Efficacy of Proton Pump Inhibitor Drugs for Inducing Clinical and Histologic Remission in Patients With Symptomatic Esophageal Eosinophilia: A Systematic Review and Meta-Analysis

Background & aims: Proton pump inhibitor (PPI) therapy might lead to clinical and histologic remission in a significant proportion of patients with symptomatic esophageal eosinophilia (>15 eos/high-power field). We aimed to evaluate systematically the efficacy of PPI therapy for these patients.

Methods: A search in MEDLINE, EMBASE, and SCOPUS databases, and the American Gastroenterological Association Institute, American College of Gastroenterology, and United European Gastroenterology meetings abstract books, was performed. Primary outcomes were clinical response and histologic remission (<15 eos/high-power field) after PPI therapy. Secondary outcomes were the influence on the response to PPIs of age group, study design/quality, PPI type, doses and interval dosing, and pH monitoring results. Data were pooled using a random-effects model.

Results: Thirty-three studies (11 prospective studies) comprising 619 patients with symptomatic esophageal eosinophilia (188 children and 431 adults) were included. PPI therapy led to a clinical response in 60.8% (95% confidence interval, 48.38%-72.2%; I(2) = 80.2) and histologic remission in 50.5% (95% confidence interval, 42.2%-58.7%; I(2) = 67.5) of patients. No differences were observed regarding the study population (children vs adults), the type of publication, or its quality. PPIs were nonsignificantly more effective in prospective studies (52.6% vs 39.1%) administered twice daily compared with once daily (55.9% vs 49.7%), and with pathologic pH monitoring (65.4% vs 49.3%). A significant publication bias in favor of studies reporting histologic responses to PPIs was observed.

Conclusions: PPI therapy induces clinicohistologic remission in half of patients with symptomatic esophageal eosinophilia. This finding should be interpreted with caution because of poor-quality evidence, heterogeneity, and publication bias.