Oesophageal cancer and gastro-oesophageal reflux: what is the relationship?

T he direct relation between reflux and adenocarcinoma of the oesophagus has recently been addressed in several well designed population based epidemiological studies, using different study designs. The studies are all in agreement that reflux is associated with an increased risk of oesophageal adenocarcinoma in a seemingly dose dependent manner. The specific role of Barrett’s oesophagus for this association was not addressed in these studies but the presence of Barrett’s oesophagus has been regarded as a potential link in the causal pathway. More speculatively, based on an interpretation of the results from a large population based study in Sweden, we hypothesised that the relation between reflux and oesophageal adenocarcinoma might be independent of Barrett’s oesophagus. This hypothesis was based on two findings of our study. Firstly, among patients with a history of severe and longstanding reflux symptoms, the association with adenocarcinoma was of similar strength as that of the association previously reported between the occurrence of Barrett’s oesophagus and the risk of oesophageal adenocarcinoma in large scale studies. Secondly, when we compared our case patients with oesophageal adenocarcinoma who had a verified Barrett’s oesophagus with those who had no such mucosa detected during endoscopy or surgery, we found that the association between reflux and oesophageal adenocarcinoma was equally strong. The possibility of tumour overgrowth covering the Barrett might explain this finding, however. Nevertheless, the hypothesis deserves further attention. In this issue of Gut, SolaymaniDodaran and colleagues address the role of Barrett’s oesophagus compared with the role of reflux per se in the aetiology of oesophageal adenocarcinoma in an interesting investigation [see page 1070]. They analysed the strength of the association with adenocarcinoma in three cohorts of patients, including a cohort of patients with Barrett’s oesophagus, a cohort of patients with oesophagitis, and third cohort of patients representing simple reflux. As a reference, a comparison cohort of patients without any of these conditions was selected. The point estimate of the relative risk of oesophageal cancer was 10 in the Barrett’s cohort while the corresponding risk estimates were approximately 2 in the two cohorts representing reflux without Barrett’s metaplasia. The authors conclude that their findings do not support the fact that gastro-oesophageal reflux in itself predisposes to oesophageal cancer. This interpretation is not necessarily true, however. There are some problems that deserve consideration in their study. Firstly, the study included oesophageal cancers of any histological type in the analysis (that is, it did not specifically address the risk of adenocarcinoma of the oesophagus, the only type of oesophageal cancer that has been linked with reflux). Secondly, the risk of chance errors was high as the study, like most other prospective studies of this relatively rare disease, was underpowered. The limited number of 43 cases of oesophageal cancer were evaluated, and among these cancers the oesophageal squamous cell carcinomas and any other types were included. Thirdly, selection bias in the comparisons between the cohorts might explain the differences between the three cohorts reported in the study. The cohort representing Barrett’s oesophagus is likely to consist of patients with the most severe and longstanding reflux disease compared with the two comparison cohorts representing oesophagitis and simple reflux. From previous studies, we know that the relationship between reflux and oesophageal adenocarcinoma is strongly dose dependent both with regard to severity and duration of reflux (the relative risk of this cancer was 43.5 among patients with long term and severe symptoms in our Swedish investigation). We also know that patients who develop Barrett’s oesophagus usually have had reflux symptoms for long periods of time, typically for decades. The other two cohorts in the study by Solaymani-Dodaran and colleagues are likely to encompass those with, on average, less severe and less longstanding reflux. Any differences between reflux cohorts in the study are likely to be strongly affected by such selection bias. Hence the study cannot reliably verify whether or not Barrett’s oesophagus is a mandatory step in the development of adenocarcinoma of the oesophagus. In summary, based on the current knowledge, there seems to be a causal link between reflux and oesophageal adenocarcinoma. Barrett’s oesophagus is very likely an important, and possibly necessary, link in the causal pathway from reflux to oesophageal adenocarcinoma but more studies are needed before the true answer to this important and interesting question can be established.