Abstract
The early detection and endoscopic treatment of patients with the dysplastic stage of Barrett's oesophagus is a key to preventing progression to oesophageal adenocarcinoma. However, endoscopic surveillance protocols are hampered by the invasiveness of repeat endoscopy, sampling bias, and a subjective histopathological diagnosis of dysplasia. In this case-control study, we investigated the use of a non-invasive, pan-oesophageal cell-sampling device, the Cytosponge™, coupled with a cancer hot-spot panel to identify patients with dysplastic Barrett's oesophagus. Formalin-fixed, paraffin-embedded (FFPE) Cytosponge™ samples from 31 patients with non-dysplastic and 28 with dysplastic Barrett's oesophagus with good available clinical annotation were selected for inclusion. Samples were microdissected and amplicon sequencing performed using a panel covering >2800 COSMIC hot-spot mutations in 50 oncogenes and tumour suppressor genes. Strict mutation criteria were determined and duplicates were run to confirm any mutations with an allele frequency <12%. When compared with endoscopy and biopsy as the gold standard the panel achieved a 71.4% sensitivity (95% CI 51.3-86.8) and 90.3% (95% CI 74.3-98.0) specificity for diagnosing dysplasia. TP53 had the highest rate of mutation in 14/28 dysplastic samples (50%). CDKN2A was mutated in 6/28 (21.4%), ERBB2 in 3/28 (10.7%), and 5 other genes at lower frequency. The only gene from this panel found to be mutated in the non-dysplastic cases was CDKN2A in 3/31 cases (9.7%) in keeping with its known loss early in the natural history of the disease. Hence, it is possible to apply a multi-gene cancer hot-spot panel and next-generation sequencing to microdissected, FFPE samples collected by the Cytosponge™, in order to distinguish non-dysplastic from dysplastic Barrett's oesophagus. Further work is required to maximize the panel sensitivity.