Introduction The current standard for Barrett’s oesophagus (BO) endoscopic surveillance is based on random quadrantic biopsies every 2 cm. This is limited by sampling error due to inconspicuous dysplasia. We have previously found that an edible lectin (wheat germ agglutinin or WGA) has lower affinity for dysplastic compared to non-dysplastic BO and in an ex vivomodel of molecular imaging we showed that WGA-fluorescein in combination with autofluorescence imaging (AFI) can target dysplasia in BO (Bird-Lieberman, Nat Med, 2012). In this study we used a pre-operative setting to investigate in vivoWGA-based molecular imaging

Method Inclusion criteria were: i. MDT referral to oesophagectomy for oesophageal adenocarcinoma (OAC) and ii.evidence of endoscopically visible BO associated to OAC on index endoscopy or staging endoscopic ultrasound. In the operating theatre prior to oesophagectomy, the oesophagus was first inspected by white light imaging to locate the tumour and measure the size of the associated BO. Imaging was then switched to AFI to locate AFI positive areas (red-purple colour). Up to two AFI negative areas (green colour) were also selected for control biopsy sampling. 20–30 mL of 5 µg/mL WGA-fluorescein were then sprayed onto the mucosal surface. After 5 min, the mucosa was gently washed with water and the oesophagus was inspected again in AFI mode. On WGA-enhanced AFI, areas of high WGA binding were located (WGA positive, green colour), as well as areas of low binding (WGA negative, red-purple and hence high risk). Targeted biopsies were taken for histopathological correlation. Histological outcome was binarised based on the presence or absence of dysplasia. Chi-squared test was used to compare diagnostic accuracy for dysplasia of AFI alone and WGA-enhanced AFI

Results 16 patients were recruited at a single centre, of which 12 completed the protocol. 4 patients were excluded due to absence of visible BO associated with the OAC at the time of resection. Across the 12 patients a total of 41 endoscopic locations were selected for biopsy of which 29 were WGA negative and 12 were WGA positive. The overall sensitivity for dysplasia of AFI alone was 78% and the specificity was 50%. When compared to AFI alone, WGA improved sensitivity from 78% to 88%, with a slight increase in the specificity to 56%. AFI and WGA-enhanced AFI had an equal false positive rate (24%). No toxicity events where recorded.

Conclusion WGA-based molecular imaging is safe and feasible in vivo. The use of WGA may improve the sensitivity of AFI for dysplasia, but larger studies in a surveillance population are required

Disclosure of interest None Declared.