- Top of page
- Patients and methods
- Acknowledgements and Disclosure
Surveillance endoscopy and random biopsy are a common practice performed to detect SIM, dysplasia and early cancer. One of the main difficulties with BE surveillance is the reliability of diagnosis across this spectrum. There is a marked discordance between the endoscopic and histologic diagnosis of BE,2 whilst at the cancer end of the spectrum, diagnosis is more alarmingly problematic. Areas of HGD and microscopic carcinoma in BE may only occupy a fraction of the total surface area of Barrett's mucosa, leading to false-negative sampling errors,3 and biopsy by the Seattle protocol will only sample 4% of the Barrett's segment.4
In view of the above dilemmas, new endoscopic techniques for Barrett's evaluation have been developed, including the combined use of chromoendoscopy to stain the Barrett's segment,5 and high resolution magnification endoscopy. Guelrud et al.6 applied the technique called enhanced magnification endoscopy (EME) with instillation of acetic acid rather than the colouring agents, methylene blue or indigo carmine, and produced a classification of mucosal surface pit patterns that predicted the histologic findings. This technique offers the possibility of sensitive targeting of a hierarchy of pit patterns within the Barrett's segment at endoscopy to confirm SIM, dysplasia and early cancer. Similarly, in a recent study, Toyoda et al. showed that EME is useful for detection of intestinal metaplasia in distal oesophagus and oesophagogastric junction.7
We examined the detection of SIM, dysplasia and adenocarcinoma by EME in patients with BE, and subsequently compared the diagnostic yield of our findings with those found at previous routine endoscopy.
The descriptive criterion for pit pattern classification means that there is a great potential for subjective interpretation. We therefore evaluated the inter- and intra-observer agreement in the assessment of pit patterns, and recognition of areas more probable to harbour intestinal metaplasia or dysplasia in BE using the EME technique.
- Top of page
- Patients and methods
- Acknowledgements and Disclosure
EME allowed clear demarcation of the Barrett's segment with high definition of mucosal pit patterns, which were selectively targeted according to a hierarchy of pit patterns (I–V). This produced a remarkably high prevalence of dysplasia (10%) and cancer (5%); pooled data from large surveillance programmes in the US estimate the combined incidence and dysplasia to be 1%/year.9
Confirmation of SIM is a prerequisite to the confirmation of BE. Excluding dysplasia and cancer, SIM rather than only CLE was found in 40 of 47 (85%) of patients, compared with 48 of 60 (80%) with their previous conventional surveillance endoscopy (Table 2). However EME found an additional four LGD, one HGD and three cancers over a short interval.
A plethora of new techniques have been applied to overcome the limitations of surveillance by random biopsy, including spectroscopy, optical coherence tomography, confocal microscopy and narrow band imaging.10 Most of these techniques require expensive instruments and are technically demanding. Narrow band imaging combined with magnification endoscopy seems to be a promising new endoscopic imaging technique, aiming to enhance the mucosal surface contrast without the use of dyes. Yet, future studies are awaited to demonstrate its efficacy in the surveillance of patients with BE.10
A more accessible technical advancement has been the combination of chromoendoscopy and high resolution magnification endoscopy. Various dyes, including Lugol's iodine, methylene blue, and indigo carmine, have been used.11–15 We have employed acetic acid to enhance the appearance of the Barrett's segment, because of its simplicity to use, safety, and long track record in the field of colposcopy. Acetic acid results in reversible alteration of the proteins in the cell, and more specifically of the bundling of cytokeratins, modifying their optical properties. A reversible alteration of the nucleoprotein wrapping after the application of acetic acid modifies the refraction index of the nucleus (the aceto-white reaction). The columnar epithelium becomes swollen and the villi and pit pattern becoming more apparent.16
Guelrud produced a classification of different mucosal surface patterns observed 6: I, round pits (representing gastric fundic columnar epithelium); II, reticular; III, villous; and IV, ridged or cerebriform. The yields for detecting SIM according to endoscopic patterns were 0%, 11%, 87%, and 100%, respectively. This technique offers the possibility of sensitive targeting of a hierarchy of pit patterns within the Barrett's segment at endoscopy to confirm intestinal metaplasia, and we have achieved comparable high yields of at least SIM.
Short segments of BE bring their own-specific problems in diagnosis and subsequent management decisions. There is a poor correlation between the endoscopists suspicion of BE and histologic confirmation.17 SSBE did not appear to be particularly benign when EME was employed in our study: 74% of patients with SSBE had SIM, and in addition there were two cancers, one with 2-cm BE and one ultra-short (<1 cm). This contrasts with a prevalent attitude that SSBE is less worthy of surveillance, indeed a recent survey of the UK gastroenterologists found that 62% considered that BE <3 cm does not warrant surveillance.18 Other groups have found a much lower incidence of SIM, with only a quarter of <3 cm (SSBE) yielding SIM in Rudolph et al's study.19 In another study17 of 146 patients with endoscopic BE, the yield of SIM fell steadily with the declining length of columnar-lined segment, from >65% (>5 cm) to 50% (3–5 cm) to 25% (<3 cm).
Evidence for cancer risk in SSBE is also equivocal, with findings that increasing length of BE has a consistent but non-significant trend for an increased cancer risk and the presence of aneuploidy.19 The proposal that only long (e.g. 8 cm) segment BE is followed up would seem to be risky,20 and the American Gastroenterology Association (AGA) workshop on BE concluded that no discrimination between SSBE and LSBE should be made in determining the surveillance strategy.21
Fifteen patients (24%) had a histological upgrade, including eight of particular clinical relevance (Table 2). There are two notable features about these ‘significant’ upgrades. The first is that they occur within a short mean surveillance interval (7 months). The screening interval for the three cancers was just 1, 6 and 3 months, with previous histology of HGD, SIM and SIM, respectively.
Our study raises the possibility that significant findings which may be amenable to life-saving intervention, such as HGD or carcinoma in situ, are missed by sampling error when conventional surveillance endoscopy is employed.3 In patients who undergo oesophagectomy for HGD, cancer may be present in over a third of the resected specimens subsequently reviewed by the pathologist.22 However, the commonly held risk for malignant transformation is 1% per year,23 so this may be because of the finding of three cancers by chance skewing the results.
The second significant finding with histological upgrades is that they all occurred in men. BE is more common in white men, among whom the incidence of oesophageal adenocarcinoma has more than quadrupled over the past few decades.24 This confirms previous findings that BE is more common in males, and they are at greater risk of subsequent adenocarcinoma24, 25 Males were more likely than females to have LSBE (65% vs. 35%), but this did not account for the increased risk.
A classification system is useful, when it can be reproducible in clinical practice, so that its application has a reliable outcome. In our study, the accuracy of predicting the pit patterns ranged from 63% to 86% between observers. The mean kappa values for inter- and intra-observer agreement in the diagnosis of pit patterns were 0.571 (0.041) (range 0.491–0.636) and 0.709 (0.038) (range 0.597–0.841), respectively, among endoscopists experienced in endoscopy. The mean kappa values for inter- and intra-observer agreement in predicting histology were 0.548 (range 0.440–0.703), and 0.701 (range 0.546–0.815), respectively, which can be regarded as moderate to substantial.
The technique of EME has several limitations. It is more time-consuming than conventional endoscopy, which may have led us to offer sedation prior to gastroscopy to more patients than we would normally. The acetic acid staining wears off after 4–5 min, so familiarity with the technique and co-ordination with the assisting endoscopy nurse is required. After the first few biopsies, a combination of bleeding and reflux obscures the view and obliterates the pit pattern. This requires frequent washing with water flush via the instrument channel and sometimes the use of antimotility agents like hyoscine. Although the biopsies are targeted, the mean number of biopsies taken was no different to conventional surveillance.
Although our study appears to show dramatic results, it has several limitations and caveats. The comparison between endoscopies is not like-for-like. The previous endoscopy was at a mean interval of 6 months. This seems to be too short for the natural history of BE to account for the change, but a synchronous controlled or cross-over study would control for confounding factors, such as any particular enthusiasm on the part of the endoscopist using EME. The high yield of dysplasia may reflect the total contribution of a specialist Barrett's clinic taking referrals and performing confirmatory endoscopies rather than a true advantage of the technique over the current surveillance techniques.
Endoscopy in BE is notoriously unreliable, from measurement of the length of Barrett's, to sampling error, and the pathologist's interpretation. The changes could have occurred by chance, as seen by the Munich group,2 although the direction of change is towards a histological upgrade rather than downgrade.
In conclusion, EME allows clear resolution of epithelial pit patterns within BE, and allows targeted biopsy with a high yield of SIM and dysplasia, even in SSBE. This may lead to a significant upgrade in histology. Additionally, the recognition of pit patterns and histology prediction in BE using the EME technique is reproducible. This indicates that a uniform description of the diverse manifestations of Barrett's mucosa after acetic acid instillation is feasible and practical in clinical practice. Therefore, EME with acetic acid has the appeal of offering a pragmatic, reliable surveillance tool that could be taken up by non-specialist or research-orientated units, the so-called ‘real world’. This technique warrants further validation in a large-controlled study.