Index of Severity for Eosinophilic Esophagitis: Growing body of evidence supports use

Introduction 

Eosinophilic esophagitis (EoE) is a chronic, immune-mediated disease characterized by esophageal dysfunction. However, assessing its severity in clinical practice has historically been variable and subjective, potentially impacting treatment decisions. To standardize this assessment, the American Gastroenterological Association (AGA) facilitated the development of the Index of Severity for Eosinophilic Esophagitis (I-SEE).¹ The practical tool integrates key clinicopathologic features from three domains (symptoms and complications, inflammatory features, and fibrostenotic features) into a single score that categorizes disease activity as inactive, mild, moderate, or severe. I-SEE aims to provide a more comprehensive assessment that goes beyond eosinophil counts to include symptom frequency, complications, and signs of remodeling. Since the assessment creation, a growing body of evidence has explored its clinical utility. Highlighted below are several of the key studies examining the application and validity of I-SEE in characterizing EoE patients and assessing treatment response. 

I-SEE in Adults

To understand how I-SEE performs in clinical practice, initial validation efforts focused on characterizing its performance in adult populations from clinical trial and real-world settings. A post-hoc analysis of a randomized comparative trial of topical corticosteroids in adults was one of the first studies to apply I-SEE retrospectively.² In this cohort, where the majority of individuals were classified as moderate or severe, every subcategory of the score decreased with treatment. Most participants shifted to a lower I-SEE category after treatment. The mean severity score decreased significantly after 8 weeks of therapy, with larger decreases observed in histologic responders (<15 eos/hpf) compared to non-responders. Furthermore, higher baseline I-SEE scores correlated with features not explicitly included in the index, such as lower body mass index, longer duration of dysphagia symptoms prior to diagnosis, and smaller esophageal diameter, supporting its construct validity. Baseline severity also predicted the need for esophageal dilation at the post-treatment endoscopy.

Subsequent real-world data provided further context. A large multi-center observational study assessed physician-perceived disease severity and calculated an adapted I-SEE score.³ In this real world cohort, only 29% of patients were classified as moderate or severe in the I-SEE category, highlighting differences between trial and real-world populations. The distribution of disease severity was similar between children and adults; however, physicians consistently overestimated severity compared with the I-SEE score, suggesting that use of I-SEE may reduce subjectivity in EoE disease assessment.

To reduce challenges associated with point-of-care use, the AGA developed an I-SEE mobile application.⁴ Usability testing across adult and pediatric gastroenterologists and allergists found the app to be user-friendly. Collectively, these descriptive analyses show that the I-SEE score behaves as expected: it decreases both numerically and categorically with treatment and demonstrates similar severity distributions between children and adults. 

I-SEE in Pediatrics 

Evaluating how I-SEE generalizes to the pediatric population is important to ensure its applicability and reliability. Recent work applying I-SEE to pediatric cohorts revealed both similarities and important age-related differences. Two retrospective pediatric studies evaluated changes in I-SEE at baseline and over time. A European, multi-center study found that most patients present with a moderate baseline I-SEE score (61%), while only 2% of patients have a severe baseline I-SEE score.⁵ Severity in this study was influenced by the use of combination therapy and esophageal perforation; the presence of malnutrition was not reported.

In comparison, a single center US study found a more even distribution of mild (43%), moderate (36%), and severe (21%) I-SEE categories at baseline with severe I-SEE driven by presence of malnourishment (low BMI percentile) and poor feeding.⁶ This difference in nutritional status may be explained by a difference in mean age (11 years old in the European cohort versus about 5 years old in the US cohort).

In both studies, food impaction represented a minority of initial presentation (17% in the European cohort and 10% in the US cohort). Esophageal strictures were not a frequent complication (4% of the European cohort at baseline and 1% of the US), which may also contribute to lower I-SEE scores.  I-SEE improved in both studies domains over time with treatment. Notably, the European study found that a higher I-SEE score was associated with treatment challenges (i.e. combination therapy). The studies emphasize the utility of I-SEE in pediatric populations for standardizing disease assessment but also highlight the challenges of applying a single, unified score across adults and children. Further high-quality studies in pediatric populations are needed to refine and validate its use in this age group. 

I-SEE and Treatment Response 

Beyond its correlation with traditional disease metrics, determining whether I-SEE can predict treatment response may offer additional value, given that it integrates multiple aspects of disease. This is increasingly important as therapeutic options for EoE expand. Evaluation of the association between I-SEE and topical corticosteroid response has shown that most patients are classified as mild to moderate pre-treatment, while histologic response is lowest in the severe category.⁷ Similarly, baseline I-SEE severity is inversely associated with post-treatment symptom response, endoscopic severity, and reduced need for dilation. A comparable pattern was seen with dupilumab, as patients with severe baseline I-SEE scores had lower post-treatment response rates than those with mild or moderate disease.⁸

Furthermore, each 1-unit increase in pre-treatment I-SEE was associated with a 1.5% decrease in histologic response to dupilumab at the <15 eosinophils per high-power field (eos/hpf) cutoff and a 3% decrease at the ≤6 eos/hpf cutoff. Although I-SEE has not been evaluated across all treatment modalities, current evidence suggests it may help stratify likelihood of treatment response by baseline I-SEE severity and provide a more objective measure of change before and after therapy. Severe baseline disease may benefit from more intensive upfront treatment, though additional studies are needed to evaluate I-SEE across different patient population and treatment strategies. 

Molecular Correlates of I-SEE 

Complementing clinical observations, research has explored the relationship between I-SEE scores and the underlying molecular pathophysiology of EoE. A multi-center study correlated I-SEE scores with the EoE Diagnostic Panel (EDP), a gene expression measure.⁹ Total I-SEE score showed a modest inverse correlation with EDP score, indicating that higher clinical severity was associated with higher molecular severity. This was driven mainly by the inflammatory and fibrostenotic domains; the symptoms/complications domain did not correlate with EDP scores. Molecular severity worsened progressively from inactive to moderate I-SEE categories. However, the severe I-SEE group had less severe molecular profiles than the moderate group, possibly because severe scores are often driven by historical complications or fibrosis. Longitudinally, changes in a modified I-SEE score (excluding complications) reflected changes in molecular activity, particularly in the inflammatory and fibrostenotic domains. This suggests that the multidimensional I-SEE captures biologically relevant tissue inflammation and remodeling. 

Conclusions and Future Directions

The development of I-SEE represents a positive step towards standardizing the assessment of EoE severity. As highlighted in the studies, I-SEE correlates with key clinical features in both adults and children, tracks with treatment response, and inflammatory and fibrostenotic domains reflect molecular disease activity. Although the drivers of severity may differ by age, I-SEE also shows potential for predicting treatment response at follow-up, with higher baseline severity appearing linked to lower response rates.

While this initial evidence is promising, several key areas for future research remain. Prospective or robust real-world validation in diverse EoE populations is crucial to confirm its performance and usability. A major long-term goal is to link specific severity levels to evidence-based treatment and monitoring recommendations, similar to guidelines for other allergic and gastrointestinal diseases. Refining the index may also be necessary, potentially by optimizing symptom assessment, clarifying scoring nuances between children and adults, and incorporating additional metrics. Further research items include correlating I-SEE with EoE endotypes, functional measures such as functional lumen imaging probe, and real-world clinical implementation will help define its validity and practical utility. Optimizing I-SEE through these efforts can enhance communication, strengthen clinical trial design, and support more tailored care for patients. 

Disclosures

Dr. Dickerson is on the speaker's bureau for Takeda and Sanofi/Regeneron. Dr. Shoda had no conflicts to disclose. Dr. Ketchem is on the advisory board of Sanofi. Dr. Cotton is a consultant for Regeneron.