371 Evaluation of Additional Spatial and Temporal Information Increases the Overall Accuracy of an Objective Risk Prediction Assay in Patients With Non-Dysplastic Barrett’s Esophagus

Abstract

INTRODUCTION: An automated, quantitative multiplex immunofluorescence assay has previously been shown to objectively identify non-dysplastic Barrett’s esophagus (NDBE) patients with an increased risk of malignant progression, independently of current clinical variables. These studies evaluated esophageal biopsies from a single level from one endoscopy. We investigated the results produced by this risk prediction assay at multiple endoscopic levels and multiple endoscopies in NDBE patients to determine whether assessment of additional spatial and temporal information can increase the accuracy of the assayin predicting progression from NDBE to high grade dysplasia (HGD) or esophageal adenocarcinoma (EAC). METHODS: A blinded, nested case-control cohort was derived from a community based BE cohort with and without progression. 76 patients with a diagnosis of NDBE of whom 38 progressed to HGD/EAC (progressors) and 38 patients did not progress (non-progressors) were matched for age, sex and maximal BE length. Baseline was defined as the endoscopy 2-5 years prior to the endpoint for progressors or last follow up for non-progressors. All random biopsy levels from the baseline endoscopy and all available prior endoscopies were retrieved and tested by the risk predictionassay in a blinded manner. The assay classifies patients into low-risk (LR), intermediate-risk (IR) and high-risk (HR) for progression to HGD/EAC within 5 years. RESULTS: 76 BE patients (58 male), mean age 63 ± 1 years, were studied. Median time between baseline endoscopy and progression was 3 years, and median surveillance time in non-progressors was 6 years. Sensitivity of the risk prediction assay (HR and IR combined vs. LR) at the baseline endoscopy was 39.5 % when a single biopsy level was selected for risk assessment (biopsy closest to the gastroesophageal junction (GEJ)). Sensitivity increased to 65.8% when all available baseline biopsy levels were assessed, and further increased to 73.7% when all available baseline and pre-baseline biopsies were assessed by the risk prediction assay (Table 1). CONCLUSION: Evaluation of spatial and temporal information increases the sensitivity and overall accuracy of an automated, quantitative risk prediction assay in patients with NDBE. The assay allows for objective risk stratification which may be a practical and effective solution to aid decision-making on surveillance intervals and therapeutic interventions in the management of NDBE patients.