Redefining Histological Cell Counts Using a Standardized Method: The Leuven Intestinal Counting Protocol

Abstract

INTRODUCTION: The diagnosis of eosinophilic gastrointestinal diseases is largely based on mucosal eosinophil counts, but thresholds and normal ranges beyond the esophagus are debated, calling for much-needed methodological standardization. We aimed to develop a standardized workflow for duodenal cell quantification and estimate duodenal eosinophil and mast cell numbers in healthy controls. METHODS: Software-based histological cell quantification using free-sized or fixed-sized regions was developed and applied to digitized hematoxylin and eosin (H&E)-stained slides from 58 individuals (healthy controls [HCs] and patients with functional dyspepsia). Intraclass correlation coefficients (ICCs) compared inter-rater reliability between software-based and microscopic quantification. Reproducibility of the software-based method was validated in an independent cohort of 37 control and functional dyspepsia subjects. Eosinophil identification on H&E staining was compared to immunohistochemistry (IHC). Normal eosinophil (H&E) and mast cell (cKit) ranges were determined in 70 adult HCs. RESULTS: Eosinophil quantification on digitized slides demonstrated excellent (ICC = 0.909) and significantly improved reproducibility over microscopic evaluation (ICC = 0.796, P = 0.0014), validated in an independent cohort (ICC = 0.910). Duodenal eosinophils were more abundant around crypts than in villi (P < 0.0001), while counts were similar on matched H&E- and IHC-stained slides (P = 0.55). Mean ± SD (95th percentile) duodenal eosinophils and mast cells in HC were 228.8/mm2 ± 94.7 (402.8/mm2) and 419.5/mm2 ± 132.2 (707.6/mm2), respectively. DISCUSSION: We developed and validated a standardized approach to duodenal histological cell quantification, generalizable to various mucosal cell types. Implementation of software-based quantification identified 400 eosinophils/mm2 and 700 mast cells/mm2 as thresholds for abnormal duodenal infiltration.