Background:
Eosinophilic esophagitis (EoE) is a chronic immune mediated esophageal condition leading to dysphagia, strictures, and food impactions. Endoscopic biopsies remain the gold standard for diagnosis. While minimally invasive devices have shown encouraging results, an unmet need for non-invasive diagnostic methods remains. Measurement of exhaled breath volatile organic metabolome compounds (VOCs) has shown promise as a diagnostic biomarker in other inflammatory disorders. We aimed to characterize the pattern of exhaled VOCs for the diagnosis of EoE.
Methods:
We performed a prospective pilot study of adult patients with symptoms consistent with EoE undergoing standard of care esophagogastroduodenoscopy (EGD). Symptoms included dysphagia, heartburn, and history of food impaction. Exclusion criteria included pregnancy, known esophageal disorder, current therapy used for EoE, and comorbidities or medications that impact VOC analysis (malignancy, enteric infections, tobacco use, antibiotics, systemic steroids, uncontrolled comorbidities). Samples were analyzed with a standardized protocol using selective ion flow tube mass spectrometry. Chart was reviewed: demographics, EoE endoscopic reference score (EREFS), index of severity for EoE (I-SEE), and eosinophil per high power field (eos/hpf). EoE was defined as ≥15 peak eos/hpf from ≥6 esophageal biopsies with no other cause of eosinophilia. Fibrostenosis (FS) was defined as esophageal luminal diameter <14 mm, determined by functional lumen imaging probe. Analysis included descriptive statistics, principal component analysis and receiver operating characteristics (ROC) curve.
Results:
21 patients were enrolled: 11 with EoE and 10 with non-EoE. Mean age was 36.7±14.7 years, 61.9% were male, and all patients self-identified as White and non-Hispanic. Demographics were not different between cohorts. Of the non-EoE cohort, 4 patients were diagnosed with gastroesophageal reflux disease and all had 0 eos/hpf. Of the EoE patients, mean eos/hpf was 57.3±19.5, and all patients had mixed inflammatory and fibrostenotic EREFS features. Median I-SEE score was 7 [range 5-34]: 45.5 % with mild and 45.5% with moderate active EoE. Hydrogen cyanide VOC was significantly reduced in EoE compared with controls. 20 VOCs explained the majority of differences between cohorts. ROC analysis to distinguish EoE and non-EoE resulted in an AUC of 0.7, suggesting a robust discriminatory ability (Figure 1). Repeat analysis comparing FS and non-FS showed differences in three VOCs: increased acetylene & tridecane and decreased dodecane. ROC analysis showed an AUC of 0.76, suggesting a robust discriminatory ability for FS (Figure 2).
Conclusion:
Exhaled VOCs are a promising method to identify EoE and fibrostenotic phenotype. Ongoing studies with a larger cohort size will be used to validate findings and evaluate longitudinal samples.

