Minimally Invasive Molecular Approaches for the Detection of Barrett’s Esophagus and Esophageal Adenocarcinoma

/ ABSTRACT Esophageal adenocarcinoma (EAC) is a lethal cancer with poor outcomes (5-year survival <20%), when diagnosed after the onset of symptoms,1but survival is excellent when diagnosed early.2 Esophageal intestinal metaplasia, or Barrett’s esophagus (BE), is the only known precursor of EAC and progresses to EAC via the development of dysplasia. EAC can be prevented by endoscopic therapy of dysplasia.3, 4 Hence endoscopic screening for BE and endoscopic surveillance to detect dysplasia and EAC, followed by endoscopic therapy to prevent EAC in those with dysplasia are recommended.5 Barriers to BE screening include the lack of accurate and user-friendly BE risk assessment tools and the invasiveness and high cost of endoscopy.7 Further, screening is recommended only in those with chronic gastroesophageal reflux (GERD), despite 50% of BE/EAC patients not reporting GERD symptoms. Endoscopic surveillance also misses 33% of prevalent EAC & dysplasia, due to the patchy distribution of dysplasia/EAC, and inadequate biopsy sampling. 8Hence the overall effectiveness of the current EAC interception strategy is severely compromised.9 In the current grant cycle, we have made substantial progress in identifying solutions to these barriers. We have developed an artificial intelligence powered automated BE risk tool, which integrates clinical data from the electronic health record and is substantially more accurate (AUC 0.84) than current guideline endorsed risk criteria (AUC 0.5-0.6). We have also developed and validated an office based, nurse administered non-endoscopic BE screening test, which combines a swallowable cell collection device (sponge on a string: SOS) with three methylated DNA markers (NDRG4, VAV3, ZNF682). The test has good sensitivity and specificity for BE detection compared to endoscopy, particularly for BE with high grade dysplasia and EAC. We have also discovered and validated additional DNA markers that are highly accurate (AUC 0.92) in differentiating dysplasia/EAC from non-dysplastic BE (hereafter “BE surveillance test”).15 These biomarkers are assayed from endoscopically collected whole-field brush cytology specimens, which are anticipated to reduce sampling error and increase dysplasia detection. In specific Aim 1, we will measure and compare BE screening rates with and without a multicomponent intervention in a pragmatic cluster randomized trial. In specific Aim 2, we will select final DNA markers, then train and test the final algorithm of the BE surveillance test. In specific Aim 3, we will perform an economic analysis of biomarker enhanced non-endoscopic BE screening and BE surveillance, using microsimulation modeling to assess the optimal thresholds under which the BE risk tool followed by the SOS BE screening test increases benefits and reduces harm. The model will incorporate inputs from the literature, and completed trials in the current grant cycle, to determine optimal thresholds for cost-effective screening. Utilizing innovative, non-endoscopic, molecular, and modeling approaches, this proposal will favorably impact BE detection and surveillance, enabling effective treatment, and improved EAC outcomes.