The impact of Helicobacter pylori on the presence of Barrett's esophagus in Azerbaijan, a high-prevalence area of infection

SUMMARY

There is a strong evidence that Helicobacter pylori infection is inversely associated with Barrett's esophagus. In a high-prevalence region of H. pylori, low rates of esophageal cancer and its precursor BE may indicate its preventive effect. The aim of this study is to determine the impact of H. pylori on characteristics of Barrett's esophagus.

A total of 3317 outpatient upper endoscopy reports from 2013 to 2015 from an urban center in Azerbaijan from all patients with dyspepsia were retrospectively analyzed for patients with Barrett's esophagus. This was matched in a 1:2 ratio to age and gender matched control patients without Barrett's esophagus. The prevalence of H. pylori on Barrett's esophagus and the randomly selected control group were compared.

There were 83 patients with BE and 167 control group cases. Biopsy-proven BE was diagnosed in 83 patients: 39 (47%) females, with mean age 43.1 ± 13.3 years. Of these, 13 (15.7%) had long segment and 70 (84.3%) had short segment Barrett's esophagus. A control group included 167 patients: 78 (46.7%) females, with mean age (45.8 ± 13.9). All patients were Caucasians. The rates of gastric inflammation, the presence of atrophy, and intestinal metaplasia in gastric specimens did not differ in patients versus controls. The prevalence of H. pylori was determined as 63.2% in male and 61.5 in female groups (odd ratio (OR) = 0.99 95%CI 0.97, 1.01; P = 0.22). Inflammation of gastric mucosa was strongly associated with the infection (67% vs. 33%; OR = 4.46 95% CI: 2.01, 9.92, P < 0.001). Atrophy was noted in majority of H. pylori-positive cases (OR = 1.43, 95% CI: 0.36, 5.65; P = 0.61). Gastric intestinal metaplasia was observed in 55.6% of H. pylori-positive patients and in 44.4% of negative individuals (OR = 0.74, 95% CI: 0.28, 1.94; P = 0.54). There was not a significant difference in the prevalence of HP in BE and control groups; 63.9% were positive for infection in BE cases and 61.7% of controls (OR = 1.10, 95% CI: 0.64, 1.90; P = 0.74). We found that neither presence of erosive esophagitis, length of BE nor dysplasia (45.5% of H. pylori-positive group, whereas 54.5%) was associated with the presence of the H. pylori infection (Table 1).

In a predominantly Caucasian nation with a high prevalence of H. pylori gastritis, the presence of H. pylori was not inversely associated with the presence of Barrett's esophagus. These data challenge the mechanistic implications of this association.

INTRODUCTION

Esophageal adenocarcinoma (EAC) is a deadly disease, which is felt in most patients to be preceded by Barrett's esophagus. It is postulated that after years of gastroesophageal reflux, the esophageal squamous mucosa undergoes metaplastic change to an intestinal type mucosa with an incident cancer risk of approximately 0.4%.^1,2 Of note is that both the prevalence and incidence of esophageal adenocarcinoma have been rising steadily in the Western population over the past several decades³ making it one of the most common gastrointestinal cancers but with a 5-year expectancy of still <20%.⁴

The cause of the rising incidence of EAC is unclear. Several factors have been proposed. These include increasing epidemic of obesity,⁵ changes in the esophageal microbiome,⁶ and incidental or purposeful eradication of Helicobacter pylori infection.⁷ The inverse pathophysiologic association of H. pylori gastritis to EAC is based on reduction of gastric acid due to atrophy and therefore avoidance of acid-rich gastric refluxate in those patients predisposed to gastroesophageal reflux disease.^8,9 Data that have supported this include an inverse correlation of presence of H. pylori in decreasing order to erosive esophagitis, Barrett's esophagus, and EAC.^8,10,11 On the other hand, not all studies have confirmed the relationship between this infection and gastric hyposecretion secondary to gastric atrophy and EAC.¹²

To date, however, most studies documenting this inverse correlation of H. pylori to EAC and Barrett's esophagus have been performed in Western countries. For example, studies have come from the US, such as done by Wang et al., provided the strongest evidence that H.pylori is inversely associated with BE.⁷ This, however, may represent biased populations with the existence of other factors that predispose to Barrett's esophagus, such as obesity, making it difficult to determine if H. pylori absence is causative or an epiphenomenon. In contrast, there are far more limited data on this subject in areas where H. pylori is endemic. For example, in a meta-analysis by Shiota et al., the prevalence of Barrett's esophagus in Asian countries, EAC was not associated with a decrease in H. pylori.¹³ In Asian countries, bias may also exist for such a study given the relatively low prevalence of Barrett's esophagus. As a result, the aim of this study is to further define the prevalence of H.pylori and to determine if a correlation exists with H. pylori in Azerbaijan, a country with both a high prevalence of H. pylori, Caucasian race, and Barrett's esophagus.

Methods

Records of patients who underwent endoscopy from 2013 till 2016 in Baku Medical Plaza Hospital for the symptom of dyspepsia were analyzed retrospectively. This medical center is one of the largest private hospitals in Baku; around 3500 upper endoscopies are performed yearly in the GI department by three gastroenterologists. Routinely, one or two gastric biopsies are obtained from antrum for H.pylori presence using rapid Campylobacter-like organism test. In case of treatment with bismuth-containing compounds, H2-receptor blockers, proton pump inhibitors, and antibiotics for the past month, biopsy with further pathology examination is used for assessing H.pylori status. In 3095 or 3117 patients, H. pylori testing was performed after at least 2 weeks off acid suppressing or bismuth therapy. In 22 patients, endoscopy performed for more emergent indications precluded cessation of these medications. Gastric biopsies are also assessed for evidence of chronic gastritis, atrophy, and intestinal metaplasia based on endoscopic findings. The presence of Barrett's esophagus is defined as finding intestinal metaplasia in an esophageal segment defined by visualization of columnar type mucosa in the distal esophagus. BE is defined as long segment if the tongues are 3 cm or more in length and short segment BE when less than 3 cm. Patients were divided into short and long segment based on the length of flame-like elongation of the esophageal gastric junction (EGJ) and all patients with Barrett's were classified by Prague criteria (8). Endoscopy was performed by experienced GI physicians. Images were reviewed. The control group consisted of computer-generated randomly chosen patients with dyspepsia without Barrett's esophagus matched for age and gender distribution in a 2:1 ratio to subjects with H. pylori.

Statistical analysis

Descriptive statistics are reported as mean ± standard deviation or frequencies and percentages. Comparisons between those with BE and controls were made using a t-test for age and either a chi-square or Fisher's exact test for the other patient characteristics. Associations between H.pylori and patient characteristics were made using logistic regression. A multivariable logistic regression model was used to assess the association between H.pylori and BE after adjusting for other patient characteristics. All tests were two-sided, and a P-value ≤0.05 was considered to be statistically significant.

RESULTS

A total of 3117 patients who underwent upper GI endoscopy for dyspepsia performed by one of the authors (SG) were identified. The total number of endoscopies performed over this time period in the hospital was 10,246. Biopsy-proven BE was diagnosed in 83 patients: 39 (47%) females, with mean age 43.1 ± 13.3 years. Of these, 13 (15.7%) had long-segment and 70 (84.3%) had short-segment Barrett's esophagus (Table 1). Prague classification ranged from C0M1 to C6M7. There was only one male patient, aged 73, smoker, who presented with esophageal adenocarcinoma. A control group included 167 patients: 78 (46.7%) females, with mean age 45.8 ± 13.9). These groups did differ in age or gender (p > .05). All patients were Caucasians, 33 (39.8%) in BE group and 59 (35.3%) in control group were smokers (P = 0.49).

The rates of inflammation, the presence of atrophy, and intestinal metaplasia in gastric specimens did not differ in patients versus controls (Table 2). The prevalence of H.pylori was determined as 63.2% in male and 61.5 in female patients for both groups (OR = 0.99 95%CI 0.97, 1.01; P = 0.22). Inflammation of gastric mucosa was strongly associated with the infection (67% vs. 33%; OR = 4.46 95% CI: 2.01, 9.92, P < 0.001). Atrophy was noted in majority of H.pylori-positive cases (OR = 1.43, 95% CI: 0.36, 5.65; P = 0.61). Gastric intestinal metaplasia prevalence did not differ between H. pylori-negative and -positive individuals (OR = 0.74, 95% CI: 0.28, 1.94; P = 0.54).

Erosive esophagitis was evaluated endoscopically according to the Los Angeles classification in control group only and was seen in 58.0% of HP group versus 42.0% of HP-negative patients (OR = 0.62, 95% CI:0.31, 1.23; P = 0.17, Table 3). The percentage of smokers was similar in BE and control groups (35.3% vs. 39.8%; P = 0.49). There was not a significant difference in the prevalence of HP in BE and control groups; 53 (63.9%) were positive for infection in BE cases and 61.7% of controls (OR = 1.10, 95% CI: 0.64, 1.90; P = 0.74). After adjusting for age, sex, inflammation, atrophy, gastric intestinal metaplasia, smoking status, and screening date, we still did not see a difference in H. pylori for BE versus controls (adjusted OR = 1.10, 95% CI: 0.61, 1.96; P = 0.75). We did however see that those with inflammation were still more likely to have H. pylori (adjusted OR = 5.50, 95% CI: 2.39, 12.7; P < 0.001, Supplementary Table).

We found that the length of BE was not associated with the presence of the infection. In H.pylori-positive group, 6 (46.2%) had long-segment BE and 47 (67.1%) had short involvement. In H.pylori-negative group, 7 (53.8%) had long segment and only 23 (32.9%) had short segment BE (OR:2.38, 95% CI: 0.72–7.91; P = 0.16). Dysplasia was noted in 45.5% of H.pylori-positive group, whereas 54.5% of Barrett's cases with dysplasia did not have H. pylori (OR = 0.42 95% CI: 0.12–1.51, P = 0.18). Table 3 shows the association of H.pylori infection with esophagitis, presence and length of hiatal hernia and dysplasia of BE.

DISCUSSION

In this study, it has been demonstrated that although the prevalence of H. pylori was high in the Azerbaijan population, the presence of this infection was not inversely associated with BE and therefore not protective. Specifically, there was a prevalence of H. pylori gastritis in both control non-Barrett's patients with dyspepsia and Barrett's patients of >60%. Furthermore, the prevalence of Barrett's esophagus in patients with and without H. pylori was similar. As a result, these data question the protective effect of H. pylori in development of Barrett's esophagus.

Multiple studies have reported in Western populations that the presence of H. pylori is inversely associated with Barrett's esophagus and therefore implies a protective mechanism. In one study, a group from Hungary performed extensive meta-analysis providing additional evidence that HPI is associated with reduced risk of BE¹⁴ Demonstrating that H. pylori infection reduced the risk of BE with an OR of 0.68 and for dysplasia 0.37. They have also suggested that HP was associated with significantly lower risk of dysplastic, nondysplastic, and long segment BE. Notably, however, the subgroups studies were Australia, Europe, and North America, and Asia, in other words, predominantly Western or Asian populations. Notably, not all studies demonstrated a protective effect and in fact the studies in Africa demonstrated an increased risk of EAC with H. pylori.¹⁴ In fact, similar to our research, a study from Canada showed that limited evidence suggests that there is a clear association between H.pylori and BE.⁷ A report from China also defined that here is a protective role of H.pylori infection to GERD but there may be no relationship between H.pylori infection of stomach and BE.¹⁵ The heterogeneous results in these studies likely underscore variable effects of H. pylori in patients, the presence or absence of more virulent strains such as Cag A,¹¹ and the varying penetrance of H. pylori in different populations. In our study, we examined a specific population with a Caucasian population thereby predisposed to Barrett's esophagus but with a high H. pylori prevalence and gastritis, different from most previous studies.

This study has several limitations. First, it is in a hospital with several gastroenterologists so the reliability in distinguishing short segment Barrett's from an irregular Z line may be of concern. Nevertheless, endoscopic images from each of the Barrett's were reviewed to provide some security in the diagnosis. We also did not measure Cag A status, which has been further inversely associated with Barrett's esophagus.¹¹ On the other hand, one of the important strengths of this study that not only was H. pylori assessed but was also associated with a high prevalence of gastritis. Furthermore, we demonstrated a close association of H. pylori to these histologic findings underscoring that the presence of H. pylori was not incidental but had histologic validity in the studied population. We could not be sure, however, that these patients had hypochlorhydria as a putative protective mechanism against Barrett's given the occurrence of atrophy and/or intestinal metaplasia found on biopsies in not all patients. Another limitation is the assurance that the hospital population studied, as only 0.1% of the population is representative of the population. As Azerbaijan is predominantly a country of Caucasian race and the prevalence of H. pylori in this population was high in spite of using a private hospital for study, we suspect use of this group of patients for study was justified.

We conclude that the inverse association of H. pylori infection with the development of Barrett's esophagus is tenuous and depends to a large degree on the population studied. Whether this reflects heterogeneity of the studies and populations or that the mechanistic implications of H. pylori as protective against Barrett's and EAC are not certain merits further explanation.

Specific author contributions: Collection of data, analysis and writing: Sevda Agayeva, David A. Katzka; Data analysis: Kristin C. Mara; Conflicts of Interest: David A. Katzka.

Advisory Board Shire and Celegene.

References