Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Gastroenterology. Author manuscript; available in PMC 2013 October 1.
Published in final edited form as:
PMCID: PMC3458136

Esomeprazole and 325 mg/d Aspirin Reduce Tissue Concentrations of Prostaglandin E2 in Patients with Barrett’s Esophagus


Background & Aims

Proton pump inhibitors (PPIs) and nonsteroidal anti-inflammatory drugs might prevent esophageal adenocarcinoma in patients with Barrett’s esophagus (BE), but there are limited data from clinical trials to support this concept. We conducted a randomized, double-blind, placebo-controlled phase II trial to assess the effects of the combination of aspirin (3 different doses) and esomeprazole on tissue concentrations of prostaglandin E2 (PGE2) in patients with BE with no dysplasia or low-grade dysplasia.


Participants were recruited through the multi-center Cancer Prevention Network and randomly assigned to groups that were given esomeprazole (40 mg, twice daily) in combination with an aspirin placebo (once daily) (Arm A; n=42), with 81 mg aspirin (once daily) (Arm B; n=63), or with 325 mg aspirin (once daily) (Arm C; n=63) for 28 days. We collected esophageal biopsies before and after the intervention period, to determine the absolute change in mean concentrations of PGE2 (the primary endpoint).


Based on data from 114 patients, baseline characteristics were similar among groups. The absolute mean tissue concentrations of PGE2 was reduced by 67.6±229.68 pg/mL in Arm A, was reduced by 123.9±284.0 pg/mL in Arm B (P=.10 vs Arm A), and was reduced by 174.9 ±263.62 pg/mL in Arm C (P=.02 vs Arm A).


In combination with esomeprazole, short-term administration of higher doses of aspirin, but not lower doses or no aspirin, significantly reduced tissue concentrations of PGE2 patients with BE with either no dysplasia or low-grade dysplasia. These data support further evaluation of higher doses of aspirin and esomeprazole to prevent esophageal adenocarcinoma in these patients.

Keywords: esophageal cancer, NSAIDs, inflammation, esophagus


Barrett’s esophagus (BE) is an acquired condition that affects about 1 in 20 United States adults.1 Esophageal adenocarcinoma, which has become increasingly common in developed countries over the past four decades, is strongly associated with BE.2 Based on currently available data, the annual incidence of esophageal adenocarcinoma in Barrett’s esophagus patients is estimated to range from approximately 0.3 to 0.5 percent.3-5 Unfortunately, the five-year survival rate for esophageal adenocarcinoma remains dismal (approximately 17%).6

Current strategies for improving clinical outcomes among patients with esophageal adenocarcinoma are largely focused on identifying malignant disease at an early, potentially curable stage. This could be accomplished through widespread adoption of endoscopic screening and/or surveillance programs. However, existing early detection guidelines endorsed by major subspecialty societies are inherently expensive, with a relatively limited evidence base.5 Thus, further evaluation of novel approaches to esophageal cancer risk reduction among BE patients is urgently needed.

Extensive preclinical data suggest that non-steroidal anti-inflammatory drugs (NSAIDs), particularly aspirin, may play a protective role against the development of esophageal adenocarcinoma by inhibiting the cyclooxygenase 1 and 2 (COX-1 and COX-2) enzymes, which regulate prostaglandin E2 (PGE2) production7-12. Upregulation of PGE2 has been associated with resistance to apoptosis, increased angiogenesis, and enhanced invasion in Barrett’s mucosa and other gastrointestinal neoplasia, thereby providing a strong biologic rationale for assessing PGE2 modulation as a surrogate endpoint biomarker for phase II trials.7-12

One proposed mechanism for Barrett’s-associated carcinogenesis is acid and/or bile salt induced COX-2 activation (with accompanying high levels of PGE2 production), which would support the chemopreventive potential of a proton pump inhibitor (PPI) given in combination with aspirin. However, to date, the chemopreventive potential of PPI and/or NSAID therapy for Barrett’s-associated esophageal adenocarcinoma remains inconclusive,5 and the combination approach has not been evaluated in a randomized clinical trial. For this multi-center, double-blind, placebo controlled phase II clinical trial, we hypothesized that a PPI plus higher dose aspirin would be more effective for downregulating PGE2 than either a PPI alone or a PPI plus lower dose aspirin in BE patients.


All aspects of the study protocol were reviewed and approved by the appropriate Institutional Review Board for human research at each participating site. Mayo Clinic in Rochester, MN, served as the coordinating research base. The Data and Safety Monitoring Board of the Mayo Clinic Cancer Center reviewed safety data every 6 months and efficacy data within 6 months of the planned interim analysis.

General Overview

Participants were recruited at 12 Cancer Prevention Network (CPN) member organizations: Cleveland Clinic, Cleveland, OH; Fox Chase Cancer Center, Philadelphia, PA; Hines Veterans Administration, Hines, IL; Kaiser Permanente Medical Group, San Francisco, CA; Kansas City VA Medical Center, Kansas City, MO; Massachusetts General Hospital, Boston, MA; Mayo Clinic, Scottsdale, AZ,, Jacksonville, FL, and Rochester, MN; St. Michael’s Hospital, Toronto, Canada; University Hospitals Case Medical Center, Cleveland, OH; and the University of Puerto Rico Cancer Center, San Juan Puerto Rico). In total, 120 eligible participants were enrolled between May 2007 and January 2011.

Study Participants

The target population included adult patients (≥ 18 years) with histologically-confirmed BE, defined as the presence of specialized columnar epithelium anywhere in the tubular esophagus with ≥ 2 cm of circumferential involvement and no evidence of high-grade dysplasia or cancer based on both clinical surveillance and additional research biopsies. Participants were required to have documented intestinal metaplasia with goblet cells in ≥ 4 out of 8 research biopsy samples (with ≥ 50% intestinal metaplasia) and could not have used aspirin, NSAIDs, or selective COX-2 inhibitors for ≥ 7 days during the 3 month period preceding a 28 day run-in phase. Women of childbearing potential were required to document a negative pregnancy test prior to enrollment. General exclusion criteria were: inability to abstain from aspirin, NSAID, or selective COX-2 inhibitor therapy for the duration of the trial; current or planned use of anticoagulants; prior fundoplication, bariatric surgery or any other major upper GI surgery (except cholecystectomy); history of confirmed esophageal high-grade dysplasia or cancer; history of cancer diagnosis ≤ 24 months (except non-melanoma skin cancer or cervical carcinoma-in-situ) or cancer treatment ≤ 12 months (except hormone therapy) prior to run-in; receipt of other investigational agents ≤ 3 months prior to run-in (except compounds such as multivitamins, topically-applied skin creams, etc. with no known interaction with the study agents); history of allergic reactions attributed to compounds of similar chemical or biologic composition to the study agents or rescue medication; history of endoscopically or radiographically diagnosed peptic ulcer disease (bleeding or nonbleeding); pregnant or lactating women; or comorbidities that might limit adherence to the study protocol.

Baseline Evaluation and Run-In Phase

Following informed consent, willing participants completed a focused interview, brief questionnaire, physical exam, and peripheral blood draw for eligibility testing. Participants who met enrollment criteria entered a 28 day run-in phase, which included self-administration of esomeprazole 40 mg bid (i.e., before breakfast and dinner). Eighty percent adherence with the run-in agent, based on an agent diary and follow-up phone call at 14 (+/- 3) days, was required for advancement to randomization. A repeat blood draw and baseline endoscopy were scheduled for 28 (+/- 3) days after initiating the run-in phase.

Pre-Intervention Endoscopy

The pre-intervention endoscopy exam was performed with a therapeutic endoscope (Olympus GIF-1T140 videoendoscope or equivalent). Endoscopic landmarks including the diaphragmatic hiatus, end of the tubular esophagus as marked by the proximal margin of gastric folds, and new squamocolumnar junction were recorded. Extent of the circumferentially involved BE segment was determined using the Prague classification system.13 Hiatal hernia size was measured by linear distance between the end of the tubular esophagus and the diaphragmatic hiatus. Short segment BE was defined as specialized columnar epithelium lining < 3 cm of the distal esophagus, whereas long segment BE was defined as specialized columnar epithelium lining ≥ 3 cm of the distal esophagus.

Four quadrant endoscopic surveillance biopsies were obtained using Boston Scientific Radial Jaw® 3 forceps at 2-cm intervals along the entire length of the Barrett’s epithelium. Additional biopsy specimens were obtained from any ulcer, erosion, plaque, nodule, stricture, or other luminal irregularity. All surveillance biopsies were assessed by local pathology review. Endoscopic research biopsies (up to 8) were obtained with the same forceps from a 1 cm zone of Barrett’s epithelium 1 to 2 cm above the level of the lower esophageal sphincter, just above the proximal margin of the gastric folds. Research biopsies were washed for 5 seconds in PBS and divided into two aliquots, which were snap frozen in liquid nitrogen or placed in 10% neutral buffered formalin at room temperature, respectively. All formalin fixed research biopsies were assessed by the study pathologist (TCS). If high-grade dysplasia or cancer was detected in any research biopsy, two additional gastrointestinal pathologists at Mayo Clinic in Rochester, MN, were consulted. If the consensus diagnosis (including input from the local pathologist) was high-grade dysplasia or cancer, the participant was classified as a screen failure.

Findings of ulcer, erosion, plaque, nodule, stricture, or other luminal irregularity within the Barrett’s segment; erosive esophagitis > LA class A; high-grade dysplasia or cancer; or inadequate Barrett’s mucosa to satisfy the study endpoints (defined as < 4 out of 8 research samples with ≥ 50% intestinal metaplasia by central pathology review) excluded further study participation.

Intervention Assignments and On-Study Monitoring

Willing, eligible participants were randomly assigned to receive a 28 day intervention with one of three study agent combinations (using a 1:1.5:1.5 schema to permit increased access to an active aspirin intervention): Arm A (aspirin placebo + PPI) = aspirin 81 mg placebo qd + aspirin 325 mg placebo qd + esomeprazole 40 mg bid (n = 42); Arm B (lower dose aspirin + PPI) = aspirin 81 mg qd + aspirin 325 mg placebo qd + esomeprazole 40 mg bid (n = 63); and Arm C (higher dose aspirin + PPI) = aspirin 81 mg placebo qd + aspirin 325 mg qd + esomeprazole 40 mg bid (n = 63). Esomeprazole was self-administered once in the morning and once in the evening, at least one hour before a meal. Aspirin or aspirin placebo (indistinguishable from the activate agent) was administered once in the morning with a meal. Adherence was defined as ≥ 80% ingestion of the protocol-defined doses. Antacid tablets (200 mg aluminum hydroxide, 200 mg magnesium hydroxide, and 25 mg simethicone per each tablet) were provided as rescue medication for acute GERD symptoms. Use of > 12 antacid tablets per day and/or for > 14 days was noted as a protocol deviation. Adverse events were classified and graded using NCI Common Toxicity Criteria, version 3.0 (available at Attribution of agent-related adverse events was performed by the designated medical monitor (GDZ), who was blinded to the intervention assignments.

Post-Intervention Evaluation

Participants returned at 28 (+/- 3) days post-randomization to assess adherence, concomitant medication use, and adverse events. A focused physical exam and peripheral blood draw were also performed. Post-intervention endoscopy, research biopsies, and tissue handling were performed according to the standardized protocol applied at baseline.

Tissue Biomarkers

Research biopsy samples were shipped to Mayo Clinic Rochester for biomarker analyses. Tissue PGE2 levels were assessed by a standardized method using EIA.12 The intra-interassay coefficient of variance is <10%, with a specificity of 100%, and a limit of limit of detection 15 pg/mL. PGE2 levels were corrected for tissue weight and represented as ng/gm of tissue. Samples were run in two batches, with the pre- and post-intervention biopsies obtained from a given participant run on the same plate.

Statistical Analyses

The primary endpoint was defined as absolute change in mean tissue PGE2 concentration (post-intervention mean PGE2 from up to 3 biopsy samples – baseline mean PGE2 from up to 3 biopsy samples). Study participants were considered evaluable for the primary endpoint if they met eligibility criteria and PGE2 data were available from both the pre- and post-intervention evaluations. The 1:1.5:1.5 randomization schema was based on the Pocock-Simon dynamic allocation procedure,14 which stratified by gender, length of Barrett’s segment (< 5 cm versus ≥ 5 cm of circumferential involvement), baseline dysplasia status (none versus low-grade or indefinite), and participating site. Assuming equal standard deviations of 50 pg/mL for all 3 intervention arms, the sample sizes were determined to be n = 38, 57, and 57 in the aspirin placebo + PPI, lower dose aspirin + PPI, and higher dose aspirin + PPI arms, respectively (n = 152 total). These sample sizes yielded 80% power to detect a difference of at least 2/3 of a standard deviation in the mean PGE2 concentration comparing each active aspirin arm vs. PPI alone, with a 2-sided alpha of 0.025 (adjusted for multiple comparisons). At least a 2/3 change in the standard deviation was selected as the minimum threshold for a clinically significant change because it was consistent with the goals for endpoint modulation of phase II chemoprevention trials, and this value represented the consensus goal defined by the investigative team in conjunction with the study sponsor (NCI Division of Cancer Prevention) to strike a reasonable balance between subject recruitment and data interpretation.

An unblinded pre-planned variability assessment (without halting accrual) in the first 40 participants was performed. This analysis showed slightly higher than expected variation in the PGE2 values, where the standard deviation of the absolute change in PGE2 values within each arm varied from a low of 63 pg/mL to a high of 112 pg/mL. Despite the increased variability observed, we retained the original sample size due to the slower than anticipated accrual rate. In addition, due to concerns of slow accrual, we added interim futility and efficacy analyses so that the study could potentially close early at the time of the interim analyses. This interim analysis was planned after 77 participants were evaluable for PGE2 (n = 19, 29, and 29 in the aspirin placebo + PPI, lower dose aspirin + PPI, and higher dose aspirin + PPI arms), during which each active aspirin arm was compared to the aspirin placebo arm using a 2-sample t-test (or nonparametric equivalent). Interim efficacy and futility boundaries were developed based on the O’Brien-Fleming method from the Lan-DeMets family15 using EAST software developed by Cytel.16 Each pair wise comparison decision was considered individually. If the 2-sided p-value was found to be < 0.0008 in favor of the active treatment arm(s), then significant efficacy would be declared. If the 2-sided p-value was > 0.634, then significant futility would be declared. The alpha and beta spent at the interim look were 0.001 and 0.036, respectively, and at the final analysis, a p-value < 0.0247 for either pair wise comparison in favor of the active treatment arm(s) would reject the null hypothesis and indicate that the active aspirin arm(s) were effective compared to the aspirin placebo arm, as per standard convention.15

Secondary endpoints included comparison of the percent change in PGE2, baseline participant characteristics, agent adherence, and adverse events between intervention arms. Summary statistics and frequency tables were used to describe baseline subject characteristics, agent adherence, and adverse events. For participants that received more than the expected dose per protocol for any of the agents, we capped the percentage adherence numbers at 100%. Fisher’s Exact and Kruskal-Wallis tests were used to test for associations between intervention arms and categorical and continuous data, respectively. Absolute and percent changes in PGE2 were compared within each arm by the Wilcoxon signed rank procedure. The between-arm comparisons were carried out using the Wilcoxon rank-sum test. All statistical tests were performed 2-sided, unless otherwise noted using SAS version 9.2 (SAS Institute, Inc.).


Cohort Description

A CONSORT overview of participant recruitment is shown in Figure 1. One hundred eighty-one unique participants provided informed consent and were pre-registered for the baseline evaluation. A total of 59 participants were deemed screen failures. Reasons for screen failure included: inadequate Barrett’s mucosa (n = 27), < 2 cm of circumferential BE (n = 20), participant or physician decision (n = 7), non-allowed concomitant medications (n = 2), and failure to meet other eligibility criteria (n = 3). Registered, ineligible participants, and intervention cohort participants were similar with respect to age, sex, smoking history, prior NSAID use, and lower dose aspirin use (p ≥ 0.32) (Table 1).

Figure 1
CONSORT Diagram that shows the number of participants that were pre-registered, the number of screen failures, along with the number of participants that were randomized and evaluable for the primary endpoint.
Table 1
Baseline demographics by trial participation status

Due to slower than anticipated accrual, randomization was closed early prior to reaching the a priori accrual goal of 152 participants. In total, 122 participants were randomized. Two participants were withdrawn post-randomization due to an inadequate number of surveillance biopsies and a finding of high-grade dysplasia at the pre-intervention endoscopy exam, respectively. With respect to race/ethnicity, 118/120 randomized participants (98%) were self-identified as Caucasian (race status was missing/unknown for 2 additional participants), while 109/120 (91%) and 6/120 (5%) were self-identified as non-Hispanic and Hispanic/Latino, respectively (ethnicity status was missing/unknown for 5 additional participants). Within the randomized cohort, 6 participants were not evaluable for the analyses of PGE2 analyses due to sample-related issues (e.g., improper temperature, lost or delayed samples), leaving 114 evaluable participants for the primary endpoint.

Intervention Arms, Adverse Events and Agent Adherence

By design, intervention arms were evenly balanced with respect to sex (p = 0.95), length of BE (p = 0.95), and dysplasia status (no dysplasia vs. low grade dysplasia) at the pre-intervention evaluation (p = 1.00). Other baseline variables were similar between the intervention arms except for smoking history (p = 0.005) (Table 2).

Table 2
Baseline demographics, intervention adherence, and adverse events by intervention arm

Adverse events were reported in 26/120 (22%) trial participants after starting the assigned study intervention. Adverse event rates were similar across intervention arms (Table 2). Hiatal hernia erosions developed in one aspirin placebo participant and one lower dose aspirin participant. Antral erosions developed in one participant in the aspirin placebo arm. One higher dose aspirin participant experienced an erosion at the GE junction. No participants in any of the arms developed an ulcer. Agent adherence was excellent, with all participants receiving the majority of the assigned study doses, and was similar across the randomization arms (p ≥ 0.57) (Table 2).

Tissue PGE2 Concentration

The primary objective was to compare pre- and post-intervention biopsies within and between treatment arms based on PGE2 concentrations in Barrett’s mucosa. An interim analysis was performed after 77 participants were randomized (n=19, 29, and 29 in aspirin placebo, lower dose aspirin, and higher dose aspirin arms, respectively). The data were highly skewed, so the Wilcoxon Rank-Sum test was used to compare the active aspirin arms to the aspirin placebo arm. The 2-sided p-value was 0.0121 for the higher dose aspirin arm vs. aspirin placebo, which did not meet the interim analysis criteria to stop the trial for improved efficacy (2-sided p-value < 0.0008 required) or futility (2-sided p-value > 0.634 required). The 2-sided p-value was 0.0550 for the lower dose aspirin arm vs. aspirin placebo, which did not meet the interim analysis criteria to stop the trial for improved efficacy (2-sided p-value < 0.0008 required) or futility (2-sided p-value > 0.634 required). As such, the study continued on until it was decided to close early due to slow accrual.

Of the 120 participants in the intervention cohort, 114 were evaluable for the primary endpoint (n = 29, 42, and 43 in the aspirin placebo + PPI, lower dose aspirin + PPI, and higher dose aspirin + PPI arms, respectively). All evaluable participants had at least two PGE2 values available from the pre- and post-intervention endoscopy exams (94% had three PGE2 values). There was no statistically significant difference between the pre-intervention PGE2 values across the three intervention arms (p = 0.27), but there was a statistically significant difference for the post-intervention PGE2 values (p < 0.0001) (Table 3). Both the lower dose aspirin and higher dose aspirin arms showed significant decreases in PGE2 from baseline for both absolute and percent change (Signed Rank p ≤ 0.0004), while the aspirin placebo arm showed no significant change from baseline for absolute and percent change in PGE2 values (p ≥ 0.27). Based on the primary endpoint (absolute change in PGE2), the higher dose aspirin + PPI intervention resulted in a significant decrease in PGE2 values as compared to the aspirin placebo + PPI arm (p = 0.02) (Table 3 and Figure 2). Higher dose aspirin + PPI was also associated with a significant decrease in the percent change in PGE2 values (secondary endpoint) compared to aspirin placebo + PPI (p < 0.0001) (Table 3 and Figure 3). Lower dose aspirin + PPI vs. aspirin placebo + PPI showed weaker results for absolute change (p = 0.10) (Table 3 and Figure 2), although there was still a trend toward a decrease in PGE2 values. Based on percent change in PGE2, lower dose aspirin + PPI showed a significant decrease in the PGE2 values as compared to aspirin placebo + PPI (p = 0.0007) (Table 3 and Figure 3). No significant difference was observed between the two active aspirin arms for absolute (p = 0.26) or percent change (p = 0.45) in PGE2 values. The primary endpoint analyses were not appreciably altered after excluding outliers (data not shown).

Figure 2
Boxplots for the Absolute Change in PGE2 by Intervention Arm. After adjusting for the interim analysis and multiple comparisons, the nominal significance level was 0.0247. As shown, the Higher Dose Aspirin + PPI intervention resulted in a significant ...
Figure 3
Boxplots for the Percent Change in PGE2 by Intervention Arm. As shown, the Higher Dose Aspirin + PPI and Lower Dose Aspirin + PPI interventions resulted in significant decreases in PGE2 values as compared to Aspirin Placebo + PPI (p < 0.001).
Table 3
PGE2 Summary in all 114 evaluable participants


To our knowledge, we report the largest prospective chemoprevention trial in BE patients conducted in North America. Data from this clinical trial provide the first demonstration of a statistically significant effect from the combination of esomeprazole 40 mg bid and aspirin 325 mg qd for 28 days on tissue PGE2 values in participants with BE, as compared to esomeprazole 40 mg bid without active aspirin. Although the lower dose aspirin (81mg qd) + PPI did not significantly alter tissue PGE2 concentration, this finding was likely influenced by our relatively small sample size. Of note, a significant difference relative to the aspirin placebo + PPI group was observed when the percent decrease in PGE2 levels was compared. These results support further evaluation of 325 mg aspirin administered together with twice daily esomeprazole as a potential chemoprevention strategy for BE patients in larger, more definitive phase III trials.

Esophageal adenocarcinoma is a lethal disease with a 5-year survival rate of approximately 17% 6 Survival is stage dependent and early spread prior to the onset of symptoms is characteristic of this tumor. Currently, the best hope for improved survival of patients with esophageal adenocarcinoma is detection of cancer at an early and potentially curable stage. In order to accomplish this, current strategies have focused on identifying more patients at risk, namely those with BE, in conjunction with widespread endoscopic surveillance of patients with known BE. Both strategies, however, are inherently flawed, expensive, and impractical. Furthermore, there are no prospective clinical data to confirm either approach is effective. Although observational studies have demonstrated surveillance to be associated with cancer downstaging with improved 5-year survival, the relevance of these studies is markedly limited by the risk of lead-time bias.17, 18 In this context, alternative intervention strategies, such as chemoprevention, are worthy of consideration, specifically during the early stages of neoplastic transformation.5

It is widely accepted that chronic injury and resulting inflammation play a major role in the development of several cancers, including esophageal adenocarcinoma.9, 19-23 The catalysis of arachidonic acid to PGE2 by COX enzymes is thought to be critically involved in injury- or inflammation-induced esophageal carcinogenesis. Low pH, as well as exposure to caustic bile salts, can induce COX-2 expression in both ex-vivo culture models of BE and in esophageal adenocarcinoma cell lines.24, 25 COX-2 expression increases concomitantly with neoplastic progression in BE, thereby supporting the association of the arachidonic acid pathway with the development of esophageal adenocarcinoma.26 Moreover, higher PGE2 levels have been reported in Barrett’s epithelium compared to normal squamous mucosa.27 PGE2 induces the proliferation of Barrett’s epithelial cells, while PGE2 inhibition reduces their proliferation.19 In addition, PGE2 is known to diminish tumor surveillance by inhibiting natural killer cell activity.28 Recent animal studies further demonstrate that inhibition of PGE2 production with the use of selective and non-selective COX-2 inhibitors can decrease the rate of esophageal adenocarcinoma development.19 Thus, chronic induction of PGE2 appears to stimulate esophageal carcinogenesis through multiple mechanisms.

A number of candidate chemoprevention agents have been proposed for BE patients including PPIs, selective and non-selective COX inhibitors (including aspirin), lyophilized black raspberries, antioxidants, green tea, retinoids, ursodeoxycholic acid, statins and curcumin.10, 29-31 Most attention has been directed toward the use of aspirin and other NSAIDs, which appear to lower esophageal cancer risk by favorably affecting COX-related pathways that are upregulated by excess exposure to acid and/or bile salts. While observational studies suggest that PPI therapy is associated with a decreased risk for the development of high-grade dysplasia and adenocarcinoma, the chemopreventive potential of PPIs, by suppressing acid and bile salt reflux, appears to be marginal when used as monotherapy.32-34 There are a variety of possible explanations for this marginal effect, including the inability of PPIs to completely eliminate acid/bile salt reflux or activation of gastrin-cholecystekinin (CCK)-COX-2 mediated pro-carcinogenic signaling pathways. It is therefore logical to use a PPI agent in combination with NSAIDS, to determine if the combination more effectively targets intermediate markers of Barrett’s-associated carcinogenesis.

Unfortunately, selective COX-2 inhibitors have been shown to increase the risk of thrombotic cardiovascular events making their widespread use for cancer chemoprevention problematic. An alternate approach is to use aspirin with a PPI, as was performed in our study. Aspirin decreases the risk of thrombotic cardiovascular events and is frequently recommended for patients at increased risk for myocardial infarction. In addition, PPIs may make aspirin a more useful and safer chemopreventive agent by minimizing aspirin-induced mucosal injury. Given the age-related increase in risk for gastrointestinal bleeding associated with NSAID use, this is a real concern for any chemoprevention strategy among BE patients, who tend to be of older age.35 While PPIs, by targeting the root cause of the problem, may prevent chronic esophageal injury and decrease inflammation, aspirin use could control any COX-2 activation that may result from breakthroughs in acid suppression or gastrin activation by PPIs.

Clinical trial data regarding the potential chemopreventive effects of aspirin and other NSAIDs in BE patients are currently limited. One clinical trial examined the effect of celecoxib given for 48 weeks in participants with low-grade and high-grade dysplasia, and found no difference in the proportion of biopsy samples with dysplasia between participants treated with celecoxib compared to placebo.36 Another small crossover study demonstrated that esomeprazole 40 mg twice daily in conjunction with aspirin at a dose of 325 mg daily administered for 10 days resulted in lower esophageal mucosal PGE2 content in BE participants, whereas esomeprazole alone or in combination with rofecoxib did not reduce PGE2 production.34 To our knowledge, no additional clinical trial data have yet been reported on the potential effects of different doses of aspirin, in combination with PPI therapy, as a potential esophageal cancer chemopreventive strategy. The rationale for the use of the lower and higher dose of aspirin was to define if the smaller dose of aspirin was equally effective in decreasing tissue PGE2 in Barrett’s patients compared to the higher dose. Another reason to undertake such a dose ranging study is that the higher dose of aspirin in one colon cancer prevention study appeared to be less effective than the lower dose aspirin.37

Strengths of the current study design included the avoidance of recruitment of participants with intestinal metaplasia of the cardia, the requirement for confirmation of intestinal metaplasia with goblet cells, and detailed interview to make sure patients were not unwittingly using aspirin or other NSAIDS prior to randomization. Thus, careful steps were taken to ensure a homogenous group of study participants with histologically-confirmed BE. Although our trial was conducted at multiple, geographically disparate sites, the relatively limited racial/ethnic diversity of our randomized cohort makes it unlikely that ethnopharmacology had any appreciable influence on our findings. While a significant difference was noted in the higher dose aspirin arm vs. the aspirin placebo arm, no such difference could be detected in the lower dose aspirin arm, despite a trend in that direction. The lack of a significant effect in the lower dose aspirin + esomeprazole arm may have been due to a lack of power and not a true lack of efficacy. The inability to achieve the a priori sample size, the lack of a sample size adjustment for the increased variability in PGE2, and the use of planned non-parametric tests (instead of the parametric tests that the sample size was based on) all may have contributed to the study being under-powered to detect a significant effect for the lower dose aspirin + esomeprazole arm. As such, we cannot conclude from our data that a combination of lower dose aspirin + esomeprazole does not provide chemopreventive benefits for BE patients.

There are a number of factors that make BE studies challenging to conduct. Importantly, many patients with BE already take aspirin or NSAIDs, which makes any such longer term clinical trial difficult to perform. Also, PGE2 levels represent a surrogate end point for more direct clinical endpoints, such as progression to high-grade dysplasia and adenocarcinoma. There is no simple study design for assessing these more clinically relevant, yet substantially less common endpoints without considerable resource investment.

Our rational to use PGE2 as a surrogate biomarker of NSAID-related chemoprevention is based upon its direct role in carcinogenesis in Barrett’s mucosa through upregulation of proliferation, resistance to apoptosis, and angiogenesis. Furthermore, PGE2 biosynthesis is a downstream target of several oncogenic signals and it can be downregulated by tumor suppressors.38-40 The findings that carcinogenic bile salts in pH dependent manner upregulate PGE2 biosynthesis along with increased expression of key regulators of PGE2 by DNA damage and by aberrant p53 expression41 further support that the effect of genetic or epigenetic gains or losses that promote neoplasia could be inferred through a mechanistically relevant composite biochemical readout such as PGE2 synthesis.

We believe that data from the current phase II trial provide strong support for a larger, more definitive phase III trial to further clarify the role of aspirin and esomeprazole in esophageal cancer chemoprevention among BE patients. Indeed, our study represents the largest chemoprevention trial in BE to date and has the advantage of examining a geographically diverse patient population. The ongoing multi-center phase III AspECT trial, which is being conducted in the United Kingdom and elsewhere in Europe, is also examining aspirin and esomeprazole as a combination chemopreventive intervention for BE patients,42 with results that should serve to complement these findings.

In conclusion, data from this multi-center, prospective phase II trial demonstrate that twice daily esomeprazole in combination with higher dose (325 mg) daily aspirin therapy can favorably affect tissue PGE2 levels in BE patients. Although the lower dose aspirin (81 mg) was not found to significantly reduce PGE2 levels as compared to esomeprazole alone based on the primary endpoint analysis, data from our secondary analyses suggest that esomeprazole twice per day in combination with 81 mg aspirin per day may also provide some chemopreventive benefits for BE patients. As such, a larger trial is recommended to more fully investigate the aspirin dose effect to see what aspirin dose is optimal. Given the importance of PGE2 and related molecular pathways in Barrett’s-associated carcinogenesis, this dual regimen of aspirin and esomeprazole warrants further evaluation as a novel chemoprevention strategy for an increasingly common disease with potentially devastating clinical outcomes.

Supplementary Material



The authors gratefully acknowledge the staff of the Mayo Clinic Clinical Research Unit (supported by grant M01-RR00585); Drs. Wilma Lingle, Schuyler Sanderson, Tsung-Teh Wu and Kenneth Wang; and Mary Fredericksen, Deb Geno, Colleen Garvey, Sharon Kaufman, and Maria Resner for their assistance with study design, administration, and manuscript preparation. The authors also gratefully acknowledge the dedicated study coordinators at the member organizations: Denise Buonocore-Sassano; Trish O’Brien; Sally Melloy; Jolanda DeBoer; April Higbee; Zehra Omer; Michael Lee; Barbara Berg; Maria Cirocco; Nancy Bassett; Beth Bednarchik; and Jessica Hernandez. This work was sponsored by the National Cancer Institute, Division of Cancer Prevention, contract N01-CN-35000.

Research Support: Supported by a contract from the National Cancer Institute (N01CN35000). Research support (equipment and drug supplies) also provided by Astra-Zeneca, Bayer, and Boston Scientific.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author Involvement:

Gary W. Falk, Navtej S. Buttar, Paul J. Limburg: Concept and design, acquisition of data, analysis and interpretation, drafting of manuscript, critical revision, study supervision

Nathan R. Foster, Katie L. Allen Ziegler, Catherine DeMars, Yvonne Romero, Norman E. Marçon, Thomas Schnell, Douglas A. Corley, Prateek Sharma, Marcia R. Cruz-Correa, Chin Hur, David E. Fleischer, Amitabh Chak, Kenneth R. DeVault, David S. Weinberg, Thomas C. Smyrk, Sumithra J. Mandrekar: Concept and design, acquisition of data, analysis and interpretation, drafting of manuscript, critical revision

Gary Della’Zanna, Ellen Richmond: Concept and design, analysis and interpretation, drafting of manuscript, critical revision, study supervision

Conflicts of Interest: Dr. Limburg served as a consultant for Genomic Health, Inc. from 8/12/08-4/19/10. Mayo Clinic has licensed Dr. Limburg’s intellectual property to Exact Sciences and he and Mayo Clinic have contractual rights to receive royalties through this agreement. Dr. Yvonne Romero receives funding from AstraZeneca for Mayo Clinic Barrett’s esophagus registry and tissue bank.

Corresponding Author: The corresponding author had full access to all of the data and takes full responsibility for the veracity of the data and statistical analysis


1. Hayeck TJ, Kong CY, Spechler SJ, Gazelle GS, Hur C. The prevalence of Barrett’s esophagus in the US: estimates from a simulation model confirmed by SEER data. Dis Esophagus. 2010;23:451–7. [PMC free article] [PubMed]
2. Reid BJ, Li X, Galipeau PC, Vaughan TL. Barrett’s oesophagus and oesophageal adenocarcinoma: time for a new synthesis. Nat Rev Cancer. 2010;10:87–101. [PMC free article] [PubMed]
3. Wani S, Falk G, Hall M, Gaddam S, Wang A, Gupta N, Singh M, Singh V, Chuang KY, Boolchand V, Gavini H, Kuczynski J, Sud P, Reddymasu S, Bansal A, Rastogi A, Mathur SC, Young P, Cash B, Lieberman DA, Sampliner RE, Sharma P. Patients with nondysplastic Barrett’s esophagus have low risks for developing dysplasia or esophageal adenocarcinoma. Clin Gastroenterol Hepatol. 2011;9:220–7. quiz e26. [PubMed]
4. Schouten LJ, Steevens J, Huysentruyt CJ, Coffeng CE, Keulemans YC, van Leeuwen FE, Driessen AL, van den Brandt PA. Total Cancer Incidence and Overall Mortality Are Not Increased Among Patients With Barrett’s Esophagus. Clin Gastroenterol Hepatol. 2011;9:754–61. [PubMed]
5. Spechler SJ, Sharma P, Souza RF, Inadomi JM, Shaheen NJ. American Gastroenterological Association technical review on the management of Barrett’s esophagus. Gastroenterology. 2011;140:e18–52. quiz e13. [PMC free article] [PubMed]
6. Howlader N NA, Krapcho M, Neyman N, Aminou R, Waldron W, Altekruse SF, Kosary CL, Ruhl J, Tatalovich Z, Cho H, Mariotto A, Eisner MP, Lewis DR, Chen HS, Feuer EJ, Cronin KA, Edwards BK, Institute NC, editors. SEER Cancer Statistics Review, 1975-2008. Bethesda, MD: 2011.
7. Corley DA, Kerlikowske K, Verma R, Buffler P. Protective association of aspirin/NSAIDs and esophageal cancer: a systematic review and meta-analysis. Gastroenterology. 2003;124:47–56. [PubMed]
8. Vaughan TL, Dong LM, Blount PL, Ayub K, Odze RD, Sanchez CA, Rabinovitch PS, Reid BJ. Non-steroidal anti-inflammatory drugs and risk of neoplastic progression in Barrett’s oesophagus: a prospective study. Lancet Oncol. 2005;6:945–52. [PubMed]
9. Anderson LA, Johnston BT, Watson RG, Murphy SJ, Ferguson HR, Comber H, McGuigan J, Reynolds JV, Murray LJ. Nonsteroidal anti-inflammatory drugs and the esophageal inflammation-metaplasia-adenocarcinoma sequence. Cancer Res. 2006;66:4975–82. [PubMed]
10. Nguyen DM, Richardson P, El-Serag HB. Medications (NSAIDs, statins, proton pump inhibitors) and the risk of esophageal adenocarcinoma in patients with Barrett’s esophagus. Gastroenterology. 2010;138:2260–6. [PMC free article] [PubMed]
11. Jayaprakash V, Menezes RJ, Javle MM, McCann SE, Baker JA, Reid ME, Natarajan N, Moysich KB. Regular aspirin use and esophageal cancer risk. Int J Cancer. 2006;119:202–7. [PubMed]
12. Buttar NS, Wang KK, Anderson MA, Dierkhising RA, Pacifico RJ, Krishnadath KK, Lutzke LS. The effect of selective cyclooxygenase-2 inhibition in Barrett’s esophagus epithelium: an in vitro study. J Natl Cancer Inst. 2002;94:422–9. [PubMed]
13. Sharma P, Dent J, Armstrong D, Bergman JJ, Gossner L, Hoshihara Y, Jankowski JA, Junghard O, Lundell L, Tytgat GN, Vieth M. The development and validation of an endoscopic grading system for Barrett’s esophagus: the Prague C & M criteria. Gastroenterology. 2006;131:1392–9. [PubMed]
14. Pocock SJ, Simon R. Sequential treatment assignment with balancing for prognostic factors in the controlled clinical trial. Biometrics. 1975;31:103–15. [PubMed]
15. Lan KDD. Discrete sequential boundaries for clinical trials. Biometrika. 1983;70:659–663.
16. Cytel . In: EAST: A software package for the design and interim monitoring of group sequential clinical trials. Corporation CS, editor. Cytel Software Corporation; Cambridge: 2000.
17. van Sandick JW, van Lanschot JJ, Kuiken BW, Tytgat GN, Offerhaus GJ, Obertop H. Impact of endoscopic biopsy surveillance of Barrett’s oesophagus on pathological stage and clinical outcome of Barrett’s carcinoma. Gut. 1998;43:216–22. [PMC free article] [PubMed]
18. Corley DA, Levin TR, Habel LA, Weiss NS, Buffler PA. Surveillance and survival in Barrett’s adenocarcinomas: a population-based study. Gastroenterology. 2002;122:633–40. [PubMed]
19. Buttar NS, Wang KK, Leontovich O, Westcott JY, Pacifico RJ, Anderson MA, Krishnadath KK, Lutzke LS, Burgart LJ. Chemoprevention of esophageal adenocarcinoma by COX-2 inhibitors in an animal model of Barrett’s esophagus. Gastroenterology. 2002;122:1101–12. [PubMed]
20. Buttar NS, Wang KK. Mechanisms of disease: Carcinogenesis in Barrett’s esophagus. Nat Clin Pract Gastroenterol Hepatol. 2004;1:106–12. [PubMed]
21. Fox JG, Rogers AB, Whary MT, Ge Z, Ohtani M, Jones EK, Wang TC. Accelerated progression of gastritis to dysplasia in the pyloric antrum of TFF2 -/- C57BL6 x Sv129 Helicobacter pylori-infected mice. Am J Pathol. 2007;171:1520–8. [PubMed]
22. Rogers AB, Boutin SR, Whary MT, Sundina N, Ge Z, Cormier K, Fox JG. Progression of chronic hepatitis and preneoplasia in Helicobacter hepaticus-infected A/JCr mice. Toxicol Pathol. 2004;32:668–77. [PubMed]
23. Wild CP, Hardie LJ. Reflux, Barrett’s oesophagus and adenocarcinoma: burning questions. Nat Rev Cancer. 2003;3:676–84. [PubMed]
24. Kaur BS, Ouatu-Lascar R, Omary MB, Triadafilopoulos G. Bile salts induce or blunt cell proliferation in Barrett’s esophagus in an acid-dependent fashion. Am J Physiol Gastrointest Liver Physiol. 2000;278:G1000–9. [PubMed]
25. Zhang F, Altorki NK, Wu YC, Soslow RA, Subbaramaiah K, Dannenberg AJ. Duodenal reflux induces cyclooxygenase-2 in the esophageal mucosa of rats: evidence for involvement of bile acids. Gastroenterology. 2001;121:1391–9. [PubMed]
26. Wilson KT, Fu S, Ramanujam KS, Meltzer SJ. Increased expression of inducible nitric oxide synthase and cyclooxygenase-2 in Barrett’s esophagus and associated adenocarcinomas. Cancer Res. 1998;58:2929–34. [PubMed]
27. Kaur BS, Khamnehei N, Iravani M, Namburu SS, Lin O, Triadafilopoulos G. Rofecoxib inhibits cyclooxygenase 2 expression and activity and reduces cell proliferation in Barrett’s esophagus. Gastroenterology. 2002;123:60–7. [PubMed]
28. Nishigaki Y, Ohnishi H, Moriwaki H, Muto Y. Ursodeoxycholic acid corrects defective natural killer activity by inhibiting prostaglandin E2 production in primary biliary cirrhosis. Dig Dis Sci. 1996;41:1487–93. [PubMed]
29. Ogunwobi OO, Beales IL. Statins inhibit proliferation and induce apoptosis in Barrett’s esophageal adenocarcinoma cells. Am J Gastroenterol. 2008;103:825–37. [PubMed]
30. Bozikas A, Marsman WA, Rosmolen WD, van Baal JW, Kulik W, ten Kate FJ, Krishnadath KK, Bergman JJ. The effect of oral administration of ursodeoxycholic acid and high-dose proton pump inhibitors on the histology of Barrett’s esophagus. Dis Esophagus. 2008;21:346–54. [PubMed]
31. Kubo A, Corley DA. Meta-analysis of antioxidant intake and the risk of esophageal and gastric cardia adenocarcinoma. Am J Gastroenterol. 2007;102:2323–30. quiz 2331. [PubMed]
32. Hillman LC, Chiragakis L, Shadbolt B, Kaye GL, Clarke AC. Proton-pump inhibitor therapy and the development of dysplasia in patients with Barrett’s oesophagus. Med J Aust. 2004;180:387–91. [PubMed]
33. Nguyen DM, El-Serag HB, Henderson L, Stein D, Bhattacharyya A, Sampliner RE. Medication usage and the risk of neoplasia in patients with Barrett’s esophagus. Clin Gastroenterol Hepatol. 2009;7:1299–304. [PMC free article] [PubMed]
34. Triadafilopoulos G, Kaur B, Sood S, Traxler B, Levine D, Weston A. The effects of esomeprazole combined with aspirin or rofecoxib on prostaglandin E2 production in patients with Barrett’s oesophagus. Aliment Pharmacol Ther. 2006;23:997–1005. [PubMed]
35. Hawkey CJ. Nonsteroidal anti-inflammatory drug gastropathy. Gastroenterology. 2000;119:521–35. [PubMed]
36. Heath EI, Canto MI, Piantadosi S, Montgomery E, Weinstein WM, Herman JG, Dannenberg AJ, Yang VW, Shar AO, Hawk E, Forastiere AA. Secondary chemoprevention of Barrett’s esophagus with celecoxib: results of a randomized trial. J Natl Cancer Inst. 2007;99:545–57. [PMC free article] [PubMed]
37. Baron JA, Cole BF, Sandler RS, Haile RW, Ahnen D, Bresalier R, McKeown-Eyssen G, Summers RW, Rothstein R, Burke CA, Snover DC, Church TR, Allen JI, Beach M, Beck GJ, Bond JH, Byers T, Greenberg ER, Mandel JS, Marcon N, Mott LA, Pearson L, Saibil F, van Stolk RU. A randomized trial of aspirin to prevent colorectal adenomas. N Engl J Med. 2003;348:891–9. [PubMed]
38. Yen CJ, Izzo JG, Lee DF, Guha S, Wei Y, Wu TT, Chen CT, Kuo HP, Hsu JM, Sun HL, Chou CK, Buttar NS, Wang KK, Huang P, Ajani J, Hung MC. Bile acid exposure up-regulates tuberous sclerosis complex 1/mammalian target of rapamycin pathway in Barrett’s-associated esophageal adenocarcinoma. Cancer Res. 2008;68:2632–40. [PMC free article] [PubMed]
39. Song S, Guha S, Liu K, Buttar NS, Bresalier RS. COX-2 induction by unconjugated bile acids involves reactive oxygen species-mediated signalling pathways in Barrett’s oesophagus and oesophageal adenocarcinoma. Gut. 2007;56:1512–21. [PMC free article] [PubMed]
40. Buttar NS, DeMars CJ, Lomberk G, Rizvi S, Bonilla-Velez J, Achra S, Rashtak S, Wang KK, Fernandez-Zapico ME, Urrutia R. Distinct role of Kruppel-like factor 11 in the regulation of prostaglandin E2 biosynthesis. The Journal of biological chemistry. 2010;285:11433–44. [PMC free article] [PubMed]
41. Benoit V, de Moraes E, Dar NA, Taranchon E, Bours V, Hautefeuille A, Taniere P, Chariot A, Scoazec JY, de Moura Gallo CV, Merville MP, Hainaut P. Transcriptional activation of cyclooxygenase-2 by tumor suppressor p53 requires nuclear factor-kappaB. Oncogene. 2006;25:5708–18. [PubMed]
42. Das D, Ishaq S, Harrison R, Kosuri K, Harper E, Decaestecker J, Sampliner R, Attwood S, Barr H, Watson P, Moayyedi P, Jankowski J. Management of Barrett’s esophagus in the UK: overtreated and underbiopsied but improved by the introduction of a national randomized trial. Am J Gastroenterol. 2008;103:1079–89. [PubMed]