Detection of Barrett's neoplasia with vibrational spectroscopy

• Main Author: Old, Oliver
• Other Authors: Shore, Angela ; Stone, Nicholas ; Kendall, Catherine ; Almond, Max ; Barr, Hugh
• Published: University of Exeter 2015
• Subjects: 610
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.681893

Early detection of Barrett’s oesophagus and associated neoplasia is key to preventing progression to oesophageal adenocarcinoma. Improving surveillance and introducing population screening for Barrett’s are major goals of current research: this project aimed to apply emerging techniques in vibrational spectroscopy to these problems. Fourier transform infrared (FTIR) mapping was used to develop an automated histology tool for detection of Barrett’s and Barrett’s neoplasia in tissue biopsies. 45 FTIR maps were measured from 22 tissue samples from 19 patients. Principal component analysis (PCA) fed linear discriminant analysis (LDA) was used to build classification models based on spectral differences, tested using leave one sample out cross validation (LOSOCV). Classification of normal squamous samples versus ‘abnormal’ samples (any stage of Barrett’s) was performed with 100% sensitivity and specificity. Using a 3-group model to differentiate normal squamous, non-dysplastic Barrett’s and neoplastic Barrett’s (dysplasia or adenocarcinoma), neoplastic Barrett’s was identified with 95.6% sensitivity and 86.4% specificity. Non-endoscopic cell collection devices have recently been developed for population screening for Barrett’s oesophagus. A further aim of this project was to evaluate FTIR for classification of oesophageal cells. Cytology brushings were collected at endoscopy, cytospun onto slides and FTIR maps measured. Cytology review and contemporaneous histology was used to inform a training dataset containing 141 cells from 17 patients. A classification model was constructed using PCA-fed LDA. Applying this training model to the entire dataset of 115 FTIR maps from 66 patients, whole samples were classified with sensitivity and specificity respectively as follows: normal squamous 79.0% and 77.0%, non-dysplastic Barrett’s 31.3% and 100%, and neoplastic Barrett’s 83.3% and 54.2%. Raman spectroscopy was also evaluated as a tool for tissue diagnosis, but several strands of enquiry were limited by instrument problems. FTIR mapping could be used as an accurate, automated tool for processing biopsies in Barrett’s surveillance. Analysis of oesophageal cell samples can be performed using FTIR with reasonable sensitivity for Barrett’s neoplasia, though poor specificity with the current technique.