The Prevalence of Barrett’s Esophagus in Asian Countries: A Systematic Review and Meta-analysis

Abstract

Background & Aims

The prevalence and risk factors of Barrett’s esophagus (BE) in Asian countries are unclear. Studies report a wide range of BE prevalence in Asian countries. We conducted a systematic review and meta-analysis to examine the prevalence of BE, its temporal changes and risk factors in Asian countries.

Methods

Two investigators performed independent literature searches using PubMed and EMBASE databases, and subsequent data abstraction for studies had to meet several set inclusion and exclusion criteria. Pooled BE prevalence was calculated using a random-effect model. Estimates of relative risk for possible risk or protective factors were also calculated.

Results

A total of 51 studies (N = 453,147), mainly from Eastern Asia, were included. The pooled prevalence of endoscopic BE was 7.8% (95% CI = 5.0 – 12.1; 23 studies) and of histologically confirmed BE was 1.3% (95% CI = 0.7 – 2.2; 28 studies). Most (82.1%) of histologic BE was short-segment BE (<3 cm). There was a trend toward an increase in prevalence of BE over time from 1991 to 2014, especially in Eastern Asian countries. Within BE cohorts, pooled prevalence of low-grade dysplasia, high-grade dysplasia and esophageal adenocarcinoma was 6.9%, 3.0% and 2.0%, respectively. Reflux symptoms, male sex, hiatus hernia, and smoking were associated with a significantly increased risk of histologic BE in patients with BE compared with patients without BE. However, half of the patients with histologic BE did not have reflux symptoms.

Conclusions

BE is not uncommon in Asian countries and seems to share similar risk factors and potential for neoplastic progression to that seen in Western countries.

Introduction

Barrett’s esophagus (BE) is commonly defined as a change of any length in the distal esophageal epithelium that can be recognized as columnar-type mucosa at endoscopy and is confirmed to have intestinal metaplasia by biopsy of the tubular esophagus.¹ Other BE definitions do not require the presence of specialized intestinal metaplasia in histology.^{2, 3} BE is a precancerous lesion for esophageal adenocarcinoma,⁴ with a relative risk of adenocarcinoma among patients with BE compared with the general population of 11.3 (95% confidence interval [CI] = 8.8 – 14.4).⁵

The prevalence of BE and the incidence of esophageal adenocarcinoma have been increasing in Western countries.^{6, 7} In these populations, BE is highly prevalent in people with gastroesophageal reflux disease (GERD)⁴; and the rising prevalence of GERD and obesity, combined with the decline in Helicobacter pylori infection, are thought to underlie the increase of the incidence of BE and esophageal adenocarcinoma.⁸ Similar trends have recently emerged in Asian countries, where the prevalence of GERD symptoms and reflux esophagitis, as well as obesity, has increased.^{9, 10} However, the effect of these trends on the prevalence of BE is unknown.

The prevalence and risk factors of BE in Asian countries are unclear. Studies report a wide range of BE prevalence (0.06 – 43.0%) in Asian countries.^{9, 11, 12} This variation may be due to the different time periods, study populations, and BE definitions.^{13, 14} A few studies also examined possible risk or protective factors for BE in Asia, but the overall patterns are unclear. There is no systematic review or meta-analysis focused on the prevalence or risk factors of BE in Asian countries. In this study, we examined the temporal changes in BE prevalence, regional variations and risk factors for BE in Asian countries.

Materials and Methods

This systematic review and meta-analysis was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.¹⁵

Data Sources

A literature search was performed using the PubMed and EMBASE databases for articles published from inception through September 2014, which included the terms Barrett’s esophagus, Barrett esophagus, or columnar-lined esophagus combined with epidemiology’, epidemiological, prevalence, incidence or population and Asia or each of 47 Asian countries, as China, Japan, Korea, Mongolia, Brunei, Cambodia, Indonesia, Lao, Malaysia, Myanmar, Philippines, Singapore, Thailand, Timor-Leste, Viet Nam, Afghanistan, Bangladesh, Bhutan, India, Iran, Kazakhstan, Kyrgyzstan, Maldives, Nepal, Pakistan, Sri Lanka, Tajikistan, Turkmenistan, Uzbekistan, Armenia, Azerbaijan, Bahrain, Palestine, Georgia, Iraq, Israel, Jordan, Kuwait, Lebanon, Oman, Qatar, Saudi Arabia, Syrian Arab Republic or Syria, Turkey, United Arab Emirates, or Yemen in the title, abstract or list of medical subject heading terms with no language restriction.

Inclusion and Exclusion Criteria

All eligible studies had to satisfy the following inclusion criteria: 1) BE defined at least by endoscopic criteria in a cross-sectional or cohort study but not case-control study where it is not possible to calculate the prevalence; and 2) sample size of the examined study population >200 eligible individuals participating in the study⁸. We excluded studies based on the following criteria: 1) study was not conducted in an Asian country; 2) prevalence of BE was studied only in patient groups limited to those with erosive esophagitis, cancer, or renal failure or the obese; and 3) conference abstracts because of their insufficient information. We defined endoscopic BE as abnormal columnar esophageal epithelium suggestive of a columnar-lined distal esophagus without histologic confirmation; histologic BE was defined as the presence of specialized intestinal metaplasia in endoscopic BE. When similar data were identified in multiple reports, we included only the report with the most complete and recent relevant data.

Selection of Studies

Two authors independently reviewed the titles and abstracts of studies identified in the search, based on prespecified inclusion and exclusion criteria. Disagreements were resolved by consensus and, when not possible, then in consultation with the senior investigator (HE). We also performed recursive search of bibliographies of existing narrative reviews and published studies.

Data Extraction

Two investigators independently extracted data from each study and entered them into a Microsoft Access^® database. The extracted information included citation information; study location (geographic location, country); study population (e.g., symptomatic patients, health screening mainly for gastric cancer, population-based); study period; study design (prospective, retrospective); BE definition (endoscopic BE or histologic BE); BE extent, based on Prague Circumferential and Maximal (C & M) criteria¹⁶; way histologic diagnosis was performed; landmark for esophago-gastric junction (EGJ); definition of short-segment BE; sample size; demographic data for study population (e.g., age, sex distribution, etc.); frequency of erosive esophagitis; frequency/proportion of endoscopic and/or histologic BE from patients/population examined; demographic data for patients with BE (e.g., age, sex); presence of esophageal dysplasia or adenocarcinoma in BE; and associated risk factors for BE (e.g., obesity, H. pylori, smoking, hiatus hernia, or reflux symptoms).

The methodological quality of each study was assessed, based on a predefined scale, according to 8 items defined as yes/no, as: 1) random sample or whole population; 2) unbiased sampling frame; 3) adequate sample size (>200 subjects); 4) standard BE measures; 5) outcome (i.e., BE), measured by unbiased assessors; 6) adequate response rate (70%) and refusers described; 7) confidence intervals and subgroup analysis presented; and 8) description of study subjects.¹⁷

Outcome measures

Primary outcomes were defined as the prevalence of histologic or endoscopic BE. Subgroup analyses were performed to find any relevant factors (e.g., age, sex distribution, study period) for the prevalence. Geographic areas were categorized as Eastern, South Eastern, South Central, and Western Asia, according to the International Agency for Research on Cancer (GLOBOCAN2012, http://globocan.iarc.fr/). The study period in which the researchers conducted their studies (distinct from publication year) was analyzed as a categorical variable (1991–1999, 2000–2004, 2005–2009, and 2010–2014), using the median year of the study period when a study was carried out over 2 years. When the study period was unavailable, the year of study publication was used instead. Studies were categorized according to the sampling frame as population-based, health check-up or screening (screening studies), or symptomatic patients (symptomatic studies), or unspecified. When studies included both screening and symptomatic subjects, they were categorized into symptomatic subjects. Secondary outcomes included risk factors for histologic BE. Proportion of age, male sex, obesity, H. pylori infection, smoking, hiatus hernia, and reflux symptoms were abstracted or calculated in the group between histologic BE and non-BE.

Statistical Analysis

The analyses were performed using Comprehensive Meta-Analysis software version 2 (Biostat, Englewood, NJ), SAS software version 9.3 (SAS Institute, Cary, N.C.) and STATA software version 13.1 (Stata Corp, College Station, TX). Calculation of pooled BE prevalence was performed in random-effect models. Heterogeneity across studies was assessed by I² statistic (i.e., <30%, 30% – 60%, 60% – 75%, and >75%), suggestive of low, moderate, substantial, and considerable heterogeneity, respectively).^{18, 19} As part of a sensitivity analysis, we assessed the impact of each study on the combined prevalence by running "Remove-One" analysis (i.e., the meta-analysis is run multiple times each with a different and single study removed). We assessed temporal trends in prevalence of BE in a Poisson regression model and binomial regression analyses. Estimates of relative risk for possible risk or protective factors were calculated as prevalence odds ratios (ORs) with 95% CIs and pooled, using a random-effect model. P values were calculated, using the Wald χ2 test.

Results

Search Results

Our search strategy identified 297 articles in PubMed and 167 articles in EMBASE. Based on the prespecified inclusion and exclusion criteria, 55 and 6 potentially eligible articles were selected for further review from PubMed and EMBASE, respectively (Figure 1). After excluding duplicate reports on the same population,^20–24 we included 51 studies (1 study reported 2 distinct cohorts) (N = 453,147) in our analysis (Supple Table 1).

Figure 1.

Flowchart of systematic literature searches.

Overview of the Included Studies

The number of subjects examined in each study ranged from 253²⁵ to 139,416.²⁶ Twenty-eight studies (total 298,850 subjects) reported histologic BE prevalence,^{11, 26–52} and twenty-three studies (154,297 subjects) reported endoscopic BE prevalence only.^{3, 12, 25, 53–71} Forty-seven studies reported the study period, whereas the remaining four studies did not. The oldest study was conducted between 1991 and 1992 and published in 1997.²⁷ Seven studies (N = 63,996 subjects) were conducted from 1991–1999, 16 studies (N = 132,271) were conducted from 2000–2004, 24 studies (N = 243,411) were conducted from 2005–2009, and four studies (N = 13,469) were conducted from 2010–2014. There was only 1 population-based study (N = 1,029),⁶⁸ 11 screening studies (N = 89,651), 37 symptomatic studies (N = 336,933), and 2 studies with an unspecified sampling frame (N = 25,534). Most studies (38 studies, N = 424,371) were conducted in Eastern Asia, 4 studies (N = 5,413) were conducted in South Eastern Asia, 5 studies (N = 2,928) were conducted in South Central Asia, and 4 studies (N = 20,435) were conducted in Western Asia.

Prevalence of Histologic BE

Histologic BE was reported in 28 studies (N = 298,850). The lowest prevalence was 0.06% (10/16,606),¹¹ and the highest was 8.9% (25/278).⁴⁵ Overall, the pooled prevalence of histologic BE was 1.3% (95% CI = 0.7 – 2.2) in the random-effect model (I² = 99%) (Figure 2 (a)). In 2 studies in which biopsy specimens were taken only from pink mucosa 3cm or more above the squamocolumnar junction, the prevalence of histologic BE was 0.3% and 0.2%, respectively.^{28, 36} In the sensitivity analysis excluding these 2 studies, the pooled prevalence of histologic BE was 1.4% (95% CI = 0.7 – 2.2). The Remove-One analysis did not show any variation in the pooled BE prevalence (Supple Figure 1 (a)).

Figure 2.

Prevalence of histologic BE (N = 28 studies) (a) and endoscopic BE only (23 studies) (b)

As shown in Table 1(a), the prevalence of histologic BE was not associated with age or sex distribution. There was a trend toward an increase in prevalence of histologic BE over time from 0.8% for 1991–1999, to 1.5% for 2000–2004, to 1.2% for 2005–2009 and to 2.2% for 2010–2014, despite there being an absence of a trend in increase in the negative binomial model (P = 0.97). Histologic BE prevalence was highest in studies from South Central Asia, followed by South Eastern, Western, and Eastern Asia. In a sensitivity analysis limited only in East and Southeast Asian countries (21 studies), the prevalence of histologic BE was 0.9% (95% CI = 0.9 – 1.0). The prevalence of histologic BE was higher in symptomatic studies than that in screening studies, although these differences did not reach statistical significance (P = 0.12).

Table 1.

The prevalence of histologic BE (a) and endoscopic BE only (b)

(a) Histologic BE
	Number of studies	BE	Total	Prevalence of BE (%) (95% CI)
Average age
≥50	7	408	32897	1.5 (0.7 – 3.3)
<50	9	185	12705	1.6 (0.8 – 3.0)
Not specified	12	682	253248	1.0 (0.4 – 2.8)
% Male
≥0.5	15	824	177257	2.0 (0.9 – 4.6)
<0.5	7	152	8668	1.8 (0.9 – 3.4)
Not specified	6	299	112925	0.3 (0.2 – 0.6)
Study period (enrolled year or published year)
1991–1999	4	226	37812	0.8 (0.1 – 5.1)
2000–2004	7	339	76252	1.5 (0.3 – 6.3)
2005–2009	15	634	180048	1.2 (0.6 – 2.6)
2010–2014	2	76	4738	2.2 (0.5 – 9.1)
Geographic area
Eastern Asia	17	889	270617	0.9 (0.5 – 1.9)
South-Eastern Asia	4	159	5413	1.7 (0.6 – 5.3)
South-Central Asia	4	102	2659	4.1 (2.2 – 7.6)
Western Asia	3	125	20161	1.6 (0.2 – 10.1)
Study population
Screening	3	79	10454	0.7 (0.3 – 1.4)
Symptomatic	25	1196	288396	1.4 (0.7 – 2.5)

(b) Endoscopic BE only
	Number of studies	BE	Total	Prevalence of BE (%) (95% CI)
Average age
≥50	13	6657	61861	13.9 (8.5 – 21.8)
<50	5	1171	34353	5.3 (1.8 – 14.5)
Not specified	5	2373	58083	2.4 (0.7 – 7.4)
% Male
≥0.5	12	8255	109363	9.4 (5.1 – 16.7)
<0.5	7	1557	11976	13.2 (9.1 – 18.8)
Not specified	4	389	32958	1.7 (1.0 – 2.8)
Study period (enrolled year or published year)
1991–1999	3	380	26184	2.6 (0.4 – 14.2)
2000–2004	9	3556	56019	8.9 (5.4 – 14.3)
2005–2009	9	5688	63363	10.0 (4.6 – 20.7)
2010–2014	2	577	8731	7.1 (0.2 – 70.6)
Geographic area
Eastern Asia	21	10111	153754	7.4 (4.6 – 11.7)
South-Central Asia	1	71	269	26.4 (21.5 – 32.0)
Western Asia	1	19	274	6.9 (4.5 – 10.6)
Study population
Population-based	1	19	1029	1.8 (1.2 – 2.9)
Screening	8	3680	79197	5.9 (2.9 – 11.6)
Symptomatic	12	6224	48537	14.4 (8.7 – 23.1)
Not specified	2	278	25534	1.0 (0.8 – 1.4)

BE, Barrett’s esophagus; CI, confidence interval

Only 14 of 28 studies defined and reported EGJ; most (12 studies) used the proximal margin of the gastric folds, 1 used the distal end of the palisade vessel zone at the distal esophagus, and 1 used both landmarks. The prevalence of histologic BE in 12 studies that defined EGJ as the proximal margin of the gastric folds was 1.5% [95% CI = 0.6 – 4.0]. The distribution of short-segment BE and long-segment BE in histologic BE was reported in 14 of 28 studies. The average weighted proportion of short-segment histologic BE (i.e., <3 cm) was 82.1% (95% CI = 73.7 – 88.2) (Supple Figure 2). Fifteen studies examined the presence of histologic BE in patients with erosive esophagitis. Erosive esophagitis was found in 3.4 – 48.3% of the total study population.^{36, 44} The prevalence of histologic BE in patients with erosive esophagitis ranged from 0.4%³⁶ to 31.9%,⁵¹ and the pooled prevalence was 5.2% (95% CI = 2.8 – 9.5) (Supple Figure 3). On the other hand, the prevalence of histologic BE in patients without erosive esophagitis was 0.9% (95% CI = 0.4 – 1.8). The prevalence of histologic BE was significantly higher in patients with erosive esophagitis than in those with nonerosive esophagitis (pooled OR = 5.94 [95% CI = 3.39 – 10.40]).

Among 28 studies, 14 studies also reported the prevalence of endoscopic BE separate from histologic BE. In these studies, histologic BE was confirmed in 7.9% to 67.7% of endoscopic BE. Overall, histologic BE was confirmed in 31.7% (95% CI = 22.2 – 43.0) of patients with endoscopic BE in the random-effect model (Supple Figure 4).

Prevalence of Endoscopic BE Only

Twenty-three studies reported the prevalence of endoscopic BE only (N = 154,297). Endoscopic BE prevalence ranged from 0.3% (56/19,812)⁶² to 43.0% (374/869).¹² Overall, the average weighted, pooled prevalence of endoscopic BE was 7.8% (95% CI = 5.0 – 12.1) in the random-effect model (I² = 99%) (Figure 1 (b)). In a sensitivity analysis excluding 1 study with higher prevalence of endoscopic BE,¹² the prevalence of endoscopic BE was 7.1% (95% CI = 4.6 – 11.1). The Remove-One analysis did not show any variation (Supple Figure 1 (b)).

Average age and sex distribution in the study population were not associated with the prevalence of endoscopic BE only (Table 1(b)). There was a nonsignificant trend toward a temporal increase in prevalence of endoscopic BE (P = 0.58). Geographically, most of studies were conducted in Eastern Asian countries. In a sensitivity analysis limited only in East and Southeast Asian countries (21 studies), the prevalence of endoscopic BE only was 7.4% (95% CI = 4.6 – 11.7). The prevalence of endoscopic BE was higher in the symptomatic studies than in the screening studies (P = 0.22). The prevalence of endoscopic BE only was 1.8% (95% CI = 1.2 – 2.9) in the population-based study. EGJ was defined in 13 of 23 studies; 6 studies used the proximal margin of the gastric folds; 7 studies used the distal end of the palisade vessel zone at the distal esophagus. The prevalence of endoscopic BE was 11.8% (95% CI = 4.0 – 29.9) for the proximal margin of the gastric folds, and 18.6% (95% CI = 13.6 – 25.1) for the distal end of the palisade vessel zone at the distal esophagus (P = 0.38).

Time Trends

The relationships between study period and the prevalence of histologic or endoscopic BE in Eastern Asian countries are shown in Figure 3. Binomial regression analyses did not show any increasing trend, either for histologic BE or endoscopic BE, despite the fact that the prevalence was gradually increasing. When 1 study examined the histologic BE conducted in 2010–2014 was combined with studies from 2005–2009, the prevalence of histologic BE was 0.3% (95% CI = 0.0 – 9.8) for 1991–1999, 0.8% (95% CI = 0.1 – 11.2) for 2000–2004, and 1.1% (95% CI = 0.5 – 2.5) for 2005–2014, respectively (P = 0.67). The prevalence of endoscopic BE only was 2.6% (95% CI = 0.4 – 14.2) for 1991–1999, 7.8% (95% CI = 4.4 – 13.4) for 2000–2004, 10.0% (95% CI = 4.6 – 20.7) for 2005–2009, and 7.1% (95% CI = 0.2 – 70.6) for 2010–2014, respectively (P = 0.57).

Figure 3.

Prevalence of histologic BE (a) and endoscopic BE (b) categorized by Eastern Asian countries.

Dysplasia in Histologic BE

The prevalence of low-grade dysplasia, high-grade dysplasia, and esophageal adenocarcinoma in patients with histologic BE was reported in 8, 8, and 10 studies, respectively (Figure 4 (a–c)). The average weighted, pooled prevalence was 6.9% (95% CI = 4.2 – 11.3) for low-grade dysplasia (8 studies), 3.0% (95% CI = 1.4 – 6.6) for high-grade dysplasia (8 studies), and 2.0% (95% CI = 0.9 – 4.3) for esophageal adenocarcinoma (10 studies), respectively. Two studies reported the prevalence of unspecified dysplasia in histologic BE; and they were 1.3% and 3.3%, respectively.^{31, 42}

Figure 4.

Prevalence of low (a) or high grade dysplasia (b) and esophageal adenocarcinoma (c) in the patients with histologic BE

Risk Factors for Histologic BE

Several studies reported on possible risk factors of BE, comparing patients with BE with control groups (Supple Table 2). Ten studies examined differences in age between patients with histologic BE and controls; six studies showed a significantly higher average age in histologic BE, and only one showed a significantly lower average age in histologic BE than that of controls, while the remaining three studies also showed higher average age in patients with BE than in controls. The proportion of men was significantly higher in patients with histologic BE than it was in controls when 16 studies were included (pooled OR = 1.50 [95% CI = 1.11 – 2.03]) (Figure 5 (a)). Hiatus hernia was significantly associated with histologic BE when 15 studies were included (pooled OR = 4.88 [95% CI = 2.93 – 8.13]) (Supple Figure 5 (a)). Smoking was significantly associated with increased risk of histologic BE when 9 studies were included (pooled OR = 1.26 [95% CI = 1.01 – 1.56]) (Supple Figure 5 (b)). The presence of obesity (i.e., defined as body mass index >25) was not associated with histologic BE when 3 studies were included (pooled OR = 3.94 [95% CI = 0.53 – 29.33]) (Supple Figure 5 (c)). Only 1 study examined waist circumference and waist-to-hip ratio and showed an association between large waist circumference, but not waist-to-hip ratio, and increased risk of endoscopic BE.⁷⁰ H. pylori infection was not associated with histologic BE when 6 studies were included (pooled OR = 1.00 [95% CI = 0.81 – 1.23]) (Supple Figure 5 (d)). Ten studies compared reflux symptoms between subjects with histologic BE and controls (Figure 5 (b), and the pooled OR was 3.15 (95% CI = 1.61 – 6.17). The proportion of patients with reflux symptoms was 45.8% in histologic BE patients. Six studies reported the results of multivariate analysis to find independent risk factors for BE. Higher age, male sex, reflux symptoms, and the presence of hiatus hernia were independently associated with BE in their multivariate analyses (Supple Table 3).

Figure 5.

Association between the risk of histologic BE and gender (N = 16 studies) (a) and reflux symptoms (N = 10 studies) (b) in Asian studies

Small-Study Effect “Publication Bias”

The relationship between the sample size and BE prevalence is shown in Supple Figure 6. Higher BE prevalence was generally reported in studies with smaller rather than larger sample size. This effect, however, was not statistically significant in Egger’s test (P = 0.25 for histologic BE, P = 0.46 for endoscopic BE only, Supple Figure 7). Furthermore, Egger’s tests were no significant for low-grade dysplasia, high-grade dysplasia, or esophageal adenocarcinoma (P = 0.85, 0.74, and 0.22, respectively, Supple Figure 7).

Assess the quality of each study

We assessed the quality of each study and found that: 1) only 1 study was population based⁶⁸; 2) no study had an unbiased sampling frame; 3) all studies had >200 subjects); 4) all studies used standard measures for BE definition; 5) no study used outcomes evaluated by unbiased assessors; 6) only 2 studies reported the response rate, and in both it was >70%^{41, 51}; 7) all studies reported or calculated the 95% CI, and performed subgroup analyses; and 8) all but 2 studies⁵⁶ described study subjects. The overall study quality measure ranged from a score of 3 to a score of 5 (of a maximum of 8), with a mean of 4.0.

Discussion

This is the first systematic review and meta-analysis to evaluate the prevalence of BE in Asian countries, its temporal trends and risk factors. Our data show that the prevalence of histologically confirmed BE pooled in 28 studies was 1.3% (95% CI = 0.7 – 2.2) and that of endoscopic-only BE was 7.8% (95% CI = 5.0 – 12.1) from 23 studies. These estimates are similar to those in Western countries, where population-based estimates are 1.6% for histologic BE and 10.3% for endoscopic BE⁷², and studies in symptomatic patients report 6 – 12% prevalence.^{73, 74} This entity seems similar in the prevalence of neoplasia and risk factors, including reflux symptoms, male sex, hiatus hernia, and smoking.

GERD in Asian countries may have resulted in an increase in BE prevalence in these countries. There was no clear trend of increasing BE prevalence among studies conducted between 1991 and 2014. Two studies reported BE prevalence in the same population over 5 years. Chen et al. from China reported that prevalence of histologic BE increased from 0.6% in 2000 to 1.2% in 2007 in referral endoscopy.⁴² Zhang et al. also from China, did not find any increasing trend for histologic or endoscopic BE (0.2% in 2005, 0.2% in 2011 for histologic and 1.5% in 2005, 1.4% in 2011 for endoscopic BE, respectively).²⁶

Different definitions for EGJ endoscopic landmarks can influence the reported prevalence of BE. The proximal margin of the gastric mucosal folds is generally accepted as an endoscopic marker of the EGJ.¹⁶ On the other hand, the distal margin of the palisade-shaped longitudinal capillary vessels is sometimes considered the EGJ, especially in Japan.³ It is possible that gastric intestinal metaplasia can be considered as BE in the case EGJ was defined as the distal margin of palisade vessels or undefined. Most studies that examined histological BE used the proximal margin of the gastric mucosal folds as the EGJ and diagnosed by the presence of specialized intestinal metaplasia with goblet cell. Importantly, 10 of 23 studies for endoscopic examination only and 14 of 28 studies for histological examination did not show the definition of the EGJ. Endoscopic landmarks for the EGJ need to be stated in the future studies.

The term endoscopic suspected esophageal metaplasia but not endoscopic BE is recommended in the Montreal Definition and Classification of Gastroesophageal Reflux Disease.⁷⁵ To confirm the presence of histologic BE, 4-quadrant biopsies every 1–2 cm for circumferential metaplastic segments is recommended.⁷⁵ In our study, the overall prevalence of histologic BE in endoscopic BE was 31.7%. The number of biopsies affects the likelihood of detection⁷⁶ and could have affected the estimates of BE; 10 studies performed 4 biopsies, mostly every 2 cms, but several did not mention the number of biopsies, and none performed 8 biopsies. Further, the diagnostic criteria for endoscopic BE varied among the studies. The Prague C & M criteria were introduced and validated in 2006¹⁶; however, only 4 studies with histological examination and 3 studies with endoscopic examination used these criteria.

GERD appears to be the main risk factor for BE in Asian countries, just as it is in Western countries. In our study, the prevalence of histologic BE was approximately 6-fold higher in patients with erosive esophagitis than in patients without erosive esophagitis. Moreover, the prevalence of BE was approximately 2-fold higher in the symptomatic studies than in screening studies. Lastly, GERD symptoms were associated with a 3.15-fold increase in the odds of having BE. Nevertheless, the proportion of the patients with reflux symptoms in histologic BE was only 45.8%. Therefore, a screening strategy to detect BE for symptomatic subjects only might be insufficient.

Our findings suggest that histologic BE in Asian countries has a similar potential to develop into esophageal adenocarcinoma as that found in Western countries. The prevalence of low-grade dysplasia, high-grade dysplasia, and esophageal adenocarcinoma in cases with histologically proven BE was comparable with those reported in US studies (1.3 – 9.8% for low-grade dysplasia, 0 – 5.3% for high-grade dysplasia, and 1.3 – 5.7% for esophageal adenocarcinoma, respectively).^77–81 In our study, approximately 80% of BE was short-segment BE; and, among the short-segment cases, most were < 1cm.^{49, 53, 63, 70} It remains unclear whether BE <1cm has a potential risk for the development of esophageal adenocarcinoma. This uncertainty is related to the low interobserver reliability^{16, 82} and the exclusion of subjects with BE <1cm from total BE in some studies.^{32, 43} Further studies are necessary to clarify the clinical implication of BE <1cm.

In studies conducted in Western regions, old age; white race; male sex; chronic heartburn; early onset of GERD symptoms; hiatus hernia; erosive esophagitis obesity with intraabdominal fat distribution; metabolic syndrome; smoking; dietary consumption of fats and red or processed meat; and family history of GERD, BE, or esophageal adenocarcinoma were shown to be risk factors for BE.^83–87 On the other hand, H. pylori infection was reported as a protective factor.⁸⁸ Our study in Asian countries showed that male sex, hiatus hernia, and smoking were associated with histologic BE but not obesity and H. pylori infection. Lack of a positive relationship between obesity and histologic BE in our study might be due to the small number of studies. Regarding H. pylori infection, the prevalence of BE was low, even in Malaysia, where the prevalence of H. pylori infection is also low,⁴⁶ which does not support the inverse relationship between H. pylori infection and BE.

In addition to the limitations of the individual studies, our meta-analysis has several limitations. First, studies with smaller sample size showed higher BE prevalence than those with larger sample size. Publication bias is a possible explanation for this observation.⁸⁹ Furthermore, studies had different time periods, study populations, and BE definitions. However, even when we selected studies of only histologic BE in which EGJ was defined as the proximal margin of the gastric folds, the prevalence of histologic BE was 1.5%, which support the evidence that it is not uncommon compared with that of Western countries. In addition, most studies targeted symptomatic patients and, therefore, it is difficult to generalize for asymptomatic or general populations. Second, most studies (38 of 51 studies) were conducted in Eastern Asian countries, with only a few studies from other Asian areas (4 for South Eastern Asia, 5 for South Central Asia, and 4 for Western Asia). Therefore, the findings cannot be generalized to all Asian countries. Moreover, there was only one population-based study that examined the prevalence of endoscopic BE only. There was no population-based study examining the prevalence of histologic BE. Diagnosis of histologic BE depends on accurate endoscopic recognition of suspected BE. The Prague C & M criteria were used only in 7 of 51 studies, which suggests that the prevalence of histologic BE in other studies may differ according to inter- and intraobserver variations. Finally, our study showed that the prevalence of low-and high-grade dysplasia were comparable with those of Western regions, however different pathological criteria could have contributed to errors in the high prevalence of dysplasia.⁹⁰ Despite similar prevalence of histologic BE and dysplasia between Asia and Western, the incidence of esophageal adenocarcinoma is considerably lower in Asian than Western. BE prevalence alone does not explain the differences in esophageal adenocarcinoma, however the preponderance of short segment BE may partly explain these differences as it is known that the carcinogenic potential of BE is proportional to segment length. It is also possible that there is a time lag of several decades between the formation of BE and the development of adenocarcinoma. Lastly, it is possible that BE is necessary but not sufficient for the development of adenocarcinoma and other factors that are different in Asian regions such as severity or composition of reflux, obesity and or diet play a role.

In conclusion, endoscopic and histologic BE in Asian countries is not uncommon. The prevalence of low-grade, high-grade dysplasia, and esophageal adenocarcinoma in histologic BE was similar with that of Western countries. GERD symptoms, male sex, hiatus hernia, and smoking were associated with increased risk of histologic BE. These findings may be a prequel for an increase in esophageal adenocarcinoma in these countries. However, targeted strategy only for symptomatic subjects may be insufficient because more than half of the subjects with histologic BE did not have any reflux symptoms.

Abbreviations

BE

Barrett’s esophagus

CI

confidence interval

EGJ

esophago-gastric junction

GERD

gastroesophageal reflux disease

OR

odds ratio

PRISMA

the Preferred Reporting Items for Systematic Reviews and Meta-Analyses