The patchy nature of esophageal eosinophilia in eosinophilic esophagitis: Insights from pathology samples from a clinical trial

Background: Eosinophilic infiltrates can be unevenly distributed in eosinophilic esophagitis (EoE), but the degree of variability in distribution is not well described. Aim: To describe variability of eosinophil counts in patients with EoE, and to determine whether original eosinophil counts performed for clinical purposes correlate with dedicated research eosinophil counts. Methods: This is a sub-analysis of a single-center, randomized, open-label clinical trial. Subjects ≥ 18 yrs with an incident diagnosis of EoE per 2007 guidelines were randomized to receive budesonide 1 mg BID for 8 weeks in 1 of 2 forms: nebulized, and then swallowed, or orally in a viscous slurry. At baseline and post-treatment, distal and proximal esophageal biopsies were taken for clinical purposes, and then research protocol biopsies were obtained from the distal, mid, and proximal esophagus. Clinical biopsies were read per individual pathologist practices. Research biopsies were read following a previously validated protocol in which pathologists were masked to treatment allocation, 5 hpfs were examined per biopsy fragment (hpf area=0.24mm2), and the presence of associated EoE findings (degranulation, microabscesses, spongiosis, basal zone hyperplasia, and lamina propria fibrosis) were recorded. For analysis, clinical and research eosinophil counts were compared for all subjects pre- and post-treatment, and then research counts for each pretreatment hpf were reviewed to determine variability throughout the esophagus. Results: Samples were available from 25 and 22 patients and at baseline and follow-up, representing 526 and 533 hpfs examined for research purposes, and 47 samples read for clinical purposes. Pre-treatment, there was marked variation in eosinophil counts. Only 38% of hpfs had ≥15 eos/hpf (Figure), there was a mean 56-fold difference between the lowest and highest eosinophil count in a set of biopsies, and 80% of biopsy fragments had at least one hpf with <15 eos/hpf . This variation was seen in all locations in the esophagus (Figure), and was also seen in patients post-treatment who had persistent esophageal eosinophilia. There was very good correlation between clinical and research reads for the maximum eosinophil count (Spearman's rho=0.87; p<0.001), and agreement was perfect at baseline for determining whether a sample had a count ≥15 eos/hpf (kappa=1.0; p<0.001). Conclusions: Eosinophil counts in EoE varied substantially throughout the esophagus, and even within different hpfs of the same biopsy sample. Research and clinical eosinophil counts correlated well. Given that only one third of hpfs had a count high enough to meet diagnostic guidelines for EoE, multiple biopsies from several locations in the esophagus are essential to diagnose EoE.