Abstract B05: MRI provides a more reliable method of esophageal tumor detection than endoscopy in a rat model

Background: The surgical rat model of end-to-side esophagojejunal anastomosis has proven utility in terms of its ability to replicate Barrett's carcinogenesis by inducing gastroduodenoesophageal reflux (GDER). Previously, visual endoscopic analysis has been shown as a valuable tool for monitoring natural history of disease, with biopsy allowing for analysis of molecular markers while avoiding animal sacrifice. However, due to the inherent model limitation of the extraluminal presentation of the majority of tumors, as well as the small size of endoscopic biopsies, adenocarcinoma histology has only been validated post-mortem, and therefore treatment efficacy studies have been limited to chemoprevention. The current study aims to test and establish MRI and / or endoscopic visualization for diagnosis and monitoring of esophageal cancer in the rat model, in order to enhance utilization of this promising de novo model to study treatment response of novel agents in established esophageal cancer. Methods: Chronic bile and acid reflux was induced in 24 Sprague-Dawley rats through a modified Levrat's model. At 40 weeks post-surgery, all animals underwent an endoscopy, T2-weighted turbo spin-echo (TSE) MRI scan and post-mortem histological analysis. A total of five esophageal surgeons and gastroenterologists participated in a blinded study and watched endoscopy videos to determine if esophageal tumor was present. Additionally, three radiology experts participated in a blinded study designed to assess correlated coronal and axial MRI scans and determine whether tumor was present. All videos and scans were studied in triplicate, and histology was used as the gold standard to assess accuracy of each participant's analysis. Results: The accuracy of MRI and endoscopic analysis to correctly identify histology was 69.6% and 57.4%, respectively. The ROC curve analysis showed average MRI rating (AUC = 0.77, p-value = 0.032) was a better discriminator of tumor status in rats than average endoscopy rating (AUC = 0.595, p-value = 0.450). Additionally, false positive rates of MRI and endoscopy were 10.1% and 16.5%, respectively. Inter-observer variability was in substantial agreement for MRI participants (average Kappa = 0.768); whereas endoscopic analysis did not show greater agreement than that what would be expected by chance. All cases where MRI participants agreed on the presence of tumor, histology confirmed the positive findings. Conclusions: MRI is a more reliable diagnos