Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Epidemiology of Barrett Esophagus and Risk Factors for Progression
Prevalence and Incidence of Barrett Esophagus
Barrett esophagus (BE) is the disease in which the normal squamous lining of the distal esophagus is replaced by a metaplastic columnar cell epithelium (termed intestinal metaplasia [IM]) in response to chronic severe gastroesophageal reflux disease (GERD). Its clinical importance is as the precursor lesion and major risk factor for esophageal adenocarcinoma (EAC). The epidemiology of EAC has been investigated thoroughly because of this cancer's high lethality and recent extraordinary increase in incidence rates. Less extensive and less consistent data are available for the epidemiology of BE, which is a difficult disease to study in populations, because the majority of individuals with BE are undiagnosed as they have not undergone upper gastrointestinal endoscopy. Access to endoscopy is influenced by demographic and other factors such as the availability of public health care. BE epidemiology studies may thus suffer from population selection bias; for example, the possibility that diagnosed BE cases have a different profile to undiagnosed cases or other confounding factors.
Considering these difficulties, it is not surprising that, as shown in Table 32.1 , the reported estimates of BE prevalence vary widely. Studies designed to reduce the risk of bias include an autopsy study and a study that screened for BE in colonoscopy patients with and without heartburn ; and the estimated prevalence of BE in patients without reflux symptoms in these studies is 0.4% to 6%. A widely cited estimate of BE prevalence is from an endoscopic screening study of 1000 unselected residents who underwent upper gastrointestinal endoscopy in two communities in northern Sweden. Endoscopic findings of possible BE were present in 10.3% of individuals, with BE (IM) confirmed by histopathology in 1.6% of cases. Thus a standard estimate is that Barrett disease affects approximately 1% to 2% of Western populations.
TABLE 32.1
Summary of Studies Estimating the Prevalence of Barrett Esophagus in Different Populations
| Author | Year | N | Study Design | Study Population | Ethnicity | BE Prevalence Estimate | BE Prevalence in GERD Patients | BE Prevalence in Non-GERD Patients | Potential Source of Bias | Comments |
|---|---|---|---|---|---|---|---|---|---|---|
| Winters et al . | 1987 | 97 | Prospective, observational study | Patients with ≥1×/week symptoms of GERD | — | 12.40% | 12.40% | — | Selection bias | BE specialized epithelium was histologically confirmed in 50% of the patients determined as having BE. Columnar gastric and junctional epithelium was also regarded as “BE” if it was confirmed to be in the tubular esophagus and/or ≥5 cm proximal of the gastroesophageal junction. |
| Mann et al . | 1989 | 180 | Prospective, cohort study | GERD patients with or without reflux esophagitis | “Vast majority Caucasian” | 11.00% | 11.00% | — | Selection bias | — |
| Cameron et al. | 1990 | 959 | Prospective, observational study including autopsy data | Population-based study + prospective search of Mayo Clinic autopsy material | — | 0.34% | — | — | — | While retrospective in nature, this paper is regarded as one of the first articles to suggest that most Barrett esophagus patients are unrecognized. |
| Clark et al. | 1997 | 241 | Prospective, observational study | GERD patients undergoing upper endoscopy | — | 29.00% | 29.00% | — | Selection bias | Biased population as patients have GERD symptoms and 68% of study population males. |
| Voutilainen et al. | 2000 | 1128 | Prospective, observational study | Patients referred for upper endoscopy from primary care setting with GERD/dyspepsia and other upper-GI symptoms warranting endoscopic investigation | — | 1.00% | 4.40% | — | Selection bias | BE prevalence estimate is only applicable to patients presenting with upper GI symptoms. |
| Gerson et al. | 2002 | 110 | Prospective, observational study | “Asymptomatic” individuals presenting for screening sigmoidoscopy | 73% Caucasian, 14% African American, 10% Hispanic, 4% Pacific Islander/Asian | 25.00% | — | — | Selection bias | Only 53% of participants had no GERD symptoms; biased population as 92% males with mean age of >60 years. |
| Rex et al. | 2003 | 961 | Prospective, observational study | Patients undergoing colonoscopy | 78% Caucasian, 20.3% black, 1.6% Latin-American/Asian | 6.80% | 8.30% | 5.60% | Selection bias | Biased population as presenting for colonoscopy, 78% white and predominantly male (~60%). |
| Malfertheiner et al. | 2005 | 6215 | Prospective, observational study | Patients undergoing upper endoscopy for GERD symptoms | — | 8.39% | 8.39% | Selection bias | In patients who had erosive reflux disease prevalence of Barrett was up to 14%, whereas in patients with NERD BE prevalence was 2.3%. This study population is also biased toward GERD patients. | |
| Ronkainen et al. | 2005 | 1000 | Prospective, multicenter cohort study | Random population sample invited to participate in screening upper endoscopy | — | 1.60% | 2.30% | 1.20% | — | Regarded as a robust assessment of the true prevalence of BE in a general population. Problems exist with regard to generalizability as only two northern Swedish communities were studied. |
| Westhoff et al . | 2005 | 378 | Prospective, cohort study | GERD patients presenting for first-time upper endoscopy | 86% white | 13.20% | 13.20% | — | Selection bias | 86% white and 95% male, with median age of 56 years, all of whom had GERD. |
| Veldhuyzen van Zanten et al. | 2006 | 1040 | Prospective, cohort study | Dyspepsia patients recruited from primary care setting | 95% Caucasian, 2% black, 1% Asian, 1% Aboriginal/Metis | 2.40% | 2.40% | — | Selection bias | Analyses patients who were promptly referred for endoscopy from a primary care setting; selection bias exists for true BE population prevalence estimates, as these are “symptomatic” patients. |
| Corley et al. | 2008 | 4205 | Observational study | Evaluation of BE diagnosis among all patients with a membership in an integrated health services delivery organization (Kaiser Permanente, Northern California; n ~3.3 million) | Study population included Caucasians, African Americans, Asians and Hispanics. Specific percentages are not provided, but it is noted that the endoscopy volume/rate did not vary between demographic groups | 0.13% | — | — | Potential diagnostic bias as incidence of a new diagnosis of BE increased over time | This study documents that, in a fairly unbiased population, the frequency of a diagnosis of BE is rare. It also demonstrates a linear increase in the diagnosed prevalence of disease. It also finds that white non-Hispanic males aged 61 to 70 have the highest risk of BE. Note: Prevalence defined as number of active members with a new diagnosis of BE divided by number of person-years in the membership for the respective year. |
| Abrams et al. | 2008 | 2100 | Retrospective, cross-sectional study | Patients undergoing upper endoscopy for any indication | 37.7% Caucasian, 11.8% African American, 22.2% Hispanic, 28.3% unknown | 4.40% | — | — | Selection bias | Main indications for endoscopy were GERD symptoms (23.5%), nonreflux dyspepsia (34.7%), and overt/occult GI bleeding (and/or anemia; 16.5%). Consequently, this is a highly preselected cohort of patients. Main focus of the study was to study racial disparities in BE incidence rates, limited by that over 1/4 of the study population had an “unknown” ethnicity. |
| Zagari et al. | 2008 | 1033 | Prospective cohort study | Population sample invited to participate in completion of reflux questionnaire and screening upper endoscopy | — | 1.30% | 1.50% | 1.00% | — | This study was performed similarly to the Scandinavian study and provides a fairly robust estimate of the true prevalence of BE in the Italian population. Also shows that frequent reflux symptoms are a risk factor for BE, but almost half (46.2%) of patients with BE report no reflux symptoms whatsoever. |
| Fan et al. | 2009 | 4457 | Retrospective study | Racial distribution in GERD group was 68% Caucasian, 19% African American, 11% Hispanic, 2% other. In the non-GERD group 65% Caucasian, 21% African American, 11% Hispanic and 3% other. | 1.72% | 4.39% | 1.56% | Selection bias | Selection bias as all patients had to be referred for endoscopy; thus not generalizable to the general population. | |
| Hayeck et al. | 2010 | — | Computer simulation | Computer simulation model with SEER data verification | — | 5.57% | — | — | Computer modeling aligned to available (published) data | A computer-modeling study that aimed to estimate the population prevalence of BE by adjusting the model inputs based on published data and aligning results to EAC incidence rates in the SEER data. |
| Shiota et al. | 2015 | 453,147 | Meta-analysis of observational studies | Various: Inclusion criteria (and subgroup analyses) were performed according to whether population sampling was population-based, from health checkups, screening, or for symptomatic patients | Asian as categorized according to International Agency for Research on Cancer: Eastern, South Eastern, South Central and Western Asia | 1.30% | 1.40% | 0.70% | Included studies own selection biases, methodologic limitations inherent to meta-analyses | This meta-analysis mainly included studies from Eastern Asian countries (38/51), and therefore the prevalence estimates cannot be generalized to the rest of Asia. Indicates that BE may be more frequent in Asian countries than originally estimated. |
BE , Barrett esophagus; EAC , esophageal adenocarcinoma; GERD , gastroesophageal reflux disease; GI , gastrointestinal; SEER , Surveillance, Epidemiology and End Results Reporting database.
Some data indicate that BE affects white non-Hispanic Westerners more than Asian or Hispanic people, but a recent meta-analysis including 51 studies with over greater than 450,000 patients from Asia suggests that the histologically proven pooled-prevalence of BE is 1.3% (95% confidence interval [CI], 0.7 to 2.2%; 28 studies, and 298,850 subjects) and thus comparable to Western estimates. Allowing for the limitations of pooled analyses, such as varying study time periods, BE definitions, and study populations (the meta-analysis only included one population-based study with 1029 participants), this suggests that BE is not uncommon in Asian (particularly Eastern Asian) countries.
Although puzzlingly much lower than the incidence rate increase for EAC, studies from several countries have reported an increase in the incidence of BE. This seems to be a true increase as only some of this rise in incidence can be attributed to increased use of endoscopy.
Risk Factors for Barrett Esophagus
Factors Associated With Increased Risk of Barrett Esophagus
Sex and Age
Most epidemiologic data show a 2 : 1 male preponderance among diagnosed cases of BE. BE patients are also typically slightly older than non-BE GERD patients at between 50 and 65 years at the time of diagnosis. BE is rare in pediatric populations, being confirmed by histologic presence of IM in approximately 0.12% of patients less than 20 years of age referred for upper endoscopy for any indication. There are no reports of IM containing BE in a child under the age of 5 years, and within children/adolescent study cohorts, the risk of BE is increased after the age of 12 years, supporting a timeline of some years for BE development.
Gastroesophageal Reflux
Chronic GERD is the main cause of BE, with the risk and length of Barrett disease correlating with the amount and duration of reflux exposure in the distal esophagus. The exact mechanism by which GERD triggers Barrett formation remains elusive, but some have postulated that the refluxate causes an erosive episode during which there is a denuding of the normal squamous epithelium, allowing it to be subsequently repopulated by columnar cells. Where these cells are derived from remains a matter of ongoing scientific debate.
Clinical studies have confirmed that severe GERD is present in patients with BE and that these patients have more esophageal dysmotility and lower distal esophageal sphincter pressures compared with patients with erosive or nonerosive esophagitis. Equally, other studies have shown that BE patients have long exposure to gastric content with very low pH levels (pH < 2.0 to 3.0) and a high frequency of hiatal hernia (76% in BE vs. 36% in GERD patients). In addition to gastric acid, duodenal juices in the refluxate are thought to be important contributors to the formation and propagation of BE. The combined exposure of esophageal cells to bile acids and low pH results in DNA damage and oxidative stress, which in turn propagates BE formation and progression. Similarly, in vitro treatment of esophageal cells with both bile and acid leads to the expression of intestinal and/or columnar cell markers.
While GERD can affect up to 20% of Western populations, only 5% to 10% of patients with GERD develop BE. A higher frequency (e.g., ≥weekly) of GERD symptoms is associated with a greatly (10 times) increased risk of BE in population control studies. A weaker association between symptoms and BE is found in endoscopy studies of patients with typical symptoms of GERD. It thus remains unclear if BE patients have considerably more frequent or only slightly more frequent symptoms of reflux compared with non-BE GERD patients. Frequency and number of years of reflux symptoms (i.e., chronicity of reflux) are better predictors of BE. Patients with BE may report that their GERD symptoms have improved in recent years, which is postulated to be related to BE development and perhaps reduced esophageal sensitivity.
Obesity
There is a strong reproducible association between obesity and the risk for EAC, indicating that patients with a body mass index (BMI) ≥ 30 kg/m 2 bear a 2 to 3 times higher risk of developing this malignancy, but the associations with BE have been less consistent. A recent meta-analysis concluded that there was no significant association with BMI and BE when comparing BE and GERD patients. However, when using the pooled estimates from three studies comparing BE patients with general population controls, a significant association between BMI and risk for BE could be determined (pooled odds ratio [OR], 1.02 per kg/m 2 ; 95% confidence interval [CI], 1.01 to 1.04; I 2 = 0%). Other data from population-based studies suggest that there is no more than a 50% increase in risk for BE with BMIs 30 kg/m 2 or greater.
Central visceral adiposity may be a more important risk factor for BE formation than BMI itself. A case-control study found that visceral adipose tissue was 1.5 times higher in patients with BE compared with controls and population-based studies have identified a significant association between BE and either increasing waist circumference or waist-to-hip ratio. Interestingly, when controlling for waist-to-hip ratio, the association between BMI and BE has been shown to be almost completely attenuated, suggesting that the obesity effect is mediated through visceral adiposity.
A mechanistic explanation for the relationship with central adiposity includes the unproven postulates that increased abdominal visceral fat elevates intraabdominal pressure or intragastric pressure. More likely is that the pressure gradient across the antireflux barrier is increased, resulting in hiatus hernia and GERD, and the metabolic and endocrine activity of visceral adipose tissue in those with central obesity (who interestingly are more likely to be male) may also be important. Central obesity alters expression levels of obesity-related and proinflammatory cytokines such as leptin, adiponectin, tumor necrosis factor (TNF)-alpha, interleukin (IL)-6, and insulin-like growth factor.
As leptin is upregulated in obesity and increases proliferation of EAC cells in vitro, the association between serum leptin levels and BE has been examined. Two studies found an increased risk of BE if leptin levels were elevated, but the association was stronger in men in one study and stronger in women in the other study. Another study did not show an association with leptin, but it did show a significant inverse association of elevated adiponectin levels and BE risk compared with colon screening controls (adjusted OR, 0.42; 95% CI, 0.22 to 0.80). Thus while recent studies on the interaction of central obesity and EAC have elucidated the molecular pathways by which adipocytokines contribute to EAC formation and propagation, more data from larger studies are required to further clarify the role for leptin and other obesity-related cytokines in BE development.
Smoking and Alcohol Consumption
Most, but not all, population-based studies have documented an approximately twofold increase in risk of BE in patients who have ever smoked. Contrary to some earlier findings, recent data suggest that a greater number of pack-years smoked may be associated with a greater risk of BE, but this effect may plateau at approximately 20 pack-years. Furthermore, a recent population-based study showed that smoking and GERD may exert a synergistic effect on disease development and progression. There are no convincing data supporting an increased risk of BE due to alcohol consumption. On the contrary, an inverse association of wine consumption and risk of BE has been reported and a pooled analysis of Barrett cases compared with controls found a moderate reduction of risk with consumption of wine (OR, 0.71; 95% CI, 0.60 to 1.00), although without a dose-response relationship.
You're Reading a Preview
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here