Background: Data on the role of the microbiome in adult patients with eosinophilic oesophagitis (EoE) are limited.
Aims: To prospectively collect and characterise the salivary, oesophageal and gastric microbiome in patients with EoE, further correlating the findings with disease activity.
Methods: Adult patients with symptoms of oesophageal dysfunction undergoing upper endoscopy were consecutively enrolled. Patients were classified as EoE patients, in case of more than 15 eosinophils per high-power field, or non-EoE controls, in case of lack of eosinophilic infiltration. Before and during endoscopy, saliva, oesophageal and gastric fundus biopsies were collected. Microbiota assessment was performed by 16 s rRNA analysis. A Sparse Partial Least Squares Discriminant Analysis (sPLS-DA) was implemented to identify biomarkers.
Results: Saliva samples were collected from 29 EoE patients and 20 non-EoE controls;, biopsies from 25 EoE and 5 non-EoE controls. In saliva samples, 23 Amplicon Sequence Variants (ASVs) were positively associated with EoE and 27 ASVs with controls, making it possible to discriminate between EoE and non-EoE patients with a classification error (CE) of 24%. In a validation cohort, the accuracy, sensitivity, specificity, positive predictive value and negative predictive value of this model were 78.6%, 80%, 75%, 80% and 60%, respectively. Moreover, the analysis of oesophageal microbiota samples observed a clear microbial pattern able to discriminate between active and inactive EoE (CE = 8%).
Conclusion: Our preliminary data suggest that salivary metabarcoding analysis in combination with machine learning approaches could become a valid, cheap, non-invasive test to segregate between EoE and non-EoE patients.
© 2022 The Authors. Alimentary Pharmacology & Therapeutics published by John Wiley & Sons Ltd.