• biomarkers
• Eosinophilic Esophagitis

Upper endoscopy (EGD) with esophageal biopsies is required to rule out eosinophilic esophagitis (EoE) in patients reporting dysphagia. Moreover, EoE monitoring currently requires esophageal biopsies to assess treatment response based on eosinophil counts on histology. Non-invasive screening and monitoring strategies are needed to reduce the diagnostic delay and streamline the follow-up of paotients with EoE.
In this prospective cross-sectional multicenter study conducted at three EoE-referral centers in Italy, we investigated whether absolute eosinophil count (AEC) and eosinophil percentage (EosPercent) could a) predict a diagnosis of EoE among patients referred for dysphagia, b) serve as a marker of histological remission in treated EoE patients. We enrolled consecutive patients undergoing EGD with six esophageal biopsies for suspected or known EoE. For all patients, a full blood count was obtained prior to the EGD. A diagnosis of active EoE or histological remission was performed based on current guidelines. Non-EoE dysphagia (NED) patients were identified based on the presence of 15 eosinophils/high-power field in all six esophageal biopsies while off eosinophil-depleting treatments. Demographic, clinical, AEC, and EosPercent data were collected. Kruskal-Wallis Rank Sum Test and Pearson’s Chi-squared were used for comparisons as appropriate. ROC curve analysis was used to assess the diagnostic accuracy (AUC) of AEC and EosPercent. Significance threshold was p<0.05.
Of the 92 included patients, 33 had active EoE, 18 had EoE in histological remission while on budesonide orodispersible tablets (BOT) or proton pump inhibitors (PPI), and 41 had NED. Patients with EoE were younger and more frequently male and atopic compared to NED (p<0.01 for all). Overall, 98% (50/51) of EoE had AEC, and 84.3% (43/51) had EosPercent within the normal range. Similarly, 100% (41/41) of NED had AEC, and 90.2% (37/41) had EosPercent within the normal range. There were no differences between EoE and NED in terms of proportion of patients with AEC or EosPercent above the upper limit of normal (p>0.4 for all) (Table 1). Median AEC [380 (210, 460)/mm3 vs 120 (70, 220)/mm3; p<0.001) and EosPercent [5.40 (3.60, 7.05)% vs 1.80 (0.8, 3.0)%; p<0.01] were significantly higher in active EoE compared to NED. Based on ROC curve analysis, a cut-off of 155 eosinophils/mm3 segregated active EoE from NED with AUC of 0.82, specificity of 91.0%, and sensitivity of 65.8%, while a cut off value of 2.5% of EosPercent had AUC of 0.83 with 68.2% specificity and 93.5% sensitivity for the identification of EoE among patients with dysphagia. In terms of prediction of BOT or PPI response, median AEC [380 (210, 460)/mm3 vs 150 (98, 268)/mm3; p<0.01) and EosPercent [5.40 (3.60, 7.05)% vs 1.85 (1.55, 4.47)%; p<0.01] were significantly higher in active EoE compared to EoE in histological remission. Based on ROC curve analysis, a cut-off of 375 eosinophils/mm3 had AUC of 0.75 with 94.1% specificity and 54.5% sensitivity, while a cut off value of 2.25% of EosPercent had AUC of 0.79 with 93.5% specificity and 60.0% sensitivity for the identification of histological remission among EoE patients.
AEC and EosPercent are within the normal range in most EoE patients regardless of histological disease activity. Nevertheless, AEC and EosPercent values accurately predict a diagnosis of EoE among patients reporting dysphagia, as well as histological remission among EoE patients. If these findings are replicated on o larger scale, AEC and EosPercent could be used as biomarkers in clinical practice.