Statistics from Altmetric.com
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.
See article on page 867
Several studies have reported significant positive associations between colorectal cancer (CRC) and Barrett’s oesophagus.1-5 In the initial report, 29 colon tumours, 10 of them malignant, were found in 65 patients with Barrett’s oesophagus.1 The authors suggested that the probability of detecting a similar number of malignant colorectal lesions in an age matched sample of the general population was less than one in a million. The early studies, however, were often hampered by their choice of a “control” group or the complete absence of one. Bias may also have arisen from the discovery of Barrett’s oesophagus during the investigation of patients with gastrointestinal symptoms referable to their presentation of a CRC. In addition, the existence of asymptomatic but prevalent cases of both colon cancer and of Barrett’s oesophagus is now more widely recognised than it was a decade ago.6 Despite reservations about the validity of these studies, the finding that colon tumours seemed to be more common in patients with oesophageal intestinal metaplasia than in patients with functional bowel symptoms or with positive faecal occult blood tests is interesting on two accounts.
Firstly, the implications for a colorectal cancer screening programme of a positive association between the index cancer and Barrett’s oesophagus are considerable. The prevalence of Barrett’s oesophagus in the general population has been reported to be as high as 1%7 and initial studies suggested a prevalence in patients with Barrett’s oesophagus of 8–45% for colorectal adenomas and 7–15% for colorectal carcinomas.3 If an association of this degree was definitively established then colorectal screening might be targeted specifically at individuals with confirmed Barrett’s oesophagus.
Secondly, if Barrett’s oesophagus and colorectal neoplasia are associated, what is its aetiological basis and is there a biological explanation for the apparent relation? Shared environmental factors or genetic predisposition might result in an association but have not been clearly demonstrated. Nevertheless, although there is a lack of experimental data linking the two conditions, they do have some interesting features in common. Increased dietary fat intake and reduced fruit and vegetable consumption has been associated with an increased risk of CRC and possibly with Barrett’s oesophagus.6 In addition there are genetic alterations that are common to both conditions: abnormalities such as p53 gene mutation and allelic loss of chromosome 18q are associated with the progression of the adenoma–carcinoma sequence in human colon and more recently have been identified in Barrett’s mucosa but it is unclear whether genetic changes are a cause or consequence of intestinal metaplasia in the oesophagus.8 ,9 Further insights into a putative genetic association may be gained by studying Carney syndrome, an autosomal dominant multi-system syndrome of which both Barrett’s oesophagus and colonic polyps can be features.10
It was clear from the early studies that the prevalence of colorectal neoplasia in patients with Barrett’s oesophagus was high; if colonoscoped about a third would be found to have either an adenoma or a cancer.3 Indeed the largest study using the renowned Mayo clinic database found that 14% of 175 records with a Barrett’s oesophagus diagnosis also contained a colorectal cancer diagnosis.4 What was less clear was how much higher these figures were compared with the general population of similar age who were not being colonoscoped. Two groups have compared the figures for patients with Barrett’s oesophagus with those from four small US studies of colonoscopy screening of the general population. Howden and Hornung concluded that although there was no increased risk of colorectal adenoma, there was a fivefold increase in risk of cancer which was statistically significant.3 In contrast Poormanet al reviewed all the prospective surveys and using the same colonoscopy screening data concluded that the increased risk of cancer was small and not sufficient for screening.11
An alternative epidemiological approach is to make the assumption that the development of oesophageal adenocarcinoma is a marker for previous Barrett’s oesophagus as well as its most significant consequence and then to use cancer registry data to examine the coincidence of the two cancers. This has now been done in two studies. The first used US cancer registry data to perform a nested case-control study in which the cases were patients with oesophageal adenocarcinoma and the controls patients with oesophageal squamous cell cancer, for which there is no association with colon cancer.5 Compared with patients with squamous cell cancer there was a 44% increased risk of colorectal cancer developing before or after the oesophageal adenocarcinoma in men (odds ratio (OR) 1.44, 95% confidence interval (CI) 1.03–1.72) but a reduced risk in women (OR 0.39, 95% CI 0.19–0.78). In contrast there was no increase in risks of prostate or breast cancer for the adenocarcinoma cases, indicating some specificity to these associations. The authors offer no explanation for the sex difference or for examining it other than the propensity of epidemiologists to break relations down by sex!
In the second study (see page 819) Lagergren and Nyren have used the comprehensive Swedish cancer registry to identify all patients with colon cancer registered during 1958–92. Only 11 patients subsequently developed oesophageal adenocarcinoma, giving an overall increased risk of 20% (standardised incidence ratio 1.2, 95% CI 0.6–2.1) with similar risks in men and women. What is striking is that despite having a cohort of 118 000 elderly patients accumulating over half a million person years, the expected number of oesophageal adenocarcinomas was only 9.5 emphasising how rare this cancer has been in Sweden. On the basis of recent oesophageal cancer mortality rates in the UK the expected number for a cohort of this size would have been over four times higher. This suggests that the prevalence of Barrett’s oesophagus in Sweden over this period was low compared with the UK. It is notable that over the same period there was a substantial increase in oesophageal cancer mortality in the UK (60% in men and 35% in women) with an increasing proportion due to adenocarcinoma, whereas in Sweden mortality increased only 11% in men and fell 40% in women.12 Although the rarity of this cancer in Sweden has reduced the power of their study to exclude small increases in risk there are no good reasons for believing that their findings are unrepresentative of other European countries.
So what is the “take home” message for the jobbing gastroenterologist? Current data are not sufficient to rule out a modest, less than twofold increased risk of colorectal cancer in patients with Barrett’s oesophagus or oesophageal adenocarcinoma and vice versa. This might be explained by either genetic predisposition or common environmental risk factors. None the less an association of this magnitude is not sufficient by itself to warrant screening and is unlikely to be a source of fruitful research.