Antireflux Surgery, Barrett Esophagus, and Adenocarcinoma

It is our responsibility as scientists......to teach how doubt is not to be feared, but welcomed and discussed.” This statement was made by Richard Feynman of the United States, one of the last century's most brilliant theoretical physicists and original thinkers. And the article by Chang et al in this issue of the journal passes the responsibility test because it challenges what we, as surgeons, would like to believe about the treatment of Barrett esophagus.

Even if a little simplistic, the current wisdom regarding Barrett esophagus suggests that it is caused by gastroesophageal reflux, and with continuing reflux, a proportion of patients with the condition develop an adenocarcinoma because of the Barrett's mucosa that is present. That this is an important topic is underscored by the rapidly rising incidence in the Western world of adenocarcinoma affecting the esophagus, and, a priori, it seems reasonable to believe that if we can stop the reflux, we can stop the development of cancer. As antireflux surgery is the only treatment available that stops reflux, it should therefore be the most effective way of stopping the development of cancer. Chang et al document a number of studies where the conclusions suggest just such a scenario. However, using epidemiologic methods, they demonstrate that such studies are almost certainly unconsciously biased; and if the studies used are limited to those where some attempt has been made to obtain comparability of groups, then there is no statistical difference in outcome between those patients with Barrett esophagus who are treated medically and those who are treated surgically.

Two important points arise from the paper. First, when uncontrolled data are used, it can lead easily to erroneous conclusions. This is particularly true when the conclusions are ones that accord with the beliefs of those conducting the study (which is why, as surgeons and scientists, we have to strive to keep doubting). Second, at the present time, the aim of antireflux surgery should be to cure reflux and its symptoms and not to be seen as a means of preventing cancer of the esophagus.

Nevertheless, despite the authors' findings from their review of the literature, is it still possible that surgery might prevent cancer, and do so more effectively, than any other therapy? Certainly, it remains possible. First, as patients undergoing surgery are usually regarded as being at the most severe end of the reflux spectrum, the fact that there were not more cancers in the surgical group compared with the medical group might be interpreted as being a positive outcome (a greater number of cancers were seen in patients treated surgically in the population-based study reported by Ye et al¹ in 2001). Probably the only way to overcome this difficulty would be to undertake a large randomized controlled trial; and according to the estimations of Chang et al, this would have to be a very large trial indeed and, therefore, is unlikely ever to be carried out. Second, it may be that, in any study of the effectiveness of surgery, it will be necessary to control for whether the surgery is actually working. This is because, in many studies of patients who have a Barrett esophagus and who have undergone antireflux surgery, a substantial minority of patients develop acid reflux again.^2,3 Furthermore, this recurrence of reflux is not always symptomatic.⁴ It is therefore possible that providing surgery stops reflux over the long term; it may prevent progression of Barrett's mucosa to cancer. If this was found to be so, then 24-hour pH surveillance following antireflux surgery would be much more important than is currently regarded to be the case.

However, the study by Chang et al contains a finding that raises another and rather more fundamental doubt regarding the problem of Barrett esophagus. Is it possible that Barrett esophagus has nothing particularly to do with esophageal cancer at all? The problem with Barrett esophagus being seen as an independent factor in the development of cancer of the esophagus is the near impossibility of separating it from the reflux that causes it in the first place. Best current evidence tells us that there is an association between reflux and adenocarcinoma, and this association is strong in patients with severe symptoms of reflux over a long period.⁵ It also tells us that patients who develop Barrett esophagus are at the worst end of the reflux spectrum.^6,7

Now consider the following:

1. Barrett's mucosa alone clearly does not cause adenocarcinoma of the esophagus, since the great majority of patients with the condition do not develop a carcinoma of the esophagus. That is, even if the presence of Barrett's mucosa is important, other factors must exist for cancer to develop.

2. In whites, the incidence/prevalence ratios, male to female, for reflux disease, are approximately 1:1,⁸ and for the development of Barrett's mucosa approximately 1:1,⁹ and yet the ratio for adenocarcinoma of the esophagus is more like 10:1.¹⁰

3. In any surgical series of esophagectomy for adenocarcinoma, the incidence of demonstrated Barrett's mucosa is only of the order of half of the patients.^11,12 This is usually dismissed as being because the metaplastic mucosa has been replaced by cancer. However, it is equally possible that there was no metaplastic mucosa in the esophagus in the first place.

4. It was once thought that the longer the segment of Barrett's mucosa, the greater the risk for the development of adenocarcinoma. Strong doubt was raised about this association, however.¹³

5. Now the paper by Chang et al shows that significant regression of columnar-lined esophagus occurred in the surgical group compared with the medical group of patients, without significantly changing the incidence of cancer.

It is possible that adenocarcinomas do not arise from columnar cells but from some form of stem cell, and so perhaps the true proximate cause of the cancer is not Barrett's mucosa at all, but chronic inflammation, a time-honored pathogenetic factor for cancer. Reflux is an important cause for such inflammation but may not explain the steep rise in incidence of cancer in the past 30 years. It is certainly easier to explain the 5 points above if there is some other factor at work that has yet to be identified. And so, in the spirit that I think Richard Feynman would approve, the question needs to be asked: “Is Barrett's mucosa really so important in the development of adenocarcinoma, other than as a marker of severe reflux disease?” and does our persistent concentration on it divert our attention from finding an as yet unidentified, but much more important, cause of adenocarcinoma involving the esophagus?