Study design and setting
This prospective study was conducted in the inpatient wards of two tertiary healthcare centers, National Cheng Kung University Hospital in Tainan City and Kaohsiung Medical University Hospital in Kaohsiung City, Taiwan from June 2015 to December 2020. The study design was approved by the National Cheng Kung University Hospital Institutional Review Board (approval number: A-BR-104-007 and trial registration identifier: NCT02456012, ClincalTrials.gov, 28/05/2015). After they provided written informed consent, the participants were enrolled. A schematic flow chart of the study protocol is summarized in Fig. 1.
The characteristics of participants
Patients 20 or more years of age who were diagnosed with bleeding peptic ulcers with major stigmata of recent hemorrhage (SRH) were considered eligible. The major SRH include Forrest Ia, Ib, IIa, and IIb [16]. At the index endoscopy, all patients received one or a combination of endoscopic therapies to stop bleeding. The endoscopic therapy methods were listed in our previous study in detail [9].
In this study, we used Rockall scores ≥ 6 to define patients who were at high risk of rebleeding [6,7,8,9]. The parameters of the Rockall score are listed in Supplementary Table 1 and the definitions for the co-morbidities in the score were listed in previous studies in more detail [6, 9]. After gastroscopy to confirm enrollment eligibility, only patients with Rockall scores ≥6 were enrolled. The exclusion criteria were patients with bleeding from marginal ulcers, tumors, Dieulafoy lesions, or mechanical factor-related ulcers (i.e., gastrostomy tube induction), or with hypersensitivity to esomeprazole or pantoprazole. Patients who had used or planned to use long-term antiplatelet agents (i.e., aspirin, clopidogrel, ticagrelor, or prasugrel) or used non-steroidal anti-inflammatory drugs after the first bleeding episode were also excluded because the protective effect of PPIs for such patients has been established [11, 12].
Our study group had a cohort in which patients with bleeding peptic ulcer diseases and Rockall scores ≥6 from Aug 2011 to Jan 2015 were enrolled [8, 9]. The patients in this cohort received an 8- to 16-week treatment with oral PPIs after the first peptic ulcer bleeding episode. From this cohort, patients who were age- and sex-matched for the experimental groups (Group D and S) were chosen as the controls, which were labeled as Group C.
American Society of Anesthesiologists (ASA) Physical Status classification (Supplementary Table 2) is a simple tool to summarize the preoperative health status of surgical patients; moreover, it has become a ubiquitous component of clinical studies to define and compare demographic characteristics of subjects between groups [17].
Randomization and masking
Each patient received intravenous PPI therapy with an 80 mg loading dose and then an 8 mg/h continuous infusion of either esomeprazole (Nexium®, AstraZeneca AB, Södertälje, Sweden) or pantoprazole (Pantoloc®, Takeda, Singen, Germany) for 3 days [1]. Thereafter, the patients received oral esomeprazole or pantoprazole 40 mg twice a day for the following 11 days and once a day for the remaining days within the first 16 weeks [9]. After the first 16-week PPI therapy, patients were randomized into Group D or Group S following a simple randomization procedure with a 1:1 allocation ratio based on drawing an envelope from a box of sealed opaque envelopes, each of which contained a written code designating the patient to group D or S. Patients in Group D received oral esomeprazole (Nexium®, AstraZeneca AB, Södertälje, Sweden) 20 mg twice daily, and those in Group S received 20 mg once daily, respectively, for the following 36 weeks. The investigator who generated the random allocation sequence was different from those who assigned the participants to interventions and checked the symptoms and signs of recurrent bleeding during long-term follow-up. The latter were blinded to the study group allocation.
After the 52-week therapy, the patients in groups D and S used oral PPIs or did not at the discretion of their physicians according to their clinical needs. Thus, these patients were divided into a PPI-on-demand group and the PPI-discontinued group. The definition of PPI-discontinued was PPI use ≤3 days per week and ≤ 1 week per month. The others were defined as PPI-on-demand.
Procedures
The patients with Helicobacter pylori (H. pylori) infection received standard first-line or second-line eradication and all were confirmed to have successful eradication by the 13C-urea breath test [18, 19]. During the periods of follow-up for primary and secondary outcomes, the patients in the three groups did not use gastric mucosal protective agents, including misoprostol and sucralfate. Bismuth was used for H. pylori eradication only.
Outcomes
The primary and secondary outcomes were the recurrent peptic ulcer bleeding during the 1st year and the second year-and-thereafter since the first bleeding episode, respectively. All of the patients were monitored every day during hospitalization, every 2 weeks in the following 4 weeks, and every 12 weeks for the remaining weeks for a total of 1 year. They were monitored for pill counts, hemoglobin concentrations, and clinical symptoms and signs which were relevant to gastrointestinal bleeding, diarrhea, pneumonia, and bone fracture at each outpatient visit. All enrolled patients were included in the intention-to-treat (ITT) analysis, but patients who were lost to follow-up, discontinued intervention because of any reasons, had a protocol violation, or died were excluded from the per-protocol (PP) analysis for the primary endpoint. After the first year of therapy, the patients were continued to be monitored at outpatient departments every 12 weeks.
Recurrent bleeding was defined as follows, including 1) the presence of hematemesis, melena, hematochezia, or bloody aspirates through a nasogastric tube, plus 2) the presence of hemodynamic instability, including systolic blood pressure < 90 mmHg, heart rate > 120 beats per minute, or a drop in hemoglobin concentration by more than 2 g/dL, or sudden increase in transfusion requirements. To confirm recurrent peptic ulcer bleeding, the hemoglobin concentration and gastroscopy were performed to check the drop in hemoglobin concentration, any blood or coffee-ground-like materials in the stomach, or the recurrence of stigmata of recent hemorrhage. The gastroscopy also confirmed whether bleeding was from a peptic ulcer or other non-ulcer lesions, such as varices. Additionally, the definition of refractory or recurrence of the ulcer was the size ≥0.5 cm.
Medical events, including diarrhea and pneumonia, were monitored when patients took PPIs until 2 weeks after discontinuing PPIs. The definition of diarrhea was that the presence of loose or watery stools ≥ three times a day lasted for 1 day at least. The definition of pneumonia was the presence of one of the symptoms and signs of fever, purulent productive cough, and shortness of breath plus a typical infiltrative patch on chest X-ray. Additionally, any bone fracture, including a partial or complete break in the bone, was monitored until the last follow-up date at outpatient departments.
Statistical analysis
The estimated recurrent bleeding rate during the first year in Group C was about 15% based on the previous studies [20, 21]. We proposed that the recurrent bleeding rate in the experiment groups could be reduced to near 2%, equal to that in H. pylori ulcers after eradication. The ratio of the patient number in each experiment group to the patient number in the control group was 2:5. With a two-side α value of 0.05 and power of 80% (β = 0.20), the number of patients required was 54 in each experiment group and 135 in the control group to detect a difference between the two groups. Assuming the rate of loss follow-up was 10%, sixty patients in each experimental group were enrolled. Data related to baseline characteristics and endpoints were evaluated using the one-way analysis of variance, Pearson’s χ2 test, Fisher’s exact test, or Kruskal-Wallis one-way analysis of variance by ranks. The log-rank test was used to compare the Kaplan-Meier curves among the study groups. The COX hazard regression analysis was used to evaluate the factors predicting the primary and secondary outcomes. When patients encountered death, no more chances of recurrent bleeding would be met. Thus, considering the patient’s death as a competing event, the proportional subdistribution hazards model for competing risk data according to the Fine and Gray method was used. All tests were two-tailed and P values of < 0.05 indicated significant differences. The statistical analysis was performed with SPSS software (SPSS Statistics 20.0, IBM Corp., Armonk, NY, USA) and the competing-risks regression was evaluated using SAS software (version 9.4, SAS Institute Inc., Cary, NC, USA).