The Risk of Barrett’s Esophagus Transforming to Esophageal Cancer

Introduction

Barrett’s esophagus (BE) refers to the mucosal transition from normal squamous mucosa to a genetically unstable specialized columnar-type intestinal metaplasia in the esophagus and is the only known precursor to esophageal adenocarcinoma (EAC).^1,2 The incidence of EAC has been increasing in the western hemisphere.^3,4 Currently, the standard practice in managing patients with BE is histopathologic evaluation of the BE epithelium which determines surveillance intervals and eradication therapy. In this treatise, we review the risk of BE transforming to esophageal cancer and the latest developments in risk assessment of BE progression to cancer.

Do All EAC Arise From BE?

Before we quantify the risk of progression in BE, it is important to understand whether the pathway to EAC is via BE. It is well-accepted that many EACs are diagnosed without detectable BE at the time of endoscopy. Observations in 2 large cohorts of patients with EAC showed a significant difference in prognosis of EAC patients with and without detectable BE. ⁵ These data raise the question whether all EACs arise from the well-accepted BE pathway. Now we have insights from the fields of molecular biology as well as computational biology to suggest that the majority of EACs follow the BE pathway. Using a multiomics approach that included single cell sequencing of specimens of normal esophagus, squamocolumnar junction (from controls and from BE), normal stomach, BE, and EAC, the investigators showed that the cell of origin of all BE and EAC came from gastric cardia, irrespective of whether visible BE was seen at the time of endoscopy ⁶ suggesting a common origin of BE and EAC. Another study used computational tools to model the prevalence of GERD and BE along with progression rates to EAC and showed a high degree of concordance between the predicted and observed number of EACs in the Surveillance, Epidemiology, and End Results (SEER) database. ⁷ Based on this computational model, the investigators concluded that 90% of all EACs arise from BE. The above combination of molecular biology and computational data strongly supports that BE is the predominant pathway to EAC. Therefore, we can be confident that an accurate description of BE progression risks will be an effective way to understand the risk of EAC in the general population.^6,7

Population-Based Risk of BE Progression From National and Other Large Registries

The progression rate of BE to EAC can vary from one study to another depending on the region of origin, definition of BE, and variations in data capture. In these risk estimates, we report on the combined incidence of high-grade dysplasia (HGD) and EAC. We decided to group HGD with EAC because HGD is a high-risk lesion and an actionable diagnosis. ⁸ The risks of progression to HGD and EAC are described separately in Table 1. The overall risk of progression to HGD or EAC was estimated to be 0.26%/year among 11 028 BE patients (defined by the presence of intestinal metaplasia) over a median follow-up of 5.2 years in the Danish Cancer and Pathology Registry, ⁹ whereas it was estimated to be 0.22%/year among 8522 BE patients (with and without intestinal metaplasia) over a mean follow-up of 7 years in the Northern Ireland Cancer Registry. ¹⁰ Both of these registries were national registries and excluded prevalent cases (HGD or EAC diagnosed within 1 year of BE diagnosis) to provide a true estimate of incident risk. Nationwide registries in the United States are difficult to create but an analysis of the National Veterans Cohort that included 29 536 patients (97% male) with BE from 121 facilities found the incidence rate of combined HGD/EAC to be 0.48%/year after excluding prevalent cases. ¹¹ Another US study that included high-quality data from the Rochester Epidemiology project from Olmsted County, MN reported a progression rate of 0.79%/year over a mean follow-up of 7 years. ¹² An important reason for higher progression rates in the Olmsted County study was that censoring of prevalent cases was restricted to within 6 months of index endoscopy (unlike the Danish and Irish registries that excluded prevalent cases within 12 months of index endoscopy). A multi-institutional study that included both veteran and non-veteran patients found the risk of progression to HGD/EAC to be 0.63%/year. ¹³ To summarize, the risk of progression from BE to HGD/EAC is variable across different cohorts but is less than 1% per year.^9,14,15

OPEN IN VIEWER

Table 1.

Risk of BE Transforming to HGD and EAC.

Study	N	Follow up (y)	Sex ratio (M/F)	Age (y)	BE definition	Annual risk to HGD (%)	Annual risk to EAC (%)	Annual risk to combined HGD/EAC (%)
Hvid-Jensen et al 9	11 028	Median: 5.2 (IQR 2.8-8.9)	2/1	Median: 62.7 (IQR 52.3-73.0)	IM	0.19	0.12	0.26
Bhat et al 10	8522	Mean: 7	1.4/1	Mean: 61	With and without IM	NR	0.13	0.22
Shakhatreh et al 11	29 536	Mean: 4.9 (SD 1.8)	31/1	Mean: 61.8 (SD 10.6)	IM	NR	0.35	0.48
Jung et al 12	487	Median: 7	2.5/1	Mean: 63 (SD 14)	IM	0.5%	0.29	0.79
Hamade et al 13	1883	Median: 6.4	5.4/1	Mean: 58 (SD 17.4)	IM	0.47	0.16	0.67

Note. All studies except that by Jung et al, excluded cases diagnosed within 1 year of index endoscopy to calculate true incidence risk. Jung et al, excluded prevalent cases within 6 months of the index endoscopy. We selected the above studies to highlight differences in risk based on geography and patient cohorts (veterans vs non-veterans). This table is not intended to be an exhaustive list of all published studies. BE = Barrett’s esophagus; HGD = high-grade dysplasia; EAC = esophageal adenocarcinoma; IQR = interquartile range; IM = intestinal metaplasia; NR = not reported.

How Does Exclusion of Prevalent Cases Affect Risk Estimates?

This topic was briefly discussed earlier but is important to rediscuss because it can affect the risk estimates of BE transformation. The precise natural history of BE is unknown. The duration of transformation from BE to EAC is unclear and difficult to calculate because the timing of BE initiation remains unclear. Multiple studies suggest that a significant majority of EACs may be prevalent rather than incident. In other words, a significant majority of EACs are diagnosed either on the index endoscopy or within 1 year of the index endoscopy. This can impact the risk calculations of incident EAC in BE. Revisiting the studies discussed above, by including EAC cases diagnosed within 1 year of the index endoscopy, the risk of EAC increased by approximately 50%, from 0.18% to 0.35%. ¹¹ Similarly, in the Danish registry that included the entire Danish population of 5.4 million, the risk of EAC increased by approximately 60%, from 0.12% to 0.29%. ⁹ Similar observations were made in the Irish registry. ¹⁰ These differences in the risk of incident versus prevalent EAC in BE are relevant for counseling BE patients and also emphasize the need for careful examination of the BE segment at the time of endoscopy and perhaps, a short interval endoscopy within 1 year of BE diagnosis.

Risk of Progression in Patients Without Specialized Columnar Epithelium that is, Intestinal Metaplasia

The US guidelines require the presence of goblet cells in the columnar mucosa that is, intestinal metaplasia lining the tubular esophagus to diagnose BE, whereas the British and Asian guidelines exclude this requirement. ² So does lack of intestinal metaplasia in the columnar mucosa affect the risk of progression to EAC? In the Northern Ireland Cancer Registry study, ¹⁰ lack of intestinal metaplasia on the index endoscopy was associated with a lower risk of progression to EAC when compared with its presence (0.07%/year vs 0.38%/year). However, other studies have suggested similar risk of progression with or without intestinal metaplasia.^16,17 Additionally, the extent and frequency of DNA abnormalities were found to be similar in columnar mucosa with and without intestinal metaplasia. ¹⁸ Debate about the progression risks in BE with and without intestinal metaplasia rages on. Due to conflicting data regarding risk of progression without intestinal metaplasia and concern for sampling error with regards to detection of goblet cells in columnar esophageal mucosa, especially with short segments, many experts recommend to repeat endoscopy at 12 to 18 months with more sampling as that may increase the yield of detecting intestinal metaplasia by >20%. ¹⁹

Risk of Progression Based on BE Length

The risk of transformation to EAC in BE less than 1 cm is being discussed elsewhere but we briefly discuss the issue of BE length and risk of transformation. Historically, the relationship of BE length to risk of transformation has been in question ¹⁰ but more recent publications from various groups demonstrate that BE length is a strong predictor of EAC risk.^20,21 In a large US multicenter study of ~2500 patients, the risk of BE progression increased by 12% for every 1 cm increase in length. ²¹ An analysis of the United Kingdom Barrett’s Oesophagus Registry controlled for multiple variables and showed BE length to be an independent predictor of progression risk over a mean duration of 5.7 years. ²² We performed a multi-center international study to show that the risk of progression in long-segment BE was higher than that in short-segment BE (0.91% vs 0.29%; P < .001). Taken together, these data suggest that the risk of BE transforming to cancer does depend on length. Based on these data, the British Society of Gastroenterology recommends more aggressive surveillance for BE lengths greater than 10 cm at a center of expertise. ²³

Whether individuals with BE < 1 cm in length are at risk of progression to HGD/EAC has been difficult to evaluate because of frequent confusion with intestinal metaplasia at the gastroesophageal junction (ie, intestinal metaplasia of gastric cardia). In the earlier discussed study from Olmsted County, MN, ¹² none of the 86 patients with intestinal metaplasia at the gastroesophageal junction (which could, in theory, be confused with ultrashort segment BE) developed HGD or EAC over a median 8 years of follow up. Similarly, none of the 167 patients with intestinal metaplasia at the gastroesophageal junction developed HGD or EAC in a multicenter cohort study with a median of 5.9 years of follow up. ²⁴ The concepts around cancer risk in BE lengths <1 cm are being discussed elsewhere.

Risk of BE Progression Based on the Degree of Dysplasia

The earliest risk estimates that we reported were for all BE patients, irrespective of the degree of dysplasia. The histologic grade of BE is clearly an important predictor of risk. The risk of progression has been variable in BE with low-grade dysplasia (LGD) and this in part is due to high inter-observer variability even among expert pathologists in histologic diagnosis of LGD.^25,26 The annual risk of progression from LGD to HGD/EAC in large epidemiological studies was ~1.5%,^9,10 but this was likely an underestimation. In a randomized controlled trial comparing ablation to surveillance in patients with LGD at 9 European sites, the annual progression rate from LGD to HGD/EAC in the surveillance group was 26.5%. ²⁷ A subsequent systematic review and meta-analysis, that included the above trial, showed a cumulative progression rate of 12.6% over follow up of 84 months. ²⁸ The differences in the risk of LGD progression rates between epidemiological studies and clinical trials can be attributed to the histologic interpretation of LGD. The interobserver agreement of LGD among pathologists is poor thus raising doubts about the diagnosis of LGD, when read by a single pathologist. ²⁹ In a remarkable study when LGD was confirmed by at least 2 pathologists, the risk of progression to HGD or EAC was 9.1% annually. ³⁰ When LGD was confirmed by 3 pathologists, the annual rate of progression increased to up to 20%. ³¹

HGD is the penultimate precursor to EAC and is an actionable diagnosis with a high rate of progression to EAC. The annual progression rate from HGD to EAC has been variable in the literature as well, ranging from 16% to 60%. ³² but a recent pooled analysis estimated the annual progression rate from HGD to EAC to be 28.6%. ³³

Based on above data, confirmed LGD and HGD (which is routinely confirmed by 2 pathologists) are important predictors of BE transforming to cancer.

Prediction Tools of Transformation Risk in Barrett’s Esophagus

In the era of precision medicine, applying general estimates of progression risk based on histopathologic evaluation to individual patients is fraught with inaccuracies since every patient is unique with different clinical and disease characteristics. Therefore, there is huge push that assessment of progression risk should be individualized. Furthermore, progression from NDBE to dysplasia and EAC is associated with a variety of genetical alterations and chromosomal abnormalities, even before dysplasia is visible endoscopically and detectable histologically (field effect). Moreover, most dysplastic areas are flat and patchy, rendering it difficult to recognize under white light endoscopy despite significant improvement toward high definitions lenses. Likewise, sampling error can occur as standard mucosal biopsies may miss dysplasia (false negative) and sampling is limited to small areas by following the Seattle protocol. The ideal risk assessment tool should include clinical, endoscopic, and histologic characteristics in addition to risk assessment from predictive tests to categorize individuals into risk groups and adjust therapy and surveillance periods accordingly.

Conclusion

The risk of BE transforming to EAC is less than 0.5% per year and that to HGD or EAC is less than 1% per year. Despite huge advances in biomarker analysis, histologic grade remains the most clinically used predictor of risk to guide surveillance versus ablation. LGD is an important marker for progression and in particular, LGD confirmed by multiple pathologists carries an annual progression risk of up to 20% and approaches the high-risk profile of HGD. Clinical tools to predict risk such as the PIB have been developed but will need further validation. Although many different molecular assays have shown promise, more studies are needed before clinical application. Ultimately, a predictive model incorporating clinical, endoscopic, histologic, and biomarkers in addition to molecular and genetic analysis will be needed to optimally predict progression risk in each individual patient with BE in the era of precision medicine.