Endoscopy 2012; 44(04): 362-365
DOI: 10.1055/s-0031-1291741
Endoscopy Essentials
© Georg Thieme Verlag KG Stuttgart · New York

Reflux and Barrett’s disease. Can we stop surveillance after 2011?

R. Bisschops
Department of Gastroenterology, University Hospital Leuven, Leuven, Belgium
› Author Affiliations
Further Information

Publication History

Publication Date:
27 February 2012 (eFirst)

Incidence of adenocarcinoma among patients with Barrett’s esophagus (Hvid-Jensen et al., N Engl J Med 2011 [1])
Risk factors for progression of low-grade dysplasia in patients with Barrett’s esophagus (Wani et al., Gastroenterology 2011 [2])

Hvid-Jensen et al. reported on a large population-based cohort study involving mostly Danish patients with Barrett’s esophagus between 1992 and 2009 [1]. Data on 11 028 patients were collected with a median follow-up period of 5.2 years. In total, 131 cases of adenocarcinoma were newly diagnosed within 1 year of the index endoscopy and were regarded as missed cases and not incorporated in the incidence calculation. During the subsequent years, an additional 66 new cases of adenocarcinoma were registered, yielding an incidence rate of 0.12 %. This is much lower than the assumed and previously reported risk of 0.5 %, the latter being used to substantiate surveillance guidelines. Notwithstanding this low incidence, the relative risk of developing cancer was 11.3 for Barrett’s esophagus – significantly elevated compared with the general population. However, in view of their findings the authors wonder whether surveillance of patients with Barrett’s esophagus is still warranted.

In the paper from the USA [2], new data on the risk for progression of low grade dysplasia (LGD) in Barrett’s esophagus are provided. Incident cases were defined as patients diagnosed with LGD at least 1 year after the first endoscopy that diagnosed Barrett’s esophagus. In total, 210 patients with an average Barrett’s esophagus length of 4.4 cm (113 prevalent cases) were included in the cohort and followed up for a mean of 6.2 years. In total 6 patients developed an adenocarcinoma (0.44 % incidence rate/year) and 21 developed high grade dysplasia (HGD) (1.6 % incidence rate/year).

This important study attempts to answer some of the shortcomings in the current literature on the progression of LGD. First, the authors address the difference in risk between prevalent and incident cases of LGD. It is conceivable that if the duration of LGD is unknown, the risk of progression in prevalent cases may be increased. However, there was no difference in the progression rate to HGD or adenocarcinoma, with 12 patients progressing in both prevalent and incident cases (2.33 % and 2.69 %, respectively). Also, the fact that LGD was repeatedly detected during follow-up did not increase the risk for progression. Second, the authors assessed whether the length of Barrett’s esophagus was a risk factor, but could not find a significant difference between Barrett’s esophagus < 3 cm and > 3 cm, with a progression rate of 1.7 % and 2.07 % per year, respectively. Third, the authors analyzed the extent of LGD as a risk factor, which could be assessed in 123 cases. Although the incidence rate of 3.01 % was higher in patients with multifocal LGD (n = 50) compared with unifocal LGD (n = 73, 1.13 %), this was not statistically significant. Finally, 41.9 % of the biopsies were reviewed by two central expert pathologists. The overall inter-observer agreement between these two expert pathologists was disappointingly low, with a kappa value of 0.14 % for the diagnosis of LGD. Only in 55.6 % was there an overall agreement. The study also assessed the difference in progression rate in cases of concordance between the first pathologist and one or two central expert pathologists; there was no difference in progression rate between patients with or without confirmation of LGD.

The data from both of these studies are important and controversial. In the era where treatment of early neoplasia in Barrett’s esophagus becomes less invasive, organ saving, and safer in terms of quality of life, morbidity and mortality with a combination of endoscopic resection and subsequent radiofrequency ablation (RFA) [3], the Danish authors question the need and socio-economical benefit of persistent surveillance of patients with Barrett’s esophagus. However, some methodological issues need to be addressed. Although the authors claim that the Danish registry is almost complete (98 %), this study remains a retrospective analysis of prospectively collected data and it is not clear how this compares with other prospectively collected data: for example, previous data [4] reported a combined risk of 5.6 % for adenocarcinoma and HGD in a cohort of 618 patients with endoscopically and histologically confirmed Barrett’s esophagus during a 4.1-year follow-up. The major confounder in the present study was the lack of endoscopic data. Data collection was completely based on pathology registration of intestinal metaplasia and adenocarcinoma. As such, we do not know anything about the disease extent, which has been reported to be a risk factor for developing neoplasia. It is not even clear, for example, whether intestinal metaplasia at the cardia was excluded in the Danish cohort. Additional risk factors (e. g. body mass index, smoking, alcohol use) are not identified. It is probably unlikely that the Danish data can be extrapolated to the world population. It is conceivable that as well as a possible selection bias in previous studies, a higher risk population in other countries may account for the higher reported incidence of neoplasia. Recent data from the USA show that the overall esophageal adenocarcinoma incidence increased from 3.6 per million in 1973 to 25.6 per million in 2006 [5]. Clearly, this is not reflected in this Danish cohort study. A very large Dutch cohort study including more than 16 000 patients showed an incidence of 0.5 % for developing adenocarcinoma [6].

The US study on the progression of LGD is as controversial as the Danish contribution. According to the authors, it adds further evidence that the diagnosis of LGD probably does not increase the risk of progression compared with non-dysplastic Barrett’s esophagus. This finding is important because RFA is now widely available as a relatively safe and efficacious treatment of dysplastic and non-dysplastic Barrett’s esophagus [7]. LGD is regarded by some as a straight indication for RFA; however, the procedure is not without side effects and long term outcome data are limited to 5-year follow-up [7] [8] [9]. Conversely, the US paper contradicts larger epidemiological studies, which indicate an increased risk in patients with LGD. In the Danish cohort [1], 0.5 % of the patients without LGD and 2.3 % of the patients with LGD developed cancer after the first year of follow-up. Also the incidence of adenocarcinoma among patients without LGD was 0.1 % vs. 0.51 % (relative risk of 4.8 for LGD). In an even larger Dutch cohort study including 16 365 patients with Barrett’s esophagus, 4132 patients with LGD progressed to HGD or adenocarcinoma with an incidence rate of 0.77 % per year, which was significantly higher than in patients without baseline dysplasia [6]. Additionally, other smaller studies have demonstrated an (at times alarmingly) increased risk of progression when the diagnosis made by community-based pathologists was confirmed by tertiary expert pathologists. In a retrospective study including 43 patients with LGD, Skacel et al. reported that there was no increased risk when the primary diagnosis was not confirmed. When two or three expert pathologists agreed on the diagnosis of LGD, the risk was significantly increased for developing more advanced neoplasia [10]. Similarly, in a recent study by Curvers et al., pathology slides of 147 patients with LGD were reviewed by expert pathologists; 85 % of LGD was down-staged to non-dysplastic. However, when the diagnosis was confirmed, the risk of progression to HGD or adenocarcinoma was 85 % after 109 months [11]. Finally, another Dutch trial published in 2011, identified LGD as a significant risk factor [12], together with a disease duration of more than 10 years, length of the Barrett’s mucosa, and persistent esophagitis. In a multicenter prospective cohort of 713 patients with Barrett’s esophagus, the absence of any of these risk factors in non-dysplastic Barrett’s esophagus implied a low risk (< 1 %), whereas in patients with LGD and at least one other risk factor the risk for progression was 18  – 40 %.

So where do we stand? The truth probably lies somewhere in between. Certainly in view of the previous literature, one cannot conclude to just abandon Barrett’s surveillance as a whole. This debate is socio-economic and ethical in nature rather than scientific. Even the Danish study confirms the increased risk of neoplasia in Barrett’s esophagus but also confronts us with the limitation of our daily clinical practice, with a low yield in detecting patients with neoplasia. One thing that can be concluded from the US trial and previous studies is that the gold standard upon which our daily clinical practice is based is not so shiny after all. As a matter of fact, in the LGD study, the overall agreement between all pathologists was 55 %, which we would consider unacceptable for any automated laboratory test. Another problem is that adenocarcinoma is often diagnosed without a previous history of Barrett’s esophagus [1] [2] [4]. In the US study, 249 patients – or 11 % of the patients in the database – were excluded because of HGD/adenocarcinoma. This does not mean that Barrett’s esophagus is not associated with an increased cancer risk, but it does indicate that despite a rising incidence of this condition [5], detection occurs too late. Rather than re-opening the debate on whether to survey those patients with a proven increased oncogenic risk, the current studies merely substantiate the need for a proper risk stratification and diagnostic tool in order to detect more patients with Barrett’s disease before they develop cancer and to select those who would benefit from surveillance or even early preventive treatment. Pursuing better histological diagnostic criteria should definitely be one goal. Additionally, new diagnostic developments, such as the Cytosponge test [13], look very promising, particularly if they could be combined with the identification and testing of molecular markers [14]. Further development and identification of patient-related and molecular risk factor will hopefully enable Barrett’s surveillance to become more cost-effective. Following a principle of caution, we might be rather more worried by the Dutch data indicating an increased risk of progression in expert-confirmed LGD than being reassured by these new data.