Wytske Westra

Barrett’s esophagus (BE) predisposes to the development of esophageal adenocarcinoma (EAC). Barrett’s esophagus is thought to be the result of longstanding (duodeno-)gastro-esophageal reflux disease, GERD. Unfortunately only a small proportion of patients with GERD develop BE and not all patients with BE have previously experienced GERD, which makes selection of patients at risk for development of BE and EAC a challenging process. Currently the only known risk factor for progression to EAC in BE is low-grade dysplasia (LGD) and possibly length of the BE segment according to the Prague classification. Patients are therefore enrolled in surveillance programs with the screening interval based on both histology results (IM/LGD/HGD) and length (Richtlijn Barrett-oesofagus, NVMDL 2018). This manner of risk stratification is imperfect and not cost-efficient and results in unnecessary endoscopies. Furthermore patients with LGD and/or HGD are currently recommended to undergo endoscopic treatment (i.e. radiofrequency ablation (RFA)) with EMR of visible lesions), requiring frequent endoscopies, which is a strain both on healthcare resources as well as on the individual patient and also not without complications. The main goal of this thesis was therefore to better identify those BE patients that are more prone to develop EAC (part 1), improve risk stratification in patients with BE (part 2) and find novel, non-endoscopic treatment modalities to prevent development of EAC in known BE patients (part 3).

GENERAL DISCUSSION AND FUTURE PERSPECTIVE

In Chapter 1 we first confirmed that cigarette smoking predisposed to the development of BE in GERD patients. We also showed that combined use of smokeless tobacco and cigarettes significantly increased the odds of development of BE in patients with GERD. To our knowledge this was the first study looking at the effect of smokeless tobacco on development of BE. It is an important finding in an era where the use of smokeless tobacco is seen as a less harmful alternative to smoking cigarettes and even sometimes promoted as part of smoking cessation strategies. The use of smokeless tobacco is therefore expected to increase further, thus enlarging the population at risk to develop BE. The most important limitation of this study is that it was a retrospective chart study, based on patient intake questionnaires. Also the study cohort was too small to show an independent effect of smokeless tobacco. However due to smokeless tobacco use often being used as a smoking cessation tool and the heterogeneity of the smokeless tobacco products commercially available it will remain challenging to proof a separate effect of smokeless tobacco in future studies. To enable studying the effect of smokeless tobacco forms on development of disease we would like to propose to make this a standard (separate) item on patients questionnaires or intake forms. This would also have to include ‘newer’ forms of smokeless tobacco such as the increasingly popular vaping.

In Chapter 2 we continued to look at risk factors for the development of BE and EAC. We showed that certain Y chromosome haplogroups predispose to the development of BE in GERD and EAC in BE. This finding could explain at least part of the difference in risk to develop BE between the male and female gender. Furthermore, there were certain haplogroups that seemed to be protective for development of EAC in BE, this may explain some of the observed geographical variation in the incidence of BE and EAC. Therefore defining a patients Y haplogroup may in theory also help tailor the identification of those patients at risk for developing BE beyond the currently known risk factors of male gender, GERD, waist circumference and age over 50. However, in order to implement such a tool into clinical practice, results of our retrospective cohort study would of course first need to be confirmed in a prospective cohort study.

In Chapter 3 DNA FISH was performed on samples prospectively collected from 426 patients with non-dysplastic BE. In this study it was shown that it was possible to predict future progression to HGD or EAC using a multivariate model consisting of clinical variables and genetic markers. As mentioned above most guidelines concerning BE recommend surveillance esophago-gastroduodenoscopies (EGDs), whereas in its current form there is little scientific evidence to support its effectiveness. Risk stratification tools like those investigated in Chapter 3 may help to identify those patients with BE where progression to HGD and/or EAC is unlikely and who therefore may not benefit from further surveillance endoscopies. In turn this will help decrease the burden on surveillance programs.

In Chapter 4 we investigated a novel risk stratification tool to predict progression to esophageal cancer in BE patients with HGD. The current guidelines recommend endoscopic treatment for all patients with HGD, and despite the histopathological selection criteria for HGD it seems that this group is very heterogeneous with unpredictable biological behavior both with respect to progression to EAC as in response to endoscopic treatment. Therefore it is important to develop innovative tools that can better differentiate between high-risk HGD and HGD that behaves more indolent and is less likely to progress to EAC. We found that using a single clinical marker (age) in combination with a single genetic diversity measure as derived from the entire Barrett segment, could help to predict which patients with HGD were at risk of progression and, maybe even more important, which patients were not. This could potentially help patients and doctors to make a better informed decision about different treatment options and surveillance intervals before, during or after treatment.

The genetic markers as described in both Chapter 3 and 4 have shown proof of principle that the use of genetic markers provides a more accurate risk prediction than using clinical markers alone both in non-dysplastic BE and in HGD. Especially the high negative predictive value (99 and 100 % respectively) of these markers indicate that a group previously considered to be high risk could be reclassified into groups with low risk and high risk of progression. For those patients with HGD that fall within the low risk of progression group it may be safe to extend the surveillance interval or avoid invasive treatment.

Interestingly, MYC appears to play an important role in the progression of BE, both in the non-dysplastic stage as well further along the process of progression to EAC. MYC serves as a transcription factor that is overexpressed in more than 70% of human cancers. It acts as an oncogene by promoting proliferation as well as preventing apoptosis. MYC overexpression is therefore thought to play an important role in promoting tumor growth. As MYC genetic diversity is already an important predictor for development of HGD and EAC in non-dysplastic BE, it appears that changes in MYC clonality in the progression from non-dysplastic IM to HGD, may drive the further development to (invasive) cancer.

Furthermore, an important advantage is that implementation of the models described in chapter 3 and 4 would be relatively straightforward as it can be used on different methods to collect cytologic samples (i.e. wide area tissue sampling, nasal sponge, brush cytology) and the DNA FISH technique is routinely used in many hospitals and laboratories worldwide. Finally DNA FISH on brush cytology is less hampered by sampling errors than biopsy protocols currently used. A potential caveat for the patients described in Chapter 4 is that a percentage of these patients were treated with PDT or RFA which could confound the results. However excluding these patients from the analysis did not change the outcomes. Another possible bias for the HGD patients as described in Chapter 4 could be the low percentage of progressors to begin with, however we think this may be explained by the fact that patients did receive EMR’s for nodular lesions may have taken out those regions of the BE segment that are most at risk to develop cancer. This also explains the relatively low positive predictive value of the tool. In this tool may be best used in predicting which patients with HGD will not progress to EAC and would therefore least benefit from rigorous treatment regimens.

Finally in our last chapter we attempted to develop a novel treatment strategy for prevention of cancer development in BE in a surgical animal model for reflux esophagitis. We were able to show for the first time that inhibition of BMPs using Noggin, a naturally occurring BMP antagonist prevented development of EAC in an animal model for BE/EAC. In an earlier study we identified BMPs as playing an important role in the evolution from IM towards EAC. This was not surprising since we know the BMP pathway to be involved in the development several aggressive cancer phenotypes. Noggin is a naturally occurring, but not very specific antagonist of the BMP pathway. It is however readily available as a recombinant protein. To circumvent possible systemic side effects (BMPs play important roles throughout the body), we used Sucralfate as a carrier substance. Sucralfate has a binding site for Noggin but also binds to injured mucosal surfaces. Because of the large proportion of animals that developed dysphagia due to a stenosing tumor in the distal esophagus with resulting impaction of chow, the observed effect on the development of Noggin may have been underestimated. A possible limitation was the dose of Noggin that was given, which had been shown to be effective in decreasing inflammation and BMP pathway activity in a pilot study but may not have been high enough to prevent development of BE and/or EAC. In conclusion we think that our findings are encouraging in finding a more specific treatment for BE and EAC. Therefore it seems that in the future more selective BMP inhibitors are promising for more effective, non-treatment of EAC.