Introduction Non-endoscopic detection of Barrett’s oesophagus (BO) and associated neoplasia would allow population screening and could potentially replace endoscopic surveillance. A swallowed cytology brush has been developed (‘Cytosponge’) for oesophageal cell collection. A recent study (BEST2) assessing the Cytosponge used immunostaining with Trefoil Factor 3 (TFF3) as a means of BO detection. Accurate identification of neoplasia could enable risk stratification to determine who requires endoscopy.

Infrared spectroscopy (IR) of biological samples analyses sample biochemistry: subtle spectral differences can be used to classify samples into pathological subtypes. This study aimed to evaluate IR as a diagnostic tool for automated detection of neoplasia in oesophageal cytology.

Method Brush cytology was collected from patients undergoing endoscopy for BO, with contemporaneous biopsy from the same region. Cells were fixed in formalin, centrifuged and plated onto calcium fluoride slides. IR spectra were measured across the entire sample area. Data pre-processing steps selected spectra from individual cells and removed confounding effects from water vapour and non-cellular debris. Contemporaneous endoscopy and biopsy results provided a reference for developing a predictive model. Prinicipal component analysis (PCA), linear discriminant analysis (LDA), and leave-one-out cross-validation (LOOCV) were used to develop and test classification models.

Results Samples were measured from 20 patients (7 BO, 13 dysplasia or adenocarcinoma). LOOCV testing showed a sensitivity for detecting neoplastic change (dysplasia or adenocarcinoma) of 79.9%, with specificity 75.8%.

Conclusion These results show feasibility of using IR to detect neoplasia in Barrett’s screening. Further work is required to expand our preliminary results and refine pre-processing steps to improve sensitivity, and develop trials using the Cytosponge.

Disclosure of interest None Declared.

References

1. Ross-Innes, et al. Evaluation of a minimally invasive cell sampling device, coupled with assessment of Trefoil Factor 3 expression for diagnosiing Barrett’s esophagus: a multi-centre case-controlled study. PLoS Med. 2015;12(1):e1001780

2. Kadri, et al. Acceptability and accuracy of a non-endoscopic screening test for Barrett’s oesophagus in primary care: cohort study. BMJ 2010;341:c4372

3. Townsend, et al. Infrared microspectroscopy for cyto-pathological classification of esophageal cells. Analyst 2015 doi: 10.1039/C4AN01884B