P238 Autofluorescence-targeted confocal endomicroscopy versus white-light for Barrett’s oesophagus dysplasia detection: a multi-centre randomised cross-over study

Introduction Dysplasia in Barrett’s oesophagus (BO) is often invisible at white light endoscopy (WLE) and Seattle protocol is labour intensive. There is lack of randomised evidence that advanced imaging improves dysplasia detection. Probe-based confocal endomicroscopy (pCLE) is accurate for high-grade dysplasia (HGD) and intramucosal cancer (IMC) associated to visible lesions, but due to the narrow field requires combination with a red-flag technique for long-segment inconspicuous BO. We aimed to assess the diagnostic accuracy of optical biopsy by pCLE targeted by autofluorescence imaging (AFI) for any grade of dysplasia in patients without visible lesions. Methods In this prospective multi-centre randomised cross-over trial BO patients were randomised to WLE with Seattle protocol (standard arm) https://mail.addenbrookes.nhs.uk/owa/auth/logon.aspx?replaceCurrent=1u (b) overall histology, which included (a) + biopsies within the 12 months of enrolment. Secondary outcomes included procedural time. Results 133 patients completed both arms. 27.8% of patients received a diagnosis of dysplasia (LGD; n=19, HGD/IMC; n=18). In primary analysis (trial histology), pCLE had a sensitivity and specificity for dysplasia of 73.0% and 68.8%, respectively and 72.2% and 61.7% for HGD/IMC. Seattle protocol had a sensitivity of 73.0% for dysplasia and 83.3% for HGD/IMC, with no significant difference between arms. In secondary analysis (overall histology), pCLE had a similar sensitivity to Seattle protocol for dysplasia (61.8% vs. 49.1%; p=0.09) and HGD/IMC (70.0% vs 50.0%; p=0.11). The procedural time in the experimental arm was longer than standard arm (Mean mins 22.3 vs. 16.4; p Conclusions In combination with AFI, pCLE detects inconspicuous dysplasia in approximately three quarters of cases. pCLE has equal diagnostic accuracy for dysplasia compared to Seattle protocol dispensing extensive sampling but at expense of longer procedural time.