Modelling progression risks in Barrett’s oesophagus

This is an opportunity to work with Prof Fitzgerald and her group who are internationally renowned for their work on Barrett’s oesophagus including the development of a novel, non-endoscopic capsule sponge test, and Dr. Katki’s group leaders in developing quantitative methods for risk-based approaches to cancer screening.

The incidence of oesophageal cancer has increased rapidly in the past 30 years. Oesophageal adenocarcinoma (OAC), the most common form of oesophageal cancer in the US and UK, has <20% 5-year survival, which improves with early detection. Individuals known to have Barrett’s oesophagus, an asymptomatic precursor condition to oesophageal adenocarcinoma, undergo surveillance with the goal of treating advanced (dysplastic) Barrett’s oesophagus before it develops into cancer, thereby preventing cancer. Even if a cancer is not prevented, surveillance may detect it an earlier, asymptomatic, stage, where survival is better. However, currently the vast majority of Barrett’s oesophagus cases are undetected. The recent development of the capsule-sponge (Cytosponge) by Prof Fitzgerald’s lab has made screening for Barrett’s oesophagus more accessible, since it is cheaper than endoscopy, the prior screening method, and can be performed at a GP practice. We therefore expect an increase in the number of people with detected Barrett’s oesophagus, and therefore the number of people undergoing surveillance.  

Current surveillance guidelines depend only on characteristics of the Barrett’s oesophagus, and have not been updated to reflect the use of capsule-sponge as a surveillance tool. We will model the time to progression among patients with Barrett’s oesophagus, to inform surveillance intervals. Since Barrett’s oesophagus is asymptomatic, it can only be detected at times when the oesophagus is evaluated; special statistical methods are therefore required to model this, such as prevalence-incidence models, developed by Dr Katki’s group. The estimates will focus on absolute risk estimates, since ideally individuals would have a surveillance visit when their absolute risk exceeded a threshold, following the concept of ‘equal management of equal risk’. 

The projects will use clinical data from large cohorts of individuals undergoing endoscopy surveillance in England, Scotland and Northern Ireland. We will use both capsule-sponge and endoscopy data to inform surveillance intervals, including how surveillance intervals could vary based on the number of previous surveillance visits.