Summary
Barrett’s esophagus (BE), a consequence of gastroesophageal reflux disease (GERD), is a premalignant condition for esophageal adenocarcinoma. Impaired gastric emptying leads to increased gastric volume and therefore more severe reflux. We seek to investigate the association between gastroparesis and BE and the predictors of BE among patients with gastroparesis. This is a retrospective review of patients seen at Cleveland Clinic between 2011 and 2016 who had an upper endoscopy and a gastric emptying study. Demographics, symptoms, medications, endoscopic and histological findings, and therapeutic interventions were abstracted. Risk of BE among gastroparesis group and control group was assessed, and logistic regression analysis was performed to identify predictors of BE among gastroparesis patients. Of the 4,154 patients, 864 (20.8%) had gastroparesis and 3, 290 (79.2%) had normal gastric emptying. The mean age was 51.4 ± 16.4 years, 72% were women and 80% were Caucasians. Among the gastroparesis group, 18 (2.1%) patients had BE compared to 71 (2.2%) cases of BE in the control group, P = 0.89. There were no differences in gender, race, reflux symptoms, or esophageal findings between the two groups. Among gastroparesis group, predictors of developing BE were a history of alcohol use (odds ratio [OR] 6.76; 95% confidence intervals [CI]: 1.65–27.67, P = 0.008), history of pyloroplasty (OR: 8.228; CI: 2.114–32.016, P = 0.002), and hiatal hernia (OR: 8.014; CI: 2.053–31.277, P = 0.003). Though gastroparesis is a known contributing factor for GERD, there was no increased prevalence of BE in gastroparesis. Among patients with gastroparesis, predictors of BE are history of alcohol use, hiatal hernia, and pyloroplasty.
Keywords: gastric emptying, gastroesophageal reflux disease, risk factors, duodenogastric reflux
INTRODUCTION
Gastroparesis is a chronic symptomatic disorder of the stomach characterized by delayed gastric emptying without evidence of mechanical obstruction.1 The three major causes of gastroparesis are diabetic, postsurgical, and idiopathic. Patients with gastroparesis typically present with nausea, vomiting, early satiety, postprandial fullness, and epigastric abdominal pain; symptoms can be persistent or manifest as episodic flares. Severe cases may also suffer from electrolyte disturbances, dehydration, malnutrition, weight loss, esophagitis, and Mallory–Weiss tear.1,2 Scintigraphy is the gold standard for diagnosis of gastroparesis by demonstrating a delay in gastric emptying; however, gastric emptying rate correlates poorly with symptoms and quality of life.1
WHAT YOU NEED TO KNOW.
Background: Gastroesophageal reflux symptoms are prevalent in gastroparesis patients.
It is not known if there is higher prevalence of Barrett’s esophagus in gastroparesis patients.
Findings: The prevalence of Barrett’s esophagus is similar in patients with normal and delayed gastric emptying. Risk factors for Barrett’s esophagus in patients with gastroparesis are history of pyloroplasty, alcohol use, and hiatal hernia.
Implications for patient care: Patients and surgeons need to be aware of this risk of Barrett’s esophagus prior to undergoing pyloroplasty.
Barrett’s esophagus (BE), a consequence of gastroesophageal reflux disease (GERD), is a premalignant condition characterized by a change in the lining of the normal stratified squamous epithelium of the esophagus to a metaplastic columnar epithelium with goblet cells that can progress into esophageal adenocarcinoma.3 The relationship between gastroparesis and GERD is complex and multi-factorial; it is believed that delayed gastric emptying may contribute to the esophageal reflux burden through several mechanisms including an increase in gastric content, increase in acid secretion, and relaxation of the lower esophageal sphincter (LES) mediated by gastric distention.4,5
The available data about gastric emptying in GERD patients are generally available from the 1980s to 1990s with the majority of the published studies at that time showing a significant delay in gastric emptying among patients with GERD and a few studies showing normal solid/liquid gastric emptying.6–9 Furthermore, some studies suggested that GERD patients have more pronounced postprandial fundic relaxation, and it is the slow proximal emptying that correlated with increased esophageal acid exposure but not delayed distal or total gastric emptying.7–9 In contrast, two studies showed reduced esophageal acid exposure in patients with gastroparesis.10,11 Only few studies have looked at the prevalence of delayed gastric emptying in BE patients, and among them, the emptying of both solids and liquids was uniformly within the normal range.12–15 While gastric emptying study (GES) was used for measuring gastric emptying, there was a lack of standardization of the test, including differences in caloric content/consistency of meals used, patient positioning, frequency, and duration of imaging.16 There were also differences in the data reported, e.g. half-time of emptying, rate of emptying (percent per minute), or the percent retention or emptying at different time points during the study. Normal values often have not been established for some of the protocols used, and the performance characteristics of the test with the specified meal may not have been validated.
Therefore, the aims of our study were to investigate (1) the presence of BE in patients with and without gastroparesis, (2) differences between gastroparesis patients with BE and without BE, and (3) the predictors of BE among gastroparesis patients.
MATERIALS AND METHODS
Data source and study subjects
This study was approved by the Institutional Review Board at Cleveland Clinic, Cleveland, Ohio, and informed consent was waived. DDSI Datamart, part of the larger Cleveland Clinic Health System data warehouse, is a data repository that houses data regarding all patients seen in digestive disease and surgery institute. Patients who had both an esophagogastroduodenoscopy (EGD) and a GES between the period from January 2011 and December 2016 were identified using the database. Patients with gastroparesis were classified as cases and patients with normal GES as controls. Patients ≤18 years of age or with evidence of mechanical obstruction were excluded.
Much of the medical information contained within the electronic medical record is not easily searchable. An internally validated natural language algorithm, using Structured Query Language, was used to search through the electronic medical record (Epic Systems, Verona, WI). Natural language programming code reads through the patient documents and retrieves discrete variables of interest. Prolog, a general-purpose programming language, was used to develop the method of parsing the visit notes and imaging and endoscopy reports in a stepwise manner. Each paragraph was tokenized into sentences, which were then numbered into words. This method creates an array that can be read for variables. The parser then analyzes each sentence for the variables of interest and puts them into a table. Subsequently, the accuracy of data extraction by natural language programming was confirmed by manual data chart review.
Data collection
Basic demographic data including age, gender, race, and social habits (smoking, alcohol, and illicit drug use) were collected. Other data captured included body mass index (BMI), symptoms of GERD (heartburn, acid regurgitation, chest pain, and dysphagia), as well as symptoms of gastroparesis (nausea, vomiting, early satiety, bloating, epigastric pain, and weight loss). The use of medications including proton pump inhibitors (PPI) (omeprazole, esomeprazole, lansoprazole, dexlansoprazole, pantoprazole, rabeprazole), histamine receptor antagonists (H2 RA; cimetidine, famotidine, ranitidine, nizatidine), prokinetics (metoclopramide, erythromycin, azithromycin, cisapride, domperidone), aspirin, nonsteroidal anti-inflammatory drugs (NSAIDs), statins, and opioids was extracted from the medication administration record. International Classification of Disease 9th and 10th revision, Clinical Modifications (ICD-9/10-CM) codes were used to obtain patient comorbidities such as diabetes mellitus, hypertension, hyperlipidemia, obstructive sleep apnea, and hypothyroidism. Data regarding botulinum toxin injection to pylorus and surgical interventions such as pyloromyotomy, gastrojejunostomy, and pyloroplasty were also retrieved.
GES protocol and data interpretation
Patients were asked to fast for 8 hours prior to the test and are off any medications affecting gastric motility for 48 hours prior to the test. A standard meal consisting of 4 oz. Egg Beaters, 1 piece toast, 30 g strawberry jam, and 1 mCi Tc-99 m sulfur colloid was given orally with 8 oz. water, consumed over 5 to 10 minutes.16 Patients were followed with imaging immediately after meal ingestion and again at 1, 2, and 4 hours following ingestion for assessment of gastric emptying. Criteria used to identify patients with evidence of delayed gastric emptying were gastric retention of >10% at 4 hours and/or > 60% at 2 hours. If patients were unable to tolerate a solid meal, a liquid nutrient meal containing radioisotope was used to permit the scintigraphic measurement of gastric emptying.
Endoscopic and histological data
All patients included in this study underwent at least one EGD within 1 year of gastric emptying study date at Cleveland Clinic. Endoscopic reports were reviewed for the presence of hiatal hernia, erosive esophagitis, esophageal stricture, and BE. Diagnostic criteria for BE included both endoscopic identification of columnar epithelium (≥1 centimeter) and biopsy confirmation of intestinal metaplasia with goblet cells.3
Statistical analysis
Statistics for patient demographics, symptoms, medications, interventions, and comorbidities are expressed as mean ± standard deviation for continuous variables and count and column percent for categorical variables. Pearson’s chi-square tests or Fisher’s exact tests were applied for categorical variables, and ANOVA, Kruskal–Wallis test, and independent sample t-test were used for continuous variables for comparison between groups. For multivariate analysis, variables were tested for independence from one another using the diagnostic methods of variance inflation factors, condition indices, and quasi-separation. Those variables that remained for each of the data sets were analyzed using the method of backward logistic regression. The full model (the model consisting of all of the terms that met the criteria of mutual independence) was entered, and backward regression was used to reduce the list of variables to a reduced model which consists of only those variables which are statistically significant (P < 0.05). SPSS 25.0 statistical software, Armonk, NY, USA: IBM Corp. was used for all analyses, and a P < 0.05 was considered statistically significant.
RESULTS
Patient characteristics and demographics
There were 4,154 patients who underwent both an EGD and GES at Cleveland Clinic from January 2011 to December 2016, of which 3,290 patients had a normal GES (79.2%) and 864 patients had gastroparesis (20.8%). Clinical characteristics of the study patients are present in Table 1. The study population was predominantly female (72%), white (79.9%) with a mean age of 51.4 years. Gastroparesis patients were younger, had a lower BMI, and had a higher prevalence of diabetes and hypertension than the controls. There were no significant differences in ethnicity, smoking history, illicit drug use, or other comorbidities such as hyperlipidemia, obstructive sleep apnea, and hypothyroidism. Approximately 39% of gastroparesis patients had a history of alcohol use compared to 48.7% of the control group. Patients with gastroparesis had more symptoms of delayed emptying including nausea/vomiting, early satiety, bloating, epigastric pain, and weight loss. For GERD symptoms, among the gastroparesis group (n = 864), the prevalence of typical symptoms only was heartburn in 12 (1.3%) and acid regurgitation in 180 (20.8%), whereas atypical symptoms only were chest pain in 71 (8.2%) and dysphagia in 74 (8.56%) patients. Among the control group, the prevalence of typical symptoms only was heartburn in 105 (3.2%) and acid regurgitation in 749 (22.76%), and atypical symptoms were chest pain in 216 (6.56%) and dysphagia in 197 (6%) patients. There were no differences in the rate of usage of PPIs, aspirin, NSAIDs, statins, or opiates; however, the use of H2RA and prokinetics was more prevalent in the gastroparesis group.
Table 1.
Baseline characteristics, medications use, and endoscopic and histological findings of gastroparesis patients and control group
| No gastroparesis N = 3,290 (%) | Gastroparesis N = 864 (%) | P-value | ||
|---|---|---|---|---|
| Age (years) | 51.6 ± 16.4 | 50.3 ± 16.3 | 0.032a | |
| Gender | Males vs. females | 927 (28.2) vs. 2,363 (71.8) | 237 (27.4) vs. 627 (72.6) | 0.66b |
| Race | Caucasians | 2,635 (80.1) | 686 (79.4) | 0.080b |
| African–American | 447 (13.6) | 128 (14.8) | ||
| Asians | 37 (1.1) | 2 (0.23) | ||
| Others | 171 (5.2) | 48 (5.6) | ||
| Smoking | Never/former/active | 1,821 (55.3)/363 (11.0)/1,106 (33.6) | 469 (54.3)/98 (11.3)/297 (34.4) | 0.85b |
| Alcohol | Never/former/active | 1,688 (51.3)/119 (3.6)/1,483 (45.1) | 527 (61.0)/36 (4.2)/301 (34.8) | |
| Body mass index | 28.3 ± 7.1 | 27.5 ± 7.2 | 0.002 a | |
| Symptoms | Heartburn | 318 (9.7) | 69 (8.0) | 0.13b |
| Acid regurgitation | 2,161 (65.7) | 552 (63.9) | 0.32b | |
| Sour taste | 58 (1.8) | 15 (1.7) | 0.96b | |
| Chest pain | 1,246 (37.9) | 310 (35.9) | 0.28b | |
| Dysphagia | 1,118 (34.0) | 341 (39.5) | 0.003 b | |
| Nausea/vomiting | 1,324 (40.2) | 478 (55.3) | <0.001 b | |
| Early satiety | 598 (18.2) | 119 (13.8) | 0.002 b | |
| Bloating | 1,116 (33.9) | 253 (29.3) | 0.010 b | |
| Epigastric pain | 1,414 (43.0) | 305 (35.3) | <0.001 b | |
| Weight loss | 949 (28.8) | 292 (33.8) | 0.005 b | |
| Comorbidities | Hypertension | 1,551 (47.1) | 462 (53.5) | <0.001 b |
| Diabetes | 715 (21.7) | 275 (31.8) | <0.001 b | |
| Hyperlipidemia | 942 (28.6) | 229 (26.5) | 0.22b | |
| Obstructive sleep apnea | 564 (17.1) | 144 (16.7) | 0.74b | |
| Hypothyroidism | 418 (12.7) | 126 (14.6) | 0.15b | |
| Medications | H2 blocker | 311 (9.5) | 104 (12.0) | 0.024 b |
| PPI | 498 (15.1) | 140 (16.2) | 0.44b | |
| Prokinetics | 472 (14.3) | 179 (20.7) | <0.001 b | |
| Metoclopramide | 390 (11.9) | 145 (16.8) | <0.001 b | |
| Azithromycin | 60 (1.8) | 20 (2.3) | 0.35b | |
| Erythromycin | 18 (0.55) | 10 (1.2) | 0.051b | |
| Domperidone | 2 (0.06) | 2 (0.23) | 0.19c | |
| Aspirin | 390 (11.9) | 106 (12.3) | 0.74b | |
| NSAID | 3 (0.09) | 0 (0.0) | 0.99c | |
| Statin | 240 (7.3) | 72 (8.3) | 0.30b | |
| Opioid | 1,331 (40.5) | 373 (43.2) | 0.15b | |
| Endoscopic and histological findings | Hiatal hernia | 99 (3.0) | 25 (2.9) | 0.86b |
| Erosive esophagitis | 17 (0.52) | 1 (0.12) | 0.11b | |
| Esophageal stricture | 30 (0.91) | 10 (1.2) | 0.51b | |
| Barrett’s esophagus | 71 (2.2) | 18 (2.1) | 0.89b | |
| Length of Barrett’s segment (cm) | 3.59 ± 3.26 | 2.61 ± 1.72 | 0.80a | |
| H. pylori infection | 206 (6.3) | 35 (4.1) | 0.013 b | |
| Procedures | Botox injection | 8 (0.24) | 15 (1.7) | <0.001 b |
| Pyloromyotomy | 20 (0.61) | 25 (2.9) | <0.001 b | |
| Pyloroplasty | 15 (0.46) | 28 (3.2) | <0.001 b | |
| Gastrojejunostomy | 55 (1.7) | 22 (2.5) | 0.090 b |
Statistics presented as mean ± SD, median (P25, P75), median (min, max) or N (column %). P-values:
aANOVA
bPearson’s chi-square test
cFisher’s exact test. PPI, proton pump inhibitor
Endoscopic and histological findings and therapeutic interventions
There were 18 patients (2.1%) with BE in the gastroparesis group compared to 71 patients with BE (2.2%) in the control group (P = 0.89). Both groups had similar prevalence rates of hiatal hernia, erosive esophagitis, as well as esophageal strictures (Table 1). Pyloromyotomy, pyloroplasty, and endoscopic botulinum toxin injection were more common in the gastroparesis group compared to the control group. Among patients with gastroparesis, there were 28 patients who underwent pyloroplasty, of which 26 patients had a follow-up after their procedure with overall improvement in 14 patients (54%). Fourteen patients had a follow-up GES, of which 8/14 (57.14%) showed improvement in the gastric emptying.
BE in gastroparesis group
Among patients with gastroparesis, 18 patients had BE and 846 patients did not. As expected, patients with BE were more likely to be male, white, older, and heavier than their counterparts (Table 2). Acid regurgitation was more prevalent in the BE group, but the prevalence of other symptoms of GERD and gastroparesis was similar in both groups. There were no significant differences in the comorbidities and use of PPIs, H2RA, prokinetics, aspirin, or NSAIDs between both groups. Among the therapeutic interventions, 16.7% in the BE group had pyloroplasty compared to 3% in the non-BE group (P = 0.001). The average time between pyloroplasty and an EGD follow-up was 22.15 ± 19.64 months. The prevalence rate of BE was 71/3,288 (2.16%), 3/320 (0.93%), 8/200 (4%), 4/169 (2.36%), and 3/177 (1.69%) in patients without gastroparesis, mild gastroparesis, moderate gastroparesis, severe gastroparesis, and very severe gastroparesis, respectively.
Table 2.
Baseline characteristics of patients with gastroparesis
| Factor | Non-BE N = 846 (%) | BE N = 18 (%) | P-value | |
|---|---|---|---|---|
| Age | 50.2 ± 16.4 | 55.9 ± 11.3 | 0.14a | |
| Gender | Males vs. females | 228 (27.0) vs. 618 (73.0) | 9 (50.0) vs. 9 (50.0) | 0.030c |
| Race | Caucasians | 686 (79.4) | 670 (79.2) | 0.65d |
| African-American | 127 (15.0) | 1 (5.6) | ||
| Asians | 2 (0.24) | 0 (0.0) | ||
| Others | 47 (5.6) | 1 (5.6) | ||
| Smoking | Never/former/active | 456 (53.9)/96 (11.3)/294 (34.8) | 13 (72.2)/2 (11.1)/3 (16.7) | 0.25c |
| Alcohol | Never/former/active | 518 (61.2)/33 (3.9)/295 (34.9) | 9 (50.0)/3 (16.7)/6 (33.3) | 0.026c |
| Body mass index | 27.4 ± 7.2 | 28.9 ± 6.7 | 0.002a | |
| Symptoms | Heartburn | 66 (7.8) | 3 (16.7) | 0.17a |
| Acid regurgitation | 536 (63.4) | 16 (88.9) | 0.026a | |
| Sour taste | 14 (1.7) | 1 (5.6) | 0.21a | |
| Chest pain | 302 (35.7) | 8 (44.4) | 0.44a | |
| Dysphagia | 331 (39.1) | 10 (55.6) | 0.16a | |
| Comorbidities | Hypertension | 453 (53.5) | 9 (50.0) | 0.77a |
| Diabetes mellitus | 272 (32.2) | 3 (16.7) | 0.16a | |
| Hyperlipidemia | 223 (26.4) | 6 (33.3) | 0.51a | |
| Obstructive sleep apnea | 140 (16.5) | 4 (22.2) | 0.52a | |
| Hypothyroidism | 123 (14.5) | 3 (16.7) | 0.80a | |
| Medications | H2 blocker | 102 (12.1) | 2 (11.1) | 0.90a |
| PPI | 139 (16.4) | 1 (5.6) | 0.22a | |
| Prokinetics | 177 (20.9) | 2 (11.1) | 0.31a | |
| Metoclopramide | 143 (16.9) | 2 (11.1) | 0.52a | |
| Azithromycin | 20 (2.4) | 0 (0.0) | 0.51a | |
| Erythromycin | 10 (1.2) | 0 (0.0) | 0.64a | |
| Domperidone | 2 (0.24) | 0 (0.0) | 0.99b | |
| Aspirin | 102 (12.1) | 4 (22.2) | 0.19a | |
| Statins | 69 (8.2) | 3 (16.7) | 0.20a | |
| Opioids | 364 (43.0) | 9 (50.0) | 0.55a | |
| Endoscopic findings | Hiatal hernia | 22 (2.6) | 3 (16.7) | <0.001a |
| Erosive esophagitis | 1 (0.12) | 0 (0.0) | 0.99b | |
| Esophageal stricture | 10 (1.2) | 0 (0.0) | 0.64a | |
| H. pylori infection | 34 (4.0) | 1 (5.6) | 0.74a | |
| Procedures | Botox injection | 15 (1.8) | 0 (0.0) | 0.57c |
| Pyloromyotomy | 24 (2.8) | 1 (5.6) | 0.50c | |
| Pyloroplasty | 25 (3.0) | 3 (16.7) | 0.001a | |
| Gastrojejunostomy | 22 (2.6) | 0 (0.0) | 0.49c |
N (%). P-values:
a,cPearson’s chi-square test
b,dFisher’s exact test. A significance level of 0.05 was used for pairwise ad hoc comparisons. BE, Barrett’s esophagus; PPI, proton pump inhibitor
Predictors of BE in gastroparesis
A multivariate logistic regression analysis was performed to assess factors associated with BE in gastroparesis patients (Table 3). In the full model, history of alcohol use, statin use, and pyloroplasty was significant with a trend toward significance with hiatal hernia, smoking, and PPI use. In the reduced model, predictors of BE among gastroparesis patients were history of alcohol use (odds ratio [OR] 6.76; confidence intervals [CI]: 1.65–27.67, P = 0.008), history of pyloroplasty (OR: 8.228; CI: 2.114–32.016, P = 0.002), and hiatal hernia (OR: 8.014; CI: 2.053–31.277, P = 0.003).
Table 3.
Predictors of Barrett’s esophagus in gastroparesis—full model using multivariate analysis
| Predictor | Odds ratio | 95% confidence interval | P-value | |
|---|---|---|---|---|
| Demographics | Age | 1.018 | 0.978–1.060 | 0.389 |
| Smoker (former vs. never) | 0.399 | 0.065–2.46 | 0.322 | |
| Smoker (current vs. never) | 0.266 | 0.066–1.078 | 0.064 | |
| Alcohol (former vs. never) | 11.316 | 1.836–69.758 | 0.009 | |
| Alcohol (current vs. never) | 0.848 | 0.259–2.781 | 0.786 | |
| Symptoms | Heartburn | 1.79 | 0.386–8.296 | 0.457 |
| Acid regurgitation | 2.513 | 0.493–12.816 | 0.268 | |
| Sour taste | 5.913 | 0.417–83.93 | 0.189 | |
| Chest pain | 1.012 | 0.296–3.458 | 0.984 | |
| Dysphagia | 1.425 | 0.461–4.403 | 0.538 | |
| Nausea/vomiting | 0.62 | 0.182–2.109 | 0.444 | |
| Early satiety | 0.951 | 0.17–5.325 | 0.954 | |
| Bloating | 0.819 | 0.249–2.695 | 0.742 | |
| Epigastric pain | 0.931 | 0.278–3.122 | 0.908 | |
| Weight loss | 0.364 | 0.087–1.523 | 0.167 | |
| Medications | H2 blocker | 0.906 | 0.129–6.348 | 0.921 |
| Proton pump inhibitors | 0.09 | 0.008–1.055 | 0.055 | |
| Metoclopramide | 0.482 | 0.084–2.765 | 0.413 | |
| Aspirin | 2.885 | 0.672–12.397 | 0.154 | |
| Statins | 4.147 | 0.802–21.435 | 0.09 | |
| Opioids | 1.962 | 0.626–6.151 | 0.247 | |
| Comorbidities | Hypertension | 0.755 | 0.224–2.539 | 0.65 |
| Diabetes | 0.302 | 0.065–1.411 | 0.128 | |
| Hyperlipidemia | 1.448 | 0.342–6.136 | 0.615 | |
| Obstructive sleep apnea | 1.078 | 0.253–4.587 | 0.919 | |
| Hypothyroidism | 1.108 | 0.254–4.830 | 0.891 | |
| Procedures | Pyloromyotomy | 1.333 | 0.106–16.759 | 0.824 |
| Pyloroplasty | 19.453 | 3.589–105.439 | 0.001 | |
| Endoscopic findings | Hiatal hernia | 4.998 | 0.925–27.009 | 0.062 |
| H. pylori infection | 0.982 | 0.071–13.523 | 0.989 |
Predictors of BE in controls
A multivariate analysis of factors associated with BE in controls is presented in Table 4. In the reduced model, increasing age (OR 1.03 for every 1-year increase, CI 1.01,1.04, P = 0.002), smoking (OR 1.9, CI 1.15, 3.19, P = 0.012), illicit drug use (OR 6.3, CI 1.32,30.22, P = 0.021), acid regurgitation (OR 2.31, CI 1.22,4.39, P = 0.01), nausea/vomiting (OR 0.45, CI 0.24, 0.85, P = 0.014), epigastric pain (OR 0.39, CI 0.22, 0.69, P = 0.001), and hiatal hernia (OR 3.06, CI 1.38,6.83, P = 0.006) remained significant.
Table 4.
Predictors of Barrett’s esophagus for controls—full model using multivariate analysis
| Predictor | Odds ratio | 95% confidence interval | P-value | |
|---|---|---|---|---|
| Demographics | Age | 1.032 | 1.013–1.053 | 0.001 |
| Smoker (former vs. never) | 0.855 | 0.342–2.141 | 0.739 | |
| Smoker (current vs. never) | 1.945 | 1.154–3.277 | 0.012 | |
| Alcohol (former vs. never) | 0.354 | 0.047–2.687 | 0.315 | |
| Alcohol (current vs. never) | 0.864 | 0.522–1.431 | 0.570 | |
| Symptoms | Heartburn | 1.766 | 0.906–3.440 | 0.095 |
| Acid regurgitation | 2.286 | 1.163–4.493 | 0.017 | |
| Sour taste | 1.899 | 0.425–8.484 | 0.401 | |
| Chest pain | 0.965 | 0.544–1.712 | 0.903 | |
| Dysphagia | 1.461 | 0.872–2.447 | 0.150 | |
| Nausea/vomiting | 0.504 | 0.263–0.969 | 0.04 | |
| Early satiety | 0.850 | 0.419–1.728 | 0.654 | |
| Bloating | 0.582 | 0.303–1.115 | 0.102 | |
| Epigastric pain | 0.446 | 0.242–0.824 | 0.010 | |
| Weight loss | 1.340 | 0.766–2.344 | 0.305 | |
| Medications | H2 blocker | 1.688 | 0.765–3.727 | 0.195 |
| Proton pump inhibitors | 1.202 | 0.607–2.382 | 0.597 | |
| Metoclopramide | 0.636 | 0.251–1.615 | 0.341 | |
| Azithromycin | 0.725 | 0.092–5.724 | 0.760 | |
| Erythromycin | 3.906 | 0.452–33.766 | 0.216 | |
| Aspirin | 0.940 | 0.425–2.081 | 0.879 | |
| Statins | 0.851 | 0.332–2.181 | 0.736 | |
| Opioids | 0.912 | 0.532–1.565 | 0.739 | |
| Comorbidities | Hypertension | 0.756 | 0.414–1.381 | 0.363 |
| Diabetes | 1.046 | 0.531–2.061 | 0.897 | |
| Hyperlipidemia | 0.906 | 0.474–1.734 | 0.766 | |
| Obstructive sleep apnea | 1.161 | 0.580–2.324 | 0.672 | |
| Hypothyroidism | 0.799 | 0.361–1.770 | 0.581 | |
| Procedures | Pyloromyotomy | 6.185 | 1.145–33.411 | 0.034 |
| Gastrojejunostomy | 0.901 | 0.114–7.107 | 0.921 | |
| Endoscopic findings | Hiatal hernia | 2.767 | 1.213–6.310 | 0.016 |
| Erosive esophagitis | 2.032 | 0.352–11.743 | 0.428 | |
| H. pylori infection | 1.076 | 0.371–3.119 | 0.893 |
DISCUSSION
Delayed gastric emptying can lead to severe GERD by prolonging the gastric distension and increasing the volume of gastric refluxate and the frequency of transient LES relaxations. Since BE is a manifestation of severe GERD, gastroparesis can potentially contribute to the development of BE. In this study of 4,154 patients who underwent a GES and an EGD, 3,290 (79.2%) patients had normal gastric emptying, and 864 (20.8%) patients had gastroparesis. Gastroparesis does not appear to increase the risk of BE as there was no difference in the prevalence of BE among both groups; 18 (2.1%) patients with gastroparesis had BE compared to 71 (2.2%) patients with normal gastric emptying. Among patients with gastroparesis, the risk factors for BE appear to be a history of pyloroplasty, the presence of a hiatal hernia, and a history of alcohol use. Patients with a hiatal hernia and a history of pyloroplasty were eight times more likely to have BE, and patients with alcohol use were seven times more likely to have BE.
A review of prior literature confirms this lack of association between BE and gastroparesis. There are four studies from the 1980s to 1990s involving a small number of patients and varying non-standardized techniques for evaluation of gastric emptying.12–15 Johnson et al. evaluated 17 BE patients using scintigraphy study for solid emptying, of which 12 patients (70%) had a normal gastric emptying, 2 patients showed a delayed emptying, and 3 patients displayed an accelerated rate of gastric emptying.12 In a study of 10 patients with BE, 7 patients with reflux esophagitis, and 10 age- and sex-matched controls, the liquid gastric emptying rates were similar among all groups.13 In another study of 39 patients with an increased esophageal acid exposure, of which 15 had BE, there was no difference in the gastric emptying of radiolabeled oatmeal between patients with BE compared to patients with reflux esophagitis.14 A fourth study of 26 patients with BE reached a similar conclusion.15
The lack of association between gastroparesis and BE prompts the question whether delayed gastric emptying leads to more severe reflux disease or not. Among 20 studies with a total of 670 patients, 14 studies concluded that significantly delayed gastric emptying was found in GERD patients.6 If delayed gastric emptying is thought to be a culprit in the pathogenesis of GERD, then an association should exist between gastric retention and the severity of GERD. In most published studies including those that showed a positive association, there was no clear correlation between the intensity of gastric stasis and the severity of the reflux or between the LES pressure and the severity of gastric stasis.6,17 The reason for this discrepancy can be attributed to two factors: Firstly, it is not the total gastric emptying per se but the proximal gastric retention which contributes to increased 24 hour and postprandial acid exposure as well as an increased number of reflux episodes.8 Secondly, slower gastric emptying is associated with increasing pH of the gastric refluxate. In a study of 30 subjects who underwent simultaneous gastric emptying and pH impedance study, there was no relationship between the esophageal acid exposure and the gastric emptying parameters, but the longer the emptying time (total/proximal or subcardial emptying), the higher the number of weakly acidic reflux events.11 In fact, delayed gastric emptying occurred with equal frequency in a study of 76 patients with GERD and 38 healthy controls. Of note, reflux esophagitis was less prevalent in the delayed gastric emptying group than the normal emptying group, suggesting that prolonged retention of food may have a buffering effect on gastric acidity.10 Therefore, we can conclude that gastroparesis is not a culprit factor but an associated disorder in patients with GERD or BE.
In our study, we also investigated the risk factors for BE among patients with gastroparesis and found that a history of pyloroplasty, alcohol use, and hiatal hernia were significantly associated with BE. The association between pyloroplasty and the risk of BE is intriguing. Pyloroplasty, due to the destruction of gastroduodenal barrier, increases the risk of duodenogastric reflux. The evidence demonstrating this relationship comes from studies performed in patients with peptic ulcer disease treated by vagotomy with pyloroplasty.18,19 In a study of 14 patients who had pyloroplasty, 42% of patients had enterogastric reflux and delayed gastric emptying time, leading to bile reflux into the esophagus.18 In another study, both an increased bile acid concentration in gastric aspirates and the percentage of time with pH > 7 on esophageal pH monitoring were observed in patients with a destroyed gastroduodenal barrier (status post-Billroth I or II resection or pyloroplasty).19
This duodenogastric reflux plays a crucial role in the development of BE in GERD patients. The damaging effects of bile salts depend upon the intraluminal pH. At a pH between 3 and 6, as in the esophagus in the presence of acid, bile salts are soluble and nonionized and can enter cells causing direct injury.20 However, while in the stomach, at a pH <3, bile salts precipitate out of the solution and are not deleterious, and in the duodenum, they are ionized and cannot cross the cell membrane. Bile acids can cause esophageal squamous cells to develop certain intestinal-like gene expression patterns similar to those found in BE.21 For example, bile acids cause squamous cells to express CDX2 (a gene known to play a key role in the development of intestinal epithelia), BMP4 (a member of the transforming growth factor-β family that might promote squamous-to-columnar metaplasia), and MUC2 (a mucin normally found in intestinal goblet cells). Furthermore, in esophageal cell cultures, p63 protein levels (a marker for esophageal squamous progenitor cells) declines when the cells are exposed to bile acids, suggesting that bile acids may affect the progenitor cells responsible for maintaining normal squamous epithelium. This may explain the increased risk of BE in gastroparesis patients with a history of pyloroplasty with higher esophageal exposure to bile acids in the absence of the pyloric sphincter. In contrast, we did not observe any association between gastrojejunostomy and BE. This is consistent with prior studies which showed lack of association 22 and, in fact, regression of BE in obese patients.23–26
Even though alcohol worsens GERD by reducing the LES pressure and causing direct esophageal mucosal injury, alcohol use has not been associated with a significant risk of BE, and in fact, wine consumption may be protective, with ORs ranging from 0.44 (95% CI 0.2–0.99) to 0.71 (95% CI 0.52–0.98).27 However, a U-shaped curve may exist with a protective effect at low/moderate consumption versus a higher prevalence with heavy alcohol use.28 Veugelers et al. demonstrated an increased risk of GERD, BE, and esophageal adenocarcinoma in patients who reported high liquor consumption (>40 drinks/month),29 and Ronkainen et al. showed that increased alcohol use (>50 g absolute alcohol per week) is a risk factor for BE.30 In a meta-analysis of 20 studies, alcohol consumption was associated with a higher risk of developing BE in men, in Asian population, and in patients with GERD.31 It might be that gastric stasis and the prolonged retention of alcohol could increase the risk of BE in gastroparesis patients, also the type of alcohol might have a role as well. On the other hand, hiatal hernia is a well-established risk factor for BE (OR 3.94; 95% CI 3.02–5.13).32 By lowering the LES pressure and prolonging the esophageal acid exposure, a hiatal hernia can increase the propensity of reflux, thereby increasing the risk for BE.
One of the main strengths of the study is the stringent criteria used for inclusion in the study. All patients underwent 4-hour standardized solid-phase GES as defined by the American Neurogastroenterology and Motility Society and the Society of Nuclear Medicine.16 All patients had an EGD performed within our institution, and BE was defined per ACG criteria as at least 1-centimeter extension of columnar mucosa into the esophagus with the presence of intestinal metaplasia on histology.3 Even though this large group of patients was identified by natural language programming, all cases were confirmed by a manual review. There are, however, several limitations to the study. It is a retrospective analysis of patients seen in a tertiary referral center, which may introduce an inherent bias. The cause and effect relationship between pyloroplasty and BE cannot be clearly established as this is a retrospective study design and the time between pyloroplasty and EGD showing BE was relatively short. An ideal study would be a prospective study where patients with gastroparesis are randomized to pyloroplasty versus other drainage treatments such as pyloromyotomy. Another limitation is the lack of assessment of fundic and antral emptying and their contribution to esophageal reflux. Also, we did not assess for liquid gastric emptying; however, liquid gastric emptying tests are generally not clinically useful, because normal emptying of liquids is frequently maintained despite very severe gastroparesis for solids.33 Also, it includes patients with gastroparesis of different etiologies such as diabetic, idiopathic, and postsurgical which may impact the study results. Another possibility is that the duration of the study is not long enough to detect any effect that gastroparesis may have on BE and it is difficult to ascertain the date of onset of BE or gastroparesis based on the clinical history.
In summary, in this study, there was no increased risk of BE in patients with gastroparesis compared to the normal gastric emptying group. Predictors of BE among patients with gastroparesis were alcohol use, pyloroplasty, and the presence of a hiatal hernia. Further prospective studies are needed to confirm the validity of these findings and delineate if the gastric motor pattern and proximal delayed gastric emptying would influence the development of BE compared to delayed distal or total gastric emptying.
FUNDING
Prashanthi Thota is supported by NIH grants: U54CA163060 and P50CA150964. John Goldblum received support from NIH IR44CA192416. All other authors disclose no funding support.
AUTHOR CONTRIBUTIONS
Motasem Alkhayyat, Vedha Sanghi, Thabet Qapaja, Carol Rouphael: study concept and design; acquisition of data; analysis and interpretation of data; drafting of the manuscript; Robert Butler, John McMichael: acquisition of data; analysis and interpretation of data statistical analysis; John Goldblum, Madhusudhan R. Sanaka: study concept and design, critical revision of the manuscript for important intellectual content; Prashanthi N. Thota: study concept and design; acquisition of data; analysis and interpretation of data; drafting of the manuscript; administrative, technical, or material support; study supervision.
Conflict of interest: No personal or financial conflict of interest to declare for all authors.
References
- 1. Pasricha P J, Parkman H P. Gastroparesis: definitions and diagnosis. Gastroenterol Clin North Am 2015; 44: 1–7. [DOI] [PubMed] [Google Scholar]
- 2. Rey E, Choung R S, Schleck C D et al. Prevalence of hidden gastroparesis in the community: the gastroparesis "iceberg". J Neurogastroenterol Motil 2012; 18: 34–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Shaheen N J, Falk G W, Iyer P G et al. American College of Gastroenterology. ACG clinical guideline: diagnosis and Management of Barrett's esophagus. Am J Gastroenterol 2016; 111(1): 30–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Jehangir A, Parkman H P. Reflux symptoms in gastroparesis: correlation with gastroparesis symptoms, gastric emptying, and esophageal function testing. J Clin Gastroenterol 2020; 54: 428–38. [DOI] [PubMed] [Google Scholar]
- 5. Savarino E, Bredenoord A J, Fox M et al. Expert consensus document: advances in the physiological assessment and diagnosis of GERD. Nat Rev Gastroenterol Hepatol 2017; 14: 665–76. [DOI] [PubMed] [Google Scholar]
- 6. Heading R C, Baldi F, Holloway R H et al. Prokinetics in the treatment of gastro-oesophageal reflux disease. International symposium. Paris, France, 5 September 1996. Eur J Gastroenterol Hepatol 1998; 10(1): 87–93. [DOI] [PubMed] [Google Scholar]
- 7. Penagini R, Hebbard G, Horowitz M et al. Motor function of the proximal stomach and visceral perception in gastro-oesophageal reflux disease. Gut 1998; 42: 251–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Stacher G, Lenglinger J, Bergmann H et al. Gastric emptying: a contributory factor in gastro-oesophageal reflux activity? Gut 2000; 47: 661–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Zerbib F, des Varannes S B, Ropert A et al. Proximal gastric tone in gastro-oesophageal reflux disease. Eur J Gastroenterol Hepatol 1999; 11(5): 511–5. [DOI] [PubMed] [Google Scholar]
- 10. Schwizer W, Hinder R A, DeMeester T R. Does delayed gastric emptying contribute to gastroesophageal reflux disease? Am J Surg 1989; 157(1): 74–81. [DOI] [PubMed] [Google Scholar]
- 11. Sifrim D, Verfaillie M, Flamen P et al. The acidity of postprandial gastroesophageal reflux is determined by the rate of gastric emptying [abstract]. Gastroenterology 2004; 126(Suppl): A-500. [Google Scholar]
- 12. Johnson D A, Winters C, Drane W E et al. Solid-phase gastric emptying in patients with Barrett's esophagus. Dig Dis Sci 1986; 31(11): 1217–20. [DOI] [PubMed] [Google Scholar]
- 13. Mulholland M W, Reid B J, Levine D S et al. Elevated gastric acid secretion in patients with Barrett's metaplastic epithelium. Dig Dis Sci 1989; 34(9): 1329–34. [DOI] [PubMed] [Google Scholar]
- 14. Stein H J, Hoeft S, DeMeester T R. Functional foregut abnormalities in Barrett's esophagus. J Thorac Cardiovasc Surg 1993; 105(1): 107–11. [PubMed] [Google Scholar]
- 15. Kogan F J, Kotler J A, Sampliner R E. Normal gastric emptying in Barrett's esophagus. Gastroenterology 1985; 88: 1451A. [Google Scholar]
- 16. Abell T L, Camilleri M, Donohoe K et al. American Neurogastroenterology and motility society and the Society of Nuclear Medicine. Consensus recommendations for gastric emptying scintigraphy: a joint report of the American Neurogastroenterology and motility society and the Society of Nuclear Medicine. J Nucl Med Technol 2008; 36(1): 44–54. [DOI] [PubMed] [Google Scholar]
- 17. Maddern G J, Chatterton B E, Collins P J et al. Solid and liquid gastric emptying in patients with gastro-oesophageal reflux. Br J Surg 1985; 72(5): 344–7. [DOI] [PubMed] [Google Scholar]
- 18. Stavraka A, Madan A K, Frantzides C T et al. Gastric emptying time, not enterogastric reflux, is related to symptoms after upper gastrointestinal/biliary surgery. Am J Surg 2002; 184(6): 596–9. [DOI] [PubMed] [Google Scholar]
- 19. Stein H J, Feussner H, Kauer W et al. Alkaline gastroesophageal reflux: assessment by ambulatory esophageal aspiration and pH monitoring. Am J Surg 1994; 167(1): 163–8. [DOI] [PubMed] [Google Scholar]
- 20. Schweitzer E J, Bass B L, Batzri S et al. Bile acid accumulation by rabbit esophageal mucosa. Dig Dis Sci 1986; 31(10): 1105–13. [DOI] [PubMed] [Google Scholar]
- 21. McQuaid K R, Laine L, Fennerty M et al. Systematic review: the role of bile acids in the pathogenesis of gastro-oesophageal reflux disease and related neoplasia. Aliment Pharmacol Ther 2011; 34(2): 146–65. [DOI] [PubMed] [Google Scholar]
- 22. Avidan B, Sonnenberg A, Schnell T G, Sontag S J. Gastric surgery is not a risk for Barrett's esophagus or esophageal adenocarcinoma. Gastroenterology 2001; 121(6): 1281–5. [DOI] [PubMed] [Google Scholar]
- 23. Andrew B, Alley J B, Aguilar C E, Fanelli R D. Barrett's esophagus before and after roux-en-Y gastric bypass for severe obesity. Surg Endosc 2018; 32(2): 930–6. [DOI] [PubMed] [Google Scholar]
- 24. Gorodner V, Buxhoeveden R, Clemente G, Sánchez C, Caro L, Grigaites A. Barrett’s esophagus after roux-en-Y gastric bypass: does regression occur? Surg Endosc 2017; 31(4): 1849–54. [DOI] [PubMed] [Google Scholar]
- 25. Adil M T, Al-taan O, Rashid F et al. A systematic review and meta-analysis of the effect of roux-en-Y gastric bypass on Barrett’s esophagus. Obes Surg 2019; 29(11): 3712–21. [DOI] [PubMed] [Google Scholar]
- 26. Felsenreich D M, Langer F B, Bichler C et al. Roux-en-Y gastric bypass as a treatment for Barrett’s esophagus after sleeve gastrectomy. Obes Surg 2020; 30(4): 1273–9. [DOI] [PubMed] [Google Scholar]
- 27. Thrift A P, Cook M B, Vaughan T L et al. Alcohol and the risk of Barrett's esophagus: a pooled analysis from the international BEACON consortium. Am J Gastroenterol 2014; 109(10): 1586–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Filiberti R A, Fontana V, De Ceglie A et al. Alcohol consumption pattern and risk of Barrett's oesophagus and erosive oesophagitis: an Italian case-control study. Br J Nutr 2017; 117(8): 1151–61. [DOI] [PubMed] [Google Scholar]
- 29. Veugelers P J, Porter G A, Guernsey D L et al. Obesity and lifestyle risk factors for gastroesophageal reflux disease, Barrett esophagus and esophageal adenocarcinoma. Dis Esophagus 2006; 19(5): 321–8. [DOI] [PubMed] [Google Scholar]
- 30. Ronkainen J, Aro P, Storskrubb T et al. Prevalence of Barrett's esophagus in the general population: an endoscopic study. Gastroenterology 2005; 129: 1825–31. [DOI] [PubMed] [Google Scholar]
- 31. Xu Q, Guo W, Shi X et al. Association between alcohol consumption and the risk of Barrett's esophagus: a meta-analysis of observational studies. Medicine (Baltimore) 2015; 94: e1244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Andrici J, Tio M, Cox M R et al. Hiatal hernia and the risk of Barrett's esophagus. J Gastroenterol Hepatol 2013; 28(3): 415–31. [DOI] [PubMed] [Google Scholar]
- 33. Chaudhuri T K, Fink S. Gastric emptying in human disease states. Am J Gastroenterol 1991; 86(5): 533–8. [PubMed] [Google Scholar]