From genomics to diagnostics of esophageal adenocarcinoma

From genomics to diagnostics of esophageal adenocarcinoma

Ordering of genomic abnormalities

Most esophageal adenocarcinomas (EA) arise in a metaplastic epithelium, termed Barrett’s esophagus (BE), usually as a consequence of chronic gastroesophageal reflux.⁵ Since persons with BE have about thirty-fold higher risk of developing esophageal adenocarcinoma, it is often recommended that they undergo periodic surveillance with endoscopy and multiple biopsies to detect the development of high-grade dysplasia or an early cancer which can be effectively treated.⁶ However, BE can remain in a relatively quiescent state for many years without becoming malignant; conversely, when it occasionally does “go bad” it can happen quite quickly.⁷ Consequently, the effectiveness of surveillance for reducing mortality is a topic of much debate, especially when costs are taken into account.^8,9 Identifying robust biomarkers of neoplastic progression risk would be quite helpful in directing primary and secondary prevention efforts to those most likely to benefit, while permitting the large majority of those unlikely to develop EA to avoid subsequent procedures, risks, costs and worry.

Weaver, et al.,⁴ address this need through a series of experiments designed to identify recurrently-mutated genes in EA and determine the stages of neoplastic progression at which they become mutated. They performed whole-genome sequencing of cancer cases in a discovery cohort (n=22), and identified 26 genes with recurrent mutations for targeted resequencing in an additional 90 cancer cases. Finally, targeted resequencing for this validated gene panel was carried out in 40 non-dysplastic BE and 39 high-grade dysplastic BE samples. Remarkably, they observed that many of the so-called driver mutations of EA were already present in non-dysplastic BE, expanding upon previous reports by Dulak et al.¹⁰ and Agrawal, et al.¹¹ In particular, 53% of the non-dysplastic BE biopsies and 91% of high-grade dysplastic BE samples harbored one or more mutations in the panel. The exceptions were SMAD4, which was mutated only in invasive cancers, and TP53, which was mutated only rarely (2.5%) in non-dysplastic BE but frequently in high-grade dysplasia (72%) and invasive cancer samples (69%.) It should be noted that the relatively small size of the discovery cohort, and the observation that additional genes varied in their mutation frequency between stages, but at a level of statistical significance that did not survive correction for multiple testing, suggests that increased sample size may identify additional genes useful in discrimination.

Opportunities for Translation

While the presence of mutated TP53 in high grade dysplasia and its predictiveness of subsequent EA has been known for a number of years,¹² a particularly exciting aspect of the research reported by Weaver, et al.⁴ relates to the application of this knowledge to a developing non-endoscopic clinical test – the Cytosponge.¹³ This is a dissolvable capsule which contains a compressed mesh attached to a string. After being swallowed and allowed to dissolve in the stomach, the expanded mesh is withdrawn and placed into a preservative. The authors established the ability of the Cytosponge to collect cells in which TP53 mutations can be reliably detected, despite low and variable allele fractions of mutant DNA (ranging from <1% to 36%). They further demonstrated the ability of this approach to identify the presence of high-grade dysplasia with 86% sensitivity (19 mutations detected in 22 individuals with high-grade BE) and 100% specificity (0 mutations detected in 67 individuals with non-dysplastic BE or with no BE).

How might this approach be translated into improved clinical care? In the current paradigm, there is a substantial “risk gap” between the large number of persons in the general population who may have recurrent symptoms of gastroesophageal reflux and/or other risk factors for EA, and the much smaller number in whom significant mutations and other genomic abnormalities have already occurred, marking them as having significant absolute risk of developing the malignancy. Performing an invasive endoscopy in (potentially) all persons with recurrent reflux symptoms, who comprise about 20% of the adult population, is an expensive proposition with very low yield. A non-endoscopic test, such as the Cytosponge with TP53 mutation assays, would provide the primary care provider with an important tool to bridge this risk gap (Figure 1). For example the office-based test could be offered only to selected individuals, e.g., those considered at elevated risk based on a panel of known risk factors and blood-based biomarkers, and only those with evidence of TP53 mutations would be referred to secondary care providers. This would potentially result in many fewer individuals undergoing endoscopy, but those who did would be at substantially higher risk, and an appropriate population for more invasive tests and intensive prevention and treatment efforts. This technology has not matured yet, but is getting closer. Inexpensive and reliable assays available outside of a research setting are needed, in concert with prospective studies and clinical trials. The inclusion of endpoints other than high-grade dysplasia also should be considered.⁵ However, as recent simulation modeling predicts that incidence of EA will continue to increase for two more decades, with a cumulative number of cause-specific deaths of approximately 160,000 in the U.S. alone, progress in this area cannot come fast enough.¹⁴