Validation of methods to assess potential biomarkers in pediatric patients with esophageal eosinophilia.

Aim: To validate methods for determining mast cell density, extracellular major basic protein content, and presence of fibrosis in esophageal eosinophilia.

Methods: Twenty specimens with > 20 eosinophils/high-power field (hpf) classified as high eosinophil density (HE) and 20 specimens with < 5 eosinophils/hpf classified as low esophageal density (LE) were identified. All 40 specimens underwent immunohistochemical staining and trichrome staining. Mast cell density, extracellular major basic protein (MBP) density, and presence of subepithelial fibrosis were assessed in a standardized manner. All specimens were evaluated by two separate observers and by a single observer on two separate occasions to evaluate reproducibility of the methods.

Results: A strong inter-observer correlation was noted for both peak and mean mast cell counts (r = 0.725, P < 0.0001 and r = 0.823, P < 0.0001). A strong intra-observer correlation also was noted for both peak and mean mast cell counts (r = 0.752, P < 0.0001 and r = 0.878, P < 0.0001). A very strong inter-observer correlation was noted for both peak (τ = 0.867, P < 0.0001) and mean extracellular MBP densities (r = 0.925, P < 0.0001). A very strong intra-observer correlation was noted for both peak (τ = 0.875; P < 0.0001) and mean extracellular MBP densities (r = 0.956, P < 0.0001). Excellent inter-rater reliability was found for fibrosis (κ = 0.887). Mast cell and MBP densities, as well as presence of fibrosis, were significantly increased in HE vs LE. The HE group had significantly higher intraepithelial mast cell peak (29.35 ± 21.61 vs 12.45 ± 8.26, P = 0.002) and mean (19.84 ± 15.81 vs 6.35 ± 4.5, P = 0.001) densities than the LE group. The HE group had significantly higher peak extracellular MBP (2.35 ± 0.67 vs 0.45 ± 0.61, P < 0.001) and mean extracellular MBP (1.95 ± 0.76 vs 0.20 ± 0.29, P < 0.0001) densities than the LE group. Seventy-three percent of patients with HE (11/15) had fibrosis, whereas only 10% of patients with LE (1/10) had fibrosis (P < 0.01). MBP performed the best in predicting classification of HE vs LE, with mean MBP demonstrating 100% sensitivity and 95% specificity at the optimal cut point.

Conclusion: This study provides methodology and proof-of-concept for future evaluation of these biomarkers for differentiating esophageal eosinophilic diseases such as reflux esophagitis and eosinophilic esophagitis.