Peptic ulcer bleeding is a common cause of hospitalization, Article

Transcription

1 Annals of Internal Medicine Article Intravenous Esomeprazole for Prevention of Recurrent Peptic Ulcer Bleeding A Randomized Trial Joseph J.Y. Sung, MD; Alan Barkun, MD; Ernst J. Kuipers, MD; Joachim Mössner, MD; Dennis M. Jensen, MD; Robert Stuart, MD; James Y. Lau, MD; Henrik Ahlbom, BSc; Jan Kilhamn, MD; and Tore Lind, MD, for the Peptic Ulcer Bleed Study Group* Background: Use of proton-pump inhibitors in the management of peptic ulcer bleeding is controversial because discrepant results have been reported in different ethnic groups. Objective: To determine whether intravenous esomeprazole prevents recurrent peptic ulcer bleeding better than placebo in a multiethnic patient sample. Design: Randomized trial conducted between October 2005 and December 2007; patients, providers, and researchers were blinded to group assignment. Setting: 91 hospital emergency departments in 16 countries. Patients: Patients 18 years or older with peptic ulcer bleeding from a single gastric or duodenal ulcer showing high-risk stigmata. Intervention: Intravenous esomeprazole bolus, 80 mg, followed by 8-mg/h infusion, over 72 hours or matching placebo, each given after successful endoscopic hemostasis. Intervention was allocated by computer-generated randomization. After infusion, both groups received oral esomeprazole, 40 mg/d, for 27 days. Measurements: The primary end point was rate of clinically significant recurrent bleeding within 72 hours. Recurrent bleeding within 7 and 30 days, death, surgery, endoscopic re-treatment, blood transfusions, hospitalization, and safety were also assessed. Results: Of 767 patients randomly assigned, 764 provided data for an intention-to-treat analysis (375 esomeprazole recipients and 389 placebo recipients). Fewer patients receiving intravenous esomeprazole (22 of 375) had recurrent bleeding within 72 hours than those receiving placebo (40 of 389) (5.9% vs. 10.3%; difference, 4.4 percentage points [95% CI, 0.6% to 8.3%]; P 0.026). The difference in bleeding recurrence remained significant at 7 days and 30 days (P 0.010). Esomeprazole also reduced endoscopic retreatment (6.4% vs. 11.6%; difference, 5.2 percentage points [95% CI of difference, 1.1 percentage points to 9.2 percentage points]; P 0.012), surgery (2.7% vs. 5.4%), and all-cause mortality rates (0.8% vs. 2.1%) more than placebo, although differences for the latter 2 comparisons were not significant. About 10% and 40% of patients in both groups reported serious and nonserious adverse events, respectively. Limitation: Endoscopic therapy was not completely standardized; some patients received epinephrine injection, thermal coagulation, or hemoclips alone, whereas others received combination therapy, but there were similar proportions with single therapy in each group. Conclusion: High-dose intravenous esomeprazole given after successful endoscopic therapy to patients with high-risk peptic ulcer bleeding reduced recurrent bleeding at 72 hours and had sustained clinical benefits for up to 30 days. Primary Funding Source: AstraZeneca Research and Development. Ann Intern Med. 2009;150: For author affiliations, see end of text. ClinicalTrials.gov registration number: NCT * For a list of study centers and principal investigators in the Peptic Ulcer Bleed Study Group, see the Appendix Table (available at This article was published at on 17 February Peptic ulcer bleeding is a common cause of hospitalization, exceeding hospitalizations annually in the United States (1). Despite therapeutic intervention, morbidity and mortality remain substantial, with a 30-day mortality rate of 5% to 10% (2, 3). Modern treatment of peptic ulcer bleeding aims to stabilize hemodynamic circulation, stop ongoing bleeding, and prevent recurrent bleeding. After initial resuscitation, prompt endoscopy identifies high-risk stigmata on ulcers and endoscopic hemostasis lowers the incidence of attributable rebleeding and mortality. Profound and sustained acid suppression, as achieved with high-dose intravenous proton-pump inhibitors (PPIs) (4), is thought to improve outcomes by clot stabilization at higher gastric ph. Indeed, gastric acid impairs clot formation and even favors platelet disaggregation (5, 6). Acid also facilitates clot lysis by activation of pepsin (7, 8), and acid suppression may prevent fibrinolysis (9). Although some placebo-controlled trials of PPIs, mainly in Asia, have shown improved outcomes in patients with peptic ulcer bleeding (10, 11), studies in Western Europe and North America have yielded conflicting results (12 15). Racial differences in parietal cell mass and genetic polymorphisms affecting PPI metabolism (16, 17), coupled with regional variations in Helicobacter pylori prevalence, have raised doubts about the generalizability of the favorable results observed in Asian patients (18). Thus, we performed a placebo-controlled trial in a large, multiethnic patient sample to evaluate the efficacy of See also: Print Editors Notes Web-Only Appendix Table Conversion of graphics into slides 2009 American College of Physicians 455

2 Article Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding Context Whether profound acid suppression with high-dose proton-pump inhibitors improves outcomes in patients with peptic bleeding ulcers is unclear. Contribution This multicenter trial included 764 adults with a single bleeding gastric or duodenal ulcer that had high-risk stigmata. All had successful endoscopic hemostasis and were randomly assigned to intravenous high-dose esomeprazole or matching placebo for 72 hours. Fewer esomeprazole recipients than placebo recipients experienced recurrent bleeding (5.9% vs. 10.3%) or needed endoscopic re-treatment (6.4% vs. 11.6%). Implication High-dose esomeprazole, given after successful endoscopic treatment, reduced recurrent bleeding in patients with high-risk peptic ulcer bleeding. The Editors profound acid suppression with high-dose intravenous PPI in patients with peptic ulcer bleeding. METHODS Design Overview A critical comparison of our adopted study design in relation to previous trials has been reported (19). This randomized, multicenter, parallel-group, blinded comparison of esomeprazole and placebo, given both intravenously and orally, was conducted internationally in many different ethnic groups. The first patient was enrolled on 30 October 2005, and the last patient completed the trial by 14 December The protocol conformed to good clinical practice guidelines and was conducted in each patient over 30 days. The local ethics committee at each center approved the protocol, and we obtained written informed consent for all patients. Setting and Participants The Peptic Ulcer Bleed Study was conducted at 91 centers in 16 countries (Austria, China, Denmark, Finland, France, Germany, Greece, the Netherlands, Norway, Romania, Russia, South Africa, Spain, Sweden, Turkey, and the United Kingdom). Patients 18 years or older presenting at hospital emergency departments, or already hospitalized for another reason, with overt signs of upper gastrointestinal bleeding (hematemesis, melena, or hematochezia) in the past 24 hours were eligible for randomization if they had only 1 bleeding gastric or duodenal ulcer that was at least 5 mm in diameter. We recruited patients with bleeding ulcers that showed 1 of the following endoscopic stigmata of recent hemorrhage (20): arterial bleeding (Forrest class Ia), oozing (Forrest class Ib), nonbleeding visible vessel (Forrest class IIa), or adherent clot (Forrest class IIb). In the case of Forrest class IIb ulcers, after attempts to remove the clot by using water irrigation or a cold snare, ulcers were either reclassified for inclusion as Forrest class Ia, Ib, or IIa or, if unsuccessful, included as Forrest class IIb. Other inclusion criteria were endoscopic control of bleeding using injection therapy with adrenaline, thermal coagulation, or both or application of hemoclips and an American Society of Anesthesiologists (ASA) score of 3 or less. We excluded patients who were bleeding from multiple ulcers or concomitant upper gastrointestinal sources; had another major disease; had a life expectancy less than 6 months; required treatment with nonsteroidal anti-inflammatory drugs, including cyclooxygenase-2 inhibitors, aspirin, or clopidogrel, during the first 7 days of the study; received more than 40 mg of PPI intravenously within 24 hours before enrollment; or required a drug known to interact with PPIs (phenytoin, clarithromycin, itraconazole, ketoconazole, warfarin and other vitamin K antagonists, cisapride, atazanavir, or ritonavir). We performed clinical laboratory tests, including a serum sample for H. pylori IgG antibody (measured centrally in an Immulite 2000 analyzer [Siemens Healthcare Diagnostics, Deerfield, Illinois]), before the intravenous treatment. Full details of individual inclusion and exclusion criteria are published elsewhere (19). Randomization and Interventions We randomly assigned eligible patients in a 1:1 proportion, by using a central computer-generated block randomization, to double-blind treatment with either esomeprazole, 80 mg bolus, given intravenously over 30 minutes, followed by an 8-mg/h infusion for 71.5 hours, or to a similarly administered placebo. Randomization was not stratified, and each randomization block consisted of 4 patients. We ensured blinding by using esomeprazole and placebo powders and vials that appeared identical. The treatment code was unblinded only in medical emergencies, when the appropriate management of the patient necessitated knowledge of the treatment randomization. Any such unblinding was documented, and study monitors checked regularly that code envelopes were not otherwise broken. A clinical suspicion of recurrent bleeding (see definition in the Outcomes and Measurements section) prompted another endoscopy when possible (no routine second-look endoscopy). After completion of intravenous therapy, laboratory tests were repeated and all patients began open-label oral therapy with esomeprazole, 40 mg, once daily for another 27 days. Adverse events (AEs) were recorded. Follow-up Procedures and Monitoring We recorded AEs associated with any study procedure or leading to treatment intervention, dose reduction, or April 2009 Annals of Internal Medicine Volume 150 Number 7

3 Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding Article additional treatment. We assessed their intensity as mild (awareness of sign or symptom but easily tolerated), moderate (enough discomfort to interfere with normal activities), or severe (incapacitating, with inability to perform normal activities). We defined serious AEs as all AEs that fulfilled 1 or more of the following criteria: results in death, is immediately life-threatening, requires inpatient hospitalization or prolongation of existing hospitalization, results in persistent or clinically significant disability or incapacity, is a congenital abnormality or birth defect, or is an important medical event that may jeopardize the patient or may require medical intervention to prevent 1 of the outcomes listed above. All researchers involved in the study were informed of the process required for AE reporting. Vital signs, concomitant medication, hemoglobin, and hematocrit were monitored while the patient was hospitalized. We recorded serious AEs throughout the study from the point of obtaining informed consent and any other AEs from the time of first administration of the investigational drug until the end of the study. Reporting was either spontaneous or in response to the following standardized question in the local language: Have you had any health-related problems since the previous assessments? The investigator asked this question every 24 hours during the intravenous treatment phase and at any unscheduled or follow-up visit. Symptoms of the disease being studied, such as rebleeding, were not reported as AEs unless they were considered to be serious. Similarly, abnormal laboratory values or vital signs were not reported unless they met the criteria for serious AEs or led to study discontinuation. The start and stop dates (or ongoing), action taken regarding the study drug, causality, maximum intensity, outcome, and seriousness were also recorded. If there was a reasonable possibility that the event may have been caused by the study drug, the investigator assessed causality. Investigators entered their protocol-driven assessments online by using the Clinical Operations On Line system, a Web-based data-capture system. All data were saved immediately to a central database at AstraZeneca Research and Development (Mölndal, Sweden), with any changes tracked, providing an audit trail. The database was subsequently locked after independent verification. An independent data and safety monitoring board oversaw the study and performed planned interim safety analyses when approximately one third and two thirds of the patients had completed the study. The Peptic Ulcer Bleed Study steering committee members remained blinded to treatment allocation until the end of the trial, and it took guidance from the data and safety monitoring board with regard to possible premature study termination if predefined monitoring bounds for harm or benefit were crossed. A blinded expert end point committee (consisting of 2 authors) was also established to assess all deaths occurring within 30 days and to adjudicate their relation to bleeding. Outcomes and Measurements The primary end point was the rate of clinically significant recurrent peptic ulcer bleeding, as defined below, within 72 hours of endoscopic treatment. Clinically significant recurrent bleeding was defined by hematemesis with vomiting of significant amounts of fresh blood ( 200 ml) as estimated by the investigator, or when at least 2 of the following clinical findings were noted: vomiting of fresh blood, fresh blood in the nasogastric tube aspirate, or hematochezia or melena after a normal stool; a decrease in hemoglobin greater than 20 g/l (or decrease in hematocrit 6%) or an increase in hemoglobin less than 10 g/l (or increase in hematocrit 3%) during 24 hours, despite 2 or more units of blood transfused during 24 hours; or unstable vital signs with systolic blood pressure of 90 mm Hg or less or pulse of 110 beats/min or greater (after having achieved hemodynamic stability). Regardless, we confirmed recurrent bleeding at endoscopy when blood was found in the stomach (except during the first 6 hours after initial endoscopic hemostasis) or when active ulcer bleeding was documented. Secondary outcomes were clinically significant recurrent bleeding within 7 days or 30 days; all-cause or bleeding-related mortality; and the need for surgery, blood transfusion, and additional days of hospitalization because of recurrent bleeding within 30 days. Endoscopic appraisal of bleeding ulcer stigmata results in considerable interrater variability (19). All investigators were given a DVD of educational material to standardize endoscopic assessments across centers and optimize the inclusion of patients with appropriate high-risk stigmata. This resource tool identifies clinically relevant issues in modern endoscopic and pharmacologic management of peptic ulcer bleeding, defines the Forrest classification (20), and provides video clips for self-training. Criteria for endoscopic confirmation of recurrent bleeding were also reviewed and important aspects of photodocumentation were highlighted. Refresher training sessions were conducted at a midterm international investigators meeting. The end point committee performed an independent and blinded review of all ulcer photodocumentation submitted by the investigators to better validate the uniformity of patient enrollment. Feedback was provided when appropriate, but no post hoc exclusion or reclassification of patients was done. Investigators recorded all outcomes at each site. The study team at AstraZeneca Research and Development verified that events of recurrent bleeding fulfilled the specified criteria, and the end point committee assessed bleedingrelated mortality. Statistical Analysis We estimated the incidence of recurrent bleeding during the first 72 hours to be 7% with esomeprazole and 7 April 2009 Annals of Internal Medicine Volume 150 Number 7 457

4 Article Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding Figure 1. Study flow diagram. Not enrolled (n = 2833): cause of bleeding not specified (e.g., no endoscopy, consent not obtained, lack of time) or nonulcer bleeding Patients assessed for eligibility (n = 4146) Enrollment (n = 1313) Randomly assigned (n = 767) Excluded (n = 546) Did not meet inclusion criteria: 421 No bleeding ulcer at endoscopy: 402 Endoscopic hemostasis not achieved: 14 Other reasons: 5 Met exclusion criteria: 125 Multiple bleeding ulcers or other GI bleeding source: 92 Other reasons: 33 Allocated to esomeprazole (n = 376) Allocation Allocated to placebo (n = 391) Excluded from ITT analysis: no medication taken (n = 1) Included in ITT analysis (n = 375) Analyses Excluded from ITT analysis: no written informed consent (n = 2) Included in ITT analysis (n = 389) Excluded from PP analysis (n = 83) Major deviation from inclusion/exclusion criteria: 22 Disallowed concomitant medication: 15 Insufficient intake of study medication: 23 Other: 23 Included in PP analysis (n = 292) Excluded from PP analysis (n = 73) Major deviation from inclusion/exclusion criteria: 27 Disallowed concomitant medication: 21 Insufficient intake of study medication: 13 Other: 12 Included in PP analysis (n = 316) GI gastrointestinal; ITT intention-to-treat; PP per-protocol. 15% with placebo (10). We calculated that a sample size of 380 patients per group was required to detect a betweengroup difference with 90% power at a 5% significance level, assuming that 10% would be nonevaluable. We included all patients who had received at least 1 dose of study medication and had follow-up data available in the safety analysis. We analyzed efficacy for all variables by using an intention-to-treat analysis set, which included all randomly assigned patients who received infusion of the study drug. The primary variable was also evaluated according to a per-protocol analysis. We used the log-rank test for time-dependent dichotomous variables, the Mantel Haenszel test for variables considered time-independent, and the Wilcoxon test for ordered numerical variables. The primary end point recurrent bleeding within 72 hours was stratified according to whether single or dual endoscopic therapy was administered before randomization. In a post hoc analysis, we stratified the primary outcome by both endoscopic treatment and Forrest class and by ethnic background. We analyzed blood transfusion (number of units transfused) and number of days hospitalized because of recurrent bleeding by using a Wilcoxon 2-sample test. We used a Breslow Day test to evaluate country treatment interactions for the primary variable. All statistical analyses were done with SAS software, version 8.02 (SAS Institute, Cary, North Carolina). Role of the Funding Source AstraZeneca Research and Development provided financial and logistic support for the study. The study was designed by the international steering committee (which included 4 of the authors) and was led by Dr. Sung. The funding source assisted in the initial design of the study and managed the data, and its representatives were involved in the data analysis and data interpretation (the data and safety monitoring board conducted the interim analyses independently). An independent end point committee adjudicated all deaths. The steering committee had complete scientific oversight throughout the study design and completion. The funding source and steering committee provided written comments that the authors considered for the manuscript April 2009 Annals of Internal Medicine Volume 150 Number 7

5 Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding Article Table 1. Baseline Demographic and Clinical Characteristics of the Intention-to-Treat Groups Characteristic Esomeprazole Group (n 375) Placebo Group (n 389) Men, n (%) 254 (67) 268 (69) Mean age (SD), y 62.1 (17.5) 60.2 (17.6) Age 65 y, n (%) 193 (51.5) 179 (46.0) Mean weight (SD), kg 74.2 (16.4) 76.4 (16.8) Mean height (SD), cm (9.3) (9.6) Smokers, n (%) 103 (27.5) 110 (28.3) Race, n (%) White 325 (86.7) 342 (87.9) African 4 (1.1) 5 (1.3) Asian 27 (7.2) 27 (6.9) Other 19 (5.1) 15 (3.9) ASA score, n (%) (37.1) 161 (41.4) (50.1) 178 (45.8) 3 48 (12.8) 50 (12.9) Hemodynamic shock, n (%) 19 (5.1) 19 (4.9) Hematemesis, n (%) 172 (45.9) 181 (46.5) Melena, n (%) 328 (87.5) 346 (88.9) Hematochezia, n (%) 28 (7.5) 24 (6.2) Previous complications related to 44 (11.7) 41 (10.5) gastric or duodenal ulcer, n (%) Helicobacter pylori status, n (%) Negative 92 (24.5) 119 (30.6) Positive 264 (70.4) 252 (64.8) Missing 19 (5.1) 18 (4.6) Medication before enrollment, n (%) NSAID 151 (40.3) 157 (40.4) COX-2 selective 3 (0.8) 1 (0.3) Nonselective NSAID 63 (16.8) 73 (18.8) Aspirin 103 (27.5) 103 (26.5) Clopidogrel 12 (3.2) 11 (2.8) Warfarin 9 (2.4) 13 (3.3) SSRI 9 (2.4) 14 (3.6) ASA American Society of Anesthesiologists; COX-2 cyclooxygenase-2; NSAID nonsteroidal anti-inflammatory drug; SSRI selective serotonin reuptake inhibitor. RESULTS Study Sample We considered at least 4146 patients for inclusion in the study (not all centers maintained screening logs), and we enrolled 1313 patients as potentially suitable candidates. Of these, 767 were eligible for randomization to esomeprazole (n 376) or placebo (n 391) between October 2005 and December The study was completed by 337 patients receiving esomeprazole and 349 patients receiving placebo. Figure 1 summarizes reasons for nonrandomization or exclusion from intention-to-treat and per-protocol analyses. No patient was excluded from the intention-to-treat sample because of drug interaction. The treatment groups were generally well balanced with regard to demographic and baseline characteristics. Patients were predominantly male (68%) and white (87%). Baseline use of concomitant medication was similar; 40% of all patients took nonsteroidal anti-inflammatory drugs, and 27% took aspirin before enrollment. Helicobacter pylori was detected (positive or trace) at enrollment in 68% of patients (Table 1). Most ulcers at study entry were Forrest class Ib and IIa (80% in each group), with slightly fewer Forrest class Ia ulcers in the esomeprazole group than in the placebo group (7.5% vs. 10.3%) (Table 2). From an initial classification of Forrest class IIb in 22% of all patients at first endoscopy, only 10% retained this categorization after attempts at clot removal. The groups were similar with respect to mean ulcer size and location, with slightly more large ( 2 cm) and multiple ulcers in the placebo group. Endoscopic treatment type was similar, with about half receiving single therapy in each group. Eight patients (included in the intention-totreat analysis) received triple therapy, which was not permitted by the protocol. All but 1 patient received the treatment allocated at randomization. Photodocumentation of Ulcers Photodocumentation of ulcers (photographs, electronic files, CD/DVD, or VHS tapes) was available for 88% of patients. The degree of concordance in ulcer grading between the end point committee members and the investigators was high: The investigators opinion was in agreement with both committee members in 30% of cases (full support) and with only 1 committee member in 62% Table 2. Endoscopic Findings at Baseline in the Intention-to-Treat Groups Characteristic Esomeprazole Group (n 375) Placebo Group (n 389) Forrest class, n (%) Ia 28 (7.5) 40 (10.3) Ib 166 (44.3) 163 (41.9) IIa 136 (36.3) 151 (38.8) IIb 42 (11.2) 34 (8.7) Missing* 3 (0.8) 1 (0.3) Endoscopic treatment, n (%) None 4 (1.1) 5 (1.3) Single 175 (46.7) 180 (46.3) Dual 192 (51.2) 200 (51.4) Triple 4 (1.1) 4 (1.0) Ulcer size, mm Mean (SD) 12.5 (7.2) 12.4 (7.8) Range Ulcer location, n (%) Stomach 157 (41.9) 155 (39.8) Duodenum 216 (57.6) 233 (59.9) Not recorded 2 (0.5) 1 (0.3) Patients with ulcers, n (%) Ulcer size 2 cm 346 (92.3) 349 (89.7) Ulcer size 2 cm 29 (7.7) 40 (10.3) Single ulcer 324 (86.4) 317 (81.5) Multiple ulcers 51 (13.6) 72 (18.5) * No photodocumentation in 2 patients and no record of assessment despite photodocumentation in 2 patients. In 53 patients (30 in the esomeprazole group and 23 in the placebo group), technical difficulties precluded confirmation of more than 1 ulcer. No more than 1 bleeding ulcer. 7 April 2009 Annals of Internal Medicine Volume 150 Number 7 459

6 Article Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding Table 3. Recurrent Bleeding Rates, Mortality Rates, Surgery, and Hospital Stay Variable Esomeprazole Group* Placebo Group* P Value Absolute Risk Reduction (95% CI), percentage points Recurrent bleeding, n (%) Within 72 h ITT analysis 22 (5.9) 40 (10.3) (0.6 to 8.3) PP analysis 14 (4.8) 33 (10.4) (1.5 to 9.8) Within 7 d 27 (7.2) 50 (12.9) (1.4 to 9.9) Within 30 d 29 (7.7) 53 (13.6) (1.5 to 10.2) All-cause mortality 3 (0.8) 8 (2.1) ( 0.4 to 2.9) within 30 d, n (%) Bleeding-related 2 (0.5) 3 (0.8) ( 0.9 to 1.4) mortality within 30 d, n (%) Surgery within 30 d, n (%) 10 (2.7) 21 (5.4) ( 0.0 to 5.5) Repeated endoscopic 24 (6.4) 45 (11.6) (1.1 to 9.2) treat- ment within 30 d, n (%) Blood transfused within 30 d Total units Mean units (SD) 1.6 (2.5) 2.4 (4.5) Additional hospital days because of recurrent bleeding within 30 d Total Mean (SD) 0.8 (3.2) 1.3 (3.7) ITT intention-to-treat; PP per-protocol. * Esomeprazole group: ITT sample, n 375; PP sample, n 292. Placebo group: ITT sample, n 389; PP sample, n 316. Secondary Outcomes The benefits attributable to intravenous esomeprazole were also evident after 7 days and 30 days (Table 3). A priori planned adjustment of the primary end point according to endoscopic treatment received at baseline and post hoc analysis (adjustment by both endoscopic treatment and Forrest class) also confirmed the results (P 0.026). In a planned post hoc analysis by ethnic group, the 2 main ethnic groups, white (87%) and Asian (7%) patients, both showed similar relative reductions in recurrent bleeding at 72 hours. The rate of rebleeding was lower in white patients (5.5% for esomeprazole vs. 10.8% for placebo) and in Asian patients (3.7% vs. 7.4%). Similarly, we found no country treatment effects for the primary variable (P 0.57). Furthermore, the choice of endoscopic treatment did not affect the efficacy of esomeprazole; the reduction in recurrent bleeding after single (4.3%) and dual (4.2%) endoscopic treatment was similar between the esomeprazole and placebo groups. Fewer esomeprazole recipients than placebo recipients needed repeated endoscopic therapy (6.4% vs. 11.6%; P 0.012) or blood transfusions (589 vs. 935 total units transfused; P 0.034) during the 30 days after randomization, and the additional stay in the hospital attributable to recurrent bleeding was shorter in the esomeprazole group than in the placebo group (284 vs. 500 additional days total; P 0.008) (Table 3). The number of patients requiring surgery (2.7% vs. 5.4%; P 0.059), 30-day allcause mortality rates (0.8% vs. 2.1%; P 0.22), and bleeding-related mortality rates (0.53% vs. 0.77%; P 1.00) were lower in the esomeprazole group than in the placebo group, although the differences between groups were not significant. Adverse Events The most common AEs in both treatment groups were related to recurrent bleeding (Table 4). Incidence of cardiac events was similar in both groups, including no difference in myocardial infarction at the end of the of cases (partial support). Neither member agreed with the investigator in 8% of cases, mostly because of poor technical photographic quality. Primary Outcome Clinically significant recurrent bleeding within the first 72 hours occurred in 5.9% of patients in the esomeprazole group and 10.3% of patients in the placebo group (absolute risk reduction in the intention-to-treat analysis set, 4.4%; P 0.026) (Table 3). This result was confirmed for the per-protocol cohort (4.8% in the esomeprazole group and 10.4% in the placebo group [absolute risk reduction, 5.6%; P 0.009]). Most clinically significant recurrent bleeding occurred within the first 72 hours and plateaued after about 7 days (Figure 2). Figure 2. Kaplan Meier estimate of the cumulative percentage of patients with recurrent bleeding within 30 days. Cumulative Patients With Recurrent Bleeding, % Placebo Esomeprazole Log-rank P = Days From Infusion Until Recurrent Bleeding Patients at Risk, n Esomeprazole Placebo April 2009 Annals of Internal Medicine Volume 150 Number 7

7 Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding Article trial (5 esomeprazole recipients and 7 placebo recipients). Infusion-site reactions (thrombophlebitis; phlebitis; or infusion-site erythema, reaction, or edema) were more common in the esomeprazole group (4.3% vs. 0.5%) but were mild, were of short duration, and did not result in treatment discontinuation. These reactions are unlikely to have led to inadvertent unblinding of treatment allocation, because the onset was not until 2 to 4 days after start of the study drug and most centers (9 of 13) reported infusion-site reactions for only a single patient. DISCUSSION The results of this large international trial indicate that high-dose intravenous esomeprazole administered after endoscopic hemostasis has a significant protective effect against recurrent bleeding for up to 30 days in patients with high-risk peptic ulcers. This was accompanied by a reduction in endoscopic re-treatment, blood transfusions, and hospital stays because of recurrent bleeding in the esomeprazole group compared with the placebo group. Although the esomeprazole group had fewer Forrest class Ia ulcers, the conclusions remained valid after statistical adjustment for the slight baseline imbalance. We believe that our study adds to the existing literature summarized by recent Cochrane reviews (21 23). It sets new standards in the field through its careful design, addressing previous methodological shortcomings. This involved incorporation of adequate statistical powering, as well as education of physicians and the use of ulcer photodocumentation, to ensure entry of appropriate at-risk patients and provision of adequate endoscopic therapy. To date, the protective effect of continuous intravenous infusion of high-dose omeprazole has been observed in heterogeneous populations of varying ethnic groups with different ASA scores and in centers using nonstandardized endoscopic diagnostic criteria. In addition, disparate treatments and differing outcome measurements were used, and recruitment was sometimes inadequate (see Table 5 for a summary of previous studies). Such methodological limitations may partly explain the disparities in therapeutic effect. In our study, although most patients (87%) were white, the results were robust and similar for Asian patients, thus complementing the earlier positive studies conducted in Hong Kong and other countries among predominantly Asian populations (18). We selected the active comparator on the basis of existing practice, regulatory suggestions, and meta-analytic data suggesting that H 2 - receptor antagonists are not efficacious (27, 28). Although the optimal dose threshold for the therapeutic effect of PPI remains a source of debate (29), the evidence supporting PPI use is strongest for intravenous bolus administration followed by infusion in patients who initially had endoscopic hemostasis for high-risk ulcer lesions (23). Despite a consistent reduction in recurrent bleeding and improvement in many secondary outcomes with intravenous esomeprazole, the study did not demonstrate statistically significant differences in deaths or requirements for surgery. Several factors might explain the former. First, the study was not powered to detect differences in mortality rates as low as those seen in both treatment groups (0.8% and 2.1%). They are lower than those traditionally observed (10, 12) but similar to those reported recently by Lau and colleagues (2.2% to 2.5%) (30). This finding, which may somewhat limit the generalizability of the results, is probably because we excluded patients with ASA grade 4 or 5, coupled with the practice of effective modernday resuscitative care and efficacious endoscopic hemostasis, because the age group, comorbid condition, and characteristics of ulcers in this study are similar to those in other studies. The requirement for surgery reduced from Table 4. Adverse Events During the Intravenous Treatment Phase in the Intention-to-Treat Patient Groups* Variable Esomeprazole Group (n 375) Placebo Group (n 389) Any AE 147 (39.2) 163 (41.9) Treatment discontinuations due to 31 (8.3) 39 (10.0) AEs All serious AEs 35 (9.3) 44 (11.3) Fatal AEs 2 (0.5) 3 (0.8) Gastrointestinal disorder 46 (12.3) 77 (19.8) Duodenal ulcer hemorrhage 16 (4.3) 16 (4.1) Nausea 8 (2.1) 8 (2.1) Constipation 6 (1.6) 9 (2.3) Upper abdominal pain 5 (1.3) 9 (2.3) Gastric ulcer hemorrhage 4 (1.1) 13 (3.3) Abdominal pain 1 (0.3) 11 (2.8) Dyspepsia 0 (0) 5 (1.3) General disorder and administration 27 (7.2) 20 (5.1) site condition Any infusion-site reaction AE 16 (4.3) 2 (0.5) Pyrexia 13 (3.5) 11 (2.8) Vascular disorder 24 (6.4) 16 (4.1) Phlebitis 9 (2.4) 2 (0.5) Hypertension 5 (1.3) 7 (1.8) Hypotension 4 (1.1) 4 (1.0) Cardiac disorder 13 (3.5) 14 (3.6) Myocardial infarction 4 (1.1) 4 (1.0) Infection and infestation 13 (3.5) 16 (4.1) Urinary tract infection 4 (1.1) 4 (1.0) Cystitis 1 (0.3) 4 (1.0) Respiratory, thoracic, and 12 (3.2) 13 (3.3) mediastinal disorder Cough 4 (1.1) 1 (0.3) Dyspnea 2 (0.5) 7 (1.8) AE adverse event. * A cutoff at 3% in either treatment group was set for the system organ class level, with a subsequent cutoff at 1% for the preferred term level. The AE started within 72 h, but death occurred after 72 h for 4 patients. Patients with AEs reported as phlebitis, thrombophlebitis, superficial thrombophlebitis, injection-site erythema, injection-site inflammation, or injection-site swelling. 7 April 2009 Annals of Internal Medicine Volume 150 Number 7 461

8 Article Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding Table 5. Previous Comparative Studies Using an Active PPI Treatment With High-Dose Bolus Infusion Followed by Continuous Infusion After Endoscopic Hemostasis* Author, Year (Reference) Test Treatment (ITT Sample) Control Treatment (ITT Sample) 3-Day Rebleeding Overall Rebleeding Odds Ratio (95% CI) Surgery All-Cause Mortality Comment Single-center studies with mainly Asian patients Lin et al, 1998 (24) Lau et al, 2000 (10) Zargar et al, 2006 (25) Multicenter studies with mainly non-asian patients Hasselgren et al, 1997 (12) Schaffalitsky de Muckadell et al, 1997 (13) van Rensburg et al, 2004 (14) Jensen et al, 2006 (15) Omeprazole (n 50) Omeprazole (n 120) Pantoprazole (n 102) Omeprazole (n 159) Omeprazole (n 130) Pantoprazole (n 618) Pantoprazole (n 72) Cimetidine, 300 mg IV, plus 50 mg/h for 3d(n 50) Placebo (n 120) Placebo (n 101) Placebo (n 163) Placebo (n 135) Ranitidine, 50 mg IV, plus 6.25 mg/h for3d(n 626) Ranitidine, 50 mg IV, plus 6.25 mg/h for3d(n 77) 0.05 ( ) 0.13 ( ) Nonestimable (0 in denominator) 0.19 ( ) Gastric ph 6.0 for 84% of time in those receiving omeprazole and 54% of time in those receiving cimetidine (P 0.001) 0.17 ( ) 0.25 ( ) 0.32 ( ) 0.39 ( ) Included a significant proportion of patients with shock at presentation; trial stopped prematurely at interim analysis because of magnitude of therapeutic effect 0.34 ( ) 0.34 ( ) 0.34 ( ) 0.49 ( ) Included a significant proportion of patients with shock at presentation; in 15 patients, the median time to reach a gastric ph of 6.0 was 45 min (range, min) after an initial bolus of pantoprazole and persisted at around for the remaining 72 h 1.29 ( ) 0.40 ( ) ( ) Patients age 60 y, black ulcer base, imbalance in risks at baseline, unexplained very low mortality with placebo 0.52 ( ) 0.78 ( ) 0.94 ( ) Endoscopic hemostasis not systematically administered for high-risk lesions; difficulty to extract data from published work 0.75 ( ) 0.75 ( ) 0.93 ( ) 0.57 ( ) Nonstandardized endoscopic inclusion criteria; performance of routine second-look endoscopy 0.51 ( ) 0.45 ( ) 0.51 ( ) 1.00 ( ) Trial stopped prematurely because of poor enrollment ITT intention-to-treat; IV intravenous; PPI proton-pump inhibitor. * Data from Barkun et al (26); Leontiadis et al (22); and unpublished data (Leontiadis G. Personal communication). All trials used the PPI at a dose of 80 mg followed by 8 mg/h for 3dintheexperimental treatment group, except for the trial by Lin et al (24), which used 40 mg plus 6.6 mg/h for 3 d. In the Hasselgren et al (12) trial, endoscopic hemostasis was performed only for spurting lesions. In the Schaffalitsky de Muckadell et al (13) trial, all patients with circulatory collapse were included regardless of the endoscopic lesion. This table does not discuss the follow-up oral treatment that was used in trials (single daily dose of a PPI for d) or review most trials that compared PPI treatment with a high-dose bolus followed by continuous infusion to a lesser dose of PPI. Odds ratios 1.0 favor PPI. End point data shown as recalculated by Leontiadis et al (22). Original primary outcome reported was overall outcome (5-point scale ranking the outcome from worst to best as follows: death, 5; surgery, 4; endoscopic treatment, 3; 3 units of blood transfused, 2; 0 3 units of blood transfused, 1). 5.4% to 2.7% favoring esomeprazole (just short of statistical significance), consistent with the overall results. When the study was designed, we attempted to minimize diagnostic variability among investigators by educating them in classification of bleeding ulcers. In the case of adherent clots, the use of rigorous washing, cold snare, or both to detach any removable clot is important to ensure that endoscopic therapy is targeted as precisely as possible to the underlying blood vessel in the ulcer bed. This procedure led to reclassification of 55% of the Forrest class IIb ulcers to Forrest class Ia, Ib, or IIa. Availability of photodocumentation for ulcers in 82% of cases and agreement on Forrest classification between 1 or both end point committee members and the investigator in 92% of cases vali April 2009 Annals of Internal Medicine Volume 150 Number 7

9 Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding Article dates and standardizes this important aspect of patient selection, resulting in significantly greater internal validity. Such quality control has not been implemented previously in bleeding trials and is critical, especially in multicenter, multinational trials (19), because the benefit of adjuvant PPI therapy to endoscopic hemostasis can only be demonstrated in patients with ulcers exhibiting lesions that are truly at high risk for subsequent rebleeding. The most notable AEs associated with the use of esomeprazole were injection-site reactions, including superficial thrombophlebitis. This is probably a class effect, and it has been observed in previous studies using different PPIs (14). Of importance, we did not see an increase in cardiovascular events with PPI, in contrast to suggestions in a meta-analysis (31). Another study limitation is that endoscopic therapy was not completely standardized. About half of the patients receivedepinephrineinjection,thermalcoagulation,orhemoclips alone, whereas the remainder had epinephrine injection in addition to thermal coagulation or hemoclips. Although thermal coagulation and hemoclips produce similar results (32), injection alone is less effective than combined therapy in securing hemostasis (33). The study protocol did not impose restrictions on the method of endoscopic hemostasis so that results would be generalizable to a wide range of hospitals (academic as well as community-based centers) where epinephrine injection is frequently used as a single therapy. Even with this design limitation, the observed effect of esomeprazole on recurrent bleeding was of a similar magnitude when added to single or dual endoscopic therapy. In conclusion, we believe that this is the first international trial to provide high-quality evidence supporting the use of high-dose intravenous esomeprazole as adjuvant therapy to endoscopic therapy for patients with high-risk endoscopic lesions and peptic ulcer bleeding in a predominantly white population. In our view, the data indicate that the efficacy of PPIs in preventing recurrent peptic ulcer bleeding is not race-specific and should be universally applied. From the Institute of Digestive Diseases, Chinese University of Hong Kong, Hong Kong, China; McGill University Health Center, Montreal, Quebec, Canada; Erasmus MC University Medical Center, Rotterdam, the Netherlands; Medizinische Klinik und Poliklinik II, University of Leipzig, Leipzig, Germany; David Geffen School of Medicine at the University of California, Los Angeles, and Center for Ulcer Research and Education (CURE), Digestive Diseases Research Center, Los Angeles, California; Glasgow Royal Infirmary, Glasgow, United Kingdom; and AstraZeneca Research and Development, Mölndal, Sweden. Acknowledgment: The authors acknowledge the contribution of all investigators who participated in this study (see the Appendix Table, available at and thank Dr. Madeline Frame for medical writing assistance sponsored by AstraZeneca. They also thank Dr. Grigoris Leontiadis and Dr. Colin Howden for providing data from the second update to their Cochrane meta-analysis on PPI and bleeding ulcers. Grant Support: AstraZeneca Research and Development. Potential Financial Conflicts of Interest: Employment: H. Ahlbom (AstraZeneca), J. Kilhamn (AstraZeneca), T. Lind (AstraZeneca). Consultancies: A. Barkun (AstraZeneca, Olympus), E.J. Kuipers (AstraZeneca), J. Mössner (AstraZeneca), J.Y. Lau (AstraZeneca). Honoraria: J.J.Y. Sung (AstraZeneca, Nycomed), A. Barkun (Olympus, AstraZeneca), J. Mössner (AstraZeneca), J.Y. Lau (AstraZeneca, Nycomed). Stock ownership or options (other than mutual funds): J. Kilhamn (AstraZeneca). Grants received: E.J. Kuipers (AstraZeneca). Other: AstraZeneca manufactures esomeprazole, the drug under study in this article. Reproducible Research Statement: Study protocol: Further information is available from Dr. Lind ( , tore.lind@astrazeneca.com). Statistical code: Further information is available from Mr. Ahlbom ( , henrik.ahlbom@astrazeneca.com). Data set: Further information is available from Mr. Ahlbom ( , henrik.ahlbom@astrazeneca.com). Requests for Single Reprints: Joseph J.Y. Sung MD, PhD, Institute of Digestive Diseases, Prince of Wales Hospital, Chinese University of Hong Kong, Hong Kong, China; , , joesung@cuhk.edu.hk. Current author addresses and author contributions are available at References 1. Yavorski RT, Wong RK, Maydonovitch C, Battin LS, Furnia A, Amundson DE. Analysis of 3,294 cases of upper gastrointestinal bleeding in military medical facilities. Am J Gastroenterol. 1995;90: [PMID: ] 2. Barkun A, Sabbah S, Enns R, Armstrong D, Gregor J, Fedorak RN, et al; RUGBE Investigators. The Canadian Registry on Nonvariceal Upper Gastrointestinal Bleeding and Endoscopy (RUGBE): Endoscopic hemostasis and proton pump inhibition are associated with improved outcomes in a real-life setting. Am J Gastroenterol. 2004;99: [PMID: ] 3. UK Comparative Audit of Upper Gastrointestinal Bleeding and the Use of Blood. London: British Society of Gastroenterology; December Accessed at on 15 January Netzer P, Gaia C, Sandoz M, Huluk T, Gut A, Halter F, et al. Effect of repeated injection and continuous infusion of omeprazole and ranitidine on intragastric ph over 72 hours. Am J Gastroenterol. 1999;94: [PMID: ] 5. Green FW Jr, Kaplan MM, Curtis LE, Levine PH. Effect of acid and pepsin on blood coagulation and platelet aggregation. A possible contributor prolonged gastroduodenal mucosal hemorrhage. Gastroenterology. 1978;74: [PMID: 21830] 6. Chaimoff C, Creter D, Djaldetti M. The effect of ph on platelet and coagulation factor activities. Am J Surg. 1978;136: [PMID: 28678] 7. Patchett SE, Enright H, Afdhal N, O Connell W, O Donoghue DP. Clot lysis by gastric juice: an in vitro study. Gut. 1989;30: [PMID: ] 8. Berstad A. Does profound acid inhibition improve haemostasis in peptic ulcer bleeding? Scand J Gastroenterol. 1997;32: [PMID: ] 9. Vreeburg EM, Levi M, Rauws EA, Deventer SJ, Snel P, Bartelsman JW, et al. Enhanced mucosal fibrinolytic activity in gastroduodenal ulcer haemorrhage and the beneficial effect of acid suppression. Aliment Pharmacol Ther. 2001;15: [PMID: ] 10. Lau JY, Sung JJ, Lee KK, Yung MY, Wong SK, Wu JC, et al. Effect of intravenous omeprazole on recurrent bleeding after endoscopic treatment of bleeding peptic ulcers. N Engl J Med. 2000;343: [PMID: ] 11. Khuroo MS, Yattoo GN, Javid G, Khan BA, Shah AA, Gulzar GM, et al. A comparison of omeprazole and placebo for bleeding peptic ulcer. N Engl J Med. 1997;336: [PMID: ] 12. Hasselgren G, Lind T, Lundell L, Aadland E, Efskind P, Falk A, et al. Continuous intravenous infusion of omeprazole in elderly patients with peptic ulcer bleeding. Results of a placebo-controlled multicenter study. Scand J Gas- 7 April 2009 Annals of Internal Medicine Volume 150 Number 7 463

10 Article Intravenous Esomeprazole for Preventing Recurrent Peptic Ulcer Bleeding troenterol. 1997;32: [PMID: ] 13. Schaffalitzky de Muckadell OB, Havelund T, Harling H, Boesby S, Snel P, Vreeburg EM, et al. Effect of omeprazole on the outcome of endoscopically treated bleeding peptic ulcers. Randomized double-blind placebo-controlled multicentre study. Scand J Gastroenterol. 1997;32: [PMID: ] 14. van Rensburg CJ, Racz I, Bailey RJ, Barkun A, Feu F, Ahrens H, et al. Prevention of peptic ulcer rebleeding using continuous infusion of pantoprazole (PAN) vs ranitidine (RAN): a multicenter, multinational, randomized, doubleblind, parallel-group comparison [Abstract]. Can J Gastroenterol. 2004;18(Suppl SA):Abstract Jensen DM, Pace SC, Soffer E, Comer GM; 315 Study Group. Continuous infusion of pantoprazole versus ranitidine for prevention of ulcer rebleeding: a U.S. multicenter randomized, double-blind study. Am J Gastroenterol. 2006; 101:1991-9; quiz [PMID: ] 16. Li Y, Sha W, Nie Y, Wu H, She Q, Dai S, et al. Effect of intragastric ph on control of peptic ulcer bleeding. J Gastroenterol Hepatol. 2000;15: [PMID: ] 17. Goldstein JA. Clinical relevance of genetic polymorphisms in the human CYP2C subfamily. Br J Clin Pharmacol. 2001;52: [PMID: ] 18. Leontiadis GI, Sharma VK, Howden CW. Systematic review and metaanalysis: enhanced efficacy of proton-pump inhibitor therapy for peptic ulcer bleeding in Asia a post hoc analysis from the Cochrane Collaboration. Aliment Pharmacol Ther. 2005;21: [PMID: ] 19. Sung JJ, Mössner J, Barkun A, Kuipers EJ, Lau J, Jensen D, et al; Pub Study Group. Intravenous esomeprazole for prevention of peptic ulcer re-bleeding: rationale/design of Peptic Ulcer Bleed study. Aliment Pharmacol Ther. 2008; 27: [PMID: ] 20. Forrest JA, Finlayson ND, Shearman DJ. Endoscopy in gastrointestinal bleeding. Lancet. 1974;2: [PMID: ] 21. Leontiadis GI, Sharma VK, Howden CW. Systematic review and metaanalysis of proton pump inhibitor therapy in peptic ulcer bleeding. BMJ. 2005; 330:568. [PMID: ] 22. Leontiadis GI, Sharma VK, Howden CW. Proton pump inhibitor treatment for acute peptic ulcer bleeding. Cochrane Database Syst Rev. 2006:CD [PMID: ] 23. Leontiadis GI, Sharma VK, Howden CW. Proton pump inhibitor therapy for peptic ulcer bleeding: Cochrane collaboration meta-analysis of randomized controlled trials. Mayo Clin Proc. 2007;82: [PMID: ] 24. Lin HJ, Lo WC, Lee FY, Perng CL, Tseng GY. A prospective randomized comparative trial showing that omeprazole prevents rebleeding in patients with bleeding peptic ulcer after successful endoscopic therapy. Arch Intern Med. 1998; 158:54-8. [PMID: ] 25. Zargar SA, Javid G, Khan BA, Yattoo GN, Shah AH, Gulzar GM, et al. Pantoprazole infusion as adjuvant therapy to endoscopic treatment in patients with peptic ulcer bleeding: prospective randomized controlled trial. J Gastroenterol Hepatol. 2006;21: [PMID: ] 26. Barkun AN, Cockeram AW, Plourde V, Fedorak RN. Review article: acid suppression in non-variceal acute upper gastrointestinal bleeding. Aliment Pharmacol Ther. 1999;13: [PMID: ] 27. Collins R, Langman M. Treatment with histamine H2 antagonists in acute upper gastrointestinal hemorrhage. Implications of randomized trials. N Engl J Med. 1985;313: [PMID: ] 28. Levine JE, Leontiadis GI, Sharma VK, Howden CW. Meta-analysis: the efficacy of intravenous H2-receptor antagonists in bleeding peptic ulcer. Aliment Pharmacol Ther. 2002;16: [PMID: ] 29. Andriulli A, Annese V, Caruso N, Pilotto A, Accadia L, Niro AG, et al. Proton-pump inhibitors and outcome of endoscopic hemostasis in bleeding peptic ulcers: a series of meta-analyses. Am J Gastroenterol. 2005;100: [PMID: ] 30. Lau JY, Leung WK, Wu JC, Chan FK, Wong VW, Chiu PW, et al. Omeprazole before endoscopy in patients with gastrointestinal bleeding. N Engl J Med. 2007;356: [PMID: ] 31. Khuroo MS, Khuroo MS, Farahat KL, Kagevi IE. Treatment with proton pump inhibitors in acute non-variceal upper gastrointestinal bleeding: a metaanalysis. J Gastroenterol Hepatol. 2005;20: [PMID: ] 32. Sung JJ, Tsoi KK, Lai LH, Wu JC, Lau JY. Endoscopic clipping versus injection and thermo-coagulation in the treatment of non-variceal upper gastrointestinal bleeding: a meta-analysis. Gut. 2007;56: [PMID: ] 33. Marmo R, Rotondano G, Piscopo R, Bianco MA, D Angella R, Cipolletta L. Dual therapy versus monotherapy in the endoscopic treatment of high-risk bleeding ulcers: a meta-analysis of controlled trials. Am J Gastroenterol. 2007; 102:279-89; quiz 469. [PMID: ] April 2009 Annals of Internal Medicine Volume 150 Number 7