Background: The primary pharmacologic treatment modality for EoE is swallowed topical steroids. The factors which predict failure of steroid treatment have been poorly explored. Aim: To determine the frequency and identify predictors of histologic non-response to topical steroid administration in patients with EoE.Methods: This was a retrospective cohort study of the UNC EoE Clinicopathologic database from 2001-2013. Subjects with an incident diagnosis of EoE who met consensus guidelines, including non-response to a PPI trial, were included. Patients were identified who were treated with 2 months of a topical steroid (fluticasone 880-1760mcg/day or budesonide 1-2mg/day). Demographic and pre/post-treatment symptom (patient-reported global improvement), endoscopic (endoscopist-reported improvement), and histologic data were extracted from medical records. Responders were defined by a post-treatment eosinophil count <15 eos/hpf (hpf=0.24mm2) and compared to non-responders. Multivariate logistic regression was performed to assess predictors of response. Sensitivity analysis with different response thresholds (<5 eos/hpf; 0 eos/hpf) was also performed. Results: Of 455 EoE patients, 307 were treated with topical steroids, and 189 met inclusion criteria. Those who had a steroid response (n=109, 58%) did not differ from non-responders (n=80, 42%) on age (28 vs 25, p=0.23), gender (72% vs 70% male, p=0.82), or race (87% vs 78% White, p=0.08). Responders were more likely to have abdominal pain (23% vs 8%, p=0.005), but did not differ on other symptoms, symptom duration, or rates of atopic disease (see Table). Baseline endoscopic findings were similar between groups. Non-responders were almost twice as likely to have undergone dilation at baseline (36% vs 19, p=0.008). Baseline eosinophil counts were similar (76 vs 73 eos/hpf, p=0.80), as were doses of fluticasone and budesonide. Patients with <15 eos/hpf on treatment had better symptom response (87% vs 60%, p<0.001) and EGD response (92% vs 47%, p<0.001). After multivariate analysis, baseline abdominal pain predicted steroid response (OR 3.27 [1.25, 8.57]), while non-White race (OR 0.42 [0.19, 0.96] and baseline dilation (OR 0.42 [0.21, 0.85]) predicted non-response. Results were similar on sensitivity analysis when different histologic response thresholds were used. Conclusions: In this large cohort of EoE patients treated with topical steroids, histologic non-response was common. White race, abdominal pain, and dilation at the initial endoscopy were independently associated with decreased histological response. It is possible that dilation was a marker of more severe disease or that the more open post-dilation esophagus resulted in increased clearance of medication. These response data can help guide therapeutic decisions, particularly when weighing dietary and steroid therapy. Comparison of Histologic Responders (<15 eos/hpf) to Non-Responders (≥ 15 eos/hpf)