What is the incidence of development of Barrett's esophagus?
T. Schnell, S.J. Sontag (Hines)
Barrett's esophagus is thought to be an acquired complication of gastroesophageal reflux (GER) disease with increased risk for development of esophageal adenocarcinoma. Once identified, patients with Barrett's esophagus may be enrolled in endoscopic surveillance programs -- a practice not universally accepted. The true prevalence and incidence of Barrett's esophagus and Barrett's adenocarcinoma have not been clearly defined, but almost certainly depend on the population being studied, the criteria necessitating biopsy, and the aggressiveness of the endoscopist. Thus, the varying populations studied, the indications for endoscopy, and the different definitions of what qualifies as Barrett's esophagus has resulted in a literature of uninterpretable and widely divergent prevalence and incidence rates.
Prevalence based on the population being studied
In a retrospective study of all patient records in Olmsted County, Minnesota with the biopsy-proven diagnosis of Barrett's esophagus, Cameron estimated the prevalence to be 22.6 cases per 100,000. In a prospective study by the same investigators, the prevalence in consecutive post mortem examinations was 376 cases per 100,000. This study, which defined Barrett's as the presence of at least 2 cm of intestinal epithelium, demonstrated that only one in 20 patients with Barrett's esophagus is being diagnosed, and that 95% of all patients with Barrett's in the general population are going unrecognized .
Prevalence based on the criteria necessitating biopsy
When the minimum criterion to establish the diagnosis of Barrett's esophagus includes the presence of 2 cm of red mucosa (seen endoscopically) with specialized intestinal epithelium (seen on histology), then the prevalence varies from 1% to 10% [2, 3].
Prevalence based on the aggressiveness of the endoscopist
When the 2 cm segment of Barrett's esophagus is not used as a criterion, and biopsies are taken from consecutive patients even when the squamocolumnar junction appears normal, the prevalence increases to 18%, which probably represents a more accurate reflection of disease prevalence .
It is clear, therefore, that the wide variation in prevalence is due to a combination of patient selection, differences in study populations, differences in biopsy methods, or a combination of all three factors.
For the same reasons, the rates of cancer in Barrett's esophagus also varies widely, ranging from one in 52 to one in 441 patient years (a mean of about one in 100 patient years). These data indicate a 100 times increased risk over the general population of developing esophageal cancer [5-8]. Specifically, cancer develops each year for every 100 patient-years of follow- up , or 1% of adult patients with Barrett's esophagus will develop adenocarcinomas each year.
In 1981, we began a large scale comprehensive screening program using upper GI endoscopy as an initial screening procedure for patients with gastrointestinal symptoms. At that time, our goal was to determine the prevalence and incidence of Barrett's esophagus and adenocarcinoma in the population with uncomplicated GER symptoms. Since GER symptoms often bring patients to endoscopic examination, knowledge of the prevalence and incidence rates may help to formulate an overall general endoscopic policy in regards to the feasibility of performing endoscopy in such a group. Any attempt to report the true prevalence and incidence of Barrett's esophagus must include consecutive patients with biopsy of the squamocolumnar junction, even when it appears normal, since the presence of specialized epithelium of any length in the tubular esophagus increases the risk for the development of adenocarcinoma .
Subjects were enrolled from all walk-in and outpatient clinics at Hines Veterans Affairs Hospital in Hines, IL. Ambulatory outpatients received endoscopy as the initial diagnostic procedure for the workup of GER. Patients were considered symptomatic for GER if they had:
a) typical symptoms (e.g. heartburn with or without chest pain, regurgitation, and/or pyrosis) as the primary complaints;
b) typical symptoms were not the primary complaint but were elicited during the history;
c) antacids and/or antireflux medications (H2-blockers) were used on a regular basis for GER symptoms.
Patients with GER complications, such as previous antireflux surgery, weight loss with dysphagia, gastrointestinal bleeding, or need of hospitalization, were considered to have complicated GER and were not considered as part of the healthy screening population. Occasional dysphagia spontaneously resolving without weight loss, bleeding or need of hospitalization was considered as uncomplicated GER for this populations.
Routine esophagogastroduodenoscopy and standard forceps biopsies were performed by one of two endoscopists using predefined criteria. Grading of esophageal mucosal inflammation was based on endoscopic appearance. The endoscopic findings were based on the methods of Hetzel and were graded as follows:
Normal: no visible abnormalities. If only erythema was present, the mucosa was considered normal (grade I).
Esophagitis: any break in the mucosa with or without exudate, as seen during endoscopy, which included erosions and/or ulcerations (grades II, III and IV).
Barrett's: intestinal-type of metaplastic epithelium obtained by biopsy from any level of the tubular esophagus.
Esophagitis present: esophagitis with or without Barrett's columnar epithelium.
Esophagitis absent: normal mucosa or Barrett's columnar epithelium without esophagitis.
Hiatal hernia: the presence of gastric folds seen during endoscopy to extend at least 2 cm above the diaphragmatic hiatus during quiet respirations. Hiatal hernia not seen on initial endoscopy was considered to be present if, during subsequent endoscopies, the definition of hiatal hernia was met.
All ambulatory outpatients who received endoscopy as the initial diagnostic procedure for the work-up of GER in the outpatient endoscopy unit at Hines VA Hospital between January 1, 1981 and July 1, 1994 who had biopsies taken from the squamocolumnar junction during endoscopy.
Patients had uncomplicated GER if they had:
- typical symptoms (heartburn, regurgitation, pyrosis) as a primary complaint;
- typical symptoms that were not a primary complaint but elicited during the history;
- if they used antacids or antireflux medications on a regular basis for GER symptoms.
Patients were considered to have asthma if they had discrete attacks of wheezing and a FEV, of less than 20% of predicted on pulmonary function studies or a 20% reduction in FEV, after methacholine bronchoprovocation.
Patients were considered in the NSAID group if they were enrolled in the rheumatology clinic, were taking chronic NSAID medications for osteo- or rheumatoid arthritis, and were part of our prospective study of consecutive rheumatology patients who underwent endoscopy regardless of gastrointestinal symptoms.
Atypical chest pain
Patients were considered to have atypical chest pain if they were referred for endoscopy from the cardiology clinic or cardiology department after having a negative work up for coronary artery disease.
Duodenal ulcer symptoms
Patients were considered in the ulcer symptoms group if classical ulcer symptoms were the primary presenting complaint.
Evaluable prevalent case
A patient diagnosed with Barrett's esophagus on the first endoscopy and biopsy or within the first twelve months of the initial biopsy.
Evaluable incident case
A patient diagnosed with Barrett's esophagus on a follow-up endoscopy at least twelve months after the initial squamocolumnar junction biopsy was negative for BE.
The prevalence of Barrett's esophagus in each of the five groups is shown in Figure 1. The prevalence of Barrett's esophagus in the GER and asthma groups was 20.9% and 18.4%, respectively. Every group had patients with Barrett's.
The incidence of Barrett's esophagus over a fourteen year period is shown in Figure 2. In the GER group, 3.5% of the total population of GER patients with Barrett's esophagus was detected at least twelve months after the initial negative endoscopy, indicating that the incidence of Barrett's esophagus is extremely low. Considering the asthma incidence rate of 7.3% over the fourteen years, only 0.5% of patients each year would be expected to develop Barrett's.
Figure 1. Prevalence of Barrett's esophagus.
The prevalence and incidence of adenocarcinoma in patients with Barrett's esophagus is shown in Figure 3. The prevalence of adenocarcinoma in all patients with Barrett's is shown in the first bar at year 1 to be 4.8%. Subsequent bars represent the incidence rate. Over a 10 year period, the difference between the total percentage of Barrett's patients with adenocarcinoma (6.4%) and the prevalence rate (4.8%) is merely 1.6%. Thus, the incidence of adenocarcinoma in all patients with Barrett's esophagus is 1.6% over a 10-year period or less than two cases each year for every 1000 patients with Barrett's. This incidence of adenocarcinoma is substantially lower than the usually reported 1% per year in patients with Barrett's esophagus.
Figure 2. Incidence of Barrett's esophagus over 14 years.
Figure 3. Prevalence and incidence of adeno-
carcinoma in patients
with Barrett's esophagus
(n = 1034).
The prevalence of Barrett's esophagus in patients with uncomplicated reflux symptoms is approximately 21%. This prevalence rate is similar for asthmatics. The range of prevalence rates varies between 10% and 21%, and appears to depend on the indication for endoscopic referral. No group of patients is free of Barrett's esophagus, and asthmatics have a similar prevalence of Barrett's as patients with GER symptoms.
The incidence of Barrett's esophagus is extremely low and may even approach zero. Indeed, if due to technical error, the initial biopsy failed to retrieve intestinal epithelium, then a prevalent Barrett's would later be diagnosed as incident Barrett's. If technical error causes a falsely high incident rate with missed prevalent cases, then the incidence is even lower than that we reported. Barrett's esophagus, therefore, may develop at a younger age before the first endoscopy is performed and the prevalence in children is similar to the prevalence in adults.
Finally, although adenocarcinoma does develop from Barrett's esophagus, the incidence, once benign Barrett's is recognized, is lower than previously reported. After eliminating the initial 4.8% prevalence rate, the risk of developing adenocarcinoma if benign Barrett's is initially diagnosed is extremely low - less than two cases each year for every 1000 patients with Barrett's.
1. Cameron AJ, Zinsmeister AR, Ballard DJ, Carney JA. Prevalence of columnar-lined (Barrett's) esophagus. Comparison of population-based clinical and autopsy findings. Gastroenterology 1990;99:918-922.
5. Haggitt RC, Dean PJ. Adenocarcinoma in Barrett's epithelium. In: Spechler SJ, Goyal RK, eds. Barrett's esophagus: pathophysiology, diagnosis, and management. New York: Elsevier Science Publishing Co., Inc., 1985:153-166.