Introduction Biomarkers have been proposed to improve risk stratification in Barrett's oesophagus (BO), however molecular heterogeneity of BO can hamper detection of molecular changes in random biopsies. Use of Autofluorescence Imaging (AFI) within endoscopic Trimodal Imaging (ETMI) can improve dysplasia detection, but has high false positive rate. Aims of study were (a) validate biomarkers previously published in separate patient cohorts in single study (b) assess whether AFI can increase detection of biomarkers (c) combine ETMI and biomarkers to improve risk stratification of patients with BO.
Methods Prospective European multicentre study. Each patient underwent ETMI with targeted biopsies on AFI positive (AFI+) areas and one AFI negative (AFI−) area, as well as random quadrantic biopsies. DNA content abnormalities (aneuploidy/tetraploidy); loss of heterozygosity (LOH) at 9p and 17p loci; RUNX3, HPP1 and p16 methylation; immunohistochemistry (IHC) for p53 and Cyclin A were tested on targeted biopsies. Each biomarker was correlated with the dysplasia and AFI status.
Results 111 patients with 210 biopsy areas were included in the analysis (AFI+, n=120; AFI-, n=90). Univariate per-biopsy analysis showed that all biomarkers correlated with dysplasia (p<0.05), with exception of 9p LOH. Multivariate analysis showed that aneuploidy, p53 IHC and Cyclin A (3 biomarker panel) were independently associated with dysplasia with an AUC=0.93 (95% CI 0.88 to 0.98) for any dysplasia and AUC=0.95 (95% CI 0.89 to 1) for HGD/early cancer (EC). AFI positivity significantly correlated with aneuploidy, p16 methylation, cyclin A and p53 staining (p<0.05). After excluding dysplastic areas, aneuploidy (p=0.03) and p53 (p=0.04) staining retained statistical correlation with AFI positivity. Analysis of the 3 biomarker panel in patients with dysplasia showed significant biomarker enrichment in AFI+ compared to AFI- areas (p=0.001). Finally, 3 biomarker panel was used to predict prevalent dysplasia. Using a cut-off of ≥2 biomarkers, the panel when applied to AFI+ areas alone, showed sensitivity and specificity of 88% and 90% respectively for diagnosis of HGD/EC, and 64% and 96% respectively for diagnosis of any dysplasia, compared to overall histology.
Conclusion AFI increases detection rate for molecular biomarkers. A panel of 3 molecular biomarkers on a small number of AFI targeted biopsies can efficiently predict the dysplasia status and potentially inform therapeutic management of patients with BE.
Competing interests None declared.