Esophageal adenocarcinoma (EAC) is the leading cause of cancer related deaths in the Western World with limited therapeutic options. Recently, we have developed a classification system that stratifies EAC into three groups based on their microscopic tumour appearance (morphology) and identified molecular expression profiles that are associated with these respective morphological groups. However, this method of molecular profiling is limited in its ability to profile each individual morphological pattern within each group. Furthermore, quantifying these different patterns can be laborious in the everyday clinical setting and limited to expertise of specialist centres. As such, our goal in this study is to develop a deep learning algorithm to predict the individual morphologies within each group using H&E slides from a large cohort of patient specimens and assess the prognostic implication of these. Next, we will use a relatively new technology, Xenium, which will allow us to identify the molecular profile of each individual cell on H&E slides and overlay this with the tumour morphological patterns. Finally, we will train the deep learning algorithm to predict these molecular profiles directly from the H&E slide. This is a first of kind study in EAC, which will permit us to identify new therapeutic vulnerabilities and better stratify patients for existing therapies.
As mentioned, EAC is an aggressive disease that is still poorly understood on a molecular level. As such, this study is a hitherto unexplored method of uncovering the molecular underpinnings of EAC by direct correlation with histomorphological features at single cell resolution with preservation of tissue architecture. Data from this study will facilitate better methods of stratifying patients for existing therapeutic options and identification of further therapeutic targets. The Xenium data will be made public, which will be beneficial to the esophageal and research community. Furthermore, the morpho-molecular correlation will have an impact on the pathology community by providing an opportunity to revise the current pathological classification system to be more prognostically significant.