Article Text

Download PDFPDF

Observer variation in the diagnosis of superficial oesophageal adenocarcinoma: another spanner in the works?
  1. D Alderson
  1. University Division of Surgery, Level 7, Bristol Royal Infirmary, Marlborough Street, Bristol BS2 8HW, UK; derek.alderson{at}

Statistics from

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.

The distinction between high grade dysplasia and intramucosal carcinoma is not easily made by expert pathologists, with little improvement even after agreeing on uniform criteria for the use of

The rapidly rising incidence of oesophageal adenocarcinoma in the Western world has stimulated considerable experimental and clinical research in recent years. After a long period of uncertainty, it is now generally held that the majority, if not all, of these cancers arise on a background of intestinal metaplasia (Barrett’s oesophagus), secondary to chronic gastro-oesophageal reflux. While there has been much debate on the risk of cancer development, it is again accepted that in a proportion of patients, this metaplastic epithelium becomes unstable, leading to a sequential progression through dysplasia to carcinoma. Considerable effort has gone into characterising the changes which seem to occur at the molecular level 1.

In this issue, Ormsby and colleagues2 suggest that it may be important to distinguish high grade dysplasia (HGD) from intramucosal carcinoma (IMC) in order to reach a decision about endoscopic surveillance and how this might influence the nature and timing of subsequent therapy [see page 671]. They show that the distinction between HGD and IMC is not easily made by expert pathologists, with little improvement even after agreeing on uniform criteria for the use of these terms. The biggest problem seems to be where the glandular architecture is complex, making the recognition of landmarks (basement membrane and muscularis mucosa) difficult. This seems to be a problem in at least 20% of cases where the pathologist is being specifically asked to separate HGD from IMC. The fact that they were working from resection specimens rather than small endoscopic biopsies probably means that this is an underestimate of the difficulties which most pathologists face.

Their paper raises a number of issues. Is it really important to separate HGD from IMC? Do the observations cast any doubt on the scientific validity of experimental studies which have tried to identify “biomarkers” of HGD? Is their own work open to criticism in terms of methodology and data interpretation?

It is easiest to deal with these questions in reverse order.

The study was carefully performed and the conclusion that experienced pathologists have difficulty distinguishing HGD from IMC is sound. Inevitably, this must mean that there are studies in which groups of “HGD patients” must include some patients with IMC. In fact, as far as experimental work is concerned, these always focus on markers of LGD (low grade dysplasia), HGD, or invasive cancer and none have sought to separate HGD from IMC 1. As such, the current study probably does not invalidate our present state of knowledge regarding the molecular changes which seem to characterise the neoplastic process.

The real issue here is just how relevant is this distinction between HGD and IMC. The authors state that some groups “advocate oesophagectomy for a biopsy diagnosis of high grade dysplasia, others prefer to wait until intramucosal adenocarcinoma is detected”. This is open to criticism. The two references chosen to support the latter point of view come from the same group of authors associated with the University of Washington in Seattle. Interestingly, the Seattle group have always sought to point out the difficulties in interpreting HGD with gross architectural distortion and have stated that in such cases “invasive adenocarcinoma cannot be excluded”.3 As such, the assertion made by Ormsby et al that the distinction between HGD and IMC is important for day to day practice is less of an issue than they imply. Their work however undoubtedly indicates one of the reasons why we see such wide variation in the apparent behaviour of HGD and the risk of finding a carcinoma in an oesophagectomy specimen undertaken for HGD.

At the end of the day, management of patients with HGD largely reflects the way in which clinicians choose to interpret the available evidence and adhere to clinical protocols. Those who would wish to spare every patient an unnecessary oesophagectomy must have an underlying belief that a diagnosis of HGD is correct and the risk of invasive carcinoma is negligible. The pursuit of a surveillance policy for HGD must satisfy a number of criteria.

The endoscopic examination itself must be meticulous so that small raised lesions or ulcerated areas which are associated with an increased cancer risk are not overlooked.4,5 The biopsy protocol must be systematic, involving multiple biopsies of adequate size to minimise the risk of sampling error. The clinician must understand the difference in risk of a cancer being present when HGD is diagnosed at a prevalent rather than surveillance biopsy. Case series advocating long term non-surgical management of HGD on the basis of a low incidence of cancer development need to be considered carefully. Schnell and colleagues6 reported a five year cumulative cancer incidence of only 9%, a figure well below that described by other groups using aggressive systematic biopsy protocols.4,7 Perhaps the clue to the low cancer incidence reported by Schnell et al was the high frequency with which dysplasia of any severity was identified. The inevitable inference is that many of the patients labelled as HGD would have been classified as LGD elsewhere. The problem of interobserver variation is once again raised.8,9

In the UK’s healthcare system, clinicians frequently work in less than ideal circumstances. The inevitable consequence of an inadequate number of small biopsies will be the finding of an unsuspected cancer in specimens resected for HGD. A number of series published in the 1990s indicate that this risk is still approximately 40% 10 and unless clinicians can reassure themselves that they can deliver an aggressive surveillance protocol allied to consensus histopathology reporting with high levels of agreement, they should be recommending surgical resection for healthy patients with HGD.

The distinction between high grade dysplasia and intramucosal carcinoma is not easily made by expert pathologists, with little improvement even after agreeing on uniform criteria for the use of


Linked Articles