644 Defining Cancer Risk in Barrett's Esophagus Using a 90-Gene Signature

Abstract

Introduction Barrett’s oesophagus (BE) has a highly variable outcome with 0.12–0.5% of patients per year progressing to oesophageal adenocarcinoma (EA). The histopathological grading of dysplasia is used to detect cancer risk in BE; however, considerable variability exists in the reporting of dysplasia. Molecular biomarkers that can detect BE patients with dysplasia would improve risk stratification in BE, enabling clinicians to focus on high risk patients requiring treatment and reduce endoscopic surveillance in the low risk group. The aim of this study was to identify and validate a gene expression signature as a biomarker that can objectively determine dysplastic status and thereby determine the risk of cancer progression. Methods Microarray gene expression profiling was done using 59 oesophageal samples with strict consensus diagnosis by expert pathologists (21 BE with no dysplasia, 10 BE with low grade dysplasia, 13 BE with high grade dysplasia and 8 EA). This data was used to identify a gene signature that separated non-dysplastic BE from high grade dysplasia. Gene expression data from publically available datasets were used to validate the signature. An independent set of 135 fresh frozen samples covering a spectrum of dysplastic Barrett’s stages and control tissue (40 BE with no dysplasia, 21 BE with low grade, 33 BE with high grade dysplasia, 32 EA and 9 duodenum) were used for validation using the high throughput 96:96 microfluidic Fluidigm® chip on the BioMark™ PCR system. Results A set of 90 genes was identified that separated BE with no dysplasia from BE with high grade dysplasia. This 90-gene signature was able to separate the remaining untrained samples on the microarray dataset (7 non-dysplastic, 10 low grade dysplasia and 8 EA). The signature also separated non dysplastic BE samples from EA samples on 2 external published datasets (p ≤ 0.0012). With the fresh frozen samples, the signature separated BE with no dysplasia from BE with dysplasia and EA with an area under the curve of 0.87 (95% CI, 0.80–0.93). Pathway analysis revealed that the RAN (RAs-related Nuclear protein) regulation pathway (p Conclusion The 90 gene-expression profile can reliably identify BE samples with dysplasia and cancer. This approach has the potential to provide robust risk stratification in BE samples as it overcomes the problems with variability in the reporting of dysplasia. Disclosure of Interest None Declared