Article Text


Association between colon cancer and adenocarcinoma of the oesophagus
  1. R MARIC,
  2. K K CHENG
  1. Department of Public Health and Epidemiology
  2. University of Birmingham, Edgbaston
  3. Birmingham B15 2TT, UK
    2. P NYREN
    1. Division of Surgery
    2. Karolinska Institutet
    3. Danderyd Hospital
    4. S-182 88 Danderyd, Sweden

      Statistics from

      Editor,—Recently, Lagergren and Nyrén (Gut1999;44:819–821) concluded that results from a population based cohort study in Sweden did not support a common link between colon cancer and oesophageal adenocarcinoma. However, there is consistent evidence that low intake of dietary fibre is associated with both diseases. In an analysis of 13 case control studies involving more than 5000 colorectal cancer cases, Howe and colleagues reported an inverse association between fibre intake and risk of colorectal cancer in 12 of the 13 studies, and an odds ratio of 0.53 (95% confidence interval 0.47 to 0.61) for the highest quintile of fibre intake compared with the lowest, in the combined analysis.1 Similarly, four case control studies have reported a significant inverse association between fibre intake and adenocarcinoma of the oesophagus and gastric cardia (table1).2-5 In contrast, two studies which included cases of squamous cell carcinoma found no significant link between fibre intake and squamous cell carcinoma of the oesophagus.2 5

      Table 1

      Dietary fibre intake and adenocarcinoma of the oesophagus and gastric cardia

      Clearly, the dramatic increase in the incidence of adenocarcinoma of the oesophagus in the USA and parts of Europe over past decades cannot be explained by secular trends in dietary fibre consumption. A more plausible explanation links increased rates of the disease to increases in the prevalence of obesity.6 This view is supported by evidence from observational studies that suggests that both overweight and symptomatic gastro-oesophageal reflux are linked to increased risk of oesophageal adenocarcinoma.4 7 Possible mechanisms for the observed protective effect of dietary fibre include the mechanical cleaning effect of the lower oesophageal mucosa, increased motility of potential carcinogens across the gastro-oesophageal junction, and the antioxidant effect of micronutrients in fruits and vegetables.3

      The lack of a significant link between colorectal cancer and oesophageal cancer in the Lagergren and Nyrén study is not surprising, as the average year of entry to the cohort study was 1977 and median follow up was 2.1 years. Thus, a substantial proportion of the accumulated person years relates to a time period when the Swedish population was at a very low risk of developing oesophageal adenocarcinoma. As the authors indicated, the limited power of the study meant that they were unable to exclude the possibility of a doubled risk.

      However, despite recent increases in incidence, the lifetime risk, even if elevated, of developing oesophageal adenocarcinoma after a diagnosis of colon cancer remains small, because of the late onset of colon cancer. Furthermore, case control studies are likely to continue to be the most efficient type of observational study design for the investigation of possible common links between these two diseases.



      Editor,—The influence of dietary fibre on the risk of adenocarcinoma of the oesophagus and gastric cardia is certainly interesting, but further and larger studies are needed before a link between the two can be confirmed. Although several case control studies have revealed an association between colon cancer and fibre intake, others have failed to identify such a link.1-1 Hence, the suggested link between fibre intake and both colon cancer and oesophageal cancer is debatable.

      We agree that changes in dietary fibre consumption cannot explain the increasing incidence of oesophageal adenocarcinoma. The increasing prevalence of obesity is a possible reason for this rise, but some seemingly incongruent observations need to be reconciled before this hypothesis can be verified.1-2 The apparently sudden deflection of the incidence curve for oesophageal adenocarcinoma,1-3 the rapidity of the increase,1-4 and the noticeable (6–8 fold) male predominance,1-4 do not entirely support this interpretation.

      The primary hypothesis of our population based cohort study was not that colon cancer would subsequently develop into oesophageal cancer, but rather that there might be a common underlying link between the occurrence to these two tumours, independent of the time latency between their development. Hence, the individual follow up latency after colon cancer diagnosis was of minor importance. Therefore, it would seem reasonable to assume similar latencies between exposure to the critical underlying factor—for example, insufficient dietary intake of fibre, and the development of oesophageal or colonic adenocarcinoma. Therefore, as long as selection or ascertainment biases are deemed to be small, the time period that follows immediately after diagnosis of colon cancer is the most informative. The total number of person years was more critical, and we were able to follow up more than 500 000 person years in our study. The rarity of oesophageal adenocarcinoma is a problem in any study of the aetiology of this tumour in any country, particularly if the studied exposure is relatively uncommon. This problem explains our limited power to exclude a weak association. Nevertheless, we were able to identify more than 100 000 people with verified colon cancer and to follow them for subsequent cancer development; this is a substantial number of exposed people. We agree that case control studies are generally more efficient than cohort studies when rare outcomes are to be investigated. However, in the case of our register based retrospective cohort study, a case control approach would not entail any advantage, as our cohort contained all individuals exposed to colon cancer in Sweden between 1958 and 1992, and all individuals among them who developed oesophageal adenocarcinoma during the same period. A case control study conducted in Sweden during this period would, at best, include the same number of exposed oesophageal adenocarcinoma cases as in our cohort study. Thus, the problem with low statistical power is not owing to study design, but that the study base (all residents of Sweden 1958–1992) was too small to generate a sufficient number of individuals with the combination of colon cancer and oesophageal adenocarcinoma.


      1. 1-1.
      2. 1-2.
      3. 1-3.
      4. 1-4.
      View Abstract

      Request permissions

      If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.