potential treatment outcomes that could be selected in EoE, focus on challenges in outcome assessment, present a framework to classify potential outcomes, and review future areas for investigation.

Range of potential outcomes in EoE

Because EoE is a gastrointestinal disease with an allergic etiology, there are distinct clinical, symptomatic, endoscopic, histologic, immunologic, and molecular features to consider as potential treatment outcomes.²⁸ Accordingly, clinicians and researchers have used symptomatic, endoscopic, and histologic outcomes, as well as quality of life, complications (e.g. food bolus impaction, strictures, feeding dysfunction, and nutritional status), esophageal distensibility, and biomarkers (Table 1). Over the last half decade, particularly prompted by the need for measures to be used in clinical trials for drug development,^{29, 30} researchers have developed validated symptom, endoscopic, histologic, and quality of life measures.^31–38 While it is beyond the scope of the article to discuss all of these measures in detail, this topic has been recently reviewed.^39–42

Table 1 –

Range of EoE outcomes and metrics

Outcome	Measure	Comment
Histology	Peak eosinophil count	Most commonly used outcome
	Mean eosinophil count/eosinophil load	Tends to correlate with peak count
	EoE Histology Scoring System (EoEHSS)	Validated and responsive
Symptoms	Dysphagia Symptom Questionnaire (DSQ)	Validated and responsive; used for adults and adolescents
	EoE symptom Activity Index (EEsAI)	Validated and responsive; used for adults
	Pediatric EEsAI (PedsEEsAI)	Under development and validation; used for children
	Straumann Dysphagia Index (SDI)	Not validated, but responsive in some trials; used for adults
	Pediatric EoE Symptom Score (PEESS)	Validated; options for different pediatric age ranges; patient and caregiver report
	Mayo Dysphagia Questionnaire (MDQ)	Validated for non-EoE strictures; used for adults
	Visual analogue and Likert scales	Not validated for EoE; responsive; easy to use
Endoscopy	EoE Endoscopic Reference Score (EREFS)	Validated and responsive; can be used in all ages
	Esophageal compliance (FLIP)	Measure of esophageal remodeling; primarily still a research tool
Quality of life	Adult EoE Quality of Life (EoE-QOL-A)	Validated
	Pediatric Quality of Life EoE Module (PedsQL-EoE)	Validated
Complications	Esophageal food bolus impaction	--
	Stricture development or progression	--
	Feeding dysfunction	--
Growth/nutrition	Normalized height, weight, BMI Z scores	--
Biomarkers	EoE RNA expression/EoE diagnostic panel	Responsive
	Esophageal tissue	Multiple identified, clinical/research role as outcomes not defined
	Blood, stool, urine	Few potential markers identified; clinical/research role as outcomes not defined

Histology, and in particular the eosinophil count, is the most commonly assessed outcome of EoE treatment, and has been used in observational studies and randomized trials.¹² This is because assessment of an esophageal biopsy is objective and the disease was initially characterized histologically by an eosinophilic infiltrate. The vast majority of clinical trials and cohort studies have used the peak eosinophil count (eosinophils per high-power field [eos/hpf]) as the histologic outcome,^{17, 43–66} though some have assessed the mean eosinophil count, also termed the eosinophil load.^67–72 Limitations of the use of eosinophil count alone includes variability by microscope high-power field,^73–75 lack of assessment of other important histologic features (e.g. eosinophil degranulation, eosinophil microabscesses, basal zone hyperplasia, etc),⁷⁶ unreliable assessment of subepithelial features^{77, 78} as well as the multiple other inflammatory cells involved in EoE pathogenesis.^{18, 79} Because of this, a system that assesses 8 individual histopathologic features, the EoE-Histologic Scoring Symptom (EoE-HSS), was developed and validated.³⁵ This was initially assessed in children and shown to be more predictive of treatment status than eosinophil count alone,³⁵ and in adults has been confirmed to have external reliability⁸⁰ and be highly responsive to treatment.^{68, 71}

Symptoms have also frequently been assessed in EoE trials. In addition to overall symptoms (either patient-reported or investigator-elicited) and inventories of individual symptom, other specific metrics have been used. Validated measures include the Dysphagia Symptom Questionnaire (DSQ),^{31, 81} EoE Symptom Activity Index (EEsAI),³² and Pediatric EoE Symptom Score (PEESSv2);³³ a pediatric version of EEsAI is under development. Other instruments that have been used include the Straumann Dysphagia Index (SDI),^{57, 69} the Mayo Dysphagia Questionnaire (MDQ),^{52, 53, 67, 82, 83} a symptom scoring tool (SST),⁴⁵ a clinical symptom score (CSS),⁴⁶ a dysphagia symptom diary (DSD),⁷¹ and visual analogue scales (VAS) or Likert Scales.⁸⁴ There are also validated quality of life metrics for both adults (Adult EoE Quality of Life instrument [EoE-QoL-A])³⁶ and children (Pediatric Quality of Life EoE Module [PedsQL-EoE]),^{37, 38} but these have not been extensively used in trials. Outside of clinical trials, symptom evaluation in routine clinical practice is non-standardized, especially in children who may present with a variety of symptoms including “incidental” EoE where EoE is encountered in an otherwise asymptomatic child.

Endoscopic findings in EoE have been well described and are observed in the vast majority of EoE patients, thought the type and frequency vary by age.⁸⁵ A validated measure of endoscopic severity, the EoE Endoscopic Reference Score (EREFS) assesses 5 key features: edema, rings, exudates, furrows, and strictures.³⁴ This system has been externally validated,⁸⁶ is responsive in both children and adults,^{87, 88} has been used in multiple clinical trials to date,^{49, 50, 57, 68, 71} and several weighted scoring systems have been assessed.^{87, 89} A novel outcome measure related to endoscopic findings is the assessment of esophageal compliance (or distensibility) with a functional lumen imaging probe (FLIP).⁹⁰ This technique has shown that esophageal compliance is much decreased in EoE, that decreased compliance is associated with food impaction, and that successful treatment results in improved distensibility and increased esophageal luminal caliber.^91–93 FLIP has also been assessed as a secondary outcome in a clinical trial.⁹⁴ A recent study reported on use of FLIP in assessing treatment response in children with EoE,⁹⁵ building on data from adults with EoE.

Other clinically important outcomes that could be evaluated involve EoE disease complications, such as acute esophageal food bolus impaction requiring a visit to an emergency department and endoscopy,^{96, 97} and development or progression of an esophageal stricture or narrowing,^{7, 98} which may be better assessed on barium esophagram than endoscopically.^99–101 In children, feeding dysfunction, abnormal growth, and malnutrition are particular of concern.^102–104

The final set of potential outcomes is related to biomarkers. Multiple surrogate biomarkers of EoE disease activity have been examined over the years, and this topic has been reviewed recently.^{105, 106} Numerous studies have identified promising esophageal tissue biomarkers, including differential expression of cytokines, chemokines, mast cells, and other factors in esophageal mucosal biopsies from patients with EoE compared to controls.^107–111 Mucosal impedance, measured with a variety of techniques including stationary, probe- and balloon-based, is also promising and may distinguish EoE from controls and track with histologic disease activity.^112–116 Non-invasive markers have also been assessed in the blood, stool, and urine, though in general results have not been strong enough to be applied clinically.^117–122 Esophageal gene expression has also been assessed. The EoE Diagnostic Panel (EDP) consists of a 94 gene signature that is strongly associated with a diagnosis of EoE,¹²³ can be assessed on a single biopsy,¹²⁴ and is responsive to successful treatment.^{125, 126} In general, these biomarker outcomes are still preliminary, do not yet have a clear place in clinical practice, and are typically secondary or exploratory outcomes in trials.

With such a range of outcomes, and with no consensus as of yet as to which outcomes are preferred, it is no surprise that there has been substantial variability in the literature.^{26, 127} As can be seen in Table 2, different measures have been selected in almost every randomized trial conducted; there is even more variability in observational and retrospective studies. At least 11 different histologic response thresholds have been reported as primary outcomes, with similarly divergent symptoms metrics. There is somewhat more consistency on endoscopic severity, with recent studies all using EREFS, and older studies using either individual endoscopic finings or a score based on the individual findings. This heterogeneity makes it difficult to compare outcomes between trials and for different therapeutic agents. Consensus on outcomes is needed in the field, and efforts to define a COS, as has been previously done for IBD,^{27, 128} are underway for EoE.

Table 2 –

Heterogeneity in histologic, symptom, and endoscopic outcomes used in randomized clinical trials for EoE

	Author/year	Histology (primary reported)	Symptom	Endoscopy
Topical steroids
Fluticasone
	Konikoff, 2006	≤ 1 eos/hpf	Individual symptoms	Individual findings
	Alexander, 2012	>90% decrease in mean count from 5 hpfs	Single question from MDQ	Individual findings
	Butz, 2014	≤ 1 eos/hpf	Individual symptoms from PEESS	--
	Hirano, 2017*	Change in median eosinophil count	Patient global assessment of disease severity	EREFS
Budesonide
	Dohil, 2010	≤ 6 eos/hpf	Symptom scoring tool (SST)	Endoscopy scoring tool
	Straumann, 2010	Change in mean eosinophil load; <5 eos/hpf	SDI	Individual finding severity
	Gupta, 2015	≤ 6 eos/hpf	≥ 50% reduction in CSS	--
	Miehlke, 2016	≤ 16 eos/mm2	SDI	Endoscopic intensity score
	Dellon, 2017	≤ 6 eos/hpf	DSQ	EREFS
	Lucendo, 2017*	≤ 16 eos/mm2	Dysphagia <2 on a 10 pt scale; EEsAI <20	EREFS
Comparative
	Schafer, 2008	Histologic grading system	Individual symptoms	--
	Peterson, 2009	≤ 5 eos/hpf	Dysphagia scale	Individual findings
	Dellon, 2012	Change in peak eosinophil count; <15 eos/hpf	MDQ	Individual findings and global assessment of severity
	Moawad, 2013	<7 eos/hpf	MDQ	Individual findings
	Dellon, 2019*	Change in peak eosinophil cout;	DSQ; EEsAI	EREFS
		<15 eos/hpf
Diet elimination
SOFEED**	--	<15 eos/hpf	n/a	n/a
PCORI**	--	<15 eos/hpf	n/a	n/a
Biologics and other
Mepolizumab	Straumann, 2010	<5 eos/hpf	SDI	Endoscopic findings score
Mepolizumab	Assa’ad, 2011	<5 eos/hpf	Individual symptom severity	Individual findings
Reslizumab	Spergel, 2012	Change in peak eosinophil count	Physician’s global assessment score; Patient’s predominant symptom score	--
QAX576	Rothenberg, 2014	>75% decrease in peak eosinophil count	MDQ	--
RPC4046	Hirano, 2017	Change in mean eosinophil count	DSD; EEsAI	EREFS
Dupilumab	Hirano/Dellon, 2017	Change in peak eosinophil count	SDI; EEsAI	EREFS
Omalizumab	Clayton, 2014	Change in peak eosinophil count	Change in dysphagia score	--
OCT	Straumann, 2013	Change in mean eosinophil load	Visual dysphagia questionnaire + chest pain score	Global assessment of severity

^*abstract only

^**outcomes details from clinicaltrials.gov

Pitfalls and challenges in outcome assessment

Beyond the variability in endpoint selection, there are multiple other challenges encountered when considering treatment outcomes in EoE. One major issue is selecting outcomes for routine practice, as opposed to outcomes for use in clinical trials, particularly those conducted for registration purposes for new drug approval. In clinical practice, outcomes are not necessarily measured as formally as in a protocol-based clinical trial, can be individualized based on patient and provider preference, and treatment success may be able to be defined in more general terms, with patient-reported global improvement in symptoms and function, normalization of endoscopic appearance, and improvement in esophageal eosinophilia. However, because this approach may result in significant subjectivity and variation between providers, some formal outcome measures, in particular EREFS, should be used in standard practice, and certain symptom metrics could also be considered. In contrast, clinical trials need clearly defined outcomes with thresholds of response. Regulatory considerations, often dictate the exact outcomes and whether there can be composite outcomes (improvement in both symptoms and histology in the same subject) or co-primary outcomes (improvement individually in symptomatic and histologic parameters). Related to this, in clinical trials the outcome of interest is ideally linked to the mechanism of action of the treatment. For example, a therapy with anti-inflammatory activity should have a primary outcome related to an inflammatory measure. However, a therapy that, for example, might have a strong anti-fibrotic effect and is targeted to patients with esophageal strictures, an outcome related to luminal caliber, esophageal distensibility, or future requirement for dilation might be most appropriate.

A second issue is measurement of symptoms in EoE. Symptoms vary and evolve by age of presentation, likely related to the extent of esophageal fibrostenosis. In adolescents and adults, the primary symptom is dysphagia, and this has been the focus of recent clinical trials in this age group.^{47, 49, 50, 57, 69, 71} However, it has been repeatedly shown that symptom and biologic outcomes are not always concordant in EoE.^{23, 25, 46, 47, 52, 67, 86, 129, 130} In some studies, trial design issues might be the reason, with non-validated metrics, multi-symptom questionnaires, and eligibility criteria that do not include a severe symptom threshold for study entry.^{46, 52, 55, 56} Another reason is dietary and eating behavior modification. Patients with EoE often take small bites, chew thoroughly, eat slowly, drink copious fluids, and avoid food consistencies that stick; these changes may even be subconscious. Because of this a patient may not perceive a great deal of dysphagia as they are going to great lengths to minimize symptoms. These types of modifications may be missed with a dysphagia measure that asks only about trouble swallowing, but were specifically taking into consideration with the design of the EEsAI.³² Esophageal dilation is also a common reason for dissociation between symptoms and biologic activity.⁸⁴ While dilation itself does not impact the underlying EoE disease process,¹³¹ it immediately increases the caliber of the esophagus and leads to symptomatic improvement that may be long lasting.^{16, 132} In contrast, if a patient has a severe persistent stricture, they will remain symptomatic even if esophageal eosinophilia has resolved with anti-inflammatory treatment. If inflammation persists at deeper esophageal levels such as the submucosa or muscularis propria, it has been recently shown that this could explain persistent symptoms despite normalization of mucosal biopsies;⁷⁷ decreased esophageal compliance or unrecognized esophageal narrowing might provide a similar explanation.^{90, 99} For all of these reasons, there is only modest correlation between symptoms and biologic measures, further impacted by a non-linear relationship between the two.^{24, 25, 89, 133, 134} Specifically, symptoms and quality of life seem to increase only in association with the most severe endoscopic findings, such as grade 3 rings or grade 2 exudates, but do not track with the peak eosinophil count itself.

The third complexity is that the current outcomes are only supported by short term data. While there are now substantial data that EoE is chronic, and if left untreated will progress in the majority of patients from inflammation to fibrostenosis,^{7, 10, 11, 135, 136} there are scant data that demonstrate that successful control of disease activity leads to decreased clinical complications such as stricture, food impactions, need for dilation, or growth failure. Such long term outcome data are strongly needed, and are beginning to be developed. One study suggests long-term treatment with topical steroids and control of esophageal eosinophilia leads to decreased food impactions,¹³⁷ while another shows this reduces the number of dilations required for strictures.¹³⁸ Given the importance of endoscopic findings on impacting symptoms and quality of life,^{24, 133} and the correlation of inflammatory endoscopic features with esophageal eosinophilia on histology,^{34, 87–89} endoscopic healing (akin to mucosal healing in IBD) could conceivably be an important primary endpoint in the future.

A fourth issue involves special considerations in pediatrics. In comparison to adults where symptoms are dysphagia predominant, symptoms in children are non-specific and vary by age.^{8, 9, 14} Infants and toddlers have failure to thrive and food refusal, while elementary age and older children can have abdominal pain, vomiting, regurgitation, heartburn, and chest pain. In relation to this, it is very challenging to measure patient-reported outcomes in children, and especially very young children with this diversity of symptoms. A metric like the PEESSv2 has several specific age ranges, with different scales, age-appropriate language, and both patient and caregiver/parent/observer report.^{33, 44} While regulatory agencies have traditionally preferred only patient-reported outcomes, new data from the multi-site Consortium of Eosinophilic GI Disease Researchers (CEGIR) has confirmed excellent correlation between child and observer reported outcomes, potentially allowing for more flexibility in outcome assessment.¹³⁹ As with adults, children too develop compensatory mechanisms which over time become normal for the child as he/she does not know what true-normal is. In addition, several of the symptoms seen in children with EoE can overlap with other etiologies including abnormal feeding/aspiration, celiac disease, malabsorption syndromes, and gastroesophageal reflux disease. Another challenge in children, especially the very young, is that endoscopic findings can be unremarkable in setting of intense esophageal eosinophilia; in such patients EREFS would be of limited value.

Framework for approaching treatment outcomes in EoE

In order to formally develop a framework for treatment outcomes in EoE, steps would include deciding which outcome measures to include, how to define these, and whether to combine them into a global measure of EoE disease activity as was done recently for development of antifibrosis drugs in Crohn disease.¹⁴⁰ Measures should be selected with a Delphi or similar process after systematic review of the literature,^{26, 39, 40} and involve experts across multiple disciplines (gastroenterology, allergy, pathology, nutrition, adult medicine, pediatrics), patient advocacy groups, and other stakeholders. This process would result in a COS to be used in EoE trials, and such efforts are currently underway. An ideal treatment endpoint in EoE would incorporate resolution of clinical symptoms, endoscopic features, epithelial and eosinophilic changes, and esophageal remodeling, as captured accurately by a PRO and within boundaries of safety and quality of care.

In the meantime, it is still possible to develop a conceptual approach to measuring response that can currently be applied to both clinical practice and research. For this, we propose focusing on histology (specifically peak eosinophil count), symptoms (though methodology needs refinement), and endoscopic findings (using EREFS) (Figure 1). We also proceeded with the understanding that that while many outcome measures exist, they are also incompletely studied, there is heterogeneity in publications, and an overall composite score for all outcomes has not been developed. In addition, special considerations need to be given to differences in patient age (especially younger children) and the reason for assessing outcomes; routine clinical practice, investigator-initiated research projects, or industry-driven pharmaceutical trials will also have different needs.

Figure 1.

A framework for classifying treatment response in EoE.

As a general definition, a patient with EoE can be considered a treatment responder if he or she has improvement or resolution of esophageal eosinophilia, symptoms, and visible esophageal abnormalities. On one side of the spectrum, a “complete” response can be clearly defined with normalization of esophageal biopsies, as measured by <1 eos/hpf, symptom resolution, and endoscopic normalization. This fits with the concept of “deep remission” that has been recently described.¹⁴¹ However, it is important to note that a minority of patients achieve a response of this level, and that this level of response is not persistent, so this may be neither a realistic goal for clinical practice nor an achievable endpoint for a trial.^141–144 Additionally, while histologic evaluation presents a clear threshold for complete response, data are less clear for formal symptom and endoscopic parameters.^{31, 81} To date, only a score of <20 points on the EEsAI has been associated with clinical remission.^{32, 50} While complete response on other symptom scales has not been defined, a decrease of >90% in a symptom score would be a reasonable initial definition for complete response. Similarly, there are some data that suggest an EREFS score <2 distinguishes EoE cases from controls,⁸⁷ so this may be considered to be a threshold for complete response. Ideally, treatments that achieve this complete response would also be safe, cost-effective, increase quality of life, improve esophageal distensibility and other biomarkers, but this has yet to be studied.

Non-response has been conceptually defined as persistent esophageal eosinophilia, symptoms, and endoscopic abnormalities.¹⁴⁵ On esophageal biopsy, an eosinophil count of <15 eos/hpf has been assessed as a histologic response threshold in EoE. In both a retrospective and prospective study, this threshold yielded a substantial associated symptomatic and endoscopic improvement, and pushing the response threshold lower only yielded minimal additional gains.^{146, 147} There is also a symmetry with this threshold, as it is the level required for diagnosis and has been shown to be sensitive and specific for EoE.^{1–3, 148} A threshold change for symptom and endoscopic findings has not been studied. However, it seems reasonable that an improvement of <30% in either a symptom score or in EREFS would be consistent with non-response.

Between complete normalization and non-response, there is a spectrum of response that has been variably defined. Other GI conditions, including IBD, have established frameworks for “response”, typically defined as improvement beyond a set threshold, and “remission”, a terms associated with quiescent disease.²² However, these gradations of treatment response have not been studied in EoE. As can be seen in Table 2, intermediate eosinophil count thresholds have included <7, ≤6, ≤5, and <5 eos/hpf, as well as ≤16 eos/mm² (~4 eos/hpf), and some studies have also used >75% or >90% decrease in peak or mean counts. This heterogeneity is directly due to lack of data, though one study suggests that <5 eos/hpf may be an optimal response threshold for clinical trials, as it identified most patients who had a combined symptomatic and endoscopic response.¹⁴⁷ In this context, we have suggested acknowledging a gradation of response between non-response and complete normalization, which could be further subcategorized with eosinophils counts of 7–14 and 1–6 eos/hpf, and symptom and endoscopic improvement of >30%, but <90%.

With this framework (Figure 1), practitioners and researchers can calibrate response to either an overall response threshold or to complete normalization. As additional data addressing these issues come to light, these responses levels can be revisited. In particular, formal COS exercises that are underway may come to different conclusions or provide additional insight.

Future research directions and conclusions

As the above discussion makes clear, substantial work is required to generate data supporting definitions of treatment outcomes in EoE. One key area is to understand what level of residual esophageal eosinophil is associated with future clinical complications, including stricture development/progression to fibrosis, food impactions, and optimal growth and development. These data would help inform definitions for gradation of response, and how low eosinophil counts should be pushed to minimize the chance of these adverse outcomes, while also balancing long-term safety of treatments. Similar analyses need to be performed for the EoEHSS, as additional pathologic features (e.g. basal zone hyperplasia and lamina propria fibrosis) may be linked to these clinical outcomes more closely than eosinophil count alone, and in the subepithelial tissue compartment. PROs in children and over a range of non-dysphagia symptoms must also be further developed and tested to demonstrate responsiveness to therapy and understand changes in multiple symptoms over time. Preferably, all symptom metrics would be open source and freely available to clinicians and the research community, thus helping to prompt more wide-spread use, including incorporating these metrics into electronic health record systems. Extensive work is also needed to understand the role of biomarkers of disease, be they tissue-based or from the blood, urine, or stool compartments, and the role of FLIP in measuring esophageal distensibility. How these metrics correlate with more traditional outcomes, and where their role in clinical care will land, remain major questions. Another key area of study is whether mucosal healing, as measured by EREFS, may be able to supplant eosinophil count or symptoms as a primary EoE outcome. Related to this, it will be important to investigate whether inflammatory and fibrotic measures should be assessed separately on both EREFS and HSS, and in what settings the focus on fibrosis as an outcome is most pertinent. Children require additional consideration. Size of instruments such as that of FLIP could be an issue in the younger child, as would be the lack of verbal skills where a parent-driven PRO would be required. In addition, in the young child one may have unremarkable endoscopic appearance of the esophageal mucosa but with significant histological changes; how EREFS can be employed in such situations needs study including the possibility of a pediatric-specific version of EREFS. Finally, developing data in all of these areas will support a standard COS that can be implemented for uniform reporting across clinical practice and trials. This knowledge will also help support the burgeoning field of eosinophilic GI diseases, and help to streamline outcome development for other related conditions such as eosinophilic gastritis, gastroenteritis, and colitis.¹⁴⁹

In conclusion, great strides have been made in the area of outcome metrics for EoE, but the need for validated outcome measures that will push the field forward and facilitate drug development is widely recognized. In that context, multiple validated measures have been developed over the last half-decade, spanning histologic findings, symptoms, and endoscopic features, and this is helping to bring a wealth of treatment options forward into research and clinical practice. This work, however, has made the need for thresholds of response even more evident. While it is easy to describe qualitative concepts like complete normalization and non-response, specifically defining and operationalizing response thresholds in EoE is challenging especially since in clinical practice a complete normalization is not necessarily the bar. For now, we propose that complete normalization of esophageal biopsies (<1 eos/hpf) includes near-elimination of symptoms with a decrease in symptoms of >90% (or an EEsAI score <20), and normalization of the esophagus an EREFS score of <2. We proposed non-response to be persistent esophageal eosinophilia with >15 eos/hpf, persistent symptoms with <30% decrease in a PRO, and persistent endoscopic findings with <30% decrease in EREFS. The middle ground of response is still wide, likely covering graded response thresholds such as (7–14 eos/hpf) or (1–6 eos/hpf), but additional work is needed to understand important cut-points in this range. Moreover, the treatment setting and individual patient’s circumstance may impact on how thresholds are interpreted and whether these should lead to treatment changes. In practice, a patient with symptom resolution and endoscopic improvement, but some residual eosinophilia on biopsy may be best served by observation rather than intensification of treatment, depending on their individual situation, as not all patients may be able to achieve full response for all outcomes. This is particularly important when considering the relative lack of data supporting many absolute outcome thresholds. In the future, these thresholds will need to be further tested and linked to clinical metrics and long-term safety and complication data, and the role of other outcomes beyond symptoms, endoscopy, and histology should be investigated. We also await the formal development of a COS in EoE, to help bring standardization to an area of EoE with substantial heterogeneity and variability.